This action might not be possible to undo. Are you sure you want to continue?
Principles of Quantum Mechanics
SECOND EDITION
R. Shankar
Yale University New Haven, Connecticut
Kluwer Academic / Plenum Publishers
New York, Boston, Dordrecht, London, Moscow
Library of Congress Cats Tog1nginPublIcatIon D«t»
Shankar , Ramamurt1.
P r i n c i p l e s of q u a n t u m m e c h a n i c s / R . S h a n k a r . — 2 n d p. cm. Includes b1b11ograph1ca1 references and index. ISBN 0  3 0 6  4 4 7 9 0  8 1. Q u a n t u m t h e o r v . I. T i t l e . QC174.12.S52 1994 530.1'Z—dc20 Bd.
9426837
CIP
1098
ISBN 0306447908
©1994, 1980 Plenum Press, New York A Division of Plenum Publishing Corporation 233 Spring Street, New York, N.Y. 10013 All rights reserved No pari of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording* or otherwise, without written permission from ihe Publisher Printed in the United States of America
Preface to the Second Edition
Over the decade and a half since I wrote the first edition, nothing has altered my belief in the soundness of the overall approach taken here. This is based on the response of teachers, students, and my own occasional rereading of the book. I was generally quite happy with the book, although there were portions where I felt I could have done better and portions which bothered me by their absence. I welcome this opportunity to rectify all that. Apart from small improvements scattered over the text, there are three major changes. First, I have rewritten a big chunk of the mathematical introduction in Chapter 1. Next, I have added a discussion of timereversal invariance. I don't know how it got left out the first time—I wish I could go back and change it. The most important change concerns the inclusion of Chaper 21, "Path Integrals: Part II." The first edition already revealed my partiality for this subject by having a chapter devoted to it, which was quite unusual in those days. In this one, I have cast off all restraint and gone alt out to discuss many kinds of path integrals and their uses. Whereas in Chapter 8 the path integral recipe was simply given, here I start by deriving it. I derive the configuration space integral (the usual Feynman integral), phase space integral, and (oscillator) coherent state integral. I discuss two applications: the derivation and application of the Berry phase and a study of the lowest Landau level with an eye on the quantum Hall effect. The relevance of these topics is unquestionable. This is followed by a section of imaginary time path integrals— its description of tunneling, instantons, and symmetry breaking, and its relation to classical and quantum statistical mechanics. An introduction is given to the transfer matrix. Then I discuss spin coherent state path integrals and path integrals for fermions. These were thought to be topics too advanced for a book like this, but I believe this is no longer true. These concepts are extensively used and it seemed a good idea to provide the students who had the wisdom to buy this book with a head start. How are instructors to deal with this extra chapter given the time constraints? I suggest omitting some material from the earlier chapters. (No one I know, myself included, covers the whole book white teaching any fixed group of students.) A realistic option is for the instructor to teach part of Chapter 21 and assign the rest as reading material, as topics for takehome exams, term papers, etc. To ignore it,
viii
PREFACE TO THE SECOND EDITION
I think, would be to lose a wonderful opportunity to expose the student to ideas that are central to many current research topics and to deny them the attendant excitement. Since the aim of this chapter is to guide students toward more frontline topics, it is more concise than the rest of the book. Students are also expected to consult the references given at the end of the chapter. Over the years, I have received some very useful feedback and I thank all those students and teachers who took the time to do so. I thank Howard Haber for a discussion of the Born approximation; Harsh Mathur and Ady Stern for discussions of the Berry phase; Alan Chodos, Steve Girvin, Ilya Gruzberg, Martin Gutzwiller. Ganpathy Murthy, Charlie Sommerfeld, and Senthil Todari for many useful comments on Chapter 21. I am most grateful to Captain Richard F. Malm, U.S.C.G. (Retired), Professor Dr. D. Schliiter of the University of Kiel, and Professor V. Yakovenko of the University of Maryland for detecting numerous errors in the first printing and taking the trouble to bring them to my attention. I thank Amelia McNamara of Plenum for urging me to write this edition and Plenum for its years of friendly and warm cooperation. Finally, I thank my wife Uma for shielding me as usual from real life so I could work on this edition, and my battery of kids (revised and expanded since the previous edition) for continually charging me up. R. Shankar New Haven, Connecticut
Preface to the First Edition
Publish and perish—Giordano Bruno
Given the number of books that already exist on the subject of quantum mechanics, one would think that the public needs one more as much as it does, say, the latest version of the Table of Integers. But this does not deter me (as it didn't my predecessors) from trying to circulate my own version of how it ought to be taught. The approach to be presented here (to be described in a moment) was first tried on a group of Harvard undergraduates in the summer of '76, once again in the summer of '77, and more recently at Yale on undergraduates (HI78) and graduates ('78 *79) taking a yearlong course on the subject. In all cases the results were very satisfactory in the sense that the students seemed to have learned the subject well and to have enjoyed the presentation. It is, in fact, their enthusiastic response and encouragement that convinced me of the soundness of my approach and impelled me to write this book. The basic idea is to develop the subject from its postulates, after addressing some indispensable preliminaries. Now, most people would agree that the best way to teach any subject that has reached the point of development where it can be reduced to a few postulates is to start with the latter, for it is this approach that gives students the fullest understanding of the foundations of the theory and how it is to be used. But they would also argue that whereas this is all right in the case of special relativity or mechanics, a typical student about to learn quantum mechanics seldom has any familiarity with the mathematical language in which the postulates are stated. I agree with these people that this problem is real, but I differ in my belief that it should and can be overcome. This book is an attempt at doing just this. It begins with a rather lengthy chapter in which the relevant mathematics of vector spaces developed from simple ideas on vectors and matrices the student is assumed to know. The level of rigor is what I think is needed to make a practicing quantum mechanic out of the student. This chapter, which typically takes six to eight lecture hours, is filled with examples from physics to keep students from getting too fidgety while they wait for the "real physics." Since the math introduced has to be taught sooner or later, I prefer sooner to later, for this way the students, when they get to it, can give quantum theory their fullest attention without having to
X
PREFACE TO THE FIRST EDITION
battle with the mathematical theorems at the same time. Also, by segregating the mathematical theorems from the physical postulates, any possible confusion as to which is which is nipped in the bud. This chapter is followed by one on classical mechanics, where the Lagrangian and Hamiltonian formalisms are developed in some depth. It is for the instructor to decide how much of this to cover; the more students know of these matters, the better they will understand the connection between classical and quantum mechanics. Chapter 3 is devoted to a brief study of idealized experiments that betray the inadequacy of classical mechanics and give a glimpse of quantum mechanics. Having trained and motivated the students 1 now give them the postulates of quantum mechanics of a single particle in one dimension. I use the word "postulate" here to mean "that which cannot be deduced from pure mathematical or logical reasoning, and given which one can formulate and solve quantum mechanical problems and interpret the results." This is not the sense in which the true axiomatist would use the word. For instance, where the true axiomatist would just postulate that the dynamical variables are given by Hiibert space operators, 1 would add the operator identifications, i.e., specify the operators that represent coordinate and momentum (from which others can be built). Likewise, 1 would not stop with the statement that there is a Hamiltonian operator that governs the time evolution through the equation ifid\yf) /dt Hly/), I would say the H is obtained from the classical Hamiltonian by substituting for J and p the corresponding operators. While C the more general axioms have the virtue of surviving as we progress to systems of more degrees of freedom, with or without classical counterparts, students given just these will not know how to calculate anything such as the spectrum of the oscillator. Now one can, of course, try to "derive" these operator assignments, but to do so one would have to appeal to ideas of a postulatory nature themselves. (The same goes for "deriving" the Schrodinger equation.) As we go alongs these postulates are generalized to more degrees of freedom and it is for pedagogical reasons that these generalizations are postponed. Perhaps when students are finished with this book, they can free themselves from the specific operator assignments and think of quantum mechanics as a general mathematical formalism obeying certain postulates (in the strict sense of the term). The postulates in Chapter 4 are followed by a lengthy discussion of the same, with many examples from fictitious Hiibert spaces of three dimensions. Nonetheless, students will find it hard. It is only as they go along and see these postulates used over and over again in the rest of the book, in the setting up of problems and the interpretation of the results, that they will catch on to how the game is played. It is hoped they will be able to do it on their own when they graduate. I think that any attempt to soften this initial blow will be counterproductive in the long run. Chapter 5 deals with standard problems in one dimension. It is worth mentioning that the scattering off a step potential is treated using a wave packet approach. If the subject seems too hard at this stage, the instructor may decide to return to it after Chapter 7 (oscillator), when students have gained more experience. But I think that sooner or later students must get acquainted with this treatment of scattering. The classical limit is the subject of the next chapter. The harmonic oscillator is discussed in detail in the next. It is the first realistic problem and the instructor may be eager to get to it as soon as possible. If the instructor wants, he or she can discuss the classical limit after discussing the oscillator.
We next discuss the path integral formulation due to Feynman. Given the intuitive understanding it provides, and its elegance (not to mention its ability to give the full propagator in just a few minutes in a class of problems), its omission from so many books is hard to understand. While it is admittedly hard to actually evaluate a path integral (one example is provided here), the notion of expressing the propagator as a sum over amplitudes from various paths is rather simple. The importance of this point of view is becoming clearer day by day to workers in statistical mechanics and field theory. I think every effort should be made to include at least the first three (and possibly five) sections of this chapter in the course. The content of the remaining chapters is standard, in the first approximation. The style is of course peculiar to this author, as are the specific topics. For instance, an entire chapter (11) is devoted to symmetries and their consequences. The chapter on the hydrogen atom also contains a section on how to make numerical estimates starting with a few mnemonics. Chapter 15, on addition of angular momenta, also contains a section on how to understand the "accidental" degeneracies in the spectra of hydrogen and the isotropic oscillator. The quantization of the radiation field is discussed in Chapter 18, on timedependent perturbation theory. Finally the treatment of the Dirac equation in the last chapter (20) is intended to show that several things such as electron spin, its magnetic moment, the spinorbit interaction, etc. which were introduced in an ad hoc fashion in earlier chapters, emerge as a coherent whole from the Dirac equation, and also to give students a glimpse of what lies ahead. This chapter also explains how Feynman resolves the problem of negativeenergy solutions (in a way that applies to bosons and fermions).
xi
PREFACE TO THE FIRST EDITION
For Whom Is this Book Intended? In writing it, I addressed students who are trying to learn the subject by themselves; that is to say, I made it as selfcontained as possible, included a lot of exercises and answers to most of them, and discussed several tricky points that trouble students when they learn the subject. But I am aware that in practice it is most likely to be used as a class text. There is enough material here for a full year graduate course. It is, however, quite easy so adapt it to a yearlong undergraduate course. Several sections that may be omitted without loss of continuity are indicated. The sequence of topics may also be changed, as stated earlier in this preface. 1 thought it best to let the instructor skim through the book and chart the course for his or her class, given their level of preparation and objectives. Of course the book will not be particularly useful if the instructor is not sympathetic to the broad philosophy espoused here, namely, that first comes the mathematical training and then the development of the subject from the postulates. To instructors who feel that this approach is all right in principle but will not work in practice, I reiterate that it has been found to work in practice, not just by me but also by teachers elsewhere. The book may be used by nonphysicists as well. (I have found that it goes well with chemistry majors in my classes.) Although I wrote it for students with no familiarity with the subject, any previous exposure can only be advantageous. Finally, I invite instructors and students alike to communicate to me any suggestions for improvement, whether they be pedagogical or in reference to errors or misprints.
xii
P R E F A C E TO THE FIRST EDITION
Acknowledgments As I look back to see who all made this book possible, my thoughts first turn B R O TH E T R Rajaraman and friend Rajaram Nityananda, who, around the TQ same time, introduced me to physics in general and quantum mechanics in particular. Next come my students, particularly Doug Stone, but for whose encouragement and enthusiastic response I would not have undertaken this project. I am grateful to Professor Julius Kovacs of Michigan State, whose kind words of encouragement assured me that the book would be as well received by my peers as it was by my students. More recently, I have profited from numerous conversations with my colleagues at Yale, in particular Alan Chodos and Peter Mohr. My special thanks go to Charles Sommerfield, who managed to make time to read the manuscript and made many useful comments and recommendations. The detailed proofreading was done by Tom Moore. I thank you, the reader, in advance, for drawing to my notice any errors that may have slipped past us. The bulk of the manuscript production cost were borne by the J. W. Gibbs fellowship from Yale, which also supported me during the time the book was being written. Ms. Laurie Liptak did a fantastic job of typing the first 18 chapters and Ms. Linda Ford did the same with Chapters 19 and 20. The figures are by Mr. J. Brosious. Mr. R. Badrinath kindly helped with the index.t On the domestic front, encouragement came from my parents, my inlaws, and most important of all from my wife, Uma, who cheerfully donated me to science for a year or so and stood by me throughout. Little Umesh did his bit by tearing up all my books on the subject, both as a show of support and to create a need for this one. R. Shankar New Haven, Connecticut
t It is a pleasure to acknowledge the help of Mr. Richard Hatch, who drew my attention to a number of errors in the first printing.
Prelude
Our description of the physical world is dynamic in nature and undergoes frequent change. At any given time, we summarize our knowledge of natural phenomena by means of certain laws. These laws adequately describe the phenomenon studied up to that time, to an accuracy then attainable. As time passes, we enlarge the domain of observation and improve the accuracy of measurement. As we do so, we constantly check to see if the laws continue to be valid. Those laws that do remain valid gain in stature, and those that do not must be abandoned in favor of new ones that do. In this changing picture, the laws of classical mechanics formulated by Galileo, Newton, and later by Euler, Lagrange, Hamilton, Jacobs, and others, remained unaltered for almost three centuries. The expanding domain of classical physics met its first obstacles around the beginning of this century. The obstruction came on two fronts; at large velocities and small (atomic) scales. The problem of large velocities was successfully solved by Einstein, who gave us his relativistic mechanics, while the founders of quantum mechanics—Bohr, Heisenberg, Schrddinger, Dirac, Born, and others—solved the problem of smallscale physics. The union of relativity and quantum mechanics, needed for the description of phenomena involving simultaneously large velocities and small scales, turns out to be very difficult. Although much progress has been made in this subject, called quantum field theory, there remain many open questions to this date. We shall concentrate here on just the smallscale problem, that is to say, on nonrelativistic quantum mechanics. The passage from classical to quantum mechanics has several features that are common to all such transitions in which an old theory gives way to a new one; (1) There is a domain Dn of phenomena described by the new theory and a subdomain D0 wherein the old theory is reliable (to a given accuracy). (2) Within the subdomain D 0 either theory may be used to make quantitative predictions. It might often be more expedient to employ the old theory. (3) In addition to numerical accuracy, the new theory often brings about radical conceptual changes. Being of a qualitative nature, these will have a bearing on all of D„. For example, in the case of relativity, D„ and Dn represent (macroscopic) phenomena involving small and arbitrary velocities, respectively, the latter, of course, xiii
It is hoped that. the postulates are invoked to formulate and solve a variety of quantum mechanical problems. read it to get acquainted with the notation. starting from its postulates. This book introduces you to this subject. I am not saying it is an easy subject. In addition to giving better numerical predictions for highvelocity phenomena. Good luck. a review of classical mechanics. you will be able to do the same yourself. Even if you know the math. especially the ones marked * or whose results carry equation numbers. but also conceptual changes that rock the very foundations of classical thought. etc. The answer to each exercise is given either with the exercise or at the end of the book. In the rest of the book. . by the time you get to the end of the book. Note to the Student Do as many exercises as you can. The first chapter is very important. absolute length. In a similar manner. Do not rush through it. unlimited velocities for particles. Between you and the postulates there stand three chapters wherein you will find a summary of the mathematical ideas appearing in the statement of the postulates. and a brief description of the empirical basis for the quantum theory. such as absolute time. But I hope this book makes it seem reasonable. quantum mechanics brings with it not only improved numerical predictions for the microscopic world.xiv PRELUDE being bounded by the velocity of light. relativity theory also outlaws several cherished notions of the Newtonian scheme.
1. 2.8.4. Active and Passive Transformations The Eigenvalue Problem Functions of Operators and Related Concepts Generalization to Infinite Dimensions .1. . Particles and Waves in Classical Physics An Experiment with Waves and Particles (Classical) The DoubleSlit Experiment with Light Matter Waves (de Broglie Waves) Conclusions XV . . 2. .5. 2. 1.5. Poisson Brackets. I 7 11 17 18 20 29 30 54 57 75 78 83 85 Review of Classical Mechanics 2. Mathematical Introduction 1. Subspaces Linear Operators . . . 2. 1.8.1.2.2.7.5.1. .3.4.6. The Principle of Least Action and Lagrangian Mechanics The Electromagnetic Lagrangian The TwoBody Problem How Smart Is a Particle? The Hamiltonian Formalism The Electromagnetic Force in the Hamiltonian Scheme Cyclic Coordinates.10. 1. 2. and Canonical Transformations Symmetries and Their Consequences 86 86 90 91 98 107 107 108 U0 112 112 3.2. Matrix Elements of Linear Operators . 3.9. . 1. 1. Linear Vector Spaces : Basics Inner Product Spaces Dual Spaces and the Dirac Notation .4. 2. . . 3. 2.3. 2. 3.6. . .7. 1. 1. All Is Not Well with Classical Mechanics 3.Contents 1. 1. 3.3.
Some Theorems 6.2.3. N Particles in One Dimension More Particles in More Dimensions Identical Particles . . The Harmonic Oscillator 7. 4. Simple Problems in One Dimension 5. The Path Integral Formulation of Quantum Theory 8. . The Free Particle 5. Systems with N Degrees of Freedom 10.3. The Particle in a Box 5.4.2. Passage from the Energy Basis to the X Basis 8.3. 9.1.2. The Classical Limit 7. The Postulates Discussion of Postulates IIII The Schrodirtger Equation (Dotting Your f s and Crossing your 5. Introduction Derivation of the Uncertainty Relations The Minimum Uncertainty Packet Applications of the Uncertainty Principle The EnergyTime Uncertainty Relation 10. Equivalence to the Schrodinger Equation Potentials of the Form a + bx + cx2 + dx + exx The Heisenberg Uncertainty Relations 9. 115 115 116 143 151 151 157 164 167 175 176 179 185 185 188 189 202 216 223 223 224 225 226 229 231 237 237 237 239 241 245 241 241 25$ 26( 4. The DoubleSlit Experiment 5. 9. .1. 9.6.2. 9. The Postulates—a General Discussion 4 J. .5.xvi CONTENTS 4. . The Continuity Equation for Probability 5.2.3. The Path Integral Recipe .5.4. 5.1. 8. .5. The SingleStep Potential: a Problem in Scattering .3. Review of the Classical Oscillator 7.4. 8.1. 10. Quantization of the Oscillator (Coordinate Basis) 7. Why Study the Harmonic Oscillator? 7.1.6. 10.4. Analysis of the Recipe An Approximation to 17(f) for the Free Particle Path Integral Evaluation of the FreeParticle P r o p a g a t o r .5. The Oscillator in the Energy Basis 7. 8.3. 8. 9. 8.2.
Degenerate Perturbation T h e o r y .3. . . Multielectron Atoms and the Periodic Table 14. 12. . 16. 11. .1.2. 16. 12.11. 11.6. The Hydrogen Atom 13. 15. The WentzelKramersBrillouin Method 17. Spin 14. 14. .2.2. Variational and WKB Methods 16. 11. The Eigenvalue Problem The Degeneracy of the Hydrogen Spectrum Numerical Estimates and Comparison with E x p e r i m e n t . 13. . Rotational Invariance and Angular Momentum 12.1.1. 14. 12.2.4. 15.4.1. Symmetries and Their Consequences 11. 11.4.5.2.2. Introduction What is the Nature of Spin? Kinematics of Spin .3. Some Examples 17.5.5. TimeIndependent Perturbation Theory 17.4. 15. 12. The Formalism 17. .1. 373 373 374 385 397 403 403 408 416 421 429 429 435 451 451 454 464 15. . . Translations in Two Dimensions.3. 13.1. 12. 14. . . Addition of Angular Momenta 15.2. . A Simple Example The General Problem Irreducible Tensor Operators Explanation of Some "Accidental" Degeneracies. . 13.3. Overview Translational Invariance in Quantum Theory Time Translational Invariance Parity Invariance TimeReversal Symmetry 279 279 279 294 297 301 305 305 306 313 318 321 339 . 14.4. 353 353 359 361 369 373 xvii CONTENTS Hi ^ ^ 12.3. The Variational Method .1. Rotations in Two Dimensions The Eigenvalue Problem of L z Angular Momentum in Three Dimensions The Eigenvalue Problem of L 2 and Lz Solution of Rotationally Invariant Problems 13.3. Spin Dynamics Return of Orbital Degrees of Freedom . . .
18. 19.3. Scattering Theory . 19. Derivation of the Path Integral Imaginary Time Formalism Spin and Fermion Path Integrals Summary Appendix A. .6. .5. 20.4. 1. .3. 19. 19.1. . . 18.4. 19. . Introduction Recapitulation of OneDimensional Scattering and Overview .2. The Problem FirstOrder Perturbation Theory Higher Orders in Perturbation Theory A Genera] Discussion of Electromagnetic Interactions Interaction of Atoms with Electromagnetic Radiation 473 473 474 484 492 499 523 523 524 529 534 545 555 563 563 566 574 581 582 613 636 652 655 .1. The Dirac Equation 20.3. 18.4.4. 21. .2. . The Partial Wave Expansion TwoParticle Scattering 20. . . Matrix Inversion Gaussian Integrals Complex Numbers The is Prescription 655 659 660 661 665 669 671 ANSWERS TO SELECTED EXERCISES TABLE OF CONSTANTS INDEX . 19. 21. Born Again (The TimeIndependent A p p r o x i m a t i o n ) . . The FreeParticle Dirac Equation Electromagnetic Interaction of the Dirac Particle More on Relativistic Quantum Mechanics . 21. 18. .XVlii CONTENTS 18. . A. . 21. The Born Approximation (TimeDependent Description) . . A.2.1.2.2. 19. . TimeDependent Perturbation Theory 18. Path Integrals—II 21.1.3.3. 20. A.5.
But note also that conspicuously missing are the requirements that every vector have a magnitude and direction. you will have ample time to appreciate the wisdom behind this choice as . The following is the list of properties the mathematicians have wisely chosen as requisite for a vector space. the baby has been thrown out with the bath water.1. and say that any set of objects obeying the same forms a linear vector space. you know that scalar multiplication is distributive: the multiple of a sum of two vectors is the sum of the multiples. All the math you will need is developed here. as with any other chapter. However. The effort you put into this chapter will be well worth your while: not only will it prepare you for this course. You know how to add them and multiply them by scalars and the rules obeyed by these operations. Numerous examples and exercises related to classical mechanics are given. displacement. etc. It is the aim of this chapter to equip you with the necessary mathematical machinery. you must. both to provide some relief from the math and to demonstrate the wide applicability of the ideas developed here. starting from its axioms. What we want to do is abstract from this simple case a set of basic features or axioms. If you keep too many. For example. Linear Vector Spaces: Basics In this section you will be introduced to linear vector spaces. starting from some basic ideas on vectors and matrices that you are assumed to know.Mathematical Introduction The aim of this book is to provide you with an introduction to quantum mechanics. To really learn this chapter. 1. force. which was the first and most salient feature drilled into our heads when we first heard about them. torque. The cleverness lies in deciding which of the properties to keep in the generalization. work out the problems. So you might think that in dropping this requirement. As you read them. there will be no interesting results to develop from the axioms. there will be no other examples. You are surely familiar with the arrows from elementary physics encoding the magnitude and direction of velocity. if you keep too few. but it will also unify many ideas you may have learned piecemeal. please compare them to the world of arrows and make sure that these are indeed properties possessed by these familiar vectors.
For the second start with 0> = { 0 + l) +  . denoted  V) + \ W) 2.2 CHAPTER l you go along and see a great unification and synthesis of diverse ideas under the heading of vector spaces. K>) = 0 F> = 0>.. Definition 1. 1.  — Vy is the unique additive inverse of  Vy. Verify these claims. Let us note that the above axioms imply • • • • 0> is unique. for which there exists jl>.. do what comes naturally.. we have a complex vector space. a{b\ V"y) = ab\ V>.. i.. • Scalar multiplication is distributive in the scalars: (a + b) V}=a\ V') + b\ V).. .. m There exists a null vector \ 0> obeying  V) +10) =  V). .  W>. For the last. While you should be duly impressed with all this..b. denoted a\ V) with the following features: • The result of these operations is another element of the space. For the first consider 0> +10') and use the advertised properties of the two null vectors in turn. . such that \v> + \v> = \oy. . if they are complex. A definite rule for multiplication by scalars a.1. this means \ Vy + \wy=\Vy + \Vy. space is defined. A definite rule for forming the vector sum. You don't have to know the proofs.. Definition 2.  — V). There is a good way to remember all of these... a feature called closure: \ V) + \W)eV. 0F>=0>. but you do have to know the statements.V). begin with  r > + ( .. • Scalar multiplication is distributive in the vectors: a( V) + \ W)) = a\V} + a\W). • Scalar multiplication is associative. m For every vector  V) there exists an inverse under addition.e. For the third. we have a real vector space. if 0') has all the properties of 0>. The proofs are left as to the following exercise.. You will see examples of vector spaces that involve entities that you cannot intuitively perceive as having either a magnitude or a direction. called vectors. then 0> = 0'>..1. are called the field over which the vector If the field consists of all real numbers. let  w y also satisfy \vy + \W) = 0>. The numbers a. A linear vector space V is a collection of objects  2 > . Since 0> is unique. Exercise 1.. . Take it from here. • Addition is commutative:  V) +1 wy = \ W) +1 V}.  V). The vectors themselves are neither real nor complex.b. the adjective applies only to the scalars. • Addition is associative:  V) + ( W > + Z>) = ( V) + \W)) + \Z). remember that it does not hurt at all to think of these generalizations in terms of arrows and to use the intuition to prove theorems or at the very least anticipate them. \—vy=—\ vy..
while the inverse of a vector is the vector reversed in direction. This object is called ket V and this nomenclature is due to Dirac whose notation will be discussed at some length later. c) + (d. we may want to refer to it as. ab. Addition and scalar multiplication clearly have all the desired associative and distributive features. The point is that while the arrows have these qualities. You were asked to verify that the set of arrows qualified as a vector space as you read the axioms. Addition and scalar multiplication are defined as follows: (a. b. e. For example.2.1.1. b. we interpret it as changing the direction of the arrow as well as repealing it by a. Scalar multiplication by a corresponds to stretching the vector by a factor a. typical ones being Vand V\ The rule for addition is familiar: take the tail of the second arrow. b. So the set of all arrows qualifies as a vector space. the set of all arrows with positive zcomponents do not form a vector space: there is no inverse. The vector space consists of arrows.) Since these operations acting on arrows give more arrows.1. and distributive requirements. You are however not discouraged from associating with \ V) the arrowlike object till you have seen enough vectors that are not arrows and are ready to drop the crutch. associativity.Figure 1. The corresponding rules obey closure. c). 1) do not form a vector space. (If a is negative. ac). c) = (aa. This statement is pointless unless I can give you examples. But we cannot tamper with it. Exercise 1. When we want to highlight the fact that the matrix M is an element of a vector space. Consider the set of all entities of the form (a. We know how to add them and multiply them by scalars (multiply all four matrix elements by that scalar). and so on as in Fig. Consider the set of all 2 x 2 matrices. Show that vectors of the form (a. Note that so far. The null matrix has all zeros in it and the inverse under addition of a matrix is the matrix with all elements negated. Note that it obeys axioms MATHEMATICAL INTRODUCTION (O(iii). You must agree that here we have a genuine vector space consisting of things which don't have an obvious length or direction associated with them. so here are two. This is a real vector space since stretching by a complex number makes no sense. Write down the null vector and inverse of (a. we have closure. Here are some of the key ideas you should have gone over. / ) = (a + d. say. no reference has been made to magnitude or direction. c) where the entries are real numbers. The rule for vector addition. members of a vector space need not. b. put it on the tip of the first. c + f ) a(a. b + e. The null vector is the arrow of zero length. Observe that we are using a new symbol  V} to denote a generic vector. We do not purposely use the symbol V to denote the vectors as the first step in weaning you away from the limited concept of the vector as an arrow. ket number 4 or: 4>r . 1. b.
\n). such a relation will exist. There is no way to write one as a multiple of the other. or equivalently.1.1. This is. We define scalar multiplication by a simply as af(x) and addition as pointwise addition: the sum of two functions/and g has the value / ( x ) + g(x) at the point x.)'. Exercise 1. As a concrete example.1. we can clearly write one as a multiple of the other or equivalently play them against each other to get 0. it could be shifted to the right. On the other hand. So let us suppose it is not. (1.12) thereby expressing 3> in terms of the others. if the set of vectors is linearly dependent. as a linear combination of the other two. Equation (1.1) tells us that it is not possible to write any member of the linearly independent set in terms of the others. The null function is zero everywhere and the additive inverse of / is .4 CHAPTER 1 As a second example. Let us say 0. we say they are linearly dependent. If it is parallel to either of the first two. It is.) Definition 3. To find the combination.' How about functions that obey / ( 0 ) = 4 ? If the functions do not qualify.f . The next concept is that of linear independence of a set of vectors 11 >.. consider two nonparallel vectors 11) and 2) in a plane. incorrect since a set of vectors can only add up to a vector and not a number. This is because we can write one of them. .3. consider all functions f(x) defined in an interval 0<x<Z. The set of vectors is said to be linearly independent if the only such linear relation as Eq. common to represent the null vector by 0. draw a line from the tail of 'i3> in the direction of 1>..1) is the trivial one with all at=0. 2> . list the things that go wrong. Notice I said 0 and not 0>. But even now the three of them are linearly dependent. and combined with the 0> there to give 0> once more. Next draw a line antiparallel to 2> from the tip of 13). On the other hand. Suppose we bring in a third vector 3) also in the plane. If the set of vectors is not linearly independent.1. no way to combine them to get the null vector. First consider a linear relation of the form X «f/> = 0> i=*i (1. we already have a linearly dependent set. strictly speaking. These lines will intersect since 11) and 12> are .1) We may assume without loss of generality that the lefthand side does not contain any multiple of 0>. and it must contain at least two nonzero coefficients. if the vectors are parallel. Then we could write 3>= £ d. however. These form a linearly independent set. say 3>. for if it did. (We are using the fact that any multiple of 0> equals 0>. D o functions that vanish at the end points Jf = 0 and x = L form a vector space? How about periodic functions obeying / ( 0 ) =/(/.
1). which is not possible in an «dimensional space by definition. So the space is at least fourdimensional. Are they linearly independent? Support your answer with details. Consider three elements from the vector space of real 2 * 2 matrices: "o .not parallel by assumption. if they are complex we have a complex fourdimensional space. the plane is twodimensional and the set of all arrows not limited to the plane define a threedimensional vector space. 0 .4. Could it be bigger? No. In view of the earlier discussions. 0. and ( 0 . 3> = "2 .0).0 f 0. The intersection point P will determine how much of 1) and 2> we want: we go from the tail of 3> to P using the appropriate multiple of 11> and go from P to the tip of 3> using the appropriate multiple of 2>.) Exercise 1. t\ d are real. b. Here is a proof. ( 1 . (t. Any vector  V") in an ^dimensional space can be written as a linear combination of n linearly independent vectors ' ! > • • . How about 2 x 2 matrices? They form a fourdimensional vector space. it would join the given set of vectors and form a set of n + 1 linearly independent vectors. It will be denoted by V"(/i) if the field is real and by V"(C) if the field is complex. we have a real fourdimensional space. 2. . 2> = "l _Q 5 MATHEMATICAL INTRODUCTION \ 1. since any arbitrary 2 x 2 matrix can be written in terms of them: a c b d = a  l > + &2> + c3> + </4> If the scalars a. A vector space has dimension n if it can accommodate a maximum of n linearly independent vectors.5.1. 1. 1 . 4> — 0 0 0" 1 since it is impossible to form linear combinations of any three of them to give the fourth any three of them will have a zero in the one place where the fourth does not. 0 1 2. !«>• The proof is as follows: if there were a vector  K> for which this were not possible. Theorem 1. 1 ) . Show that the following row vectors are linearly dependent: (1. (Notice we are calling these matrices vectors and using kets to represent them to emphasize their role as elements of a vector space. Definition 4.1). 1 . Exercise 1. 2> = 0 0 1 0 3> = 0 1 0 0. I).1. and (3. 0 ) .0 1> = 0 0. Show the opposite for ( 1 . The following vectors are linearly independent: 1 .
3) is unique. Imagine for example three arrows in the plane. having an existence of its own and satisfying various relations involving other vectors.1. (1.6) (1.1.1. .4) from Eq. (1. The expansion in Eq.3) (i. Note that given a basis the components are unique.1. (1. Suppose the expansion is not unique. If we choose a different basis. Now we choose a basis and write each vector in terms of the components.e. the components will change in numerical value.. Thus we can write.4)!i> i which implies that vi=vi (1. on the strength of the above l O = £ !>iU> i=\ where the vectors [/> form a basis. A. So far no basis has been chosen and we do not need a basis to make the statement that the vectors from a closed triangle. but the relation between them expressing the equality of C to the sum of the other two will still hold between the new set of components. the components will change. A set of n linearly independent vectors in an «dimensional space is called a basis.4) Subtracting Eq. multiplying the second by the scalar — 1 and adding the two equations) we get 0>=2>.6 CHAPTER 1 Definition 5. We refer to  V} as the vector in the abstract. B.5) since the basis vectors are linearly independent and only a trivial linear relation between them can exist. but if we change the basis. The components will satisfy Cl = At+Bi.3) Definition 6. The coefficients of expansion y. 2. C satisfying A + BC according to the laws for adding arrows. (1. i= 1.1.1. We must then have a second expansion: \V}=t i= 1 (1. Theorem 2.1. of a vector in terms of a linearly independent basis (/>) are called the components of the vector in that basis. When we choose a basis the vectors assume concrete forms in terms of their components and the relation between vectors is satisfied by the components.
Recall that the dot product has a second .2. Inner Product Spaces The matrix and function examples must have convinced you that we can have a vector space with no preassigned definition of length or direction for the elements.1.9) where we have used the axioms to carry out the regrouping of terms.1) we can read off the length of say A as J\A\ • \A\ and the cosine of the angle between two vectors as A • B/\A\\B\.8) \ r y + \ir> = il(vl + w l m ( l .w. from the dot product AB=\A\\B\ cosfl (1. If 7 MATHEMATICAL INTRODUCTION  J O = Ii>. To multiply a vector by a scalar. add their components. Here is the conclusion: To add two vectors. we can add them either using the tail and tip routine or by simply adding their components in a basis.In the case of nonarrow vectors. we can make up quantities that have the same properties that the lengths and angles do in the case of arrows. if the dot product itself requires knowledge of the lengths and angles? The answer is this. The first step is to define a sensible analog of the dot product.li> i I Wy = Y. There is no reference to taking the tail of one and putting it on the tip of the other. In the same way. = Lavt\0 i (11. Now you might rightfully object. multiply all its components by the scalar. etc.7) then (1.\0 J and (1. since in general the vectors have no head or tail.. i . Of course. we have: a\n=aZvt\0 t In other words. adding them in terms of components proceeds as in the elementary case thanks to the axioms. However. how can you use the dot product to define the length and angles.1.2. for in the case of arrows.10) 1. if we are dealing with arrows.
To this end we recall the main features of the above dot product: 1.V\Z') (linearity in ket) Definition 7. With a view to finding such a rule. A B = B A (symmetry) 2. We denote it by the symbol < V\ W). If we are going to define the length of the vector as the square root of its inner product with itself (as in the dot product) this quantity had better be real and positive for all nonzero vectors. 1.2. Geometrical proof that the dot product obeys axiom (3) for an inner product. (1. let us note that this axiom ensures that < V\ V> is real. equivalent expression in terms of the components: A • B = AXBX + AyBy + AZB. Notice that we have not yet given an explicit rule for actually evaluating the scalar product. with the two choices leading to complex conjugates. A.A>0 0 i f f A = 0 (positive semidefiniteness) 3. We want to invent a generalization called the inner product or scalar product between any two vectors  V) and  W).2. A vector space with an inner product is called an inner product space. The second axiom says that < V\ V) is not just real but also positive semidefinite. The first differs from the corresponding one for the dot product and makes the inner product sensitive to the order of the two factors.2. A • (bB + cC) = bA • B+ cA • C (linearity) The linearity of the dot product is illustrated in Fig. It is once again a number (generally complex) dependent on the two vectors. vanishing only if the vector itself does. The axiom requires that the projections obey Pk + P/=Pjk.l fli.2) Our goal is to define a similar formula for the general case where we do have the notion of components in a basis.jk Pj — ! vi 1 Pj* Figure 1. let us familiarize ourselves with the axioms. For the present. In a real vector space this axioms states the symmetry of the dot product under exchange of the two vectors. We demand that it obey the following axioms: • <y W) = {W\ V)* (skewsymmetry) • < V\ V} > 0 0 i f f  V) = 0> (positive semidefiniteness) • (V\{a\ W) + b\Z)) = (VlaW+bZy=a(V!W} + b<. we are merely demanding that any rule we come up with must have these properties. .
In other words. A set of basis vectors ail of unit norm. We are now ready to obtain a concrete formula for the inner product in terms of the components. Let us continue with inner products.V\aW+ bZ>* 9 MATHEMATICAL INTRODUCTION  =a*(W\V> wy+b*<y\zy + b*(Z\V) (1. Definition 9. the inner product of a linear superposition with another vector is the corresponding superposition of inner products if the superposition occurs in the second factor. the inner product between basis vectors. which are pairwise orthogonal will be called an orthonormal basis.4) j To go any further we have to know <i/>. This asymmetry. 1. i.3) which expresses the antilinearity of the inner product with respect to the first factor in the inner product. We say that two vectors are orthogonal or perpendicular if their inner product vanishes. Even though we are trying to shed the restricted notion of a vector as an arrow and seeking a corresponding generalization of the dot product.  V\ as the norm or length of the vector. Given  V) and  W) I K> = 2 >  i > i I ^ > = 1 ^ 1 j> j we follow the axioms obeyed by the inner product to obtain: = ' (1.e. unfamiliar in real vector spaces. What if the first factor in the product is a linear superposition.. Definition 8. what is (aW+bZ\ F>? This is determined by the first axiom: W+ bZ\ V} = <.2. while it is the superposition with all coefficients conjugated if the superposition occurs in the first factor.2). We will refer to J(V  A normalized vector has unit norm.The last axiom expresses the linearity of the inner product when a linear superposition a\W} + b\Zy = \ a W + bZy appears as the second vector in the scalar prod* uct. We will also frequently refer to the inner or scalar product as the dot product. We have discussed its validity for the arrows case (Fig. we still use some of the same terminology.2. Definition 10. is here to stay and you will get used to it as you go along. That depends on the details of the basis vectors and all we know for sure is that .
if we use an orthonormal basis such as i.J>_ fl io for i=j _ f e w " * ' where StJ. Since the vector  V> is uniquely specified by its components in a given basis. < V\ V) would not even be real.2. j. the dot product of any two of them will likewise be a double sum with four terms (determined by the four possible dot products between the basis vectors) as well as the vector components. But now it is given by <FF> = £  y i  2 > 0 i (1. (1.2.5) This is the form of the inner product we will use from now on. but for the complex conjugation of the components of the first vector. to give <KFF>=5>iV/ t (1.they are linearly independent. (1. If we write all vectors in terms of this basis. This situation exists for arrows as well. Consider a twodimensional problem where the basis vectors are two linearly independent but nonperpendicular vectors.5). Postponing the proof for a moment. we invoke Theorem 3. Given a linearly independent basis we can form linear combinations of the basis vectors to obtain an orthonormal basis. This makes it sensible to refer to < V\ V) as the length or norm squared of a vector. let us assume that the procedure has been implemented and that the current basis is orthonormal: _ <. For the more genera] nonarrow case.2. However.B = AXBX+AyBy depending only on the components.2. only diagonal terms like 11) will survive and we will get the familiar result A .7) . not to mention positive. we may. Consider Eq.is called the Kronecker delta symbol Feeding this into Eq. in this basis.6) and vanishes only for the null vector. You can now appreciate the first axiom. Theorem 3 (GramSchmidt).4) we find the double sum collapses to a single one due to the Kronecker delta. write it as a column vector: Vl Vt V> in this basis (12.
Dual Spaces and the Dirac Notation There is a technical point here.2.. but there is a way to make a number by matrix multiplication of a row times a column. Our trick for producing a number out of two columns has been to associate a unique row vector with one column (its transpose conjugate) and form its matrix product with the column representing the other. or transpose conjugate.W. The abstract bra associated with this will bear the same label.2. and form a row vector. The inner product is a number we are trying to generate from two kets  V") and j W").3. But one can also take the following alternate view. which are both represented by column vectors in some basis. This has the feature that the answer depends on which of the two vectors we are going to convert to the row. while the basis bra <(] is a row vector with all zeros except for a 1 in the (th column. the space of kets and a dual space of bras.e. Thus there are two vector spaces. We can also work backward and go from the column vectors to the abstract kets. i. with a ket for every bra and vice versa (the components being related by the adjoint operation).9) 1.itf v*] (1. There is a basis of vectors i> for expanding kets and a similar basis (i' for expanding bras. Column vectors are concrete manifestations of an abstract vector J V} or ket in a basis. The basis ket i> is represented in the basis we are using by a column vector with all zeros except for a 1 in the ith row. Inner products are really defined only between bras and kets and hence from elements of two distinct but related vector spaces. Let us take its adjoint. . Associated with every ket  V} is a column vector.Likewise Wi w2 wy in this basis (1. it will be called (V\. the two choices ((V\W} and (fV F}) leading to answers related by complex conjugation. But then it is similarly possible to work backward and associate with each row vector an abstract object < called bra.8) 11 MATHEMATICAL INTRODUCTION The inner product (V\ fV) is given by the matrix product of the transpose conjugate of the column vector representing \ V> with the column vector representing  W ) : Wi <V\ W>«[t>f. Now there is no way to make a number out of two columns by direct matrix multiplication. Now we can name the bras as we want but let us do the following.
5).3.4) i.3) (1. To find the components that go into the expansion we proceed as follows.3. Using this result we may write f ( Let us make sure the basis vectors look as they should.6) . If we set  V) — y) in Eq.Li> f (1.3. Expansion of Vectors in an Orthonormal Basis Suppose we wish to expand a vector  K ) in an orthonormal basis. (1.3. • The inner product of any two vectors is given in terms of these components by Eq.2) (1. however.3.2. The abstract relation i r>I». but these may be complex.5). We take the dot product of both sides of the assumed expansion with : (or <j  if you are a purist) \V)Ivt\i> r O I O = I«iO"i> = v.. exactly as with arrows. (1." There is. we find the correct answer: the ith component of the jth basis vector is Thus for example the column representing basis vector number 4 will have a 1 in the 4th row and zero everywhere else. The only thing you must remember is that in the case of a general nonarrow vector space: • Vectors can still be assigned components in some orthonormal basis.1) where «+ means "within a basis. to find the yth component of a vector we take the dot product with the jth unit vector. This product obeys all the axioms. just as with arrows.e.1.12 CHAPTER 1 All this may be summarized as follows: V\ v2 IF) (1. If you found the above discussion heavy going.3. (1.3. nothing wrong with the first viewpoint of associating a scalar product with a pair of columns or kets (making no reference to another dual space) and living with the asymmetry between the first and second vector in the inner product (which one to transpose conjugate?). 1. you can temporarily ignore it.
3. What we have found is that h JaW=<V\a* (1. you will see that this follows simply from the fact that if two columns are equal. Adjoint Operation We have seen that we may pass from the column representing a ket to the row representing the corresponding bra by the adjoint operation.3.3. so are their transpose conjugates. (1. i.becomes in this basis T "0" 1 D" 13 MATHEMATICAL INTRODUCTION v2 U] 0 • + v2 0 _0_ + • •• v„ 0 • J .11) The two equations above are said to be adjoints of each other. Let us now ask: if is the bra corresponding to the ket  V) what bra corresponds to a\ V) where a is some scalar? By going to any basis it is readily found that av i av2 a\ vy L^J It is customary to write a\ K> as  aVy and the corresponding bra as <a V\.7) 1.10) this implies another one among the corresponding bras: <V\a* = (W\b* + (Z\c* + • (1.3.3.e. Just as any equation involving complex numbers implies another obtained by taking the complex conjugates of both sides.9) Since the relation between bras and kets is linear we can say that if we have an equation among kets such as a\ vy=b\ wy + c\zy + (1. an equation between (bras) kets implies another one between (kets) bras.3. Here is the rule for taking the adjoint: . If you think in a basis..2. transpose conjugation.8) (1.
Suppose we have an expansion for a vector: W ) = z m i.12) 0\v* Recalling that v. we do the following: • Rescale the first by its own length.1 in terms of basis vectors. Let us take two nonparallel vectors.14 CHAPTER i (b ra T o take the adjoint of a linear equation relating kets (bras). replace every ket ) by its bra (ket) and complex conjugate all coefficients. so it becomes a unit vector. This simple example teUs the whole story behind this procedure. exchanging bras and kets and complex conjugating all coefficients.= <i V) and v* = (V\i). Imagine the twodimensional space of arrows in a plane. leaving behind only the part perpendicular to the first. • Subtract from the second vector its projection along the first. reverse the order of all factors. which will now be discussed in general terms in the Dirac notation. GramSchmidt Theorem Let us now take u p the GramSchmidt procedure for converting a linearly independent basis into an orthonormal one. which qualify as a basis. This will be the first basis vector. To get an orthonormal basis out of these. The basic idea can be seen by a simple example. We now have the second basis vector: it is orthogonal to the first and of unit length. We can extend this rule as follows.3. (Such a part will remain since by assumption the vectors are nonparallel.14) To take the adjoint of an equation involving bras and kets and coefficients. . it follows that the adjoint of ti313) (=i is <V\=1 i= i from which comes the rule: \o<i\v> <V\iXi\ (13. The adjoint is <V\= I i= 1 (1.) • Rescale the left over piece by its own length.
The first vector of the orthonormal basis will be l> = i ^ where  /  = V<7i7> Clearly As for the second vector in the basis. There is nothing new with the generation of the rest of the basis. Can you generate another orthonormal basis starting with these two vectors? If so. Dividing by its norm we get ]3>. produce another.) Exercise 1. the third member of the orthogonal basis. putting a stop to the whole procedure. (Think of the arrow example as you read on.  / / ) . . On the other hand.[ 1 > < 1  / / / ) . consider 3'> = [ / / / ) . Where did we use the linear independence of the original basis? What if we had started with a linearly dependent basis? Then at some point a vector like 2'> or 3'> would have vanished. Form an orthonormal basis in two dimensions starting with A = +• 4/ and B=2i6j. .1 2 ) < 2  / / / > which is orthogonal to both 11> and 2>. . . consider 2'> =  / / > . linear independence will assure us that such a thing will never happen since it amounts to having a nontrivial linear combination of linearly independent vectors that adds up the null vector. Finally. (Go back to the equations for 2'> or J3') and satisfy yourself that these are linear combinations of the old basis vectors.Let />.3.) Not surprisingly it is orthogonal to the latter: <l2'> = < l  / / > . l > < l  / / > which is //> minus the part pointing along the first unit vector. be a linearly independent basis.< l  l > < l  / / > = 0 We now divide 2'> by its norm to get 2> which will be orthogonal to the first and normalized to unity.1.
15)  V+ W\ < I V\ +1 w\ (1. S h o w how to go from the basis "3" 0 A i _2_ _ "0" }///> = 2 .3. When we first learn about dimensionality. first note that any mutually orthogonal set is also linearly independent. Suppose we had a linear combination of orthogonal vectors adding up to zero. The dimensionality of a space equals « i . The second will be left as an exercise. we associate it with the number of perpendicular directions. The following theorem connects the two definitions. greater than or lesser than n.3. the maximum number of mutually orthogonal vectors in it. 1> = 0 1/J5 2/V3. . to the orthonormal basis r f 2> = 0 0 3> = 2/V5 _ 1A/5 . while the linear independence of the perpendicular vectors rules out the penultimate option. The Schwarz Inequality \{v\wy\<\v\\w\ Theorem 6. The Gram Schmidt procedure eliminates the last case by explicit construction. T o show this.16 CHAPTER I Exercise 1. showing the linear combination is trivial. the dimensionality of the space.16) The proof of the first will be provided so you can get used to working with bras and kets. Now l can only be equal to. This can clearly be done for all the coefficients. Theorem 4. By taking the dot product of both sides with any one member and using the orthogonality we can show that the coefficient multiplying that vector had to vanish. l0. The Triangle Inequality (1.2.3. Tn this chapter we defined it in terms of the maximum number of linearly independent vectors.5. Schwarz and Triangle Inequalities Two powerful theorems apply to any inner product space obeying our axioms: Theorem 5.
Given a vector space V. Exercise 1.4. a subset of its elements that form a vector space among themselvesj is called a subspace. To prove the Schwarz inequality. This is an example which illustrates the merits of thinking of abstract vectors as arrows and guessing what properties they might share with arrows.3. Subspaces Definition 11. Using <w\ vy*=(v\ we find wy <w\ vy(v\ wy wy Crossmultiplying by  W\ and taking square roots. the result follows. consider axiom < Z  Z ) > 0 applied to \Z> = \ V } .3. Prove the triangle inequality starting with \V+W\2. with arrows? When will this equality be satisfied? D o e s this agree with your experience Exercise 1.18) where we have used the antilinearity of the inner product with respect to the bra. 1. Show that the final inequality becomes an equality only if  F> = a\ wy where a is a real positive scalar. We will denote a particular subspace i of dimensionality by V"'.^ ^ \ W > We get <zjz>=<K 17 MATHEMATICAL INTRODUCTION <W\Vy (1.3.• Before proving anything. You must use Re< fV)\ and the Schwarz inequality.3.3.4. t Vector addition and scalar multiplication are defined the same way in the subspace as in V. The proof will of course have to rely on just the axioms. note that the results are obviously true for arrows: the Schwarz inequality says that the dot product of two vectors cannot exceed the product of their lengths and the triangle inequality says that the length of a sum cannot exceed the sum of the lengths. .17) <w\vy ' / w\ v\w\2 <w\vy ' / wy \w\2 < w\ vy *< w\ vy  wy =<v\vy <w\ K>< v\ wy \w\2 vy(w\wy 4 + (w\ vy*(w\ j i>0 \w\ (1.
Exercise 1.} Sl{a F/> + p\ Vj>} = aQ\ F. Suppose VJ" and V"2 are two subspaces such that any element of V] is orthogonal to any element of V 2 .Example 1. } .  F'>. the following are some examples of subspaces: (a) all vectors along the x axis. we would get the correct answer.4..2. (Hint: Theorem 4.4.1* In a space V .> + < F f  a n = <F / ncr «V(\a + <Vj\P)Q = a{Vi\Q + p{VJ\n F. contained only vectors along the x and y axes. one from each direction. Given two subspaces V"' and Vp. prove that the set of all vectors { V±).5.4b) . . • Definition 12.  F i > .> (1. Example 1. (b) all vectors along the y axis. form a subspace V" . we could.) 1. Operators can also act on bras: <F'fi*=<F" (1. Show that the dimensionality of V'i©V 2 is n]+n2. the space V*1. orthogonal to any  F> # 0>. if  F> is an element of a space V. (c) all vectors in the xy plane.1. i.5. But for the elements (3). the space V^1. if we also included all linear combinations. we define their sum V ? ® V p = V** as the set containing (1) all elements of V'. the space V ^ . The action of the operator is represented as follows: n\v)=\v) (1.. . generate one along neither.1) One says that the operator H has transformed the ket  F ) into the ket  F'>. Thus the set of all vectors along the positive J t axis alone do not form a vector space.4.3b) (1. so is ) F'> = Q F>. ones that obey the following rules: QalV^^aQlV. Notice that all subspaces contain the null vector and that each vector is accompanied by its inverse to fulfill axioms for a vector space. Linear Operators An operator fl is an instruction for transforming any given vector j F ) into another.1 .5.4.5.5.2) We will only be concerned with linear operators.5. .5.e. On the other hand. If. closure would be lost.2.e. In the space V 3 (i?).4a) (1. = • Exercise 1. by adding two elements.3a) (1. We j will restrict our attention throughout to operators fi that do not take us out of the vector space. (2) all elements of (3) all possible linear combinations of the above. for example. i.
 2 > . .5) We next pass on to a more interesting operator on V 3 (R): R(\ n\)+ Rotate vector by \ it about the unit vector i [More generally.) + ifl3>  2 > + (3> MATHEMATICAL INTRODUCTION Example 1.7a) (1.5. I r > (1.] Let us consider the action of this operator on the three unit vectors i.8) .7c) learly is linear.6) for all kets  V) (1. it is clear f r o m the same figure that i?[2) + 3)]= J R2> + /?3). then for any  V) = £ u . /> ni V)=I QVi I *>=I Vl I qI i)=I v.1. • The nice feature of linear operators is that once their action on the basis vectors is known. which in our notation will be denoted by 11). /. . Note that /?[2) + 3>] = /?2> + /?3> as expected of a linear operator. I JO = F> and P <K/=<K for all bras < V\ (1. and 3> (see Fig.7b) (1. which carries the instruction: I*Leave the vector alone! Thus.5.\n) in V".5. For instance. 1. 7?(0) stands for a rotation by an angle 9 = 19 about the axis parallel to the unit vector 0 = 8/0. . 2 > (1.19 *[2> 3>] Figure 1. 2>. Action of the operator /?(i7ri). j. and k. The simplest operator is the identity operator. .5. their action on any vector in the space is determined.5.5. (We will often Tefer to R(ixi) as R if no confusion is likely.3).3. If ni> = i o for a basis 11).5. From the figure it is clear that *(iffi)l> = l> /?(Jffi)2> = 3> *(Jffi)3> = .
The operator R(\N\) has a n inverse: it is R(—s^ri).9) where  f i V) is the ket obtained by the action of Q on  V). Q A . then R] V> = v. 0 ] + [Q. F o r example R{\ITI) not commute. Q 'fi = / . i. Prove this using the ideas introduced toward the end of .R\ 1 > + V2R\2) + V3RI3) = 1 ) + U 2 3> .5.U 3 J2> The product of two operators stands for the instruction that the instruction! corresponding to the two operators be carried out in sequence AQF) = A(QF)) = AQK) (1. denoted by Q" 1 . these rules resemble the chain rule in calculus for the derivative of a product. A 0 ] = A[Q. The order of the operators in a product is very important: in general. 0 ] = A [ Q . The inverse of Q. 1 = /. Appendix A. The condition for the existence of the inverse is given in Appendix A . Two useful identities involving commutators are [Q.. their commutator is nonzero.10) (1.II) [ A O . called its components. satisfies! QQ _ 1 =0 _ I Q = 7 (1.l. If \V) = Vi\\) + v2\2) + U3 3> is any vector.A Q = [fi.' Q " ' ) = Q A A _ 1 Q _ 1 = QQ"1 = / 1. l . 0]FI Notice that apart f r o m the emphasis on ordering. The inverse of a product of operators is the product of the inverses in reverse: (QA)1 = A 'Q1 (1.6.5. 0 ] + [ A .e.12) N o t every operator has an inverse.13) for only then d o we have ( Q A ) ( Q A ) _ 1 = ( Q A X A .5.5. 1. A] called the commutator of Q and A isn't zero.» f i f i Theorem A. Matrix Elements of Linear Operators We are now accustomed to the idea of an abstract vector being represented in a basis by an ntuple of numbers.1.This is the case in the example £} = i?(57ri).5. in terms of which all vector I In V"(C) with n finite.. A ] 0 and R{\/rj) do (1.
operations can be carried out.6.2) Equation (1.6. they facilitate the computation of all basisindependent quantities. A mnemonic: the elements of the first column are simply the components of the first transformed basis vector  l f > = Q 1> in the given basis. by rendering the abstract operator more tangible. are basis dependent. written as an n x n matrix. just like the vector components.3) y'n. and called its matrix elements in that basis. Although the matrix elements. . Likewise. the elements of the }th column represent the image of the jth basis vector after Q acts on it. We shall now see that in the same manner a linear operator can be represented in a basis by a set of n2 numbers.  Q  0 = Q// (1.i'> = < . are the matrix elements of Q in this basis. Qy.6. then any vector in this space undergoes a change that is readily calculable: i l i When we say /'> is known.1) are known. The n2 numbers. we mean that its components in the original basis <.6.>) j = Z OGVJ (1. that the action of a linear operator is fully specified by its action on the basis vectors. If the basis vectors suffer a change 21 MATHEMATICAL INTRODUCTION mono (where  i') is known). Our starting point is the observation made earlier. If Q\V) = \ V ) then the components of the transformed ket  V) are expressable in terms of the Qg and the components of  V ) : vl=<i I v>=<i v>=<i « y .2) can be cast in matrix form: vr = <1Q1> <2Q1> _<«Q1> <m> n »" v2 <«Qn>J vn (1.
1. One may also ignore the mnemonic altogether and simply use the definition 7?. A n operator Q is given by the matrix 0 1 0 0 1 0 L 1 O O J What is its action? Let us now consider certain specific operators and see how they appear in matrix form. Combining our mnemonic with the fact that the operator has the following effect on the basis vectors: * ( i * i )  l > = l> * ( } * i )  2 > = 3> *(ijri)l3> = . You should verify that our mnemonic gives the same result. Example 1.4) For instance. Consider the expansion of an arbitrary ket  V) in a basis: W ) = i i= I /><iK> . (1) The Identity Operator I.1.6. Iij=<i\I\j) = <i\j> = 5i/ (1.5) Thus I is represented by a diagonal matrix with l's along the diagonal. 2 > j ! we can write down the matrix that represents it in the j 1).  2 > .6. Let us first get acquainted with projection operators.6.1 0 (1.y= </'7?. • Exercise 1. the .') to compute the matrix.6. 3> basis: 1 0 0 1 0 0 0 .22 CHAPTER 1 Convince yourself that the same matrix acting to the left on the row vector corresponding to any <u' gives the row vector corresponding to <u" = < u '  Q . (2) The Projection Operators.1 in the third column tells us that R rotates 3> into — 12).
says that the sum of the projections of a vector along all the n directions equals the vector itself.6.. P. it is called a projection operator. by definition.  Vy is a multiple of /> with a coefficient (u.) which is the component of [ Vy along />. .7) The object P . and (2) the subsequent application of Pj(j ^ i) will result in zero. (1.projects out the part of  F ) along />. and which. (1. expresses the identity as a sum over projection operators and will be invaluable to us. The completeness relation. is linear.9) Projection operators corresponding to the basis vectors obey P.6) Since Eq. Eq. further applications of P. Projection operators can also act on bras in the same way: <KP.6.7). just wait and see.In terms of the objects i></.6) is true for all  F > . which seems to do nothing. make no difference. projects out the component of any ket  F> along the direction />. is a waste of time. = i><i is called the projection operator for the ket />.6. (If you think that any time spent on the identity. Equation (1. which are linear operators. Notice that whatever  vy is. the object in the brackets must be identified with the identity (operator) / = ! IO</ = E ^ i= 1 (1.Pj=U"><i\j><j\=S9Pj (16.6. which is called the completeness relation.6.8) Clearly P.7).10) This equation tells us that (1) once P. since a vector entirely along /> cannot have a projection along a perpendicular direction _/>. act on  F> to give /><i' Vy. we may write the above as 23 MATHEMATICAL INTRODUCTION W> = ( j t /></)P> (1.6.6. Since P . = < F U > < /  = ^ < / 1 (1.) Consider pi\vy = \iya\vy = \iyvt (1.
We may equate the action of the polarizer to that of a projection operator Py that acts on the electric field vector E. We know 0 0 0 10 1 0 and </<[0. . There are two approaches. gives us a feeling for what kind of an object /)</ is. . . The action of the polarizers on the electric field E obeys the law of combination of projection operators: The following example from optics may throw some light on the discussion. An additional polarizer Py placed in the way has no further effect on the beam. 1 . PX and PV are polarizers placed in the way of a beam traveling along the z axis.. .4. the projection E cos 6 along the y axis is transmitted. The first one. . Thus the polarizers obey the equation PtPj= SjjPj expected of projection operators. that only admits light polarized along the y axis.4). 0 . is placed in the way. 0 ] . somewhat indirect.Figure 1. If a polarizer Py. 0 . . If Pv is followed by a polarizer Px the beam is completely blocked. . 1. 0 . Consider a beam of light traveling along the z axis and polarized in the x—y plane at an angle 9 with respect to the y axis (see Fig. . Let us next turn to the matrix elements of P.
11) 0 0 0 by the rules of matrix multiplication.11)... If we form the sum over just some of the projection operators. = </QAy> = <iQ/Ay> = E</Oifc><fcAy>=Xfi*A J t / k k (1. The inner product < V\ V ) represents a bra and ket which have found each other.1. Each projection operator has only one nonvanishing matrix element.6.0. Matrices Corresponding to Products of Operators Consider next the matrices representing a product of operators.12) which is of course identical to Eq. The more direct approach to the matrix elements gives (Pt)ki=<k\iXi\l) = 5ki5il=dkJdli (1.so that 25 0 0 0 0 0 MATHEMATICAL INTRODUCTION I'Xil 1 0 [0.. has the two factors looking the other way for a bra or a ket to dot with.. These are related to the matrices representing the individual operators by the application of Eq.6.0.. a 1 at the *th element on the diagonal. while  V)(V'\.. sometimes called the outer product.  V)( V'\ = (n x l matrix) x (1 x n matrix) = («xn matrix) is an operator.13) Thus the matrix representing the product of operators is the product of the matrices representing the factors.6.. (1. says that when all the P. are added.7)..6. The completeness relation. we get the operator which projects a given vector into the subspace spanned by just the corresponding basis vectors. Eq. The same result also follows from mnemonic.0] = 0 (1. The Adjoint of an Operator Recall that given a ket a\ V) = a V) the corresponding bra is (aV\ = (Via* (andnot (V\a) .6. (1. Whereas < F  K ' ) = (1 x n m a t r i x ) x (n x 1 matrix) = ( l x l matrix) is a scalar. the diagonal fills out to give the identity. (1.7): ( Q A ) .6.
First we treat QA as one operator and get <QAK = < ( Q A ) K  = < K  ( Q A ) t Next we treat (A V) as just another vector. Just as a and a*.6. (Recall that the row vector representing < V\ is the transpose conjugate of the column vector representing j V)." is best seen in a basis: (Q%=0\&\j) = <j\W* so « J = «Jf (1. Consider now an equation consisting of kets. and operators. then Q + turns the bra <K into <K'. the adjoint operation is the same as taking the transpose conjugate.) The adjoint of a product is the product of the adjoints in reverse: (QA) + = A t Q t (1. scalars. given a ket Q  F ) = QF> the corresponding bra is <QK = < r  D t (1.16) To prove this we consider < Q A F  .6. and + fi . In a given basis. One may state this equation in words: if Q turns a ket  V) to  V ) .26 CHAPTER 1 In the same way. the matrix representing Q f is the transpose conjugate of the matrix representing Q.15) = <M\j) = <j\n\i>* In other words. pushing further out: <AFQ+ = <FAtQ+ Comparing this result with the one obtained a few lines above. we get the desired result.6. and write <QAF =<fi(AK) = < A F  Q t We next pull out A.17a) .  V) and < F  are related but distinct objects.6.14) which defines the operator Q f . called the adjoint of £2 or "omega dagger. such as ail Vxy = a21 V2> + a 3  V3)(V4\ V5> + a 4 Q A  V6> (1. so are Q and The relation between £2.
6.2* Given Q and A are Hermitian what can you say about (1) OA.17a): < Vi I a f = < V2\ at + < V5\ V4)(V3\at + < V6\ A W a t (1. and (4) A]? . Definition 13. a <• a*. kets. (3) [Q. bras. reverse the order of all factors and make the substitutions Q <» Q + .6. (1.What is its adjoint? Our old rule tells us that it is <V.) Hermitian. Hermitian and antiHermitian operators are like pure real and pure imaginary numbers.6. and Unitary Operators We now turn our attention to certain special classes of operators that will play a major role in quantum mechanics.\at = <V2\a$h<V5\V4)<Vi\af In the last term we can replace <QA V6\ by < V6\ (QA) f = < K6 A f Q + so that finally we have the adjoint of Eq. An operator Q is Hermitian if Q + = Q. > <• <. An operator Q is antiHermitian if Q + = —Q.17b) + <QAV6\at 27 MATHEMATICAL INTRODUCTION The final rule for taking the adjoint of the most general equation we will ever encounter is this: When a product of operators. and explicit numerical coefficients is encountered. (2) OA + Ail. a +a aa a we can decompose every operator into its Hermitian and antiHermitian parts: Q + Qr QQ f Q = (1.18) Exercise 1. Just as every number may be decomposed into a sum of pure real and pure imaginary parts.6. Definition 14. The adjoint is to an operator what the complex conjugate is to numbers. (Of course. A]. AntiHermitian. there is no real need to reverse the location of the scalars a except in the interest of uniformity.
[i?(^7ri) is an example. Consequently. Exercise 1.y and I v2) = U\ V2) Then <V'2\V\) = <_UV2\UVl) . which defines an orthogonal or rotation matrix. An operator U is unitary if (1. from Eq.21) (Q.) Unitary operators are the generalizations of rotation operators from V^(R) to V ( C ) .20) Following the analogy between operators and numbers. If one treats the columns of an n x n unitary matrix as components of n vectors.19) This equation tells us that U and if are inverses of each other. the rotated set of vectors is also orthonormal. In fact.12). Proof 1. Proof. the j th column of the matrix representing U is the image of the y'th basis vector after U acts on it.( V2\ lflJ\V\) — ( V2\ V\) (1. lfv=l (1. these vectors are orthonormal. According to our mnemonic. Since U preserves inner products.28 CHAPTER 1 Definition 15.6.E. We now use the fact that if is also a rotation.5. u = e'0. (1. they preserve the lengths of vectors and their dot products. Theorem 7.6. (How else can it neutralize U to give 1" t IT ( / = / ? ) Since the rows of U are the columns of U (but for an overall complex .D. In the same way. Just as u*u= 1. unitary operators are like complex numbers of unit modulus. the rows may be interpreted as components of n orthonormal vectors. Unitary operators preserve the inner product between the vectors they act on. for just like rotation operators in three dimensions. the unitarity condition becomes U~] = i f (T means transpose).6. Consider next the rows.6.] Theorem 8.3* Show that a product of unitary operators is unitary. on a real vector space. so is lfU=I. Let \v\y = u\v.
Q.Lconjugation which does not affect the question of orthonormality). Exercise 1.* Verify that R(\ iz\) is unitary (orthogonal) by examining its matrix.2) . Proof 2. Exercise 1.7. (2) that det Q r = detfl {T denotes transpose).6.22) k which proves the theorem for the columns.] Prove that the determinant ofaunitary matrix is acomplex number of unit modulus. Active and Passive Transformations Suppose we subject all the vectors  V) in a space to a unitary transformation I vy^u\ vy (i.4.1 1— i i 1 +i_ IVerify that the determinant is of the form e 8 in each case.7. A similar result for the rows follows if we start with the equation ULF = I.5.7.i) Under this transformation. the result we lready have for the columns of a unitary matrix tells us the rows of U are jrthonormal. Since l f u = l 29 MATHEMATICAL INTRODUCTION I k = i k uiuhj=z u *uKJ (1.E. the matrix elements of any operator Q are modified as follows: < v\q\ vy^{ uv\o\ uvy = < r\t/Q t/ vy (1. * It is assumed that you know (1) what a determinant is.D.6. Note that LFU=I and ULFISLTE not independent conditions.6. Are any of the above matrices Hermitian? 1.6. Exercise 1. Verify that the following matrices are unitary: 1 2'/2 r j i_ ' i "l+i 2 . (3) that the determinant of a product of matrices is the product of the determinants. verify these properties for a twodimensional case 'a \7 /r 8j n= withdetO=(a^j57).6. [If you do not.
The Eigenvalue Problem Consider some linear operator Q acting on an arbitrary nonzero ket \ F ) : Q\V) = \ V ) (1. one can systematically determine all its eigenvalues and eigenvectors.8. In this chapter we will see how. Show that the determinant of a matrix is unaffected by a unitary change j of basis. such as the identity or its multiple.8. the ket will suffer a nontrivial change.3) The first case is called an active transformation and the second a passive transformation. i. .1) Unless the operator happens to be a trivial one. called its eigenkets.7.8. [Equivalently show det Q = det(f7 t Q[/). eigen.2) Equation (1. show Tr Q = Tr( l/tat/). So much for an arbitrary ket. and once we get to quantum mechanics proper. The situation is exactly the] opposite f r o m the point of view of the operators.2) is an eigenvalue equation: \ V) is an eigenket of O with eigenvalue co. The present nomenclature is in reference to the vectors: they are affected in an] active transformation and left alone in the passive case. and that active and passive transformations provide us with two equivalent ways of describing the same physical transformation.7..It is clear that the same change would be effected if we left the vectors alone ani subjected all operators to the change i Q^irnu (17. it will be eigen.  V ) will not be simply related to  V).8.1* The trace of a matrix is defined to be the sum of its diagonal matrix j elements Tr £2 = £ Show that (1) Tr(QA) = Tr(AQ) (2) Tr(QA0)=Tr(A0Q) = Tr(0QA) (The permutations are cyclic).e. [Equivalently. eigen all the way.] Exercise 1. however. Later we will see that the physics in q u a n t u m theory lies in the matrix elements  of operators. given an operator Q. Each operator. (3) The trace of an operator is unaffected by a unitary change of basis i>•£/)/>. Exercise 1.] 1. on which its action is simply that of rescaling : QP0 = fi>*0 (1. has certain kets of its own.2. H o w such an equation enters physics will be illustrated by a few examples f r o m mechanics at the end of this section.
e. This is a general feature of the eigenvalue equation and reflects the linearity of the operator: if then Qa\ Vy = aQ V> = aa\ V) = <oa\K> . After this unqualified success. (2) all vectors are its eigenvectors with this eigenvalue. we are encouraged to take on a slightly more difficult case: O = Py. is an eigenket with eigenvalue 0: Pv\V^ = \VXV\V±> = 0 = 0\V±> are simply not (3) kets that are neither. parallel to  V) is an eigenket with eigenvalue 1: Py\aV)> = \V)>(V\aV)> = a\V)\ V\2=l\aV) (2) any ket  V±).Example 1.. kets of the form ct\V) + p\Vxy.8. the basis vector 11) along the x axis: Are there others? Of course. eigenkets: Pv(a\Vy + P\V±y) = \aVy^y(a\ F> + /? F±>) Since every ket in the space falls into one of the above classes. any vector aj 1) along the x axis is also unaffected by the x rotation. the projection operator associated with a normalized ket  V).3. we will begin with a case that will be completely solved: the case Q = /. To illustrate how easy the eigenvalue problem really is. we conclude that (1) the only eigenvalue of / is 1. perpendicular to  K>. • Example 1.1. Consider now the operator i?(=:/ri).2. Clearly (1) any ket a V} = \a V). we have found all the eigenvalues and eigenvectors. • Example 1. Since 31 MATHEMATICAL INTRODUCTION / \vy=\vy for all  V).8. i. We already know that it has one eigenket.8.
It follows that our assumption that the operator (Q — ml)"1 exists (as a finite operator) is false. let us ask if has any eigenvectors besides 11). The trouble with this conclusion is that it is wrong! R{\N\) has two other eigenvectors besides 1). we get <iiQ«/F> = 0 . • The Characteristic Equation and the Solution to the Eigenvalue Problem We begin by rewriting Eq. Since the eigenvalue equation fixes the eigenvector only up to ' l s c a ^ e factor.ml) = 0 (1. we have fully solved the eigenvalue problem.6) I M = I This equation will determine the eigenvalues m. Since every vector is either parallel to 11) or isn't. we will not treat the multiples of an eigenvector as distinct eigenvectors.3) onto a basis. (1. for any vector not along the x axis necessarily gets rotated by and cannot possibly transform into a multiple of itself.32 CHAPTER l a n overa for any multiple a .3)j Now.8.8. we project Eq. The condition for nonzero eigenvectors is therefore det(Q .4). (1. l ) that the inverse of any matrix M is given by _. Basic matrix theory tells us (see Appendix A . (1. cofacrf det M N o w the cofactor of M is finite if M is. for these vectors are in V 3 ( C ) and not VI(R).8. This is impossible.8. We now turn our attention to this very question.8. we get lIO^Q^/r'lO) (1. Dotting both sides with a basis bra </. With this understanding in mind.d a / ) . we are trying to get something for nothing out of Eq.2) as ( Q . To find them. So we ask when this situation will obtain. It is clear from this example that we need a reliable and systematic method for solving the eigenvalue problem in V"(C).1 .8.4) (1. assuming it exists. It therefore seems that in asking for a nonzero eigenvector  V ) .d y / ) ] V) = ]0) Operating both sides with ( Q . Our intuition says no. But our intuition is not to be blamed. Thus what we need is the vanishing of the determinant. any finite operator (an operator with finite matrix elements) acting on the null vector can only give us a null vector.
8.. are basis independent.] Example 1.8.6. the eigenvectors may be found.8.8. [Operators on V"(C) that are not of the above variety may not have n eigenvectors—see Exercise 1. a fundamental result in analysis is that every wthorder polynomial has n roots. which are its roots.e. the eigenvalues.4.and upon introducing the representation of the identity [Eq. for they are defined by the abstract Eq. to the left of .3). we get the following image of Eq.8. not necessarily distinct and not necessarily real. at least for Hermitian and unitary operators. Now.7) 33 MATHEMATICAL INTRODUCTION Setting the determinant to zero will give us an expression of the form t cmcom = 0 m=0 Equation (1.8.8) s m=0 (1.4. (1. Let us use the general techniques developed above to find all the eigenvectors and eigenvalues of Recall that the matrix representing it is 1 0 0 0 0 1 0 1 0 R&Ki) Therefore the characteristic equation is \co dettR£tf/) = 0 0 i.8. which makes no reference to any basis. (1. Theorems 10 and 12 establish that Hermitian and unitary operators on V"(C) will have n eigenvectors. Although the polynomial is being determined in a particular basis.3): j:(Qym5y)VJ=0 (1. (1. (l6))(tf)2+l) = 0 (1. Thus every operator in V"(C) has n eigenvalues.8) is called the characteristic equation and (1. using a procedure illustrated by the following example.10) 0 co 1 0 1 = 0 co .8.8.7)]. Once the eigenvalues are known.9) is called the characteristic polynomial.
l . as shown in Appendix A . (1. Note that of the three simultaneous equations above. Consider next the equations corresponding to a> = i. = 0 0 0 •X2 = X3 = 0 Thus any vector of the form *.e. The components of the eigenvector obey the equations (l/)xi=0 — ix2 ~ x 3 = 0 x2/x3 = 0 (i. It is conventional to use the freedom in scale to normalize the eigenvectors. Thus in this case a choice is fl» = l> = l> = I say a choice. There is no universally accepted convention for eliminating this freedom.xi = 0) (i. and not the choice.8..e. Feeding o) = \ into Eq..1> 0 0 is acceptable. and x 3 of the corresponding eigenvector must obey the equations 1 . ±i. x2 = ix3) Notice once again that we have only n 1 useful equations. Let us see this come out of the formalism. This is the reason the norm of the vector is not fixed and. A properly normalized solution to the above is o >1/2 .7) we find that the components xu x2. except perhaps to choose the vector with real components when possible.1 X.1 0 0 0 01 0 1 1 0 ..e. x2 = ix3) (i.34 CHAPTER 1 with roots o) = 1. the reason the determinant vanishes. since the vector may be multiplied by a number of modulus unity without changing the norm. x2 x. In general. as expected. the first is not a real equation. We know that co = 1 corresponds to 11). there will be only (n — 1) LI equations.
Though this is not always the case.11) O=(f«)<0*)<>£0> to = &1* Q. (1. the ket corresponding to a> = 6)i is labeled  co = <y.there is just one vector labeled by it. This notation presumes that to each co.8.D. Imagine that instead of R(Ixi) we were dealing with another operator Q on V 3 (i?) with roots coi and o)2=co3. We now turn our attention to two central theorems on Hermitian operators. The phenomenon of a single eigenvalue representing more than one eigenvector is called degeneracy and corresponds to repeated roots for the characteristic polynomial. but also the procedure used in the example above for finding the eigenvectors. Proof. These play a vital role in quantum mechanics. It appears as if we can get two eigenvectors.A similar procedure yields the third eigenvector: MATHEMATICAL INTRODUCTION • 1 In the above example we have introduced a popular convention : labeling the eigenvectors by the eigenvalue. The eigenvalues of a Hermitian operator are real. Let Dot both sides with <©: (1.8. In the face of degeneracy.) or simply jco. For instance. Theorem 9.11) Take the adjoint to get <fflQ t a»> = ®*<ffl®> Since Q=Q + ..). we need to modify not just the labeling. one for each distinct 00. How do we get a third? Or is there no third? These questions will be answered in all generality shortly when we examine the question of degeneracy in detail. . only a slight change in this notation will be needed to cover the general case. this becomes <£0fi£0> = £0*<G)G)> Subtracting from Eq. by the method described above.E.
Corresponding to co{ there must exist at least one nonzero eigenvector <»i).E. [The analogy between real numbers and Hermitian operators is further strengthened by the fact that in a certain basis (of eigenvectors) the Hermitian operator can be represented by a matrix with all real elements. [If not.. Suppose c o ^ c o 2 = 0). call it . • . the first row follows from the Hermiticity of Q. This happens if there is any degeneracy. Given the first column.D.] Consider the subspace V U 1 of all vectors orthogonal to <»i>.12) The first column is just the image of after Q has acted on it. It is diagonal in this eigenbasis and has its eigenvalues as its diagonal entries. tf>2>> • • • » l ^ f . (o2. It must have at least one root. Theorem (A. (Notice that nowhere did we have to assume that the eigenvalues were all distinct. Then we have two orthonormal vectors obeying . in the basis fi>i). the basis of eigenvectors is orthonormal. and a normalized eigenvector  o) 2 ). the matrix Q becomes. To every Hermitian operator Q. Define the subspace V"7. Proof... The characteristic equation now takes the form {(D\ — co) • (determinant of boxed submatrix) = 0 n 1 (CO^CO) £ CMG>M=(G)IG))Pn~ V ) =0 0 N o w the polynomial Pn~] must also generate one root.i ) . Let us choose as our basis the vector  ft). Icon).] In stating Theorem 10.) (normalized to unity) and any n . Finally.) Q.. V U.l orthonormal vectors 2 ] ] {Fx. .1) would imply that (Q — a j ) is invertible. ~co] 0 Cl)2 0 0 0 0 0 G>3 0 0 0 Since every j fO(> was chosen from a space that was orthogonal to the previous ones.8. Let us start with the characteristic equation.. VTi } in V"u • In this basis Q has the following form: Q< (1. it was indicated that there might exist more than one basis of eigenvectors that diagonalized Q. 1.2 of vectors in V ^ 1 orthogonal to tt)2) (and automatically to &>i)) and repeat the same procedure as before.36 CHAPTER 1 Theorem 10.. there exists (at least) a basis consisting of its orthonormal eigenvectors. \a> 2 ).
•!©/> G). from which we may choose any m. there exists an infinity of orthonormal pairs o>l). orthonormal vectors to form the basis referred to in Theorem 10. Let us begin with two eigenvectors: (1.16) (1. ft>2>. there will be an eigenspace V™. if the characteristic equation has nij of its roots equal to some . Since the vectors  ce?i> and  CD2> are orthogonal (and hence LI). and the second with <foiQG)i> = a)i<G)yft)i> . Q}2y.) = G>ct)i) QG)2> = G>G)2> It follows that + f3(o\(02y = G)[a\<n\y + Plcoi}] 37 MATHEMATICAL INTRODUCTION Q[a{(0iy + p\a)2y] = aw\(oly for any a and ft. = G)f (1.8. times.  ©.8. — G *) < G).8. In general.13b) we get (1.14b) Taking the adjoint of the last equation and using the Hermitian nature of Q.15) QcOi) = a>to)j) Q\(0Jy=(0j\(0jy Dotting the first with <o>. since <©. obtained by a rigid rotation of o)i>.8.. > > If i=j. we can prove Theorem 9 and 10 very easily. that is. In the absence of degeneracy.14a) (1. One refers to this space as an eigenspace of Q with eigenvalue Besides the vectors ©i> and ® 2 >.14a).13a) (1. we get 0 = (©. we get (o)j\£i\(oiy = Q)*(Q)J\Q)iy Subtracting this equation from Eq. the elements of which are eigenvectors of Q with eigenvalue w. we find that there is a whole twodimensional subspace spanned by ©i> and o) 2 >.8. if an eigenvalue occurs m.8.Q£0. (1. we get. from which we may select any pair in forming the eigenbasis of Q.8.
a) = 0 . The characteristic equation is (ct) — 2)2CD = 0 i.38 CHAPTER 1 If i ^ j . we get <G)ICOJ> = 0 (1. but an infinity of orthonormal eigenbases. our general analysis of Theorem 10 showed us that in the face of degeneracy.8. 2 ..1/2 Xi — x 3 = 0 . That the proof of orthogonality breaks down for = coj is not surprising.17) since ©.e. Let us see through an example how this variety manifests itself when we look for eigenvectors and how it is to be handled. 2 The vector corresponding to co = 0 is found by the usual means to be 1 0 1 The case co = 2 leads to the following equations for the components of the eigenvector: — X] KY3 = 0 0= 0 a> = 0> . Example 1. for two vectors labeled by a degenerated eigenvalue could be any two members of the degenerate space which need not necessarily be orthogonal.5. Consider an operator fl with matrix elements 1 Q 0 1 0 2 0 1 0 1 in some basis. Degeneracy We now address the question of degeneracy as promised earlier.8.a>* = a)/ —a?y/0 by assumption. The advantage in the way Theorem 10 was proved first is that it suffers no modification in the degenerate case. The modification of this proof in this case of degeneracy calls for arguments that are essentially the ones used in proving Theorem 10. we have not one. Now.
> no longer refers to a single ket but to a generic element of the eigenspace V™. i. Theorem 12. 2 T Clearly each distinct choice of the ratio. To refer to a particular element. The eigenvalues of a unitary operator are complex numbers of unit modulus.e.1/2 The third vector is now chosen to lie in this plane and to be orthogonal to the second (being in this plane automatically makes it perpendicular to the first  o = 0 ) ) : 1 O = 2. second one) 6 T. gives us a distinct doublet of orthonormal eigenvectors with eigenvalue 2. A natural choice of the label a will be discussed shortly. lie in a plane perpendicular to  o = 0). a ) .. instead of the two {n — 1) we have grown accustomed to! This is a reflection of the degeneracy. • Notice that in the face of degeneracy. Thus degeneracy permits us extra degrees of freedom besides the usual one (of normalization). We now consider the analogs of Theorems 9 and 10 for unitary operators. This is in agreement with our expectation that a twofold degeneracy should lead to a twodimensional eigenspace. it takes away one equation.. F o r every extra appearance (besides the first) a root makes. we must use the symbol o« . where a labels the ket within the eigenspace.  o . j o = 0 ) . to get a normalized eigenvector corresponding to (0 = 2:  o = 2> . The conditions 39 MATHEMATICAL INTRODUCTION x2 arbitrary define an ensemble of vectors that are perpendicular to the first.) . x2/x3. The eigenvectors of a unitary operator are mutually orthogonal. Theorem 11. Let us arbitrarily choose x2 = 1. (We assume there is no degeneracy. the ratio x2/x3) corresponds to the freedom of orientation in this plane. The freedom in x2 (or more precisely.Now we have just one equation.
8. . . .  i ) . Since mi = 1. .8. . . (If you wisl you may apply our mnemonic to U and verify its unitary nature: its columns contain the components of the eigenvectors cd.y=U\i) (1. .D. . n>..) This result is often.) If U is degenerate.1S (1. Diagonalization of Hermitian Matrices \ Consider a Hermitian operator Q on V " ( C ) represented as a matrix in some orthonormal basis 11).  o>„). . all the other elements in the first row must vanish. (1. . Whereas in Eq. <m..  cd.181 If we take the adjoint of the second equation and dot each side with the corresponding side of the first equation.8. the zeros of the first row followed f r o m the zeros of the first column and ^ = H. (Q. we get (Uj  if U\M/> = Uj U* <Uj  M. Let u\uiy=ui\uiy and U\Ujy = Uj\Uj} (1.). .40 CHAPTER 1 Proof of Both Theorems (assuming no degeneracy).181 (1.>/0. . summarized by the statement: Every Hermitian matrix on V " ( C ) m a y be diagonalized by a unitary change of basis. the matrix representing Q will become diagonal.M. .8.) ^ I t ^ M / ^ W / W * i=uiu*=>uiuf ^ 1. since <M.8. for it "rotates" one orthonormal basis into another. with just one difference.20al while if i j. If we trade this basis for the eigenbasis ct>j).20b] since  m. we get. UiUf= 1 (1.E.> that are orthonormal. . we can carry out an analysis parallel to that for the Hermitian operator H.21] is clearly unitary.M/> = 0 (1.8. here they follow from the requirement that the sum of the modulus squared of the elements in each row adds up to 1.> so that (1 uiuf)(uj\uiy = 0 If i=j. Now thei operator U inducing the change of basis \w.12).8. .
U being the matrix of eigenvectors of Q.2* Consider the matrix "0 0 r 0 Q= 0 0 J (1) Is it Hermitian? (2) Find its eigenvalues and eigenvectors.3* Consider the Hermitian matrix 0" "2 0 0 _0 3 1 1 3_ (1) Show thatftJi= o)2= 1. « 3 = 2.1. there exists a unitary matrix U (built out of the eigenvectors of £T) such that UFQU is diagonal.8. Thus the problem of finding a basis that diagonalizes Q is equivalent to solving its eigenvalue problem. Exercise 1.We may restate this result in terms of passive transformations as follows: If Q is a Hermitian matrix. (2) Show that ft) =2) is any vector of the form 1 (2« ) 1/2 (3) Show that theft)= 1 eigenspace contains all vectors of the form 1 (b +2c2y/2 either by feedingft)= 1 into the equations or by requiring that the a = 1 eigenspace be orthogonal to  < = 2). Exercise 1. (1) Find the eigenvalues and normalized eigenvectors of the matrix "1 41 MATHEMATICAL INTRODUCTION 3 2 1 r 0 4_ Q= 0 _0 (2) Is the matrix Hermitian? Are the eigenvectors orthogonal? Exercise 1.8.8. 0 2 . 0 o_ (3) Verify that ifOU is diagonal.
(2) Using the invariance of the trace under the same transformation. A n arbitrary n xn matrix need not have n eigenvectors. show that they are orthogonal.42 CHAPTER 1 Exercise 1. where V is the matrix of eigenvectors of i l Exercise 1. * (1) W e have seen that the determinant of a matrix is unchanged under a unitary change o f basis.5.8.4. By using the results on the trace and determinant from the last problem. Find the corresponding eigenvectors. = G)2 = 3. (2) By feeding in this value show we get only one eigenvector of the form 1 (2« ) W e cannot find another one that is LI. show that the eigenvalues of the matrix are 3 and — 1. . . Verify this by explicit computation.8. show that n T r Q = £ (Dj i— 1 Exercise 1.* Consider the matrix cos Q= 9 sin 9 —sin 9 cos 9 (1) (2) (3) (4) Show that it is unitary. Consider as an example n= 4 1 1 2 (1) Show that ft). N o t e that the Hermitian nature of the matrix is an essential ingredient. Argue n o w that n det = product o f eigenvalues o f Q = Y\ i= 1 for a Hermitian or unitary Q. Verify that lPQU= (diagonal matrix).2x1/2 +a —a Exercise 1.8.7.6.8. Show that its eigenvalues are e'° and e~'6.
8.] or in Dirac notation \l)=M\co> (1) Will the angular momentum and angular velocity always be parallel? (2) Show that the moment of inertia matrix My is Hermitian. located at ra and rotating with angular velocity A around a common axis has an angular momentum > l = Jm„(r a 5<y H ) a where va = o x r a is the velocity of m„. [Hint: TX{ACB)=TT(CBA). M 4 that obey ATM J +M J M''=28 ( J I. Theorem 13. of 1 is given by /f = I A/pffl. . i. M3. Exercise 1. A collection of masses ma. . By using the identity > A x (B x C) = B ( A • C ) . (ra)_. it is clear that every direction is its eigendirection for rotation.8. .) (2) By considering the relation M'M'^M'M' f o r I^J show that M' are traceless. there exists (at least) a basis of common eigenvectors that diagonalizes them both.9. What does this say about the three eigenvalues of the matrix Ml Simultaneous Diagonalization of Two Hermitian Operators Let us consider next the question of simultaneously diagonalizing two Hermitian operators. (3) Argue now that there exist three directions for w such that 1 and o> will be parallel.C ( A • B) show that each Cartesian component I. How are these directions to be found? (4) Consider the moment of inertia matrix of a sphere. . Due to the complete symmetry of the sphere.j= 1. . (Hint: go to the eigenbasis of M'.Exercise 1.where My = I ma[rl8u~ (r c ). If Q and A are two commuting Hermitian operators.] (3) Show that they cannot be odddimensional matrices.8* Consider Hermitian matrices M\ M2. 4 43 MATHEMATICAL INTRODUCTION (1) Show that the eigenvalues of M' are ±1. and use the equation for /=/.
a) where the additional label a runs from 1 to H o w does A appear in the basis? Although we made no special efforts to get A into a simple form..8.) Since [A.. Consider any one of its eigenvectors: A Q  o j I ) = ft>. Since Q is nondegenerate. i.> will diagonalize both operators. flr) = ct). it already has a simple form by virtue of the fact that it commutes with Q. = VT' corresponding to the eigenvalue (o. a ) = A^<0/.> is an eigenvector of Q with eigenvalue to a scale.22) Q A  o . (1.> is also an eigenvector of A with eigenvalue A*.e.Aft). up to a scale.44 CHAPTER 1 Proof. W h a t if both operators are degenerate? By ordering the basis vectors such that the elements of each eigenspace are adjacent. > = a)/Aa>. Consider first the case where at least one of the operators is nondegenerate. say Q.> (1.e.8.23) Thus j <y. Q] = 0. we can get one of them. there is only one basis with this property. Since every eigenvector of Q is an eigenvector of A. Act). Let us assume Q is nondegenerate. into the form (Theorem 10) ©i ct) i toi w2 (0. there is just one eigenvector. Since this vector is unique up A£Ui) = /lIft}. Let us start by mimicking the proof in the nondegenerate case: QAG>. it is evident that the basis ct).> i. there exists an infinity of bases..AG>/. Let us arbitrarily pick in V^J. a ) . (Or a N o w this basis is not unique: in every eigenspace V™. to a given eigenvalue.. a set  (Dj.
due to the degeneracy of Q. for the eigenbasis of A. in V™' that we started with. . (1..E.D.However. a ) of the eigenspace VT'. in V?'. If the eigenvalues of A. since it is indifferent to the choice of orthonormal basis in each degenerate eigenspace... P\A\co. (coj. The block diagonal form of A reflects the fact that when A acts on some element co. Let us make such a change of basis in each eigenspace. It can obviously be diagonalized by trading the basis (tit. 2 ) .. A is given by a matrix A. thereby rendering A diagonal. which appears as a block in the equation above. ft). Consider a matrix A.) then we end up with A. 45 MATHEMATICAL INTRODUCTION 0 which is called a block diagonal matrix for obvious reasons. . It is Hermitian since A is. .. Meanwhile what of Q? It remains diagonal of course.. we can only conclude that Ao>.8. . a?. . . .15)]. are Ap\ . a) lies in V" Now.(1) A]™1' 2 ( Ak k ) m Q. . or> = 0 if ot)/. 1). Xjm. in this basis.. /?> are basis vectors such that 0 Consequently. since vectors from different eigenspaces are orthogonal [Eq. Within each subspace i. a) and \<Oj. it turns it into another element of V?'.
5. one can always find. . . that commutes with both Q and A.24a) (1. there is a twodimensional eigenspace f r o m which we can choose any two orthonormal vectors for the common basis. In our study of q u a n t u m mechanics it will be assumed that such a complete set of commuting operators exists if n is infinite. these coordinates obey the following equations. if say Ai(1) = A{2). show that the following Hermitian matrices may be simultaneously diagonalized. in some detail. " 1 Q= 0 0 0 0 r 0 u . The two masses m are coupled to each other and the walls by springs of force constant k.8.10* By considering the commutator. eigenbasis. the elements of which may be labeled unambiguously as  co.6.24b) m *i m *2 Figure 1.46 CHAPTER 1 If A is not degenerate within any given subspace. y . It is important that you understand this example thoroughly. A. / . both matrices are diagonalized.8. common. In general.8. ) . for it not only illustrates the use of the mathematical techniques developed in this chapter but also contains the main features of the central problem in quantum mechanics. _k 2k X2 x 1 — x2 m m (1. If x.. and i. Example 1.8. A= "2 1 0 1 r 1 2.. T . . All masses are m. and x2 measure the displacements of the masses from their equilibrium points. } that commute with each other and that nail down a unique. the basis we end up with is unique: the freedom Q gave us in each eigenspace is fully eliminated by A. 1 J _ 1 Since Q is degenerate and A is not. It is then necessary to bring in a third operator T. /. X). Find the eigenvectors common to both and verify that under a unitary transformation to this basis. f o r finite n. . 1. and which will be nondegenerate in this subspace. derived through an elementary application of Newton's laws: . a set of operators {Q. We will now discuss. you must be prudent in deciding which matrix dictates the choice of basis. X . _ _2k k . . .X] — X] H m m . for any k. with X playing the role of the extra label a. The coupled mass problem. all spring constants are k. . The elements of this basis may be named uniquely by the pair of indices w and X as G>. Exercise 1. The mechanical system in question is depicted in Fig.5. and the displacements of the masses from equilibrium are Xi and x2. the complete solution to a problem in mechanics. If A is degenerate within an eigenspace of Q.
24) in matrix form: Xi x2_ 47 MATHEMATICAL INTRODUCTION fin _fi2I q 12 " n 2 2 Xi X 2 (1.1.8. (1. from the standpoint of finding a mathematical solution to the initialvalue problem.8. — kjm (1.27b) An arbitrary state. etc. 12) is very desirable physically. and fi is a Hermitian operator on V2(R). If we restrict ourselves to the case of zero initial velocities.8. it is not so desirable. (1.8. it C 2 is an element of V (R).25a) is l*(0>=ni*(0> (1.8.29) It is in this 1>. n12 = n2.8.y as the matrix elements of a Hermitian operator Q.25). for the components of  x> in this basis (*! and x2) have the simple interpretation as displacements of the masses.25a) is obtained by projecting Eq. in which the masses are displaced by basis by and J 2 .26) Equation (1. The abstract form of Eq.26) on the basis vectors 1>.8. we will formulate the problem in the language of linear vector spaces and solve it using the machinery developed in this chapter.27a) 2> "" o .8. for the components x{ and x2 obey the coupled . is given in this C X. which in this case consist of the initial positions and velocities. given Xi(0) and x2(0). However. The basis 11). As a first step. In what follows.25a) _ _ where the elements of the Hermitian matrix Q„ are Qn — 0. with elements —2kjm.The problem is to find jt. we rewrite Eq. which have the following physical significance: T 1) _0_ first mass displaced by unity second mass undisplaced first mass undisplaced second mass displaced by unity.(0 and x2{t) given the initialvalue data.22= ~2k/m. ^—• (1.8. k/m. "f "" o Xi + 1 X 0 7 The abstract counterpart of the above equation is  x ) =  l ) x 1 + 2)x 2 (1.25b) We now view and x2 as components of an abstract vector x>.8. (1. our problem is to find Xiit) and x2(t). (1. 2> basis that Q is represented by the matrix appearing in Eq. Since the vector  J > has two real components.8.28) (1.8. (1. 12). and Q.
The components of x> in this basis will then obey another uncoupled differential equations which may be readily solved.8. 12) basis. the components xx and x n will evolve as follows: pilj®? bciJ L 0  ° 1U1 ©yLcnJ I (1. we can return to the physically preferable 11). we can now see how to get around it: we must switch to a basis in which ft is diagonal. and in which  x ) has components xx and.8.8.29)]. 2> basis. Having found the solution.30b) We are departing here from our usual notation: the eigenvalue of Q is written as .8.8.33) conXnJ We obtain this equation by rewriting Eq.o)2 rather than as co in anticipation of the fact that Q has eigenvalues of the form — <d2.8.26) in the I>.8.31b) If we now expand the vector x(f)> in this new basis as x(/»I>Xi(0 + in>xn(0 (1.8. From our study of Hermitian operators we know that the basis that diagonalizes Q is the basis of its normalized eigenvectors. (1. is our broad strategy and we now turn to the details. with a> real.30a) (1.c a f ) and ( .24a) and (1.8. This. We are also using the symbols 11) and  II) to denote what should be called  . Having identified the problem with the 11).32) [in analogy with Eq. II> basis in which Q has its eigenvalues as the diagonal entries. . Let I> and II> be its eigenvectors defined by QI>a>?I> ftII> = ©?iII> (1.24b). then.48 CHAPTER L differential equations (1.  2 ) basis (in which the matrix elements of Q are known) and to obtain H W ^ U (1. It is a simple exercise (which you should perform) to solve the eigenvalue problem of Q in the 11).31a) (1.© « > in our convention. The coupling is mediated by the offdiagonal matrix elements = ^21 = k/m. (1.8.
Step (1). The solution to the decoupled equations Xi + CQ2Xj = 0. CI Alternately we can apply the operator 49 MATHEMATICAL INTRODUCTION to both sides of the expansion of Eq. It corresponds to the following algorithm for finding \x(t)} given x(0)>.36) into Eq.8.34) subject to the condition of vanishing initial velocities.8. (1. II> basis obey decoupled equations that can be readily solved. each coefficient is zero.8.37a) (1.8. .32) we get  x ( f ) > . and get 0> = I>(jci + ( o h i ) +  n > ( x „ + (OuXw) Since I) and II> are orthogonal. is x{(t) = x^O) cos a)it.32). II (1. Solve the eigenvalue problem of Q. the components of x> in the I>. Append to each coefficient x.8.8. Feeding Eq. II) a time dependence cos carf to get the coefficients in the expansion of x(0>Let me now illustrate this algorithm by solving the following (general) initialvalue problem: Find the future state of the system given that at f = 0 the masses are displaced by xi(0) and x 2 (0).I>x. (1. We can ignore this step since the eigenvalue problem has been solved [Eq. (1. Step (1).(0) cos<»if + II)xn(O) cos m a t = I><Ix(0)> cos t + II><IIx(0)> cos coii t (1.8. Find the coefficients x t (0) = <Ix(0)> and x„(0) = <IIx(0)> in the expansion x(0)) = I>x I (0) + II>x„(0) Step (3). Step (2).8. 1 = 1.JJ .37b) Equation (1. i = I.8.(0) (z' = I. II (1.8.31)].37) provides the explicit solution to the initialvalue problem. (1.36) As anticipated.35) (1.
(1. *i(0) xi(0) + x 2 (0) >1/2 x I (0) = <Ix(0)) = .39) .8.(0)x 2 (0) 11/2 2 l/2 x 2 (0)J Step (3).8.8. (1. \mj < l  I I >2 !/2° ^ c o s ^ . 2> basis to find x\(t) and x2(t).38b) We can rewrite Eq.* 2 ( 0 ) ] cos 3k\J 1/2 ' (1. « ) ) .m 1/2 n 1 3k t + . (0) + JC2(0)] cos w J ~ [xi(0) — x 2 (0)] cos \mj .7 5 ( l . We get (feeding in the explicit formulas for a>j and X\(t) = <lx(0) 1/2 .(0) *n(0) = <IIjc(0)> = x.[ * .Step (2). ( 0 ) + x 2 ( 0 ) ^i(O) —x2(0) \x(t)} = I> COS ©J t+ II> COSffl„t >1/2 The explicit solution above can be made even more explicit by projecting x(0> onto the 1>. i / \ \ ITX x .8. x.38a) using the fact that < l  I ) = < l  I I ) = l/2 1 / 2 It can likewise be shown that x2{t) = ^ [X.38) in matrix form as "cos [ ( k / m ) 1 /2t] + cos [(3k/m)1 Xi(t) x (t). (1. the displacements of the masses. l ) MO). 2 /2 t] cos [(k/m)1 •/zt] .cos [ ( 3 k / m ) 1 /2t] *i(0) MO).[ x i ( 0 ) + x 2 (0)] cos fvr.I f + =<1I>^!<2)±M5cos M/2 Am) 1 = .
(2) This matrix is independent of the initial state.41b) You may easily convince yourself that if we take the matrix elements of this operator in the 1>. We may view Eq.8. for given any other initial state with displacements xi(0) and x 2 (0).8.39) as the image in the 1>.39): (1) The finalstate vector is obtained f r o m the initialstate vector upon multiplication by a matrix. T o solve the equation x>=nx> . (1.8. (1. Finding the propagator is tantamount to finding the complete solution to the problem. (1. 51 MATHEMATICAL INTRODUCTION The Propagator There are two remarkable features in Eq. = <lC/l> = <1 jI><I cos + II><II cos 1/2 Km fk Km. = <1I><I1> cos If = jcos + <lII><IIl>COS I f \mj t + COS _ Am/ 1 rwl/a i H _ Notice that U(t) [Eq.8.40) By comparing this equation with Eq. we find the abstract representation of U: C/(f) = I><I cos 6?if + )II><II COS 0)ut ii £ z></ cos a it I=I (1. We may then restate our earlier algorithm as follows.37b). (1.8.41)] is determined completely by the eigenvectors and eigenvalues of Q. F o r example tfi.39).This completes our determination of the future state of the system given the initial state. we get and x2(t) by applying the same matrix to the initialstate vector.41a) (1.8. We call this matrix the propagator.8.8. (1. 2> basis of the abstract relation *(f)> = C(OI*(0)> (1. we regain the matrix appearing in Eq. m 3k \m 11) 1/2 . 2> basis.
42) Thus the system starting off in I> is only modified by an overall factor cos o)\t. If the masses are adjusted by A and —A. Suppose we have  x ( 0 ) ) = I>. these are the eigenkets I> and II>. These two modes of vibration. in which all (two) components of a vector oscillate in step are called normal modes.8. Consequently 1 cos Lcos [(k/my/2t] [(k/m)1/2tl 11(0 On the other hand. In this basis J_ ID W2 and corresponds to a state in which both masses are displaced by equal amounts. each mass feels a restoring force of 3A:A (2kA from the middle spring and kA from the end spring nearest to it). the vibrational frequency is (3&/m) 1/2 and 1 cos [ ( 3 k / m ) W 2 t ] >1/2 L . The Normal Modes There are two initial states x(0)> for which the time evolution is particularly simple.1 the masses are displaced by equal and opposite amounts. 2> basis. Not surprisingly. Since the effective force constant is keff=3kA/A = 3k. The middle spring is then a mere spectator and each mass oscillates with a frequency 6>i= (k/m)1/2 in response to the end spring nearest to it. Then the state at time t is I(f)>stf(0H> = (I)<I cos ®i? + jII)<II cos WIIOII) = I> cos cojt (1. (3)  x ( 0 > = £ / ( 0 l * ( 0 ) > . A similar remark holds with I»II. (2) Construct the propagator U in terms of the eigenvalues and eigenvectors.52 CHAPTER 2 (1) Solve the eigenvalue problem of Q.c o s [(3k/m) 1 / 2 t]_ II>(0> If the system starts off in a linear combination of I> and II> it evolves into the corresponding linear combination of the normal modes  I ( 0 > and II(0>. respectively. In this case the middle spring is distorted by twice the displacement of each mass.This . The physics of the normal modes is clear in the 11). if we start with 1 HI) >1/2 .
(1) Given that the initial state is 1>. II> basis: COS CO\1 0 0 cos cou t.8.] In most cases. after time evolution). [Since the equation is first order in t. (1.t + II)<II x(0)> cos On t = /(0><Ix(0)> +  I I ( 0 X I I  x ( 0 ) > Another way to see the simple evolution of the initial states I> and  I I ) is to determine the matrix representing U in the I>. (1. in which the first mass is displaced by unity and the second is left alone.43) You should verify this result by taking the appropriate matrix elements of U(t) in Eq.41b). and H is a Hermitian operator called the Hamiltonian. calculate 11(/)> by following the algorithm. . l v ( 0 > = t f ( 0 I V ( 0 ) > .8. You must of course wait till Chapter 4 to find out the physical interpretation of  v>. E/<—» IJI basis (1. which is determined by the Schrodinger equation to be ( . Solve the eigenvalue problem of H. The state of the system is described in quantum theory by a ket  v ) which obeys the Schrodinger equation where H is a constant related to Planck's constant h by fi = hflit. and the precise relation between U(t) and the eigenvalues and eigenvectors of H. (which is to say.8.8. Step (3). The central problem in quantum mechanics is very similar to the simple example that we have just discussed. The problem is to find  y/(t)} given  y(0)>. H is a timeindependent operator and the algorithm one follows in solving this initialvalue problem is completely analogous to the one we have just seen: Step (1).is the content of the propagator equation 53 MATHEMATICAL  x(t) > = U(t) I x ( 0 ) > INTRODUCTION = I><Ijc(0)> COSfi>.39).11. Consider the coupled mass problem discussed above.ijfi)H y/(0)>. (2) Compare your result with that following from Eq. Step (2). Find the propagator U(t) in terms of the eigenvectors and eigenvalues of H. no assumptions need be made about  vH0)>. • Exercise 1. Since each column above is the image of the corresponding basis vectors (I> or  I I » after the action of U(t). we see that the initial states I> and II> evolve simply in time. the actual form of the operator H.
etc.9. We define the same function of an operator or q number to be O O /(Q) = £ a„nn «=o (1.9. Since 10\ Q = £02 (a. consider a common example: 00 Q" «=i n\ (1. which do not generally commute with each other.8. log(x).3 Let us restrict ourselves to Hermitian H.8. Use the fact that x(0)> is arbitrary. .12. Now.3). We wish to examine the question whether functions of q numbers can be given a sensible meaning. 1. It is customary to refer to the former as c numbers and the latter as q numbers. which commute with each other and with all operators.9. and operators. solve for the elements of the matrix U in this common basis and regain Eq. we are accustomed to functions of c numbers such as sin(x). We will restrict ourselves to those functions that can be written as a power series. (1.2) This definition makes sense only if the sum converges to a definite limit. By going to the eigenbasis of Q we can readily perform the sum of Eq.9. (1.1) where x is a c number. Functions of Operators and Related Concepts We have encountered two types of objects that act on vectors: scalars.54 CHAPTER 1 Exercise 1. (1. Consider a series O O f(x)=Zanx" n—0 (1.43). To see what this means.9. Consider once again the problem discussed in the previous example. Assume x(0)> = 0. (1) Assuming that x>=ax> has a solution x(/)>=t/(/)x(0)> find the differential equation satisfied by £/(/).9. (2) Assuming (as is the case) that Q and U can be simultaneously diagonalized.
9.3. show that U=e'H is unitary. u — e'e is a number of unit modulus.9. Derivatives of Operators with Respect to Parameters Consider next an operator 0(A) that depends on a parameter A. Its derivative with respect to A is defined to be fl( A + AA)fl(A) ^ .9. Exercise 1.Q ) '. By going to the eigenbasis. A special case of 0(A) we .) Exercise 1. examine when the q number power series 00 i / ( f t ) = X or M— 0 of a Hermitian operator Q may be identified with ( 1 .2* If H is a Hermitian operator.9.6) 00 m v=o rn\m Since each sum converges to the familiar limit the operator en is indeed well defined by the power series in this basis (and therefore in any other). (Notice the analogy with c numbers: if 0 is real. If 0(A) is written as a matrix in some basis. then the matrix representing d6{X)/dX is obtained by differentiating the matrix elements of 0(A).H. Exercise 1. wi! (19.and 55 MATHEMATICAL INTRODUCTION Qm CO2 = (1.1 * We know that the series 00 m = z v" M = 0 may be equated to the function f(x) = (1 — x)™1 if x < 1. AA .5) con  00 m =0 E £1. For the case above. show that det U=en.H m dX AA^O .
even if Q. by working with the power series. by going to the eigenbasis of Q. the chain rule is — e e = Lie e +e e u (1.9.) In the above. while it is not true that eaa^0 = eaa + pe or that aS1 B ^ e~aSi = efi9 unless [Q. the order of the factors is all important. in differentiating a product. provided it exists: d  r n «! dX„=o Conversely. that (1. Likewise. we can say that if we are confronted with the differential Eq. it is true that as may be verified by a powerseries expansion. everything commutes and we can treat them as c numbers. The solution 9 = e^ x corresponds to the choice c ~ I. more than one q number is involved. c is a constant (operator) of integration. F o r example. if only one q number (or powers of it) enter the picture. 0] = O.9. We can show. on the other hand. the real difference between c numbers and q numbers is that the latter do not generally commute.7). However. If one remembers this mnemonic. In all the above operations. one can save a lot of time. we see that Q behaves as if it were just a c number.8) . (1. its solution is given by (It is assumed here that the exponential exists.7) The same result may be obtained. If.9.56 CHAPTER l are interested in is 0(1) = where O is Hermitian. Now. is not Hermitian.
. t) of a string clamped at x = 0 and x = L (Fig. . measure the displacement f ( x t ) at the 19 points x = L/20. 1.).10. . The string is clamped at x = 0 and x = L. we can specify the values of / ( x ) at a larger number of points. It is free to oscillate in the plane of the paper. . Consider a function defined in some interval. . 19L/20.6. f„{xn)} as components of a ket  /„> in a vector space V(R): ' fn(Xx)' I fn> fn(x2) _fn{xn)_ (1. 2 L / 2 0 . Let us denote by f„(x) the discrete approximation to / ( x ) that coincides with it at n points and vanishes in between.We are free to move fl through e and write the first term as MATHEMATICAL INTRODUCTION but not as unless [Q. Generalization to Infinite Dimensions In all of the preceding discussions. our friend on the moon will be able to reconstruct the approximate picture of the string shown in Fig. the dimensionality («) of the space was unspecified but assumed to be some finite number. 0]=O. Let us begin by getting acquainted with an infinitedimensional vector. Suppose we want to communicate to a person on the moon the string's displacem e n t / ^ ) . 1. Let us now interpret the ordered ntuple {/„(xi). a<x<b. 1.10.7.6). x=0 • Xi x I "Z X9 I . . We now consider the generalization of the preceding concepts to infinite dimensions.x s L . at some time t. One simple way is to divide the interval 0 — L into 20 equal parts. The string as reconstructed by the person on the moon. . and transmit the 19 values on the wireless. f(x) O' Figure 1.1) Figure 1. If we wish to be more accurate. . Given these/(x. / „ ( x 2 ) . A concrete example is provided by the displacement / ( x .7. say.
Y o u should convince yourself that if we define vector addition as the addition of the components. Vector addition and scalar multiplication are defined just as before.2) xt and zero elsewhere.58 CHAPTER 2 The basis vectors in this space are 0 0 1 0 0 corresponding to the discrete function which is unity at The basis vectors satisfy JCJ) = 8ij (orthogonality) ith place 1 (1. L and that are specified at n points in between.! (1.0 .3).10. etc.10.. ) : I / • > = ! Mxdlxd i= 1 (1. For finite n it was MxdgnM (1.6) . the inner product.>. h„(x).10. /f„>.10. and scalar multiplication as the multiplication of each component by the scalar. We next define the inner product in this space: I <fn\gn>=i i=l Two functions f„(x) and g«(. one for each point xAlong each axis is a unit vector x.x) will be said to be orthogonal if < / „  g « > . and vice versa. by giving its value at every point in the interval 0 . Each function is now represented by a ket /<»> in an infinitedimensional vector space and vice versa. there is a corresponding ket g„>.4) Try to imagine a space containing n mutually perpendicular axes.L . The function fn(x) is represented by a vector whose projection along the /th direction is / „ ( * .5) To every possible discrete approximation g„(x). Consider. In this case f00(x)=f(x) is specified by an ordered infinity of numbers: an f ( x ) for each point x. forms a vector space. however. £  x ^ <x. Let us now forget the m a n on the m o o n and consider the maximal specification of the string's displacement.10. etc..  = I (completeness) (1. then the set of all kets representing discrete functions that vanish at x = 0.
)A. a smooth limit obtains.x " ! Should we require. The orthogonality of two different axes requires that (x\x'} = 0. by the usual definition of the integral.4) to the case where the kets are labeled by a continuous index x': (1.9) What are the basis vectors in this space and how are they normalized? We know that each point x gets a basis vector x>.11) JC')<JC' dx' = I .10.10. we get. the only modification we need is in the inner product: rb <f\g> = f*(x)g(x)dx (1.6') If we now let n go to infinity. We start with the natural generalization of the completeness relation Eq. x^x' (1.10.10. W h a t we need is the redefinition of the inner product for finite n in such a way that as n tends to infinity. for practically any function.7) <f\f> = 0 f{x)dx •2 (1. as in the finitedimensional case.10. in some interval a<x<b.defined as n 59 MATHEMATICAL INTRODUCTION </«!#«> = I i= 1 fn(Xi)gn(xt) in particular </i/>= i /=i If we now let n go to infinity. The natural choice is of course < / * I S « > = I Mxi)g„(x.10.8) If we wish to go beyond the instance of the string and consider complex functions o f j t a s well.10. < x  x ) = 1? The answer is no. <f\g>= f f(x)g(x) rL dx (1. and the best way to see it is to deduce the correct normalization. (1. A = L/(n+\) (1. so does the sum.10) What if x .
let us write it as The "function. x') cannot be finite at x/=x. the projection of  / > along the basis ket \x). x') dx'=f(x) * X— £ (1.10. f ( x ' ) (for any reasonably s m o o t h / ) can be approximates by its value at x'—x. In fact 5(x.12. (1.13 In this infinitesimal region.15) Clearly 8{x.10. Let the inner product <xx') be some unknown function 5(x.x ' ) (1.x')dx' xs =1 (1.10." with the properties . as always.x') = 0. the identity is required to leave each ket unchanged. S(xx')dx'= 1. One sometimes calls . Likewise (x'\ f ) = f ( x ' ) . Since S(x.14 so that rx+ e 5(x. and pulled out of the integral: e f(x) 5(x.10.16) a<x<b is called the Dirac delta function and fixes the normalization of the basis vectors: <xx') = 5 ( x . Dotting both sides of Eq. Now. x') vanishes if x^x' we can restrict the integral to an infinitesimal region near jt' = . x^x' (1. is just f ( x ) .11) with some arbitrary ket  / > from the right and the basis bra (x\ from the left. (1.17) It will be needed any time the basis kets are labeled by a continuous index such as x.10. <x / ) . J <xx')<x' / > dx' = 0  I  / > = <x / > (1.60 CHAPTER 2 where. x') depends only on the difference xx'.10.10.x')f{x')dxf^f(x) (1. for then its integral over an infinitesimal region would also be infinitesimal. t i n E q .10.12) : rx+s 5(x.x' Since <5(x. Note that it is defined only in the context of an integration: the integral of the delta function <5(xx') with any smooth function f ( x ' ) is f ( x ) . x f ) should be infinite in such a way that its integral is unity.
x') What is the action of this function under the integral? The clue comes from the Gaussian model. 22 1 / 2 e x p (*A ) (xx')2 (1.8a. the 8 function. the derivative of the phase of g vanishes and the oscillations are suspended. its values being either infinite or zero.18) 8(xx')f{x')dx'=f(x) The delta function does not look like any function we have seen before. This may be verified as follows: 8(xx') = < x  x ' ) = <x'x> * = 8(x' — x)*~ 8(x' . First we see from Eq. up to a scale factor.x')/dx = ~dgA(xx')/dx' as a function of x'. equal to if}2. since it samples the value of the function f(x') at one pointt (1.10. However.(a) 0A(xx'> A M A (b) dfl^KX ) dx 1 61 A *7 / € V +< MATHEMATICAL INTRODUCTION I /* Figure 1.x ' ) = .5) that g has unit area. Consider next the integral of g times f(x') over a region in x' that includes x. each b u m p at ± e will become. The I We will often omit the limits of integration if they are unimportant.8. Consider a Gaussian 1_ [ g A ( x . gA becomes a better and better approximation to the delta function.x) since the delta function is real. independent of A. As gA shrinks. has width A. As A approaches zero. Consider dgA(x.10. the delta function the sampling function. § A fine point that will not concern you till Chapter 8: This formula for the delta function is valid even if A2 is pure imaginary. we get zero because / i s smooth and g is wildly oscillating as f)*0.§ It is obvious from the Gaussian model that the delta function is even. (A. The Gaussian is centered at x' = x.2. say. 1.10. (a) The Gaussian gA approaches the delta function as A»0. (b) Its derivative (dg/dx)(xxr) approaches S'ixx ) 1 as A»0. For the most part. Consider next an object that is even m o r e peculiar t h a n the delta function: its derivative with respect to the first argument x : 8(x x1) dx dx' (1. Pulling f(x' = x) out of the integral. .19) as shown in Fig. maximum height (TTA2)_1/2.20) S'(xx')=— 8(x . It is therefore useful to view it as the limit of a more conventional function. we get the desired result. at x = x'. and unit area.
62 CHAPTER 1 first S0 one will sample —f(x — e) and the second one +f(x+s). We know from basic Fourier analysis that. that again up to a scale.10.10.x') f ( x ' ) dx'ocf{x + s) .20). 8 \x .23) We will now develop an alternate representation of the delta function. In this notation we can describe the action of higher derivatives of the delta function: dx" dx' (1. An equivalent way to describe the action of the 8' function is by the equation ! S'(xx') = S( JCJC')— dx' (1. we may define its transform m (2 n) 1/2 e~lkxf(x) dx (1. 1.24) . (1.22) ii where it is understood that both sides appear in an integral over x' and that the differential operator acts on any function that accompanies the 8' function in the integrand.21) This result may be verified as follows: S'(xx')f(x')dx'= \ J _df{x) dx dx =— dx J 8(xx')f(x')dx' Note that JC') is an odd function. given a f u n c t i o n / ( x ) .8b or Eq.10.10.f ( x £) = 2e~ dx' The constant of proportionality happens to be 1/2 £ so that df(x) v dx' dx (1. This should be clear from Fig.10.
What are the matrix elements of D in the x> basis? To find out. (1./(x).2/r dk eik(x'~x} / ( x ) </x \ Comparing this result with Eq.and its inverse 63 MATHEMATICAL INTRODUCTION 1/2 dk (1. [Consider J <5(ax) d(ax).24) into Eq. namely. (1. we see that 1 2K .26) Exercise 1. In the function space we can describe the action of this operator as D\r> = \df/dx> where \df/dx> is the ket corresponding to the function df/dx.3 * Consider the theta function O(x^x') which vanishes if x — x' is negative and equals 1 if x — xf is positive.10. (1.10. which. one operator that does such a thing is the familiar differential operator. are the zeros of fix). Exercise 1. (1. acting on / ( x ) .2 * Show that r w/dx\ where x. gives / f x ) = df(x)/dx.10. keeping the first nonzero term. we get /co= . Now.] Exercise 1.10. Remember that 5(x) = 5(x). takes the function / ( x ) into another. Operators in Infinite Dimensions Having acquainted ourselves with the elements of this function space.25) Feeding Eq.1 * Show that <5(ax) = <5(x)/a. Consider the equation fll/>=l/> Since the kets are in correspondence with the functions. the kets  / ) and the basis vectors x>.18).10. Show that 8(x. let us turn to the (linear) operators that act on them. Hint: Where does <$(/(*)) blow up? Expand f(x) near such points in a Taylor series.25).10. we dot both sides of the above equation .x') = d/dx d(xx').10.10.
If. (1. and insert the resolution of identity at the right place < x  £ >  x ' ) < x '  / > dx' = — dx Comparing this to Eq.1C It is worth remembering that Dxx = 8'(x — x') is to be integrated over the second it (x') and pulls out the derivative of f at the first index (x). we deduce that d — dx' (1. Usually when a matrix acts o vector. we will continue to use it for a while to remind us that D. Thus the familiar differential operator is infinitedimensional matrix with the elements given above. however.x') . is a matrix.with <x. In fact.27) contains sue sum over the index x'. we would have But this is not the case: Dxx=8'(x~x') while D*x = 8 '(x' x)* = 8 '(x' .x) = ~8 '(x . Eq.10. i convention is more natural if one views Dx* as a matrix acting to the right on components fx=f(x') of a vector  / > .11 <x  D x'> = Dxx =8'(x x') = 8{xx') (1. like all lint operators. Some people prefe integrate 8'(xx') over the first index. there is a sum over a common index. (1.10. we feed into this equation the value of Dxx. Although we too will drop the integral over the common ind ultimately. delta function renders the integration trivial: df dx Thus the action of D is simply to apply d/dx to / ( x ) with no sum over a comm index in sight. in which case it pulls out df/dx'. If D we Hermitian. Normally one doe think of D as a matrix for the following reason. Let us now ask if D is Hermitian and examine its eigenvalue problem.21).
Thus K is Hermitian if the space consists of functions that obey Eq. Let  / ) and g> be two kets in the function space. Let us require that these functions be single . =i dx f i x ) dx 1 tegrating the lefthand side by parts gives dg*(x) dx ig*(x)f(x) a +i f (x) dx So K is Hermitian only if the surface term vanishes: ~ig*(x)f(x) = 0 (1.29). the operator K is not guaranteed to be Hermitian. One set of functions that obey this condition are the possible configurations f ( x ) of the string clamped at jc = 0.10.10.x') = Kx* 65 MATHEMATICAL INTRODUCTION It turns out that despite the above. One also needs to look dt the behavior of the functions at the end points a and b.29) In contrast to the finitedimensional case..10. But condition (1. KXX =K*X is not a sufficient condition for K to be Hermitian. whose images in the X basis are two functions f ( x ) and g(x) in the interval ab. (1. If K is Hermitian. since f ( x ) vanishes at the end points. it must also satisfy <Jg\K\fy = <g\Kfy So we ask rb (g  x> <x\K\x') O' /> dx dx' = (Kf\gy* = (f\K'\gy* = (f\K\gy '(ITidf(x) g*(x) dx dx y(x\K\x'y(x'\gydxdx' idg(x) dx dx} v r I .But we can easily convert D to a Hermitian matrix by multiplying it with a pure imaginary number. L.29) can also be fulfilled in another way. as the following analysis will indicate. 6. parametrized by r.x' ~x) = iS '{x . Consider functionssjn our own threedimensional space. and 0 ( 0 is the angle measured around the z axis). Consider K=iD which satisfies K$x=[~i5 *)]* = +i5 '(.
Let us begin with K\k) = k\ky (1. K~ ~i d/d<f> is a Hermitian operator.^ oo CHAPTER 1 1 valued. It turns out to be quite simple and you might have done it a few times in the past without giving yourself due credit.o o < x < + o o . we will be interested in functions define over the full interval . We now turn to the eigenvalue problem of K.10. since in either caa the surface term vanishes. we have. The surface term vanishes because the contribution from one extremity cancels that from the other: 2jt = ife*(2*)/(2«)g*(0)/m= 0 o In the study of quantum mechanics.. the Hermiticity hinges on whether „ikx —ik'x = 0 e e If k = k\ the contribution from one end cancels that from the other. and those that do not. there exists a way of defining a limit for such functions that cannot make uj their minds: the limit as x oo is defined to be the average over a large interval According to this prescription.e. k being a real paramet that labels these functions. The task seems very formidably indeed. When sandwiched between two functions of the secon< class. for we have now to find the roots of an infiniteorder characteristic poljl nomial and get the corresponding eigenvectors. th< answer is unclear since e i [ k ~ k ) x oscillates. if we start at a certain point and go once around the z axis returning to the original point. rL+A lim e * e . rather than approaching a limit as\x\ » oo Now. In particular. If k^k'. those that vanish *1 »• oo. / ( * ) = / ( * + 2*) In the space of such periodic functions. i.30 .* * = lim ]x—*co Ltoo A A»co x ei(k"k')xdx =0 xik^k' and so K is Hermitian in this space. the latter behaving as e*x. the function must take on its original value. It is clear that K= i d/dx is Hermitian when sandwich© between two functions of the first class or a function from each. They fall into two classes. say as x * oo.
(We use the qualifier "physical" to distinguish it f r o m the Hilbert space as defined by mathematicians. is a free parameter. F r o m now on we shall resort to this shortcut unless there are good reasons for not doing so.Following the standard procedure. The solution to the above equation is simply y/k (x) = Ae ikx (1. i.e.31).10. our space includes such functions. however. which is of interest in q u a n t u m mechanics.. The delta function normalization is the natural one when the eigenvalue spectrum is continuous. depending on the sign of the imaginary part k2. K must be classified a nonHermitian operator.) In restricting ourselves to real k we have restricted ourselves to what we will call the physical Hilbert space. vectors normalizable to unity. W e choose A to be ( l / 2 ^ ) _ .) . unspecified by the eigenvalue problem. which contains only proper vectors. So the eigenvalue problem of K is fully solved: any real number k is an eigenvalue. As usual. (1. This equation could have been written directly had we made the immediate substitution K= id/dx in the X basis.10. If. This space is defined as the space of functions that can be either normalized to unity or to the Dirac delta function and plays a central role in q u a n t u m mechanics.32) where A. and the corresponding eigenfunction is given by A eikx.31) 67 MATHEMATICAL INTRODUCTION K\x'}(x'\kydx'=k\i/k(x) dx \j/k{x) = k\j/k(x) where by definition y/k(x) = (x\k).10. <x X] <x =  A:> (1. The role of the improper vectors in quantum theory will be clear later. the freedom in scale will be used to normalize the solution.) The attentive reader may have a question at this point. / 2 so that 1 l*> (2/r) 1/2 r oo and i(kk')x <k\fcy= e — 00 dx = 5(k — k') (1.33) (Since <Jc\ ky is infinite. (The surface term n o longer vanishes since e** blows u p exponentially as x tends to either +oo or —oo. the overall scale. no choice of A can normalize /c> to unity. yes.10. "Why was it assumed that the eigenvalue k was real? It is clear that the function A with k=*kx + ik2 also satisfies Eq. there are eigenfunctions of K with complex eigenvalues." The answer is.
3' The passage back to the X basis is done as follows: eikx f (k) dk /(*) = <*!/>« x) (&/> dk = 1/2 (1. let us begin with X\f> and follow the routine: <xX / > =  < x  X  x ' > < x '  / > dx' = x f ( x ) = <x  / > = / ( x ) Rx) = x f ( x ) % Hereafter we will omit the qualifier "physical.3 = \J> . we know where the K basis came f r o m : it was generated by the Hermitia operator K.10.10. functions that were expanded in the X bas with components / ( x ) = < x  / ) must also have an expansion in the K basis." 1 (1. To fii the components. Either basis may be used to expand functions that belong to til Hilbert space.) Since K is a Hermitian operator.3! Thus the familiar Fourier transform is just the passage f r o m one complete basis to another.31 Now. namely.10. (The trouble with infinitedimensional spaces is that even if you have infinite number of orthonormal eigenvectors. you can never be sure you have thei all.10. we start with a ket  / > . since adding or subtracting a few still leaves you with an infinite number them. 1 The matrix elements of K are trivial in the K basis: (k\K\k'} = k\k\k'} = k'S(kk') (1. Which operator is responsible for the orthonormal X basis? Let us a it the operator X.3' Its matrix elements in the X basis are <x'Xjx> = x 5 ( x ' . The kets  x ) are its eigenvectors with eigenvalue x: ^ i Xx> = xx> (1. and do the following: 1 e'ikxf(x)dx x> < x  / ) dx — (2?r) 1/2 (1.10. holds in the Hilbei space. that eigenfunctions of a Hermitian operator form a complete basis.68 CHAPTER 2 We will assume that the theorem proved for finite dimensions.x ) T o find its action on functions.
Thus the effect of X is to multiply f ( x ) by x.10.39) 69 MATHEMATICAL INTRODUCTION where as usual \xf(x)) is the ket corresponding to the function xf(x). then X\g(k)) = dk / (1. one generally suppresses the integral over the common index since it is rendered trivial by the delta function. Their commutator may be calculated as follows.il\ /> t In the last step we have used the fact that S(k'k) S(kk'). There is a nice reciprocity between the X and K operators which manifests itself if we compute the matrix elements of X in the K basis: J /* 0 0 I < w > = In „ —ikx „ no xe dx \ ei{k'~k)x dx\ = / 1 iS'(kk')X Thus if \g(k)) is a ket whose image in the k basis is g(k). while in the K basis. in the X basis. Let us operate X and K in both possible orders on some ket  / ) and follow the action in the X basis: X\ />>*/(*) dx So df(x) dx KXlf)^i^xf(x) dx Therefore [X.40) In summary then. . We can summarize the action of X in Hiibert space as X\ / ( * ) > = !*/(*)> (1. Operators with such an interrelationship are said to be conjugate to each other. X acts as x and K as —id/dx [on the functions /(*)].10. As in the case of the K operator. The conjugate operators X and K do not commute. K acts like k and X like i d/dk [on f(k)]. K ] \ f ) ix f + ix dx dx = if= if .
The analog of the operator £2 in Eq. Construct the propagator U(t) in terms of the eigenvecto eigenvalues. t) as components of a vector  y ( 0 > in a Hilber the elements of which are in correspondence with possible displacements functions that are continuous in the interval 0 < x < L and vanish at the end You may verify that these functions do form a vector space.8 [see Eq. But for the change in dimensionality.10. The displacement y/(x. Equation (1. We re this to be minus the square of the operator K^id/dx.8. Solve the eigenvalue problem of — K 2 . we now have the desired result: [X. Consider of length L clamped at its two ends x = 0 and L. Since K acts on a s 2 which y/(0) = y ( L ) = 0. except f< example. where ft is a consta defined later. K] = il This brings us to the end of our discussion on Hilbert space. it is Hermitian. t) c differential equation d2if/ _d2\f/ dt2 ~ dx2 ( Given that at r = 0 the displacement is y/(x. 0) and the velocity \j/(x. Step (2).42) abstract counterpart lv>(0>—*2lv(0> (i We solve the initialvalue problem by following the algorithm developed in E 1. to determine the time evolution of the string.2( recommended that you go over that example once to refresh your memon proceeding further. I v ( 0 > = t/(0lv(0)> 0 .6: Step (1). and so is K . (1.8. Although there are many other operators one can study in this s restricted ourselves to X and K since almost all the operators we will quantum mechanics are functions of X and P = UK. (1.70 CHAPTER 2 Since  / > is an arbitrary ket.26) is the operator d2/dx2.8. the problem is identical to tha two coupled masses encountered at the end of Section 1. We first identify ii/(x.10. Step (3).1: A Normal Mode Problem in Hilbert Space. Example 1. 0) = 0.
(1.51) o Let us associate with each solution labeled by the integer m an abstract ket \m): m> > ( 2 / L ) . 3 .10. At x = 0 we find V*(0) = 0 = A while at x=L we find 0 = B sin kL If we do not want a trivial solution (A = B = 0 ) we must demand sin kL = 0.10.46) the general solution to which is yk(x) = A cos kx + B sin kx (1. not all these solutions lie in the Hilbert space we are considering.50) ~where we have chosen B=(2/L)x/2 so that \ffm(x)\f/m {x) dx=Smm' (1.The equation to solve is 71 MATHEMATICAL K 2 \y/} = k 2 \\i/) In the X basis. kL=mji.45) INTRODUCTION (1. We want only those that vanish at .47) where A and B are arbitrary.10.10.49) (1. m= 1 .10.10. this becomes 2 d2 if/k(x) = k if/k(x) dx (1. However. / 2 sin ( — ) X basis \ L / (1.48b) (1.52) . . 2 .c = 0 and x — L. The allowed eigenvectors thus form a discrete set labeled by an integer m: V«(*)=(f) s i n (^~) (1. . .10.10.s i n x}.10.48a) We do not consider negative m since it doesn't lead to any further LI solutions [sin(x) = .
2 .10.10.58) .56) that <x U(t)\x'} <x w ) </w x ' ) COS GW . ItlK com — — (1. (1.10.10. .57) It follows from Eq. . in which K is diagonal with eigenvalues {mn/L)2.10. (1. 0 = <xC/(OIV(0» ' L <x f/(0l*'> <x' y(0)> dx' (1.8. m= 1 .55) or 00 U(t) = £ l m ) < w l c o s comt.10.10. These equations may be readily solved (subject to the condition of vanishing initial velocities) as ( Consequently IV(0>= I [m><wy(0> m= 1 oo = mjit\ — J (1.53) in analogy with Eq.33).72 CHAPTER 2 If we project  i f ( t ) } on the [m) basis. the components ( m \ v ( 0 > will obey the decoupled equations ' m 2 n2s —2 <m v(0> = ~ \jtJ<m\ W(t)>. (mizx\ . (mnx'\ I { — J sm —^—J sin ^—^— J cos = (1.56) The propagator equation lv(0> = ^(0lv(0)> becomes in the  x ) basis <x V(0> = V(*.54) L m~ 1 V(0)> cos (Dmty com=— L (1. . (1.
0). (1.58). • Exercise 1.59) Given 11//(0)). one must then compute <mv/(0)>=f£) sin j Vi** 0) dx Usually we will find that the coefficients <m i//(0)) fall rapidly with m so that a few leading terms may suffice to get a good approximation. Dotting both sides with < x  .10. (1. Show that L 0<x< — L <x< L „(x. (mnx\ .n ^ (nm\ j . j s. given any y/(x'. we get 00 0 = E (x\my m=1 00 73 MATHEMATICAL INTRODUCTION (m\ y/(Q)) cos COmt (mnx\ = E (jj / 2 \1/2 sin^^Jcosfi>mr<w v(0)> (1. 2h = — (Lx).0)=—.10. we can get y{x.10. A string is displaced as follows at / = 0: .10. 0 = 1 .n ^ — j COS m a • ^ ( %h \ . we can begin with Eq. t) by performing the integral in Eq. (1.4.10. using (x\ C/(0x'> from Eq. If the propagator language seems too abstract.55). a\ 2xh yr(x. ® .57).Thus.
which we will follow f o r the most part. one or the other may be more inviting for conceptual. For example. It should be emphasized. computational. again to inch forward to t. But our principal interest in them stems from the fact that they are the ideal springboards from which to make the leap to quantum mechanics. and it will soon become apparent. The passage from the Lagrangian formulation to quantum mechanics was carried out by F e y n m a n in his p a t h integral formalism.+Af) = x(ti) + x(tt)At we can repeat the process Having updated the state variables to the time t{+At. According to Newton's Second Law. has as its starting point the Hamiltonian formulation. in that their domains of validity and predictions are identical. Using the initial velocity and acceleration [obtained from Eq.1. or simply aesthetic reasons.1)] we compute the position and velocity at a time /(+Af. The Principle of Least Action and Lagrangian Mechanics Let us take as our prototype of the Newtonian scheme a point particle of mass m moving along the x axis under a potential V(x). Heisenberg. Nonetheless. Dirac. that all three formulations of mechanics are essentially the same theory.). and Born.l u ( 2 U ) If we are given the initial state variables. (2. i xel (/. d 2 * = d v m l ? .2 k Review of Classical Mechanics  1 In this chapter we will develop the Lagrangian and Hamiltonian formulations of mechanics starting from Newton's laws. we can calculate the classical trajectory xct(t) as follows. the position JC(^) and velocity *(/. in a given context. . A more common route to q u a n t u m mechanics.+2Af and so on. These subsequent reformulations of mechanics bring with them a great deal of elegance and computational ease.1. a n d it was dis[ covered mainly by Schrodinger. 2.
and x(ti).) The Lagrangian approach is thus global.1. 2. . respectively.V.x. An equivalent way to do the sam< and one that we will have occasion to employ. what is it that distinguishes the actual trajectory xci (t) from all other trajectories or paths that connect these points? (See Fig.) The motion of the representativi point is given by d2x d v J where mj stands for the mass of the particle whose coordinate is Xj.) an The equation of motion being second order in time. (The term "configuration space' is used even if the n coordinates are not Cartesian. Thus J? = J?(x. however. and xf at times f. In the Lagrangian formalism. are needed to specify a unique x c) (t). is to specify two spacetime poin (X. given by i f = T . which concerns itself with what the particle is going to do in the next infinitesimal time interval. the problem of a single particle in a potential! V(x) is posed in a different way: given that the particle is at x. x„) to specify th positions of the particles.X2. t f ) on the trajectory. tt) and ( X f . T and 1 being the kinetic and potential energies of the particle. two pieces of data.1. If we use n Cartesian coordinates (X\. assume the absence of this t dependence.. the spatial configuration of the system may be visualize* as a point in an ndimensional configuration space.(*i.1.3 . to determine the trajectory. The above scheme readily generalizes to more than one particle and more thai one dimension. These equation can be integrated step by step. Th explicit t dependence may arise if the particle is in an external timedependent field We will. in that it tries to determine at one stroke the entire trajectory (r). tf ). in contrast to the local approach of the Newtonian scheme. The answer to the question posed above comes in three parts: (1) Define a function called the Lagrangian.x)dt J (2.. just as before. (2) F o r each p a t h x(t) connecting (JC. x(f. . and t f . The Lagrangian formalism asks what c tinguishes the actual path xcl (f) taken by the particle fra all possible paths connecting the end points /.1l> t Figure 2. calculate the actioi S[x(t)] defined by f 5tx(0]= y(x.. tt) and (x/. t).
.6) (2.77 < xf . the minimum x° is characterized by the fact that if we move away from it by a small amount T in any direction. To prove this. . If J c (f) minimizes S.. then 5Sm = 0 if we C1 go to any nearby path xcl (/) + r\(t). the firstorder change 8 f 0 ) in / vanishes.ti) We use square brackets to enclose the argument of S to remind us that the function S depends on an entire path or function x(t)..2..=] OXj then (2.1.1.. . we simply choose r\ to be along the ith direction.1.. i=\. It is. and not just the value of x at some time t. (Actually we will only require that it be an extremum. df/dxi.4) Let us now mimic this procedure for the action Let x d (t) be the path of least action and xa(t)+Tj(t) a "nearby" path (see Fig. The first step is to realize that a functional 5[x(0] is just a function of n variables as In other words..) We will now verify that this principle reproduces Newton's Second Law. customary to refer to this condition as the principle of least action. Thus V dx.tf> REVIEW OF CLASSICAL MECHANICS Figure 2. . That J is. xn) = /(x).7) . (3) The classical path is one on which S is a minimum.. <* l. for say. if we make a Taylor expansion: T]i + higherorder terms in t] /(x°+ti)=/(x°)+ i f . x(tf). one for each instant in time t in the interval ti<t<tf.. and S is a function of these variables. the function x(t) simply specifies an infinite number of values x(ti). To find its minimum we simply generalize the procedure for the finite n case.. Let us recall that i f / = / ( x i . and tf means (2. x(t).n (21.5) 8xi From this condition we can deduce an equivalent and perhaps more familiar expression of the minimum condition: every firstorder partial derivative vanishes at x°. The requirement that all paths coincide at t. One calls S a functional to signify that it is a function of a function. = 0. however... 2. .2).
1.5 To deduce this result for some specific time f 0 .Xci(t)) dse dx(t) *cl = S[xc\ (f)] + + dse dx{t) n(t) 77(0 + dt ' + higherorder terms We set <?>S(1) = 0 in analogy with the finite variable case: 0 = SS(i) = " dse / ^ d & Jx(t) m c Vl dt If we integrate the second term by parts. continue the tradition of referring to this extremuni as the minimum. Note that the condition <SS(1> = 0 implies that S is extremized and not necessari minimized. d d^ 77(f) _dt dx(t)_ •V] C dt The first of these terms vanishes due to Eq. (2. it turns into dse dx{t) Xcl 'f V(t) h « ptf t.1.1. the sum over i is replaced by an integral over f. (2. u replaced by d££ dx(t) d d££ dt d x ( t ) There are two terms here playing the role of df/dx. and df/dx. We shall. however. Since 77(f) arbitrary.1. So that W 0 = SS = r>/ dse cbc(f) d dse •c Vl dt dx(t) 77(t) dt (2. since (or equivalently S) hi both explicit and implicit (through the x terms) dependence on x(t). (2.1.7). xM + 77(f)) dt f ^(xc](t).5): the discrete variable q is replaced by 77(f). This equation is the analog of Eq.78 CHAPTER 2 Now S[xa ( f ) + 17(f)] = l J?(xd(t) + ?K0. we may extract the analog of Eq. we simply choose an 77(f) that vanish? everywhere except in an infinitesimal region around fo • .6): I d x ( f ) dt _ dx{t) =0 for t i < t < t f (2.
since FB=q\ x B. that of a magnetic field B on a moving charge is excluded by this restriction. (2.1.1. If we feed into it 79 REVIEW OF CLASSICAL MECHANICS dV dx dx that the EulerLagrange equation becomes just — (mx) = — — dt dx which is just Newton's Second Law.y j dt cXi Which is identical to Eq. Eq. q being the charge of the particle and v = r its velocity. (2. in the sense that we can find an that yields the correct [force law when Eq.1.Equation (2. We will show shortly that this force too may be accommodated in the Lagrangian formalism.1.1). An important force.1. we get dSe_dT dx dx mx equation. Thus the minimum (action) principle indeed reproduces Newtonian mechanics if we choose i f = T— V. V= V(x).10) T= 2 x i=i and V= V(x.10) is employed. and views i f as some . T= 2mi 2 .2). the same procedure yields dt \ dxi J Jow dxi (2. (2. If we consider a system described by n Cartesian coordinates. [So that Eq.9) is the celebrated EulerLagrange '= T.1.10) becomes m >(x.V. One therefore frees oneself from the notion that i f = T— V. But this i f no longer has the form T— V. (2. Notice that we have assumed that V is velocityindependent in the above proof.
at the next stage. x„.10)1 and seeing that an identical equation with x....1.1.. T o the reader who wonders why one bothers to evi deal with a Lagrangian when all it does is yield Newtonian force laws in the end present a few of its main attractions besides its closeness to q u a n t u m mechanii These will then be illustrated by means of an example.1.1. But once the principle of least action is seen to generate the correct dynamics..1. for in these coordinates the kinetic energy T and the Newtonian equations have simple forms. (2. which deals with vectors and is thus more complicated. insti of the n Cartesian coordinates Xi. are Cartesian.13 called the generalized force conjugate to q.1. . in showing that the EulerLagrange equations were equivalent to Newton's.10) have the same form if we use. for another nonCartesian set of q. (2. Cartesian coordinates were used.10) as (2. nor is I always a force (with dimensions of mass times acceleration). making a change of variables in Eq. Equation (2.11) One can either verify this by brute force. Of course. p.2) will have a different form (see Example 2.1. one must remember that neithe is pi always a linear m o m e n t u m (mass times velocity or " m u " momentum). (2. are form invariant under an arbitrary change of coordinates. replaced by q. (2. .12! F^f oq< (2.1.. if q. we can forget all about Newton's laws and use Eq. q„. This must be con trasted with the Newtonian scheme.1 at the end of this section). (1) In the Lagrangian scheme one has merely to construct a single scalar $ and all the equations of motion follow by simple differentiation. will be an angular m o m e n t u m and F( a torque. . follows.11) can be m a d e to resemble Newton's Second Law if one defines a quantity dS€ P^fr dqi called the canonical momentum conjugate to <y. and the quantity (2.1. This f o r m invariance must be contrasted with the Newtonian equation (2. which presumes that the x. ) which yields the correct Newtonian dynamics when fed into t] EulerLagrange equations.. Eq. W h a t is being emphasized is that these equations.1. which express the condition for least action.1.80 CHAPTER 2 function JSf (x. Although the rate of change of thi canonical m o m e n t u m equals the generalized force.2).11) as the equations of motion. . i . F o r example. any general set of n independent coordi nates qu q2. To remind us of this fact we will rewrite Eq. (2) The EulerLagrange equations (2. If one trades the x. or one can simply go through our derivation of the minimum action condition and see that nowhere were the coordinates assumed to be Cartesian. is a angle 0. .
If one rewrites Newton's laws in polar coordinates to exploit 3V/d(j> = 0. It is the Lagrangian formalism that allows us to choose coordinates that best reflect the symmetry of the potential.1. without altering the simple form of the equations. Example 2. and not on the polar angle <p.16) my which are identical to Newton's laws.1 below).1. in Cartesian coordinates. (2. for example. the corresponding momentum is conserved. The corresponding equations of motion are d_v dx dV dy (2. In contrast. as T depends only on 0 (see Example 2. is = imwwV(x. The Lagrangian. some careful vector analysis is needed to unearth the centrifugal and Coriolis terms . If one wants to get the same Newton's laws in terms of polar coordinates p and 0. (2. <p)= V(p).2). the corresponding equations get complicated due to centrifugal and Coriolis terms. with v = r . Eq. no obvious conservation law arises from the Cartesian Eqs. but not on the corresponding coordinate q.19) . It follows that the corresponding pt is conserved: 81 REVIEW OF CLASSICAL MECHANICS (2.(3) Conservation laws are easily obtained in this formalism.1.15) where v is the velocity of the particle. y) in two dimensions that depends only upon p = (x2+y)]/2.18) 1 dV 2mp<j> m<p — ——z — P p (2. It follows that 0 is a cyclic coordinate.17) mx = (2. a potential V(x.1. Consider a particle moving in a plane.y) (2.1.14) is more general.. + mp(0)2 mp = (2. The latter is then called a cyclic coordinate. 8V dp • .1. Consequently 3 i ? / 3 0 p<t> is conserved.1. (2.1.2) since neither x nor y is cyclic. Suppose the Lagrangian depends on a certain velocity q.1.1. r being its position vector.1. so that V(p.1. We now illustrate the above points through an example. also tells us that if a Cartesian coordinate Xj is cyclic.14) dt \ dqi J dt dqt Although Newton's Second Law. Eq. Consider.
+ (pd(l>)2]]/2 .16) and (2. Notice how easily the centrifugal and Coriolis forces came out. (2. the conservation of p$ is obvious in Eq. • dV dp 12 \mpq> dV 7 o<p (2. 2.19). •.1. both along the z axis. Both the conserved quantity an( its conservation law arise naturally in the Lagrangian scheme. In Eq.18) and (2.1.22] which are the same as Eqs. In the Lagrangian scheme one has only to recompute i f in polar coordinates. . This does not happen in Cartesian coordinates.19) on the other.1. (2. From Fig.17) on the one hand and Eqs.1.1.± x~ Figure 2. (2.1.1.1.16) and (2.1. (2.1.211 (2. Points (1) and (2) are positions of the particle at times differing by At.18) and (2. . if V(p.3 it is clear that the distance traveled by the particle in time At is dS=[(dpf so that the magnitude of velocity is dS dt and <e = \m( p2 + p2<j>2) .V(p.1. (2. <p) (2.1.22) the canonica m o m e n t u m p ^ = mp2(j) is the angular momentum and the generalized force —dV/di is the torque.3. Finally.19) only after some manipulations am is practically invisible in Eqs.17).„ (Notice that in these coordinates T involves not just the velocities p and 0 but alsc the coordinate p. (2. <f)) = V(p). Notice the difference in form between Eqs. (2.) The equations ol motion generated by this i f are d — (mp) = dt d dt . .1.20.22) The conservation of p^ follows from Eq.1.
not relying on the preceding discussion.8.3) and B=Vx A t See Section 18. Compare the equations of motion with Eqs.8.2) c is the velocity of light.2. called a harmonic oscillator.2.2* Do the same for the coupledmass problem discussed at the end of Section 1. The Electromagnetic Lagrangian} Recall that the force on a charge q due to an electric field E and magnetic field B is given by F= g E+ xB (2. while (j) and A are the scalar and vector potentials related to E and B via _ E = V0 ^ 1 dA c dt (2. 8.2.1 * Consider the following system. we must analyze the problem afresh. Exercise 2. 2.1.1. which was restricted to velocityindependent forces.8. (1.2.1) where v = r is the velocity of the particle.3* A particle of mass m moves in three dimensions under a potential V(r. 83 REVIEW OF CLASSICAL MECHANICS f V V V V V VY V J ) Y V V Y V Y VV> T ' / / / / ~7 ~7~ Write the Lagrangian and get the equations of motion.25). The block has a mass m and lies on a frictionless surface. The spring has a force constant k.2. (2. Now it turns out that if we use (2. (j>) = V(r).2) c we get the correct electromagnetic force laws. (2. Write its if and find the equations of motion.1.24) and (1. Since the force is velocitydependent. Exercise 2.4 for a review of classical electromagnetism.4) .F Exercise 2. In Eq.2.
. The rule i f = T . the total derivative d A / d t has two parts: an explicit time dependence dA/dt plus an implicit one ( v V ) A which represents the fact that a spatial variation in A will appear as a temporal variation to the moving particle.2. redefine if to be that function J£(q.2.2. To accommodate forces such as the electromagnetic. we must.2. we get d a — (m\) = q\<t) + 1 dt c dA dt + VK v A ) ( (2.Now Eq.' which is identical to Eq. Rewriting Eq. reproduces the correct dynamics.4 « a\ ty — \mxi + Ai\ = q— + at \ c j 0Xi c qdjvA) .6).8) Now. (2.2. 3 (2.V ( v A ) dt \ c J c The canonical momentum is qA (2. therefore.+ [V(rA)(rV)A] c dt c = q\<i> (2.Vbecomes just a useful mnemonic for the case of conservative forces.^ V 0 + .5 Combining the three equations above into a single vector equation we get — (m\ + —  = . only q<p can be interpreted as the electrical potential energy. when fed into the EulerLagrange equations. p = mv + (2.2. since the magnetic force FB = q(v x B ) / c never does any work.8) become^ d {rm) dt q dA q .(v'V)A Notice that is not of the form T~ V.m are d ( . q.2. First of all. Even in the special cases when the force is conservative. (2.A) . for the quantity U=q<p{q/c)vk (sometimes called the generalized potential) cannot be interpreted as the potential energy of the charged particle.1) by virtue of the identity v x (V x A) = V(v. t) which. The [—q(v A)/c] term is not a magnetic potential energy. the force due to a timedependent electromagnetic field is not generally conservative and does not admit a pathindependent work function to play the role of a potential. (2. being always perpendicular to the velocity.2.2.84 CHAPTER 2 The EulerLagrange equations corresponding to S£e. 0Xi i=l.
3.3. . The TwoBody Problem We discuss here a class of problems that plays a central role in classical physics : that of two masses m1 and m2 exerting equal and opposite forces on each other.4) f ri = rCM + (2. it follows that the potential between them depends only on the relative coordinate r = r.6) .3) m} + m 2 r 2 = rCM If one rewrites the Lagrangian ml + m 2 (2.3. Since the particles are responding to each other and nothing external.4) 2«i*i 2 + im2\r2\2in terms of rCM and r. . a coordinate missing in V is also cyclic.3.r 2 ) (2. for V depends on only three variables rather than the possible six. (2.3.3. To bring out these features.1) where rCM is called the centerofmass (CM) coordinate.r 2 and not the individual positions r. r 2 ) = V(r1r2) means in turn that there are three cyclic coordinates.) The corresponding conserved momenta will of course be the three components of the total momentum.3. and r 2 . and r 2 in favor of r = r.5) if = i ( W l + m 2 )  r C M  2 + i . (In Cartesian coordinates.3.2) to get (see Fig. One can invert Eqs.2. it is better to trade r. r  2 2 mi+ m2 2 F(r) (2.^ .2) (2.3. since T is a function only of velocities.r2 and icm= m]T] +m2T2 ni\ + m 2 (2.1) and (2. But F f r . 2. which are conserved in the absence of external forces. one gets V(r} .
which is to say. using Eqs.3. H o w Smart Is a Particle? m The Lagrangian formalism seems to ascribe to a particle a tremendous amoun of foresight: a particle at (x.5) by changing variables. h) destined for (xf.3. One clear way to do this is to go to the C M frame in which rCM = 0.3. and takes the one wit! the least action.. Since the motion of the C M is uninteresting one usuall] ignores it. r) + (RCM. classical particles do seem to follow x c] (t) is an interesting question that will be answered when we come to the path integral formalism of quantum mechanics. This in turn means just following Newton's law.3. One has just ti solve this onebody problem. But this. goes from (x.. r TCM ? *CM) = (r. Since r CM is I cyclic variable. the independent variables are the coordinates q and velocities g h The momenta are derived quantities defined by = ^ oqi (25H . the particle has to sample the potential in its immediate vicinity and accelerate in the direction of greatest change.86 CHAPTER 2 The main features of Eq.2).1.3. s < that the C M is completely eliminated in the Lagrangian. ( 2. giving equal weight to each!. (2.* Derive Eq. and r2 at the end. 1 2. r. How it is that despite this. is an illusion.4. it needs only to obey the EulerLagrange equations at each instant in time to minimize the action.3. The particle need not know it' entire trajectory ahead of time. of mass + m2. t f ) along all possible paths.6) are the following. in a sense.6) from (2. one may easily return to the coordinate r. momentum p = //r and moves under a potential V(r). of course.) to (xf. the momentum pCM = M " c m (which is just the total momentum) ij conserved as expected. The Hamiltonian Formalism In the Lagrangian formalism. because i?(r. (3) The second fictitious particle has mass n = m{m2/(mx + m2) (called tti reduced mass). If one chooses. the particle. (2.5. t f ) manages to calculate ahead ol time the action for every possible path linking these points. In other wordij the equations of motion for r do not involve rCM and vice versa. ICM )• (2) The first fictitious particle is the CM. • (1) The problem of two mutually interacting particles has been transformed! that of two fictitious particles that do not interact with each other. (2. Exercise 2. We will discover that far from following any kind of strategy.1) and (2. Our esteem for the particle will sink further when we learn quantum mechanics.
. . Uj. t We will often refer to qi.. x2. .4) u + x(u) (2.4) a Legendre transformation and / and g Legendre transforms of each other...6) It is understood in the righthand side of Eq.. .'s to be eliminated have been rewritten as functions of the allowed variables in g.] If we define a function g(u) = then dg dx — =— du du df dx dx — = x(u) du x(u)uf{x{u)) =«1\ (2. i f / = / ( j c i .5. x„). One calls Eq. . and q becomes a derived quantity. xn) = £ UiXif{x}.. (2..2) thereby completing the role reversal of the g's and the p's. which is illustrated by the following simple example.5. Suppose we have a function f ( x ) with df dx x(u) u(x) = (2.3) Let it be possible to invert u(x) to get x(u).5. More generally..5. . one keeps all the other variables in g constant. i=i ..7) where in taking the above partial derivative.p„ as p. [For example if u{x) = etc.. . xj+.In the Hamiltonian formalism one exchanges the roles of q and p: one replaces the Lagrangian q)X by a Hamiltonian J f (<y. .. There exists a standard procedure for effecting such a change. . xn) (2. .5. . one can eliminate a subset i = 1 to j} in favor of the partial derivatives M = dfjdxt by the transformation .5. It can be easily verified that dg — = x< OU. . p) which generates the equations of motion. .6) that all the x.5) That is to say. ] g(u!. q„ as q and pi. called a Legendre transformation.5. in going from / to g (or vice versa) we exchange the roles of x and w. (2.5. . (2. . dJf <tr dpi 87 REVIEW OF CLASSICAL MECHANICS (2.
q and p are independent variables. mor succinctly. n. (1) Hamiltonian formalism The state of a system with n degrees of fn dom is described by n coordinates and momenta (qx.5. and obtain Hamilton' (2...88 CHAPTER 2 (1) Table 2.•(())).] Similarly. z = 1 . Consider now dtf CPi (2. Comparison of the Lagrangian and Hamiltonian Formalisms Lagrangian formalism The state of a system with n degrees of freedom is described by n coordinates (qi. This inversion is general!. q). .•(()). . (<?. and p t = d ^ / d q f is a polynomial of ran 1 in the q's. .5. . qn) and n velocities q„).5.10)1 that is.9 dqj Ipj. .q„. .1 provides a comparison of the Lagrangian and Hamiltonian formalisms.5. by (q. (2./>..5. . [There are no (8^/8qj)(8qj/dpi) terms since q is held constant in 8Jf/8pr. easy since i f is a polynomial of rank 2 in q. .. The n coordinates evolve according to n secondorder equations. . (2) (2) (3) (4) (3) (4) Applying these methods to the problem in question. .pn) or. The state of the system may be represented by a point moving with a definite velocity in an ndimensional configuration space.. Note that we have altogether 2n firstorder equations (in time) for a system with n degrees of freedom. J ! (2. we cai integrate the equations to get (qt(i).=— d V % sse\ (2. e.7). The state of the system may be represents by a point in a 2ndimensional phase spaa with coordinates (qt....pt.12. 8jf 8q( j ^ dqj_d& } dqj__d& } % 8qt 8qj dqt 8qt (2.8 8 / OPi \ j .2. we define q) 1=1 where the q's are to be written as functions of q's and p's.p).pi(i)).. j fyi J fy fyi = qt since/>. Table 2. or in a more compact notation by (q. Eq.11) We now feed in the dynamics by replacing (8S£/8qt) canonical equations: 8jf 8Pi 4M 82*e _ z Pi 8qi by pi.g. For a given only one trajectory passes through a given point in phase space.pn The 2n coordinates and momenta obey 2 » firstorder equations. For a given f / \ several trajectories may pass through a given point in configuration space depending on q.q„ \ pu . .. Given the initialvalue data. .1.
just as may be interpreted as T— V if the force is conservative. = 2T.14) . Tjjiq)^. iHii. in terms of which 89 REVIEW OF CLASSICAL MECHANICS T= £ i= 1 2 pi= — = OXj dT ~T=mixi UXj and rt n = 2 T Z M " X i=1 1=1 so that (2.14) is a relation among scalars and thus coordinate idependent. for which — 2mx2 The canonical momentum is \kx2 p .13) (2. so there exists a simple interpretation for J f in this case. The Hamiltonian method is illustrated by the simple example of a harmonic llator. the resulting equation (2.5.q.5.5.— = mx dx It is easy to invert this relation to obtain x as a function of p : x = pjm .1.^.5. Show that if T= ' p. Exercise 2. total energy. Consider the sum ^ us use Cartesian coordinates. where q's are generalized velocities. Notice that although we used Cartesian coordinates along the ay.Now.
17) These equations can be integrated in time.17) to get .5. ~— = </—>. Using the conservation of energy. respectively.1.3. If. however.5. Exercise 2. show that the trajectories in phase space for the oscillator are ellipses of the form ( x / a ) 2 + (p/b)2 = 1.5. pi is the reduced mass.hkx2 = ^+lkx2 2m 2 The equations of motion are dJf p . q is the charge and not the coordinate. where M is the total mass. Let us therefore repeat the analysis for the electromagnetic case. the total energy. (2. given the initial q and p.m is not sensitive in any way to the velocitydependent nature of the force.5.5. () and 4>(r.2 using the Hamiltonian formalism.16) with respect to time. so will J f e m . and feed it into Eq. i 2. mx + kx = 0 Exercise 2. where a2 = 2E/k and b2 = 2 mE. we differentiate Eq. the dynamical content of the schemes being identical.5. t). As and! t Note that in this discussion. The Electromagnetic Force in the Hamiltonian Scheme The passage from to its Legendre transform J f e. .3.5.4. * Show that J f corresponding to i f in Eq. Solve Exercise 2.= x m dp m aye —=p^kx=p 8q (2. the velocity independence of the force was assumed in showing that the numerical value of J f is T+ V. p) = T+ V= l2m[x(p)]2 4.5. I Exercise 2. we want the familiar secondorder equation.90 CHAPTER 2 and obtain 3/e (x.6.m generated the correct force laws.16) (2.2.15) (2.6) is J f =  p C M  2 / 2 M + p 2 / 2/i + V(r). If i f e. pCM and p are the momenta conjugate to r C M and r. In contrast. (2. (2. The (Cartesian) coordinate r is hidden in the functions A(r.
7.1) = {my• y + q(J) — T+ qtfi low.6. Cyclic Coordinates. Making the change of variables.p = my we have em F * em qk c 91 REVIEW OF CLASSICAL MECHANICS = mvv + q vA 1 qs'K — mvv + qcp — c 2 c (2. <f>)= V(p).6. L . in terms of p and not v. that may be conserved in addition to the canonical momenta. q) be some function of the state variables. How is to generate the correct dynamics without knowing what A is? The answer is.1): the vector potential A ieems to have dropped out along the way. there will be other quantities. Let (o{p. (2.§ There exists a nice method of characterizing these in the Hamiltonian formalism. and p<h = mp24> = lz is conserved because it is the momentum conjugate to the cyclic coordinate <f>.y)=V(x2+y2).2) § Another example is the conservation of l2 — xpy—ypx when V(x. V(p. there is something very disturbing about Eq.2) 2. if we work in polar coordinates. of course. such as the energy. i \ dqi ' .1) dqi Now. we get 2m with the vector potential very much in the picture. then 0 (2. \ djf\ dpi 7 idea dJf_dm t \ dqi dpi = {(o. the J f is more than just T+qq!>. Its time variation is given by da) _ dt /8co .Jtj dpt dqi j (2. There are no cyclic coordinates here. in particular.6. and Canonical Transformations Cyclic coordinates are defined here just as in the Lagrangian case and have the same significance: if a coordinate qt is missing in J f . Of course.7. (2. it is T+q(f> written in terms of the correct variables. Poisson Brackets. with no explicit dependence on t.7.
„.* Show that {«.3) It follows from Eq.10) and (1. * (i) Verify Eqs.« (2./>/} = 0 {qi.7. \dqt dpi dm 6X\ dpidqiJ and ' (2...5. i.jr} Pi={Pt. Canonical Transformations j We have seen that the EulerLagrange equations are form invariant under an arbitrary! change of coordinates in configuration space qi^Qi(q\ > • • •» qn).4b) etc. (2.6a) J We assume the transformation is invertible.11) for commutators. Of fundamental importance are the PB between the q's and the p's.7. J . or p.5b) 1 Exercise 2. (1. = to} {to. /)].. conserved.2) that any variable whose PB with Jf vanishes is constant in time.5a)l (2. (4. (2.. may also depend on time explicitly [q = q(q. so we may write q in terms of q: q~q(q). Hamili i (2.). in Eq.1. a} Note the similarity between the above and Eqs.2. X] + {a>.7. X+a} = {GJ.5). (ii) Consider a problem in two dimensions given by Jf = pi + + ax2 + by2.7.Pj} = S 9 since (qt.2).. In particular jf itself is a constant of motion (identified as the total energy) if it has no explicit t dependence. JV) must vanish. Exercise 2. dqijdpk0. (2.7.7. X}a + X{a. _ (dm dk .7.q) AQ>. Observe] that = {/>*..7.4a) (2. Argue that if a = b. The transformation. /'=!.4) and (2. a) {ft?. but we do not consider such cases.5. . Verify by explicit computation.92 CHAPTER 2 where we have defined the Poisson bracket (PB) between two variables m(p.7.e.p„) are independent variables (Cqfdqj5lh ton's equations may be written in terms of PB as qt={qi.7.*} by setting m =q. .7..7.g)tobe . (2. ACT} = {(o.
7) The response of the canonical momenta may be found by rewriting i f in terms of (q. dq.V.8): Pr m s j ) dqi (2. q). q) »• (q.e.10) where Jf=Jf(q. however.7.7. The transformation qi^qtei.9) Notice that although i f enters Eq.. By i f (q. does change and so we should really be using two different symbols i f (q. which connects p to the old variables. (2. (2.9).7.7.8) I (2. The invariance of the EulerLagrange equations under (q. q) refer to the same physical state. (2.11) .6a): . ( q . it drops out in Eq.8). q) and taking the derivative with respect to q: i The result is (Exercise 2. q) implies the invariance of Hamilton's equation under (q.or more succinctly q q(q) (2. q) and ^(q.7. q).7. by a fixed symbol in all coordinate systems.6b) 93 REVIEW OF CLASSICAL MECHANICS Hie response of the velocities to this transformation follows from Eq. This is as it should be. q) we mean the Lagrangian (say T.7. a rotation) is a purely kinematical question.p). q) (q. Pi = I ( j ^ j P j (2. The functional form of the Lagrangian. A word of explanation about J£(q.. such as the Lagrangian. The proof is simple: we start with q). p ) obey q^dtf/dp.7. for we expect that the response of the momenta to a coordinate transformation (say. Nonetheless we follow the convention of denoting a given dynamical variable. qn). for definiteness) written in terms of q and q. Thus the numerical value of the Lagrangian is unchanged under (q. i. and use the fact that q obeys EulerLagrange equations. • • •. perform a Legendre transform.7. q) and (q.7. pf=(djr/dqi) (2. p) (q. [ dqA . for (q. T (2.p) is the Hamiltonian written in terms of q and p. q).
1 If we view Jf as a function of (q. the simple form mqt = —dVi is valid only if the q t are Cartesian). the point transformation is unnecess restrictive.7. this is the most general (timeindependent) transformation ^ preserves the form of Hamilton's equations (that we can think o f ) . Let us now ask the following question. p(q. Any set of coordinates (qi. In this space.p). (This is like saying that although Newton's laws ma} written in terms of any complete set of coordinates.7. . are canonical cooi nates. Given one set. how can we tell if they are canonical [assuming (q.= dl£/dqi). we get and (2.p) obey the canonical equati (2. p) and use the chain rule. Given a new set coordinates (q(q. p)). p) are formally adequate describing the state of the system.p). . q„). if we view the Hamiltonian formalism in its own right. by the point transformati which is a special case of the canonical transformation.10). 2/7dimensional phase space.7 p^p(q. On the ( hand.7.7. not all of them will preserve the canonical f< of Hamilton's equations. (q. (2. we can get another. If we view the Hamiltonian formalism as some derived from the Lagrangian scheme. P) (2.1 Applying this to qj(q.p) Although all sets of 2n independent coordinates (q. however. p). and the correspond momenta generated in the Lagrangian formalism (p. which is formulated in ^dimensional cc uration space.p) that (2.151 . If. One can contemplate a more general transformation of phase s coordinates: 9 q(q.is called a point transformation. . howe^ exhaust the possibilities.12) define canonical transformation.p) we find (2.7. . we say that they are canonical coordinates and that Eq. are]? Now it is true for any oo(x. the backdrop i. (q. (q. This does not.
7.16) and (2. It can similarly be established that Pj= k \dqk + 77" dpk (2. Exercise 2. (2.ycos 8 px =Px c o s 6 —py sin 6 py = px sin 6 + py cos 0 is a canonical transformation.14) we find. Exercise 2. Notice that these constraints make no reference to the specific functional form of J f : the equations defining canonical variables are purely kinematical and true for any J f { q .7.7.7.Jhe unit vector in the radial direction.17) are to reduce to the canonical equations (2.4. p ) .16) ft * V dqk dpk . 95 REVIEW OF CLASSICAL MECHANICS (2. (2.14).7. Exercise 2.18) I* These tl lese then are the conditions to be satisfied by the new variables if they are to be canonical.7.7. S h o w that the polar variables p = (x2+y2)h'2.3. upon regrouping terms. Fill in the missing steps leading to Eq. (e p is.7.7.10) for any J f (q.18) starting from Eq.7.Feeding all this into Eq.17) / If Eqs. we must have (2. (2. 0 = tan 1 (>'/*).7. p).5.) .7. Pp = e p • xpx+ypy (x2+y2y P<t> = XPy~yPx(L) are canonical. (2. Verify that the change to a rotated frame X=ATCOS 0— / s i n 6 ^ = xsin 0+.
?„) (1) Argue that if we invert the above equation to get q = q{q). v dqi . 0} q .7. (2. p C M . * Verify that the change from the variables r . j Sqj (2) Show from the above that =dqi .7. Verify that q — \n(q 1 sin p) p = q cot p is a canonical transformation. {©.5. J i Should we not also define a PB. r 2 . .7. We would like to derive here Eq. Exercise 2. r. <7} = 1 \ dqi dpi — d<j da? dpi dqj — — = {©.dqj)_ dqj (3) N o w calculate Pi = dm. q) 9 Use the chain rule and the fact that q — q(q) and not q{q.6. If (q. Exercise 2. by calculating the PB in Eq.8.7.q) d&(q. (4) Verify..7. . . we must give them both the same status. p) and (q. (See Exercise 2. p . p) are both canonical.7. which gives the transformation of the momenta under a coordinate transformation in configuration space: . (2.. e r } ^ for every canonical pair (q.7): .7. p 2 to r C M .19) (It is understood that m and a are written as functions of q and p on the righthand side.. (2. ^ (dm t {*>. p)1 Fortunately it turns out that the PB are invariant under canonical transformations: (2.9). that the point transformation is canonical.9). and p is a canonical transformation. we have defined the PB of two variables co and er in terms of (q>p) as . for Hamilton's equations have the same appearance when expressed in terms of either set. pi.) . Now.7. we can derive the following counterpart of Eq. (2.Exercise 2.7. q) to derive Eq.18).4).
p) obey canonical equations. p ) and (q.9 above) there is an alternate way to establish Eq.z) = x2+y2+z2 Ko(r. We know that since (q.7.7. Verify Eq. as a consistent mathematical scheme. Consider now a restricted class of transformations. we have viewed the transformation p=p(q. p). Use the chain rule to go from q. It is the physical requirement that the time evolution generated by coincide with what is actually observed. So (2. under a change from Cartesian to spherical coordinates. Collect terms that represent PB of the latter.20) (which is just a relation among partial derivatives.9. #ey<u> q. (2. 97 REVIEW OF CLASSICAL MECHANICS Besides the proof by direct computation (as per Exercise 2.p) as passive: both ( q .p) refer to the same point in phase space described in two different coordinate systems. Now 0) is some physical quantity such as the kinetic energy or the component of angular momentum in some fixed direction. call it a . the numerical values of all dynamical variables are unchanged (for we are talking about the same physical state). called regular transformations. J f } 4P But then (q. (f>) = r2.p derivatives to q. which preserve the range of the variables: (q. we now pull the following trick.20) Having proved the result for what seems to be the special case cr = J f . places no restriction on J f . we get the desired result. A change from one Cartesian coordinate to a translated or rotated one is . p) (<y.p) or co(q. p) and (q. so co = {a).p derivatives. Consider first cr = J f . As mentioned earlier.Exercise 2.19) by direct computation.) If we understand that Jf is not T+ V in this argument but an arbitrary function. we use the same symbol for a given variable even if its functional dependence on the coordinates changes when we change coordinates.19). p) have the same range. 0) {(0.7. Thus Jf could have been any function at all in the preceding argument and in the result Eq. Note that nowhere in the derivation did we have to assume that Jf was any particular function of q and p. In fact. a is <b. (2. i.7. For instance. (2. Under the transformation (q. 9.7. Hamiltonian dynamics. (o(x. that restricts Jf to be T+ V.p).. whether co = co(q. so its rate of change is independent of the phase space coordinates used.y. but their functional form is changed.7. Active Transformations So far.e.p) also obey canonical equations.
The circle is thus characterized by its invariance under infinitesimal rotations. Under this change.7.. p) ~*(q. p) if a>(q. the corresponding set for a circle is a continuum. which does nothing.e. Symmetries and Their Consequences Let us begin our discussion by examining what the word "symmetry" means in daily usage. p) permits an alternate interpretation: instead of viewing (q.p) as the same phase space point in a new coordinate system. but not as much: the rotation must be performed around its axis. for any finite rotation may be viewed as a sequence of infinitesimal rotations (each of which leaves it invariant).21) (This equation has content only if we are talking about the active transformations. in our example this is a rotation by an infinitesimal angle s. We may characterize the continuous symmetry of the circle in another way. A cylinder has symmetry too.18).y) = x2 + y2 . Or.98 CHAPTER 2 regular (each variable goes from — oo to +oo before and after). This leaves both the circle and the hexagon invariant. Given this property. the numerical value of any dynamical variable a>(q.p) evaluated at the new point (q = q.p) as active or passive.) Whether we view the transformation (q. its invariance under finite rotations follows. only regular canonical transformations are physically interesting. the value of the function is invariant. We say that co is invariant under the regular transformation (q. a sphere looks the same before and after it is subjected to a rotation around any axis passing through its center. The infinitesimal rotation leaves the circle invariant but not the hexagon. namely. As we shall see.p) >(q. A regular transformation (q. 2. While the rotation angles that leave a hexagon unchanged form a discrete set.p) = co(q. We say that a sphere is a very symmetric object because it looks the same when seen from many directions. It is also possible to think of functions of some variables as being symmetric in the sense that if one changes the values of the variables in a certain way.p) (2.p=p). as illustrated by the example of a hexagon and a circle. which in our examples are rotations. A symmetry can be discrete or continuous. for it is true for any a under a passive transformation. equivalently.8. multiples of 60°.) is not. etc.p) will generally change: °> (<7>/0 (<?>/>)> though its functional dependence will n o t : co(q. whereas a change to spherical coordinates (where some coordinates are nonnegative. the symmetry of an object implies its invariance under some transformations. some are bounded by lit. Consider next an infinitesimal transformation. we may view it as a new point in the same coordinate system. which is infinitesimally "close" to the identity. p) (a. p) is the same function co(q. p ) obey Eq. Generally then. Consider the identity transformation. This corresponds to an active transformation which changes the state of the system. i. it is called canonical if ( q . rotates by 0° in our example.7. (2. Consider for example f(x.
p) is any dynamical variable.p) (q. then g is conserved. In the terminology introduced earlier.£ — =Pi + Spi (2. canonical.7. One calls g the generator of the transformation. but not necessarily infinitesimal. etc. the response of any variable co to the transformation is Sco = e{a). If J f is invariant under the regular.3) Pi ^ pi=pi dqi where g(q.. (2.g} (2. i. so is the transformed (translated.8.p(t)) is a solution to the equations of motion.2) y ~*y = x sin e + j cos e = (to orders) Consider now the function J^(q. (q(t). We say that / is symmetric under the above transformation.8. and if (q(t).8. a constant of motion.2).8.p). Exercise 2. I.1) y~*y = xsinO+y in the arguments. 7 dpt .e. There are two important dynamical consequences that follow from its invariance under regular canonical transformations. If M' is invariant under the following infinitesimal transformation (which you may verify is canonical. the transformation in question is continuous: its infinitesimal version is x *x = x cos e—y sin £ = x—y£ (2. transformation (q. Consequence I. {g. \ = 0 ~*g is conserved (2. Working to first order in e.5) (More generally.6) .If we make the following change x x = x cos 9y sin 8 cos 9 99 REVIEW OF CLASSICAL MECHANICS (2. II.p). we find that / is invariant. Let us first verify that g is indeed conserved if MJ is invariant under the transformation it generates. if we equate the change in under the change of its arguments to zero.4) But according to Eq.) trajectory.8.p(t)).8. we get (2. Let us now analyze these two consequences. rotated. qt < / = q< + £ — = qt + S q.2).8.
. Exercise 2. the total momentum. as usual. The conservation of lz then follows. Exercise 2. is the generator of infintesimal translations for a twoparticle system.1. = p 2 /2m.8. (2.100 CHAPTER 2 Note that Sp and Sq in Eq. But if the potential doesn't vary from point to point. (Hint: Work.8. a particle in one dimension and the case g~p. Thus the angular momentum around the z axis is the generator of rotations around that axis.3) may also be written as PBs.) Exercise 2. to first order in £. Since p is unchanged in a translation.3).8. The physics behind this result is clear.2)]. Under the rotation of the coordinates and the momenta.7)' which we recognize to be an infinitesimal translation. (2.8. 8x= e — = e dp (2.8.8. Consequently V(x + e) = ^(x). S h o w t h a t p = p t + p 2 .8) which we recognize to be an infinitesimal rotation around the z axis. and is conserved if tff is invariant under rotations of the state around that axis. (2. Thus the linear momentum p is the generator of spatial translations and is conserved in a translationally invariant problem. * Verify that the infinitesimal transformation generated by any dynamical variable g is a canonical transformation.8. there is no force and p is conserved. (2. Here.3. From Eq. [Eq. The relation between the symmetry and the conservation law may be understood in the following familiar terms.2. so is T=p2/2m.) Consider as an example. Consider \ 2m 2 K y ' . Next consider an example from two dimensions with g = h = xpyypx. This in turn means that there is no force in the tangential direction and so no torque around the z axis. p doesn't change and so neither does 7 .8. Consequently. Fis a constant as we go along any circle centered at the origin.
Verify that this is a noncanonical transformation.g} 8pi=£{pi. p) at any future time.5. Consider now a general case of g ^ J f .whose invariance Under the rotation of the coordinates and momenta leads to the conservation of / z . i.12) . 8qi=eqi 8pi=epi (2.8. The following examples should convince you that this is possible. Let us consider for instance the case g — L which has units erg sec and the corresponding fake time e = 80. Now.8.g} (2. * Consider J f = ip2+ {x2. p) . 4} = ey = 8y = (80)x (80)y (2.. p) and obtain the transformation given by Eq.9) Jf) etc.) Given an infinitesimal canonical transformation.10) Thus the new point in phase space (q. The transformation of the coordinates is 8x~ s{x.2. i. p) is just the point to which (q. and fake "time" s. p) would move in an infinitesimal time interval e.9). We have 8qt=e{qit 8pi=e{pi. Exercise 2. But J f is also invariant under the rotation of just the coordinates. a way to generate infinitesimal canonical transformations. In other words.6).4.8. 101 REVIEW OF CLASSICAL MECHANICS The preceding analysis yields. J?) (2. Consider the transformation generated by g = JC. we can find (q. an angle. (2. (2. the motion of points in phase space under the time evolution generated by jtf is an active canonical transformation.e. Convince yourself that in this case it is not possible to write 8jif as e { j f . Clearly there should be no problem integrating these equations for the evolution of the phase space points under the "fake" Hamiltonian g. We still have s 8qi=s{qi..8.= {q. get the finite canonical transformation..(q + 8q. you know that by integrating the equations of motion.8. ( Y o u could have guessed the answer based on Exercise 2. these equations are identical to Eq.e. we can get a finite one by "integrating" it. with g playing the role of the Hamiltonian. We take any function g(q. until now we had no means of generating one. Find the generator of this transformation (after verifying that it is canonical). that n o conservation law follows.).8.11) Mathematically. So But we know from the equations of motion that q. g j for any g. (Recall that although we defined a canonical transformation earlier. p + 8p) obtained by this canonical transformation of (q. as a byproduct. which is invariant under infinitesimal rotations in phase space (the xp plane).8.
Sudharshan and N. jy). So x = A cos 9 + B sin 9 y = C sin 9 + D cos 9 We find the constants from the "initial" (0 = 0) coordinates and "velocities": A = x 0 . denotes the transformed one.p2(t)) for some time and ends u p in the state ( x [ .e. Classical Dynamics: A Modem Perspective. see H. we get —x and likewise.102 CHAPTER 2 The fake equations of motion are dx dy d0 Differentiating first with respect to 9. we may write the finite canonical transformation (a finite rotation) as x = x c o s 9—y sin 9 (2. E. Let an observer SA prepare. Mukunda.t Consequence II. p ) >• (~q. there are many that aren't. rather than (x 0 . C. >'o). F o r instance. at / = 0. x2(t). y). . labels the initial point and (x.p\) which evolves as (xi(f). We do not consider ways of generating these. a state (x?. Although a wide class of canonical transformations is now open to us. x [ . Let t For an excellent and lucid treatment of this question and many other topics in advanced classical mechanics. Consider a twoparticle system whose Hamiltonian is invariant under the translation of the entire system. —p) is a discrete canonical transformation that has no infinitesimal version. Reverting to the standard notation in which (x. rather than (x.8. Reading. Classical Mechanics.. Massachusetts (1950) . nor an active interpretation. and using the second.B— (dx/d9)Q= y0.p\. Goldstein. such as the change from Cartesian to spherical coordinates. G. C=(dy/d9)o = x 0 . which have neither infintesimal forms. p f .14) y=x sin 9+y cos 9 Similar equations may be derived for px and py in terms of px and py. There are also the transformations that are not regular. Wiley. >'). AddisonWesley. D=yo. i. both particles. New York (1974). x\. Let us understand the content of this result through an example before turning to the proof. ( q . p\{t). p [ ) at time T.
p) > (q. i. p).. for example.p)} like any (2. J4f(q. which is obtained by transforming each point (q. the evolution of the second system will appear to be identical to what SA saw in the first. Let us associate with it an image trajectory. = we can go through the very same steps that led to Eq. translational invariance of Jf implies the invariance of V(x\. pi). Assuming for the sake of this argument that SB had in fact prepared the second system. Now qj(q. and not about where the system as a whole is in space. If you do not believe this. because of the symmetry of i. you may verify it . Proof. p) by means of a regular canonical transformation.16) If (q. x2) — V(x\~x2). p)}u = dJV(q.p). a given experiment and its rotated version will give the same result (according to the observers who conducted them). p) were a passive canonical transformation. Consequence II is just a generalization of this result to other canonical transformations that leave Jtf invariant. is (x° + a. Imagine a trajectory (q(t). p)}g>p = %. (2. In this case. Thus each particle cares only about where the other is relative to it. pi. x2 + a. we may say that a given experiment and its translated version will give the same result (as seen by the observers who conducted them) if is translationally invariant.x2).e. displaced relative to SA by an amount a. Consequently the outcome of the experiment is not affected by an overall translation. We are told that as a result of the translational invariance of any other trajectory that is related to this by an arbitrary translation a is also a solution to the equations of potion. To an observer SB. (2. (2.16) from Eq. if J f is rotationally invariant. p) to the image point (q.8. we could write.7. We ask if the image point moves according to Hamilton's equation of motion. This in turn means that V(X]. The physical idea is the following. p) dynamical variable a)(q.p(t)) in phase space that satisfies the equations of motion. since the PB are invariant under such a transformation. Let us now turn to the proof of the general result. if 103 REVIEW OF CLASSICAL MECHANICS 1Jz dpj Pr dXfap) dqj (q.15) if 3/f is invariant under the transformation (q. the initial state.us call the final state the outcome of the experiment conducted by SA.14) and prove the result. However. For the usual reasons.7.8. p).P) dpj But it is an active transformation. For instance. obeys qj={qj. {q{t).. 3f{q. qj ={qj.p{t)).e.The final state and all intermediate states are likewise displaced by the same amount.
8.. /() Otf Jrw) where Sci(xf.D. Why is it that a «o«canonical transformation that leaves Jf invariant does not map a solution into another? Or.e. A Useful Relation Between S and E We now prove a result that will be invoked in Chapter 16: 3Sd(x/. yd) and (pxb. tf'yXj.5. why is it that an experiment and its transformed version do not give the same result when the transformation that leaves tf invariant is not canonical? It is best to consider an example.3.A similar argument shows that Pj zz— dqj I281' So the image point moves according to Hamilton's equations.104 CHAPTER 2 by explicit computation using ^{q..py = 0). Exercise 2. Try to understand what goes wrong in the general case. in view of the discussions on consequence II.py = 0) and you release One in the transformed state in which (x = 0. you will see that the symmetries of the Hamiltonian have similar consequences for the dynamics of the system. Convince yourself that at later times the states of the two particles are not related by the same transformation. iyand JT is the Hamiltonian at the upper end point. As you go on and learn quantum mechanics. to xf. tt) is the action of the classical path from x. tf. t j ) _ dtf jp where E is the conserved energy. _ = 0) y with (px~b. constant on the whole trajectory. t f . p) = p). Since we shall be working with problems where energy is conserved we may write C2 8 18v dSc\{xf. x«.E. xt. This is a noncanonical transformation that leaves tf invariant. At first sight you may think that since — sedt . i. Suppose I release a particle at (x==a. Q.8. Consider the potential given in Exercise 2. you rotate the coordinates but not the momenta.
dx At. .5 wherein we have set x.Af. . Consider the harmonic oscillator. Show that dSd/dxf=p(tf).7. so that the total change in action comes from the difference in paths between t = 0 and t = tf as well as the entire action due to the extra travel between tf and tf+ Atf.5. From the figure it is clear that now the particle takes a different classical trajectory x ( 0 = JCd (/) + /?(/) with r}(Q) = 0. The explanation requires Fig. t + At the right side must equal i f and not —E.8.8.. 2.= f . for which the general solution is x(t) = A cos cot + B sin cot. The derivative we are computing is governed by the change in action of the classical path due to a change in travel by At holding the end points x.105 REVIEW OF CLASSICAL MECHANICS Figure 2. Only the latter is given At. and xf fixed. The upper trajectory takes time t while the lower takes ft. dt At It is clear from the figure that J](tf) = —x(t/) At so that ¥ 8S= dse dx At=3tf{tf) <r At from which the result follows. Exercise 2. Exercise 2.6. = 0 for convenience. The correct answer is then <r SSCi = rtr dx dx dt + &(tf) tf At d_ d<£ — rj(t) dt + £e(tf) ox = 0 + — r r 7 ( 0 + &(tf) dx „ dt dx .
x2> T ) = — [ ( * i + x l ) c o s 2 sin 6)T (oT2X. Write down the energy in terms of xY. . Verify that dSc]/8T= E.106 Express the energy in terms of A and B and note that it does not depend on time.x2. and 1 Show that the action for the trajectory connecting xY and is Sci(xi.X2]. Now choos A and B such that x(0) = Xi and x(T) = x2.
t) which characterizes the disturbance at the point r at time t. It is described by a wave function y/(r.) at time /. A wave. in contrast. if they do not. The experiments to be described were never performed exactly as described here. but they contain the essential features of the actual experiments that were performed (in the first quarter of this century) with none of their inessential complications. The analogs of q and q for a wave are y/ and y/ at each point r. while in the case of electromagnetic waves. assuming y obeys a secondorder wave equation in time. 3.3 All Is Not Well with Classical Mechanics It was mentioned in the Prelude that as we keep expanding our domain of observations we must constantly check to see if the existing laws of physics continue to explain the new phenomena. In the case of sound waves. y/ is the excess air pressure above the normal. is a disturbance spread over space. We have studied the particles in some detail in the last chapter and may summarize their essential features as follows. Particles are localized bundles of energy and momentum. we must try to find new laws that do.1. y/ can be any component of the electric field vector E. They are described at any instant by the state parameters q and q (or q and p). the trajectory q(t) may be deduced for all future times from the equations of motion. and that. such as 107 . These parameters evolve in time according to some equations of motion.. Given the initial values q{ti) and <?(/. Particles and Waves in Classical Physics There exist in classical physics two distinct entities: particles and waves. In this chapter you will get acquainted with experiments that betray the inadequacy of the classical scheme.
3. the plane wave may be written as y/(x. It should be clear from th context what it stands for. 0 remains zero. which is peculiar to them and. c. while co measures the phase change per unit time at any fixed point x. is not exhibited by particles described by classical mechanics.1) At some given time t. the wave is periodic in space with a period A . t) = A exp = A exp[/0] (3. called plane waves. time averages of the energy an< momentum flow are still proportional to the intensity (as defined above) in the cas of plane waves. This phenomenon il illustrated by the following experiment (Fig. If y/ describes an electromagnetic wave.1. only the real part of y describes it. [Since the electromagnetic field i real. the pattern /1 + 2 is observed. called its wavelength. Given y/(r. (b) With both slits open. One calls k the wave vectorX 3.1.1. For any wave. t) for all future times by solving the wave equation. respectively. repeating itself every T seconds. (a) When a wave yr = e !<ky ~an is incident on the screen with either slit or S2 open. An Experiment with Waves and Particles (Classical) Waves exhibit a phenomenon called interference..] Plane waves in three dimension are written as y/(T.(k.1. 0) and yr(r. . instead of A and T. it is periodic in time. In terms of the phase 0 in Eq.1a).108 CHAPTER 2 r" Figure 3. Let a wave y/ = A el<ky~wt) b < t Unfortunately we also use k to denote the unit vector along the z axis.1). the intensity patterns and 12. 0) one can get the wave function \f/(r.2. (3. F o r a plane wave this is a constant equal to \A\2. However. This wave travels at a speed v = (o/k. Note that I\ + 2 # I\ + h • This is called interference. and likewise at a given point x. which describes waves propagating at the speed of light. note that if we start out at a point where 0 = 0 and move along x at a rate x = (co/k)t. the intensity is defined to be / =  y/\2. The overall scale A up front is called the amplitude. To check this claim.21 where each component ki gives the phase changes per unit length along the z'th axis. T being called the time period.r(0'\ G) = \k\v (3. are measured by the row of detectors on AB.t) = Ae. the related quantities k = 2n/X called the wave number and (o = 2k/T called the (angular) frequency. We will often use. k measures the phase change per unit length at any fixed time t. In one dimension. Of special interest to us are waves that are periodic in space and time. the intensity is a measure of the energy and momentum carried by the wave.
3. What if both Si and 5*2 are opened? Classical mechanics has an unambiguous prediction: Ii+2 = I\ + I2. In other words. which has the same frequency and wavelength as the incident wave. One may think of Si as the virtual source of this wave y/\. the maxima correspond to the case d2~d\= nX (n is an integer). To a particle headed . <j>2(x)~ <j>\(x) = 2nn at a maximum and <f>2(x)<pi(x) = (2n+ 1 )k at a minimum.1b. there is more energy flowing into x m i n with just one slit open than with both. The reasoning is as follows: each particle travels along a definite trajectory that passes via Si or S2 to the destination x. = i//2 is registered by the detectors.2. Although this fact may be obscured if the beam is dense. Similarly if S2 is open instead of Si. The ups and downs are due to the fact that the waves i//t and \j/2 have to travel different distances dy and d2 to arrive at some given x (see Fig. it can be easily detected as the incident flux is reduced. (b) The pattern with both slits open according to classical mechanics (/\+2 — • s \ h+hl incident normally on a screen with slits 5] and S2. when the waves are exactly out of step. If we first keep only open. the wave iff2 produces the pattern / 2 =  y 2 \ 2 . At a distance d parallel to it is a row of detectors that measures the intensity as a function of the position x measured along AB. These look very much like the corresponding patterns for the  wave. The feature to take special note of is that if x min is an interference minimum. (a) Intensity pattern when Si or Si is open. In particular. the incident wave will come out of and propagate radially outward. due to a beam of incident particles.In both cases the arrival of energy at the detectors is a smooth function of x and t. which are a distance a apart. the opening of an extra slit can actually reduce the energy flow into  Consider next the experiment with particles (Fig.2a). The only difference will be that the particles arrive not continuously. Let us define the intensity I(x) to be the number of particles arriving per second at any given x. but rather the interference pattern shown in Fig. The source of the incident plane waves is replaced by a source of particles that shoots them toward the screen with varying directions but fixed energy. but in a staccato fashion. and the minima correspond to the case d2di =(2n+ l)A/2. both waves y/i and yr2 are present and produce an intensity pattern I [ + 2 ~ \ y i + y 2 \ 2 The interesting thing is that I \ + 2 ^ I \ + I 2 . when the waves arrive exactly in step.2a). The patterns with 5*1 or S2 open are shown in (Fig. In terms of the phases 0] and <f>2. One can easily show that the spacing Ax between two adjacent maxima is Ax = Xd/a. 3. 3.1a) and thus are not always in step.109 ALL IS NOT WELL WITH CLASSICAL MECHANICS Figure 3. 3. The intensity pattern /. Let the line AB be filled with an array of particle detectors. each particle triggering a counter at some single point x at the time of arrival. Now if both Sy and S2 are opened.
etc. the envelope of this histogram. of course. called photons. Say we repeat the experiment with a change that is expected (in classical physics) to produce no qualitative effects. 3. Being localized in space it has even y knowing if S2 is open or closed. however. after many counts. for a plane wave.) We find that with the slits open one at a time we get patterns Ix and / 2 . the envelope of the histogram becomes. With enough data. but in sudden bursts.1a). 3. for 3. Each bundle carries the same momentum p. will define the smooth functions I u I 2 . it will go to some point x. if. a burst there. and thus cannot respond to it in any way. Now the conclusion ^1+2 = I\ + h is inevitable. If. A given particle will of course not produce a pattern like /. This discrete nature is obscured in intense beams. 3. As each burst occurs at some x. The following objection may be raised: although particles heading for Sx are not aware that S2 is open.2b). between counts. If we further assume E is polarized perpendicular to the page. We have made an important discovery: light energy is not continuous—it comes in bundles. However. and I\+ 2 . (Strictly speaking. the pattern I x . a burst here. it is immaterial whether S2 is open or closed. Thus the number coming via Si to x is independent of whether S2 is open or not and vice versa. that the energy flow seems continuous in space and time. This is what classical physics predicts particles and waves will do in the doubleslit experiment. B is uniquely fixed by E. 3. This objection can be silenced by sending in one particle at a time. We set up the double slit as in Fig. the electromagnetic wave must be characterized by giving the orientation of the E and B vectors in addition to co and k. or I 2 by itself.1a.3. they certainly can be deflected by those coming out of S2. It follows that I X + 2 = I\ + I 2 (Fig. for the bundles come in so fast and all over the line AB.110 CHAPTER 3 n o wa . The DoubleSlit Experiment with Light Consider now what happens when we perform the following experiment to check the classical physics notion that light is an electromagnetic wave phenomenon. we record it and plot a histogram. We find that the energy is not arriving continuously. for instance. 1 1 . We pursue our study of these bundles.) The energy arrives at the detectors smoothly and continuously as befitting a wave. We can therefore suppress the explicit reference to this constant vector and represent the field as a scalar function y/.1a and 3. with a row of lightsensitive meters along AB and send a beam y = A e. (The interference pattern is of course what convinced classical physicists that light was a wave phenomenon.iky~"n) i n a direction perpendicular to the screen. the former are heading for x{ and the latter for x 2 (see Fig.1b. 2. and with both slits open we get the interference pattern Ix + 2 as in Figs. in some detail and find the following properties: d 1. this polarization is unaffected by the double slit. we make a histogram. We start with S\ open and cut down the intensity. We now cut down the intensity further so that only one detector gets activated at a given time and there is enough of a gap. say a millisecond. A very strange thing happens. Each bundle carries the same energy E.
consider a point *min which is an interference minimum. From the famous equation E2=p2c2 + m2c4.2) 111 ALL IS NOT WELL WITH CLASSICAL MECHANICS where H=h/2n is a constant. From these facts Born drew the following conclusion: with each photon is associated a wave y/. The correct explanation of this experiment. called the probability amplitude or simply amplitude. a photon going in via Si should be insensitive to whether S2 is open or not (and vice versa). If photons followed definite trajectories. and watch the histogram take shape. That light is made of particles will. still keeping the intensity so low that only one photon is in the experimental region at a given time. 4. but is instead an interference pattern characteristic of wave number k. More photons arrive here with either Si or S2 open than with both open. If we vary the frequency of the light source we discover that E= fioo p = hk (3. wait. whose modulus squared  y ( x )  2 gives the probability of finding the particle at x. This wave interferes with itself and forms the oscillating pattern  \f/(x)  2 along AB. in terms of photons. we won't find that /. the actual experiment that revealed the granular nature of light is called the photoelectric effect. we deduce that these bundles are particles of zero mass. surprise classical physicists but will not imply the end of classical physics. They will cheerfully plunge into the study of the dynamics of the photons. which is a plane wave with (o = E/fi and k=p/fi. for physicists are used to the idea that phenomena that seem continuous at first sight may in reality be discrete. What really undermines classical physics is the fact that if we now open both slits. This result completely rules out the possibility that photons move in welldefined trajectories like the particles of classical mechanics—for if this were true.3.3. Since we are doing the experiment with one photon at a time. The constant h is called Planck's constant. E=pc.3) For those interested in history. [Strictly speaking. These subtleties can. one cannot even raise the improbable hypothesis that photons coming out of S\ collide with those coming out of S2 to modify (miraculously) the smooth pattern Ix +1 2 into the wiggly interference pattern. and the result I\+ 2 = I\ +1 2 is inescapable! To say this another way. Its value is 27 — = ^ ~ 10 lit erg sec (3.3. we must not refer to  y/(x)\2 as the probability for a given x. which gives . Every incoming photon of energy E and momentum p has a wave function y/ associated with it. trying to find the equations of motion for its trajectory and so on.1) (3.] The entire experiment may be understood in terms of this hypothesis as follows. of course. and has the dimensions of erg sec. was given by Einstein in 1905. + 2 equals I\ +1 2 as would be expected of particles. but rather as the probability density at x since x is a continuous variable. it is incomprehensible how opening an extra pathway can reduce the number coming to x m i n . which is the same as that of action and angular momentum.3. however.
A large ensemble of such photons will reproduce the phenomena expected of a classical electromagnetic wave <// and the probabilistic aspect will be hidden. But what about common sense. 26 . arrive at the same time and all along the line AB. the following example should be instructive. many photons. the number at any x will become proportional to the probability function  y/(x)2. incidentally. The precise meaning of this explanation will become clear only after we fully master quantum mechanics. which will obey I\ + 2 = I\ + h as predicted classically.6 10 cm which is 10" 13 times smaller than the radius of the proton! For any reasonable value of the parameters a and d (see Fig.112 CHAPTER 3 at s o m e the probability that the given photon will arive at x. A given photon of course arrives definite x and does not reveal the probability distribution. which one thought was a pure wave phenomenon.5. More specifically. If. should exhibit wavelike behavior. is that a wave is associated not with a beam of photons. prompted de Broglie to conjecture that entities like the electron. 3. The wavelength associated with these particles is „ 2n h X= — = k p . to get a glimpse of wha . all described by the same wave and hence the same probability distribution. should consist o photons. 3. It is now widely accepted that all particles are described by probability amplitudes and that the assumption that they move in definite trajectories is ruled out by experiment.1b). all described by the same y/. Likewise. he conjectured. besides the probability interpretation. Conclusions The main objective of this chapter was to expose the inadequacy of classica physics in explaining certain phenomena and. generally believed to be particles. Nonetheless.4. have arrived. which says that billiard balls and baseballs travel along definite trajectories? How did classical mechanics survive for three centuries? The answer is that the wave nature of matter is not apparent for macroscopic phenomena since it is so small. shortly thereafter. The intensity distribution then assumes the shape of the probability distribution right away and the energy flow seems continuous and in agreement with the predictions of classical electromagnetic theory. that particles of momentum p will produce an interference pattern corresponding to a wave number k =p/ft in the doubleslit experiment. but with each photon. Matter Waves (de Broglie Waves) 1 That light. moving at 1 cm/sec. the interference pattern would be so dens in x that our instruments will only measure the smooth average. 3. The main point to note. every photon is given by the same y/ and the same probability distribution. in analogy with photons. This prediction was verified for electrons by Davisson and Germer. we wait till several photons. If the beam is monochromatic. however. . Suppose we do the doubleslit experiment with pellets of mass 1 g. if an intense (macroscopic) monochromatic beam is incident.
in localized f o r m . such that t)\2 gives the probability of finding it at a point x at time t. summarize the results of the doubleslit experiment and many others not mentioned here. t). This is called waveparticle duality. charge. It appears that each particle has associated with it a wave function y/(x.the new (quantum) physics ought to look like. if we think of functions as vectors in a n infinitedimensional space. etc. so that when the postulates are encountered in the next chapter. t) or. and at the same time they are not particlelike in that assuming they move along definite trajectories leads to conflict with experiment. The dynamics of the particle is then the dynamics of this f u n c t i o n y/(x. W e f o u n d t h a t entities such as the electron are particles in the classical sense in that when detected they seem to carry all their energy. they will appear highly plausible. which specify what sort of information is contained in and how  if/(t)} evolves with time. m o m e n t u m . . the doubleslit experiment contains most of the central features of the theory. of the ket  In the next chapter the postulates of q u a n t u m theory will define the dynamics in terms of  y ( 0 > . Fortunately. T h e doubleslit experiment was described here to expose the inadequacy of classical physics and not to summarize the entire body of experimental results f r o m which all the postulates could be inferred.The postulates.
We consider first a system with one degree of freedom.1. . 4.e. a single particle in one space dimension. Despite your preparation you may still find the postulates somewhat abstract and mystifying on this first encounter. The straightforward generalization to more particles and higher dimensions will be discussed towards the end of the chapter.he Postulates—a General Discussion "Having acquired the necessary mathematical training and physical motivation. II. The Postulates! The following are the postulates of nonrelativistic q u a n t u m mechanics. the quantum postulates are accompanied by their classical counterparts (in the Hamiltonian formalism) to provide some perspective. Q u a n t u m Mechanics The state of the particle is represented by a vector  if/it)} in a Hilbert space. namely. Classical Mechanics The state of a particle at any given time is specified by the two variables x(t) a n d p ( t ) . as a point in a twodimensional phase space. I.p). II. In what follows. In this chapter the postulates will be stated and discussed in broad terms to bring out the essential features of q u a n t u m theory. however. The subsequent chapters will simply be applications of these postulates to the solution of a variety of physically interesting I problems. you are now ready to get acquainted with the postulates of q u a n t u m mechanics. Every dynamical variable co is a function of x and p : co = (o(x. i. { I. The independent variables x and p of classical mechanics are represented i t Recall the discussion in the Preface regarding the sense in which the word is used here. disappear after you have worked with the subject for some time.. These feelings will.
§ By this we mean that Q is the same function of X and P as c» is of x and p.p^P)§ III. IV. leaving the fourth for the next. The state vector  y ( 0 > obeys the Schrddinger equation IV. You may wish to go over that section now to refresh your memory. Likewise P=fiK. 4. i discussion of ideal quantum measurements follows. Discussion of Postulates lIII The postulates (of classical and quantum mechanics) fall naturally into two sets: the first three.116 CHAPTER 4 by Hermitian operators X and P with the following matrix elements in the eigenbasis of Xt <x\X\ x') = (x\P\x'y = x5(xx') iK8'{xx') The operators corresponding to dependent variables (o(x.  That is. M .10. in an ideal experiment consistent with the theory. where K was also discussed therein. p^P) is the quantum Hamiltonian operator and J f is the Hamiltonian for the corresponding classical problem. contains proper vectors normalizable to unity as well as J Note that the X operator is the same one discussed at length in Section 1.P) III.The state of the system will change from  </) to as a result of the measurement. the measurement  of the variable co will yield a value co(x.p) are given Hermitian operators £l(X. which tell us how the system is depicted at a given time.p).2. you will recall. If the particle is in a state given by x and p. and the last. The first postulate states that a particle is described by a ket  y/> in a Hilbert space which. The state will remain unaffected. = o)(x^X. If the particle is in a state  meas11 urement of the variable (corresponding to) Q will yield one of the eigenvalues co with probability />(ft))oc<fi> v> 2 . It is assumed you are familiar with the idea classical measurement which can determine the state of the system without disturbing it in any way. The state variables change with time according to Hamilton's equations: dt J J f dx where H(X. which specifies how this picture changes with time. P) = Jf(x>X. We will confine our attention to the first three postulates in this section.
for example. on the other hand. the state a\y/} + ft y/'> will. When we say that  y/> is an element of a vector space we mean that if  y/} and  VO represent possible states of a particle so does a\y/> + ft] y / ) . where X and P are the operators defined in postulate II. The probability P(ca) that the result co will obtain is proportional to the modulus squared of the projection of \y/) along the eigenvector that is 2 P(co)oc<ft> y/}\ . in the sense that if the variable is measured the result co(x. In classical mechanics when a state ( x . a ket in such a space has in general an infinite number of components in a given basis. We have seen in Section 1. In the case of the string. The ket  y/) of quantum mechanics is none other than the vector representing the probability amplitude y/(x) introduced in the doubleslit experiment. Let us note. no need to speak loosely. in classical mechanics. One wonders why a particle. Step 3.^ Now. which answer exactly this question. In terms of the projection operator P w = £o)<ct>. x and p. This is called the principle of superposition.> and eigenvalues co{ of £2. I The status of the two classes will be clarified later in this chapter. What is the analogous statement one can make in quantum mechanics given that the particle is in a state  y > ? The answer is provided by Postulates II and III. p) will obtain. . that if f ( x ) and g(x) [with / ( 0 ) = / ( L ) = g ( 0 ) = g ( Z .p). the state af+ fig has very different attributes from the states / and g: it will look different. for the present. and so on. the following salient points. ) = 0] are two possible displacements of a string. now needs to be specified by an infinite number of variables. have attributes that sometimes resemble that of  y/) and at other times those of  (//').•! v> Step 4. The principle by itself is not so new: we know in classical physics. In quantum theory. Expand y/> in this basis: 117 THE POSTULATES —A GENERAL DISCUSSION v>=11^><>.improper vectors. There is a tremendous amount of information contained in these steps. have a different amount of stored elastic energy. There is. 2 ?(a))oc <<»  y/}\ = {y/  to) <g>  y/) = <y/\ P „  y/} = <y/\P^PJ y/) = <P ffl y/ \ P ffl y/>. p ^ P ) . loosely speaking. p ) is given. F o r the present let us note that the doubleslit experiment has already hinted to us that a particle such as the electron needs to be described by a wave function y/(x). What is new is the interpretation of the superposed state a\ y/) + y/'}. Find the orthonormal eigenvectors g). since we have postulates II and III to tell us exactly how the state vector  y/) is to be interpreted in quantum theory. one can say that any dynamical variable co has a value co(x.10 that a function f ( x ) may be viewed as a ket  / > in a Hilbert space. which had only two independent degrees of freedom. however. normalizable only to the Dirac delta functions. Step 2. in terms of the following steps: Step 1. so is the superposition af{x) + ftg(x). Construct the corresponding quantum operator Q = a>(x>X. Let us find out. What do these variables tell us about the particle? To understand this we must go on to the next two postulates.
then an arbitrary linear combination. Even with the condition <y/ y/} = 1. A particle in such a state may be said to have a value to. If  y/) and  y/') are normalized. such as a\(oxy+p\(o2y o«r+i/r) 2sl/2 \v>~ one gets the state. (3) If  v ) is an eigenstate <0/>. a\y/} + P\y/'). y/}\2 by th sum of all relative probabilities: P( < P  )^ l < < P f l y >  2 ^ l < f l > ' l y >  ! (4.2. these eigenvalues are all real. the relative probability. corresponding to each physical state. which upon measurement of Q. Thus. y/)\2 is onl. This freedom will be exploited at times to make the components of  y/} in some basis come out real. This is the peculiar! . respectively. can yield either co} or (o2 with] probabilities\a\ 2 /(\a\ 2 + \p\ 2 ) and j8j 2 /(l« 2 +1£ 2 ). T o get the absolute probability. for in the classical sense.) Note that the relative probability distributions corresponding to the states  (//) and a \y/}. Further. the measurement of Q is guaranteed to yield the result . it assigns (relative) probabilities only for obtaining some eigen value (o of Q. (In fact the set of all normalized vectors does not even form a vector space.11)1 It is clear that if we had started with a normalized state I V> < ¥ \ ¥ > / we would have had P(a>i) = \(<Ot\Y'y\2 (4. When we speak of the state of the particle.2 below.1 y/)}2.2) If  is a proper vector.2. we have the freedom to multiply the ket by a number of the form e19 without changing the physical state. but a ray or "direction" in Hiibert space. we divide <<w. is not. The status of such vectors will be explained in Example 4. (4) When two states ®I) and CB2> are superposed to form a (normalized) state. for in this case < y/ ] y/} — (5(0) is the only sensible normalization. such a rescaling is possible and will be assumed hereafter. we usually mean the ket  y/> with unit norm. when they are renormalized to unity. The probability interpretation breaks down if  v> happens to be one of the improper vectors in the space. Note that the condition (y/1 y/} = 1 is a matter of convenience and not a physical restriction on the proper vectors. the quantity <<a. there exists not one vector. (2) Since we are told that P((ot)oc<0).118 CHAPTER 4 (1) The theory makes only probabilistic predictions for the result of a measure ment of Q. Thus the only possible values of Q are its eigenvalues. Since postulate II demands that Q be Hermitian. reduce to the same absolute probability distribution.
(b) The Q basis. (C) The Q and the A bases.. one does not expect to get the value corresponding t o / some of the time and that corresponding to g the rest of the time. finding the eigenvectors \Xi) and the eigenvalues X.> figure 4. In Fig. In summary. (5) When one wants information about another variable A. y/) in this basis: v>=E j ®/><<o/l v > Example 4. ffl2>. Suppose for example we are working in the D basis in which lv>=II«/><>il i v ) and P(£U.>). Consider the following example from a fictitious Hilbert space V (R) (Fig. 4. A. For example. (a) The normalized ket in V3(R) representing the state of the particle. we must determine the eigenbasis of the corresponding operator and find the projection of  y/> along all its eigenkets. consequence of the superposition principle in quantum theory. we find the eigenvectors of the corresponding operator and project  ijr) on that basis.1. to the kets A>. and it contains the statistical prediction for all observables. Likewise.1). find its eigenvectors  Xi) (which are column vectors with components <<s)y Xi)). referred to earlier. so does our interest go from the kets ©>. Then P(X) = <Ay>2 The bases of fl and A will of course be different in general. to another. and !<M3>.) 'ix.1a we have the normalized state y/>.y= <©. if a dynamical variable of the string in the state af+ fig is measured. A ©. To extract this information for any observable.) = <fi). instead. (6) As our interest switches from one variable Q. one repeats the whole process.1. It has no analog in classical mechanics. however. and take the inner product <A. There is. the functions/and af (a real) describe two distinct configurations of the string and are not physically equivalent.I*. we have a single ket  y/) representing the state of the particle in Hilbert space.2. To get the statistical information on a variable. with no reference i . no need to change the basis each time. 4. If we want P{Xi) we take the operator A (which is some given matrix with elements A. one expects a unique value generally distinct from both. y> 2 . ®i>.
and a>3> (Fig. 1) and 2 ) in eigenspace V m with eigenvalue (a.2. 3. we don't know if Q.1c) there is just one common eigenvector  « 3 > = A3>.1b).3 (4. for example] co — xp. then the result coi would obtain with unit probability. say o>i). 4. and other values of a> are impossible. a>2. There is no universal recipe for resolving such ambiguities. We will not encounter such cases in this book. tt>2>.3% In general. one can replace in Postulate III P(a>)oc<VP„y> . Symmetrization is the answer as long as Q does not involve products of two or more powers of X with two or more powers of P. If we want P(Xf) we construct the basis \X\}. we find its eigenbasis and express the state vector  y/> in terms of the orthonormal eigenvectors o>i>.—XP or PX since xp = px classically. i Complication 1: The Recipe Q = G)(x—>X. 4. and <a3 are expected with probabilities k. Let us say coi = co2 = a>. Pco = {co. \X2}. Notice incidentally that symmetrization also renders Q Hermitian. If it does. In the present case.y> (4. l + fl).2><fi). Then which is the modulus squared of the projection of  y/> in the degenerate eigenspace This is the result we will get if we assume that t»i and (o2 are infintesimally distinc and ask for P(a)\ or a>2)• In terms of the projection operator for the eigenspace. only experiment can decide the correct prescription. We discuss below the major ones and how they are to be surmounted. p^P) Is Ambiguous. and 2. • Returning to our main discussion. If. To get predictions on Q.120 CHAPTER 4 to any basis. there are a few complications that could arii iJ as one tries to carry out the steps 14. Wha is P(co) in this case? We select some orthonormal basis \cd. Only a particle in a state  (//) — ( « . 1 ><fi). Let us suppose This means that the values co\. Complication 2: The Operator Q Is Degenerate. respectively. and A3>.2 we have P(w)~ (ylPalyy = <P <u y'l Pa. ) has a welldefined value of £2 in the classical sense. which can in general be distinct from the Q basis. If instead  y/> were some eigenvector. In our example (Fig. the rule is to use the symmetric sum: Q = (XP + PX)/2.2.
121 where is the projection operator for the eigenspace with eigenvalue a>. one expects <fi) i//) to be a smooth function y/(co). Since the number of possible values for co is infinite and the total probability is unity. The wave function in the X basis (or the x space). the points in which are labeled by the coordinate co. in order to avoid getting into details. This definition meets the requirement that the total probability be unity. Can we interpret <fl) y/) 2 as the probability for finding the particle with a value co for D? No. if t«i = G)2~ CO (Fig. called the co space. there is just one space. One interprets i > (fu) =  < «  y/)\2 to be the probability density at co. are auxiliary manifolds introduced for the purpose of visualizing the components of the infinitedimensional vector  y/> in the Q basis. T o visualize this function one introduces an auxiliary onedimensional space. Then postulate III as stated originally would become a special case in which there is no THE POSTULATES degeneracy and each eigenspace is simply an eigenvector. and so on. N o w the time has come to become precise! .1b) then P(CO) is the square jlof the component of  y/} in the "x y" plane. since P(co) dco = J  < < o  y/}\2 dco = J (y/\ a>)(co  y/) dco = <VMV> = <VI V> = 1 (4. the A space. each single value of co can be assigned only an infinitesimal probability. the Hilbert space. As far as the state vector  <//) is concerned. that is to say. In our discussions in the last chapter. rather than as the probability density. by which one means that P(co) dco is the probability of obtaining a result between co and co+dco. In this space y/(co) will be a smooth function of co and is called the wave function in the co space. The co space. 4. is usually referred to as just the wave function. the A basis. expands  y ) as In this case one lv> = to) <a) ( y/) dco One expects that as co varies continuously.2. the operator corresponding to the position x. the state cannot be normalized to unity and P(co) must be interpreted as the relative probability density. since the X basis is almost always what one uses. etc.  y ( x )  2 was referred to as the probability for finding the particle at a given x. We are merely doing the converse of what we did in Section 1. The wave function y/(co) is also called the probability amplitude for finding the particle with Q = co. y/(x). in which it resides. An important example of a continuous spectrum is that of X. Complication 3: The Eigenvalue Spectrum of Q Is Continuous. We will discuss such improper states later.10 wherein we started with a function f ( x ) and tried to interpret it as the components of an infinitedimensional ket  y/} in the x> basis. —A GENERAL 3 DISCUSSION In our example from V (/?). so will <co  y ) .4) If <y/ y/) = £(0) is the only sensible normalization possible.
a definite position. ust t w o num Complication 4: The Quantum Variable Q Has No Classical Counterpart. The answer is now clear. the same ket 1 <//> when expanded in terms of the eigenkets  p) of the momentum operator P gives the odds through the wave function in p space. they must often be supplemented by intuition and classical ideas. . Even "point" particles such as the electron are now known to carry "spin. we continue with our study of what else the postulates of< quantum theory tell us. Since such a degree of freedom is absent in classical mechanics. but one doesn't need a new vector for specifying this. As we will see in Chapter 14. For example. This is part of the information contained in y/(x) = <x y/>. This phenomenon is called the collapse or reduction of the state vector. on the other hand. Of course. that is to say. But one can think up nonideal measurements which do change the state. Let us first note that any definitive statement about the impact of the measurement process presupposes that the measurement process is of a definite kind. at any given time. the components of y/) in the X basis." which is an internal angular momentum. In quantum theory one again gives the odds for getting different values of momenta. What makes Postulate III profound is that the measurement process referred to there is an ideal quantum measurement.122 CHAPTER 4 Earlier on we were wondering why it was that a classical particle defined by j b e r s x and p now needs to be described by a ket which has an infinite number of components. which is in general some superposition of the form Iv> = LI<B><©IV> 1 into the eigenstate a)> corresponding to the eigenvalue co obtained in the measurement. A quantum particle. which in a sense is the best one can do. in the case of the classical particle. can take on any value of x upon measurement and one must give the relative probabilities for all possible outcomes. One simply has to give this value of x in specifying the state. namely. We now illustrate the notion of an ideal quantum measurement and the content of this postulate by an example. the classical mechanics maxim that any dynamical variable can be measured without changing the state of the particle. assumes that the measurement is an ideal measurement (consistent with the classical scheme). It is worth bearing in mind that n o matter how diligently the postulates are constructed. Collapse of the State Vector We now examine another aspect of postulate III. imagine trying to locate a chandelier in a dark room by waving a broom till one makes contact. A classical particle has. that the measurement of the variable Q changes the state vector. angular momentum unrelated to their motion through space. our postulates do not tell us which operator is to describe this variable in quantum theory. Having discussed the fourstep program for extracting statistical information from the state vector. the solution is provided by a combination of intuition and semiclassical reasoning. y f { p ) = (p  <//). one needs also to specify1 the momentum p as well.
the ideal position measurement. One way to measure the momentum of the particle . Thus one can always make the change in momentum p' . then so does co'. and that the state will be the same after the measurement (since  y> =  / j ) is already an eigenstate of the operator P in question). Since \p} is a sum of position eigenkets x>. Thus even the ideal position measurement will change the state which is not a position eigenstate.fico' 123 THE POSTULATES —A GENERAL DISCUSSION cp cp Solving for co' and p' in terms of co and p.Consider a particle in a momentum eigenstate /?>. when conducted on a particle in state x>. Hereafter. (How d o we know what the photon energies are? We assume we have atoms that are known to emit and absorb photons of any f given energy.p arbitrarily small. Let us assume the particle is forced to move along the xaxis and that we send •ina rightmoving photon of energy fico that bounces off the particle and returns as ^leftmoving photon of energy fico'. when we speak of a momentum measurement. This then is the big difference between classical and quantum mechanics: an ideal measurement of any variable co in classical mechanics leaves any state invariant.) Using momentum and energy conservation : cp' = cp + fi(co + co') E' = E+ fi(co — co') it is now possible from this data to reconstruct the initial and final momenta of the particle: _ (fico + fico') (V H coco' .__(fico + fico') m2c4 ficoco' ficofico' fico . particle. this is what we will mean. while momentum measurement does not? The answer is that an ideal position measurement uses photons of infinitely high momentum (as we will see) while an ideal momentum measurement uses photons of infinitesimally low momentum (as we have seen). We will also assume that to each dynamical variable there exists a corresponding ideal measurement. The postulate tells us that if the momentum in this state is measured we are assured a result p. which. Why does a position measurement alter the state  p ) . \p >= \xy(x\pydx !e measurement will force the system into some state x>. We will discuss. Suppose now that we measure the position of a particle in a momentum eigenstate Jp}. is by Compton scattering. for example. if co >• 0. one readily sees that for any choice of p. will give the result x with unit probability and leave the state vector unchanged. in which a photon of definite momentum bounces off the . .
we could not say what the state was after the measurement. Say we had co 1 = co2 = (o. W/—'• \ r^ <P„imp«V>/2 PUvO where P ^ is the projection operator for the eigenspace V©.  co. .1b). we could only say lW? (a2 + pzY/2 where a and /? are arbitrary real numbers. The effect of measurement may be represented schematically as follows: Pco\W> Si measured. and the measure ment yielded a degenerate eigenvalue co. Note that although we do not know what a and /? are from the measurement. In other words. though we did not know  y/>. and the measurement yielded' a degenerate value co. 2 » If. Let us use an orthonormal basis  co. and the state after measurement has been normalized. Special note should be taken of the following point: if the initial state  y/} were unknown. where.  . if the initial state j y/} were known. that 1 > + 2  < » . fo 3 >. 1). / 2 ! © 3 > and the measurement gives a value co. to obtained <( \f/  P ^ i f f } 1 where f \ u is the projection operator associated with  « > . 2> + ( I ) . on the other hand. the initial state were u n k n o w n and a measurement gave a result co. If a> is degenerate. the system had a welldefined state vector  y/> before the measurement. 4. the normalized state after measurement isl known to us to be lvO = 2!/2(1<y. and has a welldefined state vector P^l y/) after the measurement. 2 ) . they are not arbitrary.124 CHAPTER 4 whereas the ideal measurement of Q in q u a n t u m mechanics leaves only the eigenstates of fi invariant. for example. as usual. except that it was some state in the eigenspace with eigenvalue (0. the extra labels 1 and 2 are needed to distinguish the basis vectors in the degenerate eigenspace. l> +  f f l . Consider our example f r o m V 3 (i?) (Fig. On the other hand. If in this basis we know. although all we know is that it lies within a subspace V ^ . the state after measurement is known to be P J y ) ( U P t 0 normalization).
(Of course it does tell us that the original state had some projection on the state <D> obtained after measurement.with p—pi at and wait at x = x/ and see if the particle arrives there with p=pf at t = 2 seconds. 2 (or worse still a A ^ a n y eigenvalue!) that is the end of the theory. iV 1. For sufficiently large N. 125 THE POSTULATES —A GENERAL DISCUSSION ffl> = i  A . It may predict for example that a particle leaving x — xi with momentum p{ in some potential V(x) will arrive 2 seconds later at x = xf with m o m e n t u m p = p f . and if say. We are not ready to discuss this problem. If we measure some variable A. so that the state could not have changed from <»>. This is consistent with the theory but does not fully corroborate it. assume we have prepared a state fy).f How to Test Quantum Theory One of the outstanding features of classical mechanics is that it makes fully deterministic predictions. we must repeat the experiment many times. say A]. measurement. say.) Anyway. I The collapse of the state vector provides us with a good way of preparing definite [States: we begin with a particle in an arbitrary state \y/} and meaure a variable Q... tells us what it is doing just after the measurement. (If co is degenerate. classical statistical mechanics. * (2) Check the probabilistic predictions at any time. makes statistical predictions about a •particle in a state  y > and claims that this state evolves in time according to Schrodinger's equation. If our measurement gives a A. since the odds for Aj could have been 1 /30 instead of 1 / 3 and we could still get k\. one can think of the ensemble as having contained N/3 particles with A = AL and the others with A = A2 before the . of the type one encounters in. instead of telling us what the system was doing before the measurement. approximately N/3 will yield a value A] and end up in the state  A]> while approximately 2N/3 will yield a value A2 and end up in a state A 2 ). immediately thereafter. F o r this purpose we require a quantum ensemble.) Notice how in quantum theory. which consists of a large number N of particles all in the same state  <D>. we have in our hands the state &»>. on the other hand. But we cannot repeat the experiment with this particle. If we get a nondegenerate eigenvalue co. If in a classical ensemble of N particles N/3 gave a result Ai and 2N/3 a result A2. We must start afresh with another particle in  co>. further measurement is needed. To test the prediction we release the particle at x = x. since after the measurement the state of the particle is A>. So let us assume we get one of the allowed values. > + ^ j A2> + 0(others) the theory predicts that k\ and k2 will obtain with probabilities 1/3 and 2/3. To test these predictions we must be able to (1) Create particles in a welldefined state  y > . respectively. The chief difference between a classical ensemble. But this information is nothing compared to the complete specifications of the state just after measurement. Therefore. and the quantum ensemble referred to above. Quantum theory. is the following. the deviations from the fractions 1/3 and 2 / 3 will be negligible. If a measurement of A is made on every one of these particles.
One could equally well send in a macroscopic. A given photon c momentum p and energy E was expected to hit the detectors with a probabilit density given by the oscillating function  y ( x )  2 . every particle is assum ^ e same state <o> prior to measurement (i. and study it at a future time to see if the final state is what the Schrodingi equation tells us it should be. Once we have an ensemble. We choose as our classical system a sixfaced die for which the probabilities P(n) of obtaining a number n have been empirically determined. In this case the intensity pattern will take the shape of the probability density  y ( x )  2 . on the other hand. In a quantum ensemble. As our quantum system we take a particle in a state 6 1 l v / > = £ C M > i= 1 Suppose we close our eyes.2. The following example is provided to illustrate the distinction between the probabilistic descriptions of systems in classical mechanics and in quantum mechanics. every particle is potential capable of yielding either result X\ or X2).126 CHAPTER 4 to measurement. If one makes tht assumption (correct to a high degree) that the photons are noninteracting. We can also prepare an ensemble. 5. The corresponding statements for the particle in the state j y/> are no doub known to you. we can measure any other variable and test tl expectations of quantum theory. mono chromatic beam of light of frequency (o = E/1\ and wave number k=p/fi.. j .e. sending one such photon at a time to see if the final number distributioi indeed was given by  y ( x )  2 . (4) If an ensemble of N such dice are thrown. NP(n) of them will give the result J (as iV>oo). whicl consists of a large number of photons of energy E and momentum p. 2. Only after that measurement are a thi of them forced into the state A) and the rest into A 2 ). One could repeat the experimen N times. and cover it with a mug.2. j (2) The only possible values of n are 1. let it evolve i time. the function P(n collapses to d n 3 . say with photons. Let us now examine some of the key differences between the statistica descriptions in the two cases. the instant the beam is turned on. (3) If the mug is lifted. toss the die. 3. sending ic the beam is equivalent to experimenting with the ensemble. 4.2. and 6. • Example 4.3. and some value—say n — 3—is obtained. Example 4. An example of an ensemble being used to test quantum theor was encountered in the doubleslit experiment. Its statistica description has many analogies with the quantum description of the state ] y/>: (1) The state of the die is described by a probability function Pin) before the muj is lifted.
and NP(a>i) particles in I®. and angular velocity at the time of release. This prediction.=EI<fflil vol2©/ < / = £ <y I 1 w}<jOi (4. .><fl>. there will be NP( 1) dice with n = 1. its position. . given the mass of the die. and so on. to predict exactly which face of the die will be on top.5) But for the factors < > multiplying each projection operator ty . however. we now define and discuss the two statistical variables that characterize an ensemble. The expectation value is just the mean value defined in statistics: <n>=IP(ffl. . .2. The ensemble after measurement is a mixture of six ensembles representing the states  f t ) .y> i Now we can use 1 < t o . we could have xi used \cDi}(cOi \ = / . If one is not interested in such detailed information on the state (or the corresponding ensemble) one can calculate instead an average over the ensemble. These will be discussed in the digression on density matrices. each can yield any of the values © i . Feeding this in and continuing.(1) It is possible. namely j y / > . In the q u a n t u m case.6) J This is an example of a mixed ensemble. the state after measurement. etc. note that Q)i<O1> = Q£O(>. . however. the only possibility in the quantum case. . we get <n> = £ < yQa>. orientation. velocity. involves solving the eigenvalue problem of the operator Q.  = / to get <fl> = < y  Q  y > (4. .) after the measurement. ) <&>. Only the ensemble before the measurement represents the state  if/}. the viscosity of air. however. say  is not the state before measurement. in the ensemble before and after the measurement. (2) If the result n = 3 was obtained u p o n lifting the mug. The statistical description is.)©. it is consistent to assume that the die was in such a state even prior to measurement. at least in principle. to 6 ) before the measurement. <Q). In contrast.2. > . the elasticity of the table top. (3) If N such dice are tossed and covered with N mugs. which follows in a while. Expectation Value Given a large ensemble of N particles in a state j q u a n t u m theory allows us to predict what fraction will yield a value co if the variable Q is measured. fl) 6 >. . even in principle. T o get around this. called the expectation value. the q u a n t u m ensemble corresponding to  y/> will contain N particles all of which are in the same state  y ) (that is. . . NP(2) dice with n = 2.J • 127 THE POSTULATES —A GENERAL DISCUSSION Having seen the utility of the ensemble concept in q u a n t u m theory.
referred to as the uncertainty in Q. . [A familiar example of this phenomenon is that of the mean number of children per couple. that is Q y> = m \ y ) . J (1) To calculate <fi>. we have concentrated on the measurement of a single variable at a time We now turn our attention to the measurement of more than one variable at a time (Since no two independent measurements can really be performed at the same time we really mean the measurement of two or more dynamical variables in rapi< succession.] The Uncertainty In any situation described probabilistically. J (2) If the particle is in an eigenstate of Q. ) ( f f l .8. In quantum mechanics. which may be 2.2. just like <Q>. There is na need to find the eigenvectors or eigenvalues of Q.1 (4. For example.) .2. respectively. (AQ ) 2 = (4. which measures the average fluctuation around the mean. is also calculable given just the state and the operatoi for Eq. then <Q) = m. So far. in some basis). P(m){mi€iyf dm Notice that AQ. (4.< n > ) 2 i and if it has a continuous spectrum.128 CHAPTER 4 There are a few points to note in connection with this formula.2.10 Usually the expectation value and the uncertainty provide us with a fairly gock description of the state. I (3) By the average value of Q we mean the average over the ensemble./2 (4.7 and often called the rootmeansquared deviation.2. it i. If Q has a discrete spectrum  (A n ) 2 = E P ( f l ) . another useful quantity to speci besides the mean is the standard deviation. Th< mean value will generally be an inaccessible value for a single measurement unlesi it accidentally equals an eigenvalue. we know that the particle is likely to be spotted near x = with deviation) of order A. A givel particle will of course yield only one of the eigenvalues upon measurement.7) means just An = [<y](Q<0>)2 y>]. one need only be given the state vector and the operator! (say as a column vector and a matrix. It is defined as AQ=<(Q<Q>)2>1/2 1 (4.2.12. although the number in a given family is restricted to be an integer. if we are given that a particle has <X> = a an< AX= A.
The exceptions are the states \co). and ALX1 (3) Find the normalized eigenstates and the eigenvalues of Lx in the Lz basis. and P(Lz = . what are the outcomes and respective probabilities? (6) A particle is in a state for which the probabilities are P(LZ = 1) = 1/4.\ ) = 1/4. to have a value co for Q. Consider the following operators on a Hilbert space 129 THE POSTULATES —A GENERAL DISCUSSION V (C): 0 1 0 I 0 1 0 Ly 0 i 0 i 0 i 0. The measurement process acts as a filter that lets through just one component of  along some o)>. Convince yourself that the most general. <L*>. and Lx is measured. (4) If the particle is in the state with Lz = 1. P{LZ = 0) = 1/2. To produce such states we need only take an arbitrary state  y > and measure Q. We consider first the question of two operators. D o e s this mean that the factors e'Si multiplying the Lz eigenstates are irrelevant? [Calculate for example P(Lx ~0). A particle in one of these states can be said. since a measurement is assured to give this result. in the L z basis. one cannot say in general that the particle has a definite value for a given dynamical variable Q: a measurement can yield any eigenvalue co for which <co  is not zero. The probability that this will happen in P(co) = We now wish to extend these ideas to more than one variable. 1 0 _0 0 0 0 0 0 1 Lx= 2 l/2 1 0 21/2 i 0 (1) What are the possible values one can obtain if L z is measured? (2) Take the state in which L z = \ . what are the possible outcomes and their probabilities? (5) Consider the state 1/2 ' \V> = 1/2 L1/21/2.1 (Very Important). If l l is measured in this state and a result + 1 is obtained. normalized state with this property is eiSl eiSl e'5* It was stated earlier on that if  y / ) is a normalized state then the state e'°  y/> is a physically equivalent normalized state. In this state what are ( L x ) .] Compatible and Incompatible Variables A striking feature of quantum theory is that given a particle in a state  y>. The extension to more than two will be .Exercise 4.2. as in classical mechanics. what is the state after the measurement? H o w probable was this result? If Lz is measured.
A pair of operators C2 and A will fall into one of the three classes: A. We immediate/) measure A and pick those particles that give a result X. This in turn requires that ©> also be an eigenstate of A The answer to the first question above is then in the affirmative only for the simulta' neous eigenstates \coX). D o we have now an ensembl^ that is in a state with Q = co and A = A? Not generally. An exception occurs when the state produced after the first measurement is unaffected by the second. (4. After th first measurement.2.2. In other words.2.11) with A. but nothing definite for A (since co) need not be an eigenstate of A). Incompatible: [Q. us first measure Q on the ensemble described by 1t//> and take the particles that yiell a result co. and taking the difference: ( Q A — A Q )  COA) = 0 (4. let us try to devise a multiple filtering scheme.11) (4. A] must have eigenkets with zero eigenvalue if simultaneous eigenkets an to exist.12) with Q. which assured a result © fa Q. This change is just the collapse of the state vector )<D) = ^  A ) < A  © ) into the eigenstate A). A ] = 0 B. The means for producing them are just as described above. What is the probability that the filtering will give such a state if \ start with the state (y>? To answer these questions. These are in a state that has a welldefined value for Q. the second filtering generally alters the state produced by the first. These kets satisfy the equations a \ c o x y = (o\(oxy A<YA) = AA>A) (4. A] = something that obviously has no zero eigenvalue C.12)* The question that arises naturally is: When will two operators admit simultaneous eigenkets? A necessary (but not sufficient) condition is obtained by operating Eq.2. Is there some multiple filtering process by which we can take ensemble of particles in some state j t//) and produce a state with welldefined vain co and A for two variables O and A? Question 2. Upoj performing the second measurement. Compatible: [Q.13 Thus [f2. The reason is clear.130 CHAPTER 4 straightforward. We ask: Question 1.. Others . the state was converted to \v') = \xy and we are now assured a result X for A. we had the system in the state  © ) .2. Eq. but nothing definite for Q (since  A) nee< not be an eigenstate of Q). (4.
There is nothing very interesting we can say about this case except to emphasize that even if two operators don't commute. we know a complete basis of simultaneous eigenkets can be found. one can still find a few common eigenkets. In the degenerate case. Any attempt to filter X is ruined by a subsequent filtering for P and vice vesa. Class B. Class C.15b) The reason for calling Q and A compatible if [Q. which obey the canonical commutation rule [X. After the easurement. to a given eigenvalue there is just one ket and this must be a simultaneous eigenket \o>X). this implies the invariance of the state •vector as well. Thus. The most famous example of this class is provided by the position and momentum operators X and P. X) = \(coX\y/}\2 = P(X.14) 131 THE POSTULATES —A GENERAL DISCUSSION I Hr Evidently we cannot ever have ifi\ y/) = 0 y/) for any nontrivial \y/). (You should 5 able to tackle case C yourself after seeing the other two cases. Suppose e measured Q first. If two operators commute.15a) •oduct of the two probabilitiesy= = X  f t ) A > 21 t >= <«A yf)\2  P(co.Class A. Each element \a>X) of this basis has welldefined values for Q and A. the probability is the same for getting the results X and co. le question doesn't arise for case B.P\ = ifi (4. (Why?) Let us now turn to the second question of the probability of obtaining a state )X) upon measurement of Q and A in a state  y/).2.) t Case A. Note the emphasis on the invariance of the eigenvalue under the second •measurement.2. We get co with a probability P(Oi) =  <&>A y/)\2. co) (4. and case C is not very interesting. The probability for obtaining co for Q and X for A is just the  > <£ A y> (4. X) <a>Ay) Notice that if A were measured first and Q next. Thus if we expand  yt) in the complete common eigenbasis as ^hen H P(co. the particle is in a state  coX). the state vector can change due to the second . though not a full basis. which will be developed as we go along. This is the origin of the famous Heisenberg uncertainty principle. This means there doesn't exist even a single ket for which both X and P are well defined. In this case there are some states that are simultaneous eigenkets. We will consider just case A . In the nondegenerate case. A] = 0 is that the measurement one variable followed by the other doesn't alter the eigenvalue obtained in the I n s t measurement and we have in the end a state with a welldefined value for both •observables. The measurement of A is certain to eld the result X. Let us first assume there is no degeneracy.2.
2. and \aJ3A3). Let us . If it is degenerate.  c o 2 X ) . The state j t//') after measurement is the projection of 11j/) in the degenerate X eigenspace: KPjvIP^)!" 2 ( f + r ) ' where. Therefore. X) =  < ( M W}\ 2 j • 1 =  <O>IA y/}\2=(32 (4. But thisj presumes co is not a degenerate eigenvalue. Suppose that the measurement of Q gave a value cox. Thus P(a>i. Let us use as a basis a>iA). Thus X ^ ^ K c o i X i l y / y f ^ a 2 .) = (/? 2 + r 2 ) • P +7 = = . except to within an eigenspace. A3) = P(A 3 . the square of the projection of  \j/} in the eigenspace. we cannot nail downi the state. ©1) independent of the degeneracy. The subsequent measurement of A will leave the state alone and yield the result X with unit probability. A) (4. The state becomes  o>3A3> and the subsequent measurement of A is assured to give a value A3 and to leave the state alone. If Q is measured now. as the following example will show. it was observed that the measurement process could itself be used as a preparation mechanism: if the measurement of Q on an arbitrary. 1)j is the product of the probabilities: ! P(cox. Consider a normalized state  y/) = aca3A3> + p\oaxX) + y  co2X> (4. The resulting state is and the probability for this outcome is \ {(0\X\ y }  2 . the eigenspace is one dimensional and the state vector itself remains invariant. We can now lift that restriction. Q = G>\. Let us say we measure Q first and get a>3. A) =P(X. we are sure we have the state \ \ f / ) = \(o). unknown! initial state given a result co. In our earlier discussion on how to produce welldefined states \y/) for testing q u a n t u m theory. ©3). the state can only be said to remain in the same eigenspace after the second measurement. Let co\ # co2 be the eigenvalues of Q in this degenerate space. Consider two operators A and Q on V 3 (/?). It was therefore suggested that we stick to variables with a nondegenerate spectrum. The probability for obtaining a>x is <G>IA^') 2 = P 2 / ( P 2 + Y2). Thus compatibility generally implies the invariance under the second measurement of the eigenvalue measured in the first. The probability for this outcome is P(X) = (32+ y 2 . ffl.2. Evidently P ( © 3 .Thus. though the eigenvalue will not.17) Let us now imagine the measurements carried out in reverse order. in the expression above.2.measurement.19) Thus P(a)\. the projected state has been normalized. is the product of the probabilities: J P{ A. But this time the state suffered a change due to the second measurement (unless by accident (1//') has no component along \(o2X)).16. both results a>\ and co2 are possible. Let the result of the measurement be X. the probability for the result A = X. If the first eigenvalue is nondegenerate. L e t !tt>3A3>be one c o m m o n eigenvector Let X\ = X2 = A.
2 . . we take an arbitrary initial state and filter it by a sequence of compatible measurements till it is down to a unique. .. If we get a result A. X)^P(X. exists./N is the probability that a system picked randomly out of the ensemble is in the state i'). and k occupancy numbers nu . (i= 1 . all by itself. they correspond to all /?. However. (2) The probability P(co. given all the simultaneous eigenvalues: \a>.The ensembles we have dealt with so far are said to be pure.. we have a definite state  co A). Ultimately we will get a state that is unique. We now measure another compatible variable A. known vector. . We must then measure a third variable T compatible with Q and A and so on.) The state produced by the two measurements is just the eigenstate of the second operator with the measured eigenvalue. A) itself is degenerate. even if the operators £2 and A are incompatible.say a degenerate eigenvalue co for the variable Q was obtained. 12). They are hard to come by in practice. Any nondegenerate operator.2. .(We restrict ourselves to the case where z> is an element of an orthonormal basis. A) is not the probability for producing a final state that has welldefined values co and A for Q and A. co) in general. To prepare a state for studying quantum theory then. More common are ensembles of N systems. A convenient way to assemble all this information is in the form of the density matrix (which is really an operator that becomes a matrix in some basis): P = 5>/i><i (42.). A) that the measurement of Q followed by that of A on a state 11/) will give the results co and A.) Thus the ensemble is described by k kets 11). THE 133 POSTULATES A GENERAL DISCUSSION The Density Matrix—a Digressiont So far we have considered ensembles of N systems all in the same state  y > . (Such a state doesn't exist by the definition of incompatibility... k) of which are in the state i>. unless the value (co. We have then some vector in the co eigenspace. It is <n>=2><iQ/> i (4. y. A. It is presumed that such a set of compatible observables. n..nk. .21) The bar on <Q> reminds us that two kinds of averaging have been carried o u t : a quantum average <7f2r') for each system in t) and a classical average over the I This digression may be omitted or postponed without loss of continuity. = 0 except one. called a complete set of commuting observables. Consider now the ensemble average of Q." Incidentally. respectively. the following should be noted: (1) P(co. . we can specify the probability Pico. .20) where pi=n. is a "complete set.. . . A general ensemble is mixed.
( x . You are urged to convince yourself of these relations.Observe that Tr(Qp)=£0"npl/> j = 1 1 <JW> W>P. the probability of obtaining a particularj value a>.22): The density matrix contains all the statistical information about the ensemble. that is.=I I <11 f> J < ' J 0W>/>/ 1 = <Q> (4.2a). Let us begin by normalizing the state: ' oo /* oo \ = (y/\y/y=\ — oo * oo <>x><* y}dx= JCQ  v(x)\ i 2 dx 1 A2 Joo dx = A2(7TA2)l/2 (see Appendix A.134 CHAPTER 4 systems in different states  z>. (4. but instead P(m). y/(x)=A e x p [ .2. p(co) = \(co\¥)\2=(¥\co)((o\w) which combined with Eq. We first note that. We now try to extract information about this state by using the postulates.22) tells us that P(©)=Tr(Pwp) The following results may be easily established : (1) Pf = P (2) Trp=l = w\pa>\¥y = <.2.4.2. 4. Example 4.pay (3) p 2 = p (4) p = (1 jk)J (5) Tr p2 < 1 for a pure ensemble for an ensemble uniformly distributed over k states (equality holds for a pure ensemble) (4.2. Suppose we want.2) . for a pure ensemble.a ) 2 / 2 A 2 ] (Fig. Let us assume y/(x) is a Gaussian. To gain more familiarity with quantum theory let us c o J sider an infinitedimensional ket j y ) expanded in the basis  x ) of the position operator X : x) (x \f/y dx = jx)y(x) dx \V> = We call y/(x) the wave function (in the X basis). not <Q>.2.
2a. Thus the particle is most likely to be found around x=a. </> V>l = 1 V(/>)l = lv(*)l.(b) *lp) 135 THE POSTULATES —A GENERAL DISCUSSION O D Figure 4. function.. <xXx')<x' — x\f/(x) Using this result.2.10.e. i.0>) The modulus of the wave So the normalized state is y/(x) (xa)2/ZA2 W ) 2.1/4 The probability for finding the particle between x and x + dx is P(x) dx =  y ( x )  2 dx: {xa)2/A2 (xtf) 2U/2 dx which looks very much like Fig. 4. and chances of finding it away from this point drop rapidly beyond a distance A. We can quantify these statements by calculating the expectation value and uncertainty for X. Now. (a) The modulus of the wave function. if <x y> = y ( x ) then. Let us do so. the operator X defined in postulate II is the same one we discussed at length in Section 1. Its action in the X basis is simply to multiply by x. the mean or expectation value of X is \ff)dx'x5(xx')y/(x') dx' <X) = ( ¥ \ X \ ¥ ) < y  x ) <xX y ) dx y * ( x ) x y ( x ) dx J .r 1 2\ 1/2 .
sa. Next. </> \j/y\2 = P(p). (1) (2) (3) (4) First we must construct the operator P in this basis.1/2 (x&l J— J rr. for if (x\yy = y/(x) .136 CHAPTER 4 If we define y = x~a.10 multiplied by ft and has the action of — i"h d/dx in the X basis. Suppose we next want to know the probability distribution for different values of another dynamical variable. Then we must find its eigenvalues p.. Finally.e. Now.  </>. ey2/&2dy —a We should have anticipated this result of course. and if p is continuous.2 e~yVAl (y2+2ya + a2)dy = ~+0 2 + a2 So AX=4~2l/2 2 So much for the information on the variable X. I* O O (y + a) 2. the P operator is just the K operator discussed in Section 1. we must take the inner product </> If p is discrete. since the probability density is symmetrically distributed around x = a. the uncertainty AX=[(¥\(X(X))2\Y)]i/2 =[<¥\x22X(X) = Kw\X2+(X)2\¥)f / 2 (X)2\ Y)]l/2 / 2 (Since < ¥ \ X \ ¥ } = ( X ) ) =[(x2y(x)2f =[<x2ya2r Now <* 2 > (7TA2) 1 1 2^1/2 . y/y\2 = P(pt). we calculate the fluctuations around \Xy = a. th« probability density. and eigenvectors  p y . the momentum P. i.
and APfi/21/2A.10) The solutions. normalized to the Dirac delta function t are 1 Jpx/ft . (Py = 0. This explains the (2Trfi) 1/1 normalization .2b) of width fi/2i/2A. Since AX=A/2i/2. we get <x\P\p)=p(x\p) or p ifi dx =PWp(x) where yP(x) = {x\p}. = 5{kk!)/h. we get the relation AXAP = fi/2 t Here we want (p\p')> = 8{pp') factor. where p = fik. if we project the eigenvalue equation P\P)=P\P> onto the X basis.then 137 THE POSTULATES —A GENERAL DISCUSSION <x\P\¥y= I — oo ' oo {x\P\x')(x'\¥)dx' [m8'{xx')]y(x') dx' (Postulate II) dx Thus. (from Section 1.1/2 (2nK) Now we can compute <P\V> </>x) foo e ~ipx/n y ) dx = e ~(xa} 2 y/*(x)\f/(x) /2A 2 dx / ^ 2 \V4 (2xtiy/2 (ttA2)1/4 dx = \KK2 It follows that The modulus of y/(p) is a Gaussian (Fig. 4.
. So if you are 110lb weakling and are taunted by a 600lb bully. say 500 GeV/c. however. so essential to the physical interpretation of the wave function. It should. For example. when we speak of a particle being in a state x 0 > we shall mean that its wave function is so sharply peaked a x = x 0 that it may be treated as a delta function to a good accuracy. vectors that can't be normalized to unity but only to the Dirac delta function) are introduced into the formalism as purely mathematical entities. Since any particle that we are likely to be interested in will definitely be known to exist in some finite volume of such large dimensions. Thus the absolute probability of finding it in any finite volume. by the physical Hilbert space. even as big as our solar system. But despite all this. because these vectors are so much more convenient to handle mathematically than the proper vectors. will be normalizable to unity and will have a Fourier transform so sharply peaked at p =po that momentum measurements will only give results practically indistinguishable from p0. O T U inability to normalize them to unity translates into our inability to associate with them a sensible absolute probability distribution. while states of welldefined momentum exist.e. a wave function that behaves as e'PoX/fl over a large region of space and tapers off to zero beyond. is in a proper normalizable state (since it is known to be located in our laboratory) and not in a plane wave state corresponding to /> = 500GeV/c>." Thus a particle coming out of some accelerator with some advertised momentum.. is zero. which also form a convenient basis. we mean the space of interest to physicists. there do exist states that are both normalizable to unity (i." but only "exact to any measurable accuracy.150 138 CHAPTER 4 The Gaussian happens to saturate the lower bound of the uncertainty relation (tj he formally derived in chapter 9): &X~AP>n/2 The uncertainty relation is a consequence of the general fact that anythi narrow in one space is wide in the transform space and vice versa. i. But. The other set of improper kets we will use in the same spirit are the position eigenkets  x > . it is really in a proper state with a momentum space wave function so sharply peaked at p=po that it may be replaced by a delta function S(pp0). be borne in mind that when we say a particle is (coming out of the accelerator) in a state [po). docs it mean that states of welldefined momentum d o not exist? Yes. .e. it is clear that no physically interesting state will be given by a plane wave. Again. for "well defined" never means "mathematically exact. in the strict sense However. In the present case we have a particle whose relative probability density is uniform in all of space. correspond to proper vectors) and come arbitrarily close to having a precise momentum. since the plane waves are eigenfunctions of P.t t Thus. Thus there is no conflict between the fact that plane waves are unphysical. just ask him to step into momeq turn space! [j This is a good place to point out that the plane waves eipx/n (and all improper vectors. we will continue to use the eigenkets [ p ) as basis vectors and to speak of a particle being in the state p}. not one whose element all correspond to physically realizable states.
Here is an example from Chapter 19 : Consider the probability that a particle coming out of an accelerator with a nearly exact momentum scatters off a target and enters a detector placed far away.) Exercise 4. which interferes with itself and produces a nice interference pattern. sure enough we get the interference pattern. which upon measuring the position of the electron (as being next to S. But this state is necessarily altered by the light source. the old prediction of interference is no longer valid.2. So much for quantum theory and its interference pattern!" But the point of course is that quantum theory no longer predicts an interference pattern! The theory says that if an electron of definite momentum p is involved. as we shall see. Once the state is changed.cy/r. The above experiment can also be used to demystify to some extent the collapse pi the wave function under measurement. Each passing electron will be exposed by the beam and I note which slit it comes out of. This prediction is valid only as long as the state of the electron is what we say it is.Occasionally. (Hint: S h o w that the probabilities for the momenta ±p are equal. the replacement of a proper wave function by its improper counterpart turns out to be a poor approximation. the corresponding wave function is a wave with a welldefined wave number k=pj%.3* Show that if momentum <JP) + p0. where y/r is real and c an arbitrary (real or complex) constant. we go back to the proper wave (which has tapered off) to represent the initial state.) Generalize this result to the case y/ . We note where these arrive. however. So we proceed as follows. we use the plane wave to approximate the initial wave function. At the detector.2.. has to do Bath the fact that ft is not zero.2* Show that for a real wave function y(x). say) changes its wave function from something that was extended in space to something localized near S i . Now there is no escape from the conclusion that the number arriving at a given x is the sum of the numbers arriving via S\ and S2. We had better. eipax/f> y/(x) has mean Example 4. (Recall that \ y/) and a\y/) are physically equivalent. has mean momentum 139 THE POSTULATES —A GENERAL DISCUSSION <JP>. The collapse of the state vector and the uncertainty principle play a vital role in explaining the following extension of the double slit experiment. . for the two are indistinguishable over the (finite and small) range of influence of the target. Then I note where it arrives on the screen. Intuition says that the answer must be zero if the target is absent. Now. "I don't believe that a given particle (let us say an electron) doesn't really go through one slit or the other. and not in the initial direction. the state not having been tampered with in these cases. In the vicinity of the target. I make a table of how many electrons arrive at each x and which slit they came from.5. but cannot classify them as coming via S\ or S2. This reasonable condition is violated if we approximate the initial state of the particle by a plane wave (which is nonzero everywhere). Exercise 4. the expectation value of momentum <JP) = 0 .2. for quantum theory predicts it. When the distribution of just these electrons is plotted. once in a while some electrons will get to the detectors without being detected by the light source. Why is it that even the ideal measurement jwoduces unavoidable changes in the state? The answer. Suppose I say. So I will set up a light source in between the slits to the right of the screen.
If we soften the blow of each photon by increasing A or narrowing the objective to better constrain the final photon momentum. 4. . (4. classical optics limits the accuracy of the position measurement by an uncertainty j AX ^ A / s i n 80 I Both classically and quantum mechanically. enters the objective O of the microscope.3. were it not for the fact that the x component of the photons entering the objective can range from 0 to p sin 80 = 2kH sin 80/A. As A decreases. If 8 0 is the opening angle of the cone of light entering the objective after interacting with the electron.3.] This is the famous uncertainty principle. and enters the eye E of the observer. the improved accuracy in the position measurement is at the expense of producing an increased uncertainty in the x component ( p ) of the electron momentum. Consider the schematic set up in Fig. the collisions between the electron and the photons become increasingly violent. Since at least one photon must reach our eyes after bouncing off the electron for us to see it. we lose in resolution^ t This would be the ideal position measurement. The reason is that light of wavelength A is not a continuous wave whose impact on the electron momentum may be arbitrarily reduced by a reduction of its amplitude. There is: no way around it. but rather a flux of photons of momentum p = 2nfi/X. we have at the end of our measurement an electron whose position and momenta are uncertain by AX and AP x such that J AXAPx~2nti~n I [The symbols AX and AP x are not precisely the quantities defined in Eq. we can reduce AX to 0 by reducing X to zero.2. enters the objective (O) of a microscope ( M ) and reaches our eye (£). Light of wavelength A illuminates an electron (e~). This in itself would not lead to an uncertainty in the electron momentum. are of the same order of magnitude. there is a minimum uncertainty in the recoil momentum of the electron given by x a > j sin 80 1 Consequently.J In the latter description however.140 CHAPTER 4 M t t t v e Figure 4. Light of frequency A bounces off the electron.7) but.
. . but this light hardly affects its momentum.. We then regain classical mechanics. there exist in quantum theory N mutually commuting operators X\. The position of a billiard ball can be determined very well by shining light on it. XN ? . that we want to find the eigenvalues and . • • • .. . are equally futile. xN)\ xdx\ ... But what we can do is to use bigger and bigger objects for our experiment so that in the scale of these objects ft appears to be negligible. Pi^ifid/dXi.. Postulate II.) are represented by operators Q = a>(xr+Xt. a single particle in one dimension.. OXi . p. .. . For example 2  . . x'N} — 8(X\ X'i) . Suppose.. Physically. x2 + dx2. . The other postulates remain the same... . Dependent dynamical variables a>(Xi.. XN.. We now extend our domain to a system with N degrees of freedom.. xN + dxN. Note that if ft were 0. The only modification is in postulate II. Corresponding to the N Cartesian coordinates xu . dxN is the probability that the particle coordinates lie between xi. x2.More elaborate schemes.. This postulate is stated in terms of Cartesian coordinates since only in terms of these can one express the operator assignments in the simple form Xi~>Xj. . xN) of these operators. . x2. one can perform any desired change of variable before solving them. 8(XN x'N) (the product of delta functions vanishes unless all the arguments vanish) we have the following correspondence: ..xN describing the classical system. x2.. Of course fi is not zero and we can't make it zero in any experiment. In the simultaneous eigenbasis \xux2.. .Pj>Pj). . This is why one imagines in classical mechanics that momentum and position can be well defined simultaneously. . .xN\Pi\\i/y = ih—y/(xl. we could have AX and APX simultaneously 0. XN\X\ . . • 141 THE POSTULATES —A GENERAL DISCUSSION Generalization to More Degrees of Freedom Our discussion so far has been restricted to a system with one degree of freedom—namely.. Once the substitutions have been made and the desired equations obtained in the coordinate basis. it means that we can increase our position resolution without increasing the punch carried by the photons. which now reads as follows. which determine the recoil of the microscope. xN and x\ + dx\.xN) Pi being the momentum operator corresponding to the classical momentum Pi. for example. Pi\y/}^(xu.. called the coordinate basis and normalized as (X\...
let us follow this popular notation and rewrite Eq. d sin 9a — sin 9 d9\ d9. Px^>—ifi etc. y. and are the three Cartesian coordinates and p. 9.2 Once we have obtained this differential equation.2.2. 1 d sin2 9 d<j>2 .142 CHAPTER 4 eigenvectors of the operator corresponding to the classical variable 2m (4. z) d_ dx . we can switch to any other set i coordinates. and z. y.x2.2. Since I 82 dx2 d2 dy2 82 dz1 1 A dr v >i dr. y.24 as \ 2m To solve the equation Q[co) = o)[(oy with Pl + Pl+P} Q =— ^ 2m we make the substitution \a>y^y/a(x.2. z) (4.2! + Y+Z . and get (d2 _ 2m \ dx n 2 dy dz  + x2 + y2 + z2 yfm(x.z) = o>ym(x. Since the coordinates aj usually called x. (4. y. X^x. the correspond! momenta of a particle of mass m in three dimensions. 2 (4. In the present case the spherical coordinates r. and <j> recommej themselves. where x\.
(4.2. There does exist a implicated procedure for quantizing in nonCartesian coordinates.2. there is agreement.26) becomes 143 2 I A (r 2m ^ dr \ + dr 1 3 r sin 0 80 \ ( sin 0 • I 80 r r sin2 61 8<j>2 2 THE POSTULATES —A GENERAL DISCUSSION + r2y/0) = ojy/a What if we wanted to go directly from <y in spherical coordinates (4.1) Setting Up the Schrodinger Equation To set up the Schrodinger equation one must simply make the substitution where Jf' is the classical Hamiltonian for the same problem.27) Eq. which specifies the change of this state with time. Except for a brief discussion of these toward the end of the program. (4. According [to this postulate. . The Schrodinger Equation (Dotting Your Ps and Crossing Your ' Having discussed in some detail the state at a given time. namely. Thus. to relativistic quantum mechanics and to quantum mechanics of systems in which particles are created and destroyed (so that the number of degrees of freedom changes!).3.3. as in say the hydrogen spectrum. There are further generalizations. tin the sense that in cases where comparison with experiment is possible. d <j> which generates Eq.2.27)? It is clear upon inspection that there exists no simple rule [such as >(# d/dr)] for replacing the classical momenta by differential operators in r. 0.27) starting from the a above. (4.2. since the recipe eventually reproduces what the Cartesian recipe (which leems to work!) yields so readily. we now turn our . 4. we will not address these matters. attention to postulate IV. but we will not :uss it. the state obeys the Schrodinger equation at Our discussion of this equation is divided into three sections: (1) Setting up the equation (2) General approach to its solution (3) Choosing a basis for solving the equation (4.
3. we must use til symmetrized form H=— ( P . which is classically described by Hamiltonian 2m 2 the Hamiltonian operator in quantum mechanics is P2 1 2 H=—bmaP'X 2m 2 (4. . 1 c c J since P does not commute with A.P — 2m V c + + (4. If the particle in one dimension is subject to a constant force / . We will discuss this question when we study spi in some detail in Chapter 14. Problems involving spin have no classical counterparts ar some improvisation is called for. For a particle of charge q in an electromagnetic field in three dimensions. which is a function of X. 2m In constructing the corresponding quantum Hamiltonian operator.3 In three dimensions. then 2 2m and P2 H= — .3.144 CHAPTER 4 if we are describing a harmonic oscillator. assuming the force constant is the same in all directions. In this manner one can construct the Hamiltonian H for any problem with classical counterpart.f X 2m (4. the Hamiltonian operator for the quantum oscillator is likeW + imft>2(x2+ 2 Y2 + Z2) H^LlZlz^Li 2m (4. and Z .3. Y. .
3.3. Schrodinger's equation is first order in time. Assume that we have solved it and found the kets  E).3.General Approach to the Solution Let us first assume that H has no explicit t dependence. and the specification of )\f/} at 7 = 0 is a sufficient initialvalue datum. If we expand  i//> as I V(')> = I\E> <£1 V ( 0 > = 1 as(t)\Ey (4. Let us now construct an explicit expression for U(t) in terms of \E}. Once we have U(t).3.1): IW)> = n . (4.10) is aE(t) =a£(0) e —iEt/% (4.8) This is called the timeindependent Schrodinger equation.3. since it is determined by Eq.9) the equation for a£(t) follows if we act on both sides with (ifi d/dt — H): ? 0 = (ifid/dtH)\\ff(t)y = YJ(ifidEEaE)\Ey ^>iUE=EaE (4. In this case the equation 145 THE POSTULATES —A GENERAL DISCUSSION is analogous to equations discussed in Chapter 1 * and describing the coupled masses and the vibrating string. the normalized eigenkets of H with eigenvalues E which obey H\ Ey = E[Ey (4. we can write M O > = t/(Olv(0)> There is no need to make assumptions about  yr(0)> here. The solution to Eq. respectively. (4.3.11a) .^ V ( 0 ) > In other words.10) where we have used the linear independence of the kets l/s). Our approach will once again be to find the eigenvectors and eigenvalues of H and to construct the propagator U(t) in terms of these.
13). We may therefore think of the time evolution of a ket y/(0> as a "rotation" in Hilbert . the sum must be replaced by an integral.146 CHAPTER 4 or <E\¥(t)) so that I v(0>£> <E\ y(0)> e~m.14) If this exponential series converges (and it sometimes does not).) Since H (the energy operator) is Hermitian.i E l / n are also called stationary states for the following reason: the probability distribution P(a>) for any variable Q is timeindependent in such a state: P(o>.3. one must first introduce an extra label a (usually the eigenvalue i a compatible observable) to specify the states. The normal modes \E(t)) = \E)e. In this case f(0 = EElE. We have been assuming that the energy spectrum is discrete and nondegenerate.3.a\ a E e iEt/n If E is continuous.11 We can now extract U(t): (4. 0) There exists another expression for U(t) besides the sum. (4. f) = K<yy(0>!2 = \<co\E(t))\2 e~mt/T = IO£> = \<a>\E)\2 = P(®. ai that is _ iHt/h U(t) = e (4.% (4.1 = (E\W(0)yeiEl/* (4.a><E. Eq. this form of U(t) can be very useful. E is degenerate. it follows that U(t) is unitary. (Convince yourself that  y/(t)} = e~'Ht/n I y(0)> satisfies Schrodinger's equation.3.3.3.
come out the same as before since which is the dot product of and is invariant under rotations. One immediate consequence is that the norm <y/(0i <K0> < 1 V(0> = <vm invariant: (4. We no longer look for normal modes. This view of quantum mechanics is called the Heisenberg picture. In such a basis the vectors would appear frozen. we can write to first order in A d IV> dt y(A)> =  ^ ( 0 ) > + A iA n = IV(0)>#(0)y(0)> iA ft which. stays normalized.3.15) 147 THE POSTULATES . where N is very large and A is very small. one can abandon the fixed basis we have been using. There are other consequences of the fact that the time evolution may be viewed as a rotation. how is 17(/) in  v ( f ) > = £/(Olv(0)> related to H i t ) ! To find out. There exists no fixed strategy for solving such problems. For example. Any physical entity. since the operator in question is changing with time.A GENERAL DISCUSSION If it) U(t) I v(0)> = <y(0) v(0)> so that a state. but the operators. What is the propagator U(t) in the timedependent case? In other words.space. By integrating the Schrodinger equation over the first interval. would. once normalized. would now appear to be time dependent. such as a matrix element. So if you think you were born too late to make a contribution to quantum theory fear not. while the one we have been using is called the Schrddinger picture. We will take up the study of various pictures in Chapter 18. we divide the interval (0 — f) into N pieces of width A = t/N. We will also study a systematic approximation scheme for solving problems with A # H(t) = H° + H\t) where H° is a large timeindependent piece and H\t) is a small timedependent piece. for you can invent your own picture. however. and adopt one that also rotates at the same rate as the state vectors. which were constant matrices in the fixed basis. In the course of our study we will encounter a timedependent problem involving spin which can be solved exactly. to this order iA ft lv(0)> = exp Hi 0) l v ( 0 ) > . Infinitely many pictures are possible. Let us now consider the case H= H(t). each labeled by how the basis is rotating.
But notice thai being a product of unitary operators. if H(t)^X2 then H(0)=X: and H(tz/2(O) = P: and [ ^ ( 0 ) . The difference between these possibilities is of order A and hence irrelevant. since there is already one power of A in front of H.] Inching forth in steps of A. F o r example. U(t) is unitary.1 H(t') dt' >= lim N—• oo n = 0 n exp[(//*)i/(«A)A] in such problems. one must use //(0) or H{A) or / / ( A / 2 ) and so on. j . H(n/2«)] #0 integral 2 cos 2 o)t + Pz2 sin2 oat H{h)]±0 It is common to use the symbol. We will not make much use of this form of U(t). U(t) = exp . called the timeordered 7A exp am N . we get N 1 re = 0 We cannot simply add the exponents to get. and time evolution continued to be a "rotation" whether or not H is time independent.0 7 * ) H ( f ) dt' since [H(tx).148 CHAPTER 4 [One may wonder whether in the interval 0 —A. in general. in the N* oo limit.
tl) = U(h.3.t2)U{t2.19) which can be solved. since HH(X. You can easily prove them by applying the U's to some arbitrary state and using the fact that U is unitary and U(t. for in going to one of them the corresponding operator is rendered diagonal. the propagator satisfies the following conditions: U(h.20) . P) the X and P bases recommend themselves. Thus one can go to the X basis in which X^x and P^iH d/dx or to the P basis in which P>p and X^iH d/dp.3.3. Although all bases are equal mathematically.3.tx)=U\h. First of all. Assuming it is of the form (in one dimension) P2 — +V(X) 2m H=T+V= (4.18) (4.Whether or not H depends on time.u)=U{tut2) It is intuitively clear that these equations are correct.16) U\t2. Since V(X ) is usually a more complicated function of X than T is of P. Thus if V " p2 H= — + 2m the equation H\E> = E\E> becomes in the X basis the secondorder equation ( n2 d2 l * y/E(x) = Ey/E(x) 2m dx2 cosh 2 x. cosh 2 (iH d/dp) _ \f/E{p) = Ey/£ip) (4.tl) (4. The choice between the two depends on the Hamiltonian. the Schrodinger equation is always solved in a particular basis. one would have ended up with the equation 1 _2m which is quite frightening. Had one gone to the P basis. one prefers the X basis.3.17) the choice is dictated by V(X). 1 r~~ cosh X (4. t) = l 149 THE POSTULATES —A GENERAL DISCUSSION Choosing a Basis for Solving Schrodinger's Equation Barring a few exceptions. some are more equal than others.
z2) is a function in a sixdimensional configuration space an cannot be visualized in our space. One may thus visualize y/(x) as a function in our space.2 whereas in the X basis one gets the secondorder equation — — f x \y/£(x) = 2m dx' Ey/E(x) (4. This has to do with the fac that the x space is the space we live in. we will restrict ourselves to problems of a single particle in one dimension in the next few chapters.3. There exists a builtin bias in favor of the X basis. But like all pictures. But. yx. y2. We now turn to the application of q u a n t u m theory to various physical problems.3. hydrogen atom. In other words. T o understam this point. z) can b visualized as a wave in our space. * This brings us to the end of our general discussion of the postulates.3J In the P basis one gets a firstorder differential equation (4. it is not a real wav< like the electromagnetic wave. for which (4. . when we speak of th probability of obtaining a value between x and x + dx if the variable X is measurec we mean simply the probability of finding the particle between x and x + dx in ou space. if we consider next a twoparticle system y/(xi. whose modulu squared gives the probability density for finding a particle near x. it has its limits.2 The harmonic oscillator can be solved with equal ease in either basis since H quadratic in X and P. etc. y. F o r pedagogical reasons. momentum. x2.150 CHAPTER 4 A problem where the P basis is preferred is that of a particle in a constant fore field / . which carries energy. Such a picture i useful in thinking about the doubleslit experiment or the electronic states in . The function \j/(x. consider a particle in three dimensions. Z]. First of all it must be borne in mind tha even though y/(x) can be visualized as a wave in our space. Thus the case of the single particle is really an exception: there is only oa position operator and the space of its eigenvalues happens to coincide with the spac in which we live and in which the d r a m a of physics takes place. It turns out to be preferable to solve it in a third basis i which neither X nor P is diagonal! Y o u must wait till Chapter 7 before you see hoi this happens.
(5.1) 2m The normal modes or stationary states are solutions of the form iEt/h (5. we get the timeindependent Schrodinger equation for £>: 2m This problem can be solved without going to any basis.Simple Problems in One Dimension Now that the postulates have been stated and explained. 5.1. they contain most of the features of threedimensional quantum mechanics but little of its complexity.1.1. artificial.3) . This problem is so important that a separate chapter has 'been devoted to its study. We begin with the simplest class of problems—concerning a single particle in one dimension. Although these onedimensional problems are somewhat . The Free Particle The simplest problem in this family is of course that of the free particle. it is all over but for the applications.1). The Schrodinger equation is (5.1.1. One problem we will not discuss in this chapter is that . First note that any eigenstate (5. of the harmonic oscillator.2) Feeding this into Eq.
So we feed the trial solution /?> into Eq. ~y — \p= (2mE)1/2> —(2m£')1/2) (5.of P is also an eigenstate of P2. E = P2/2m}.( 2 m £ ' ) 1 / 2 > (5. Now.1.+) = \p = \E.1. In terms of the ideas discussed ii the past.1.31 In other words. We sha therefore drop this redundant label. we find that the allowed values of p are p=±(lmE) 1/2 (5. The propagator is then (* O O V(t) = IP><P\ e~iE(p)l/n dp (5. The obvious choice i given by the kets \E. In other words. + ) and \E. the label p by itself is adequate. and find i or (5. Since P is a nondegenerat operator.1. there are two orthogonal eigenstates for each eigenvalue E. So what's new?" What is new is the fact that the state I E ) = P\p = ( 2m £ ) 1 / 2 > + y\p = .4 Since \p} is not a null vector. spanned by the above vectors. we are using the eigenvalue of a compatible variable P as an extra labe within the space degenerate with respect to energy. \E. there is no need to ca the state \p. — ) themselves.6) (5.7) Thus. (5.1.8 is also an eigenstate of energy E and represents a single particle of energy E that ca be caught moving either to the right or to the left with momentum ( 2 m E ) V 2 \ To construct the complete orthonormal eigenbasis of H.1. "This is exactly what happens in classical mechanics. we must pick fron each degenerate eigenspace any two orthonormal vectors. we find that to the eigenvalue E there corresponds a degenerate twodimensional eigenspace. Physically this means that a particle of energy E can be moving to the right or to the left with momentum \p\ = (2m£) l/2 .U . since the value of E=E(p) follows. you might say. given p.
x'. x'.f and not the absolute values of t and t'.9) may be rewritten as an integral over E and a sum over the ± index as m 153 SIMPLE PROBLEMS IN ONE DIMENSION m= i \E.1. x'.Exercise 5. (5. Suppose we started off with a particle localized at . s i n c e Udepends only on the time interval t .1. any initialvalue problem can be solved./:i/ndE Exercise 5.1.] Whenever we set t' = 0. (2mE) 1/2 a e .8). t) = U(x. (') = (x\ U(t .1. only time differences have physical significance. /') by considering a special case of Eq.OI*'>. t)  U'(x. t. In terms of this propagator. t.1. f ) dx' where U(x.. We start with the matrix element 1* 00 <x\pXp\x') U(x. t\ x').10) using the result from Appendix A.1.x') = <:x\U(t)\x') = gip(xx')/fi . A nice physical interpretation may be given to U(x. t')\{/{x'. In the absence of anything defining an absolute time in the problem. regain Eq. ^ip2t/2mfi dp dp I j t f ij e \1/2 _ I ) e'™<x ~ x')2/2fit \2 jtHit) (5. 0) dx' (5.] Though will satisfy the equation even if E< 0.1. [Had there been a timedependent potential such as V(t) = V0 e a{1 in / / . we could have told what absolute time it was by looking at V{t). show that the general solution of energy E is exp [i(2mEy/2x/fi] V =P (2n K) + 7 exp[i{2mE) i/2 x/h] (2nK) 1/2 i(fE(x) [The factor ( 2 n f i ) 1 / 2 is arbitrary and may be absorbed into p and 7. 0) as simply U(x.2 on Gaussian integrals.e. t. (5. t\ x')y(x'. a}(E. x'. (5. Show that Eq.1.2* By solving the eigenvalue equation (5. since y/(x.3) in the X basis. we will resort to our old convention and write U(x. t.11) Had we chosen the initial time to be t' rather than zero. t. are these functions in the Hilbert space? The propagator U{t) can be evaluated explicitly in the X basis. we would have gotten (5.12) yr(x.1.12).1. i.
One also refers to U(x. Time Evolution of the Gaussian Packet There is an unwritten law which says that the derivation of the freeparticlj propagator be followed by its application to the Gaussian packet.12) then tells us that the total amplitude for the particle's arrival at (x. x'0. f).uj This packet has mean position <^> = 0. t.12) w < get ifit mA/ 1/2 Hx. Then y/(x. t') = S(x' — xo).\/4 (5. t') (5.2/2A2 y/(x'. 0) = eipox'/f' 2. and meai momentum p0 with uncertainty fij21/2A. with \j/(x'. (5.13 x exp ipaj . 0 = A+ • exp (jcp0t/m)2 2A 2 (1 +ifit/mA2)_ (5. Let us follow thil tradition.XQ). t) A + 2 ft t /m A y 2 2 2 2 /2 P j~ [x—(po/m)tf 2 \ A + h2t2/m2A2 (5. t)= U(x. By combining Eqs. f.154 CHAPTER 4 x' = Xq. t') ends with at the spacetime point (x. [It can obviously be given such an interpretation in any basis: (co\ U(t. t) is the sum of the contributions from all points x' with a weight proportional to the initial amplitude y/{x'.13) In other words.1.1.16 The main features of this result are as follows: (1) The mean position of the particles is pot m <py m ay .] Equation (5. the propagator (in the X basis) is the amplitude that a particle starting out at the spacetime point (xo. . _PoV ft V 2m j The corresponding probability density is 1 P(x.x'0. f ) that the particle was at x' at time t'.10) and (5. that is. Consider as the initial wave function the wave packet x. t') as the "fate" of the delta function y(x\ t') = S(x' . with an uncertainty A X = A / 2 1 / 2 . t')\co') is thfl amplitude that a particle in the state  co'} at t' ends up with in the state \co) at f.1.1.1.1.
This is just one of the consequences of the Ehrenfest theorem which states that the classical equations obeyed by dynamical variables will have counterparts in quantum mechanics as relations among expectation values.1.17) for large times.1.In other words. and set A = l. Although we are able to understand the spreading of the wave packet in classical terms. the classical relation x = (p/m)t now holds between average quantities. (5. Hints: (1) Write f ( x .15). (5. (5. This and similar questions will be taken up in greater detail in the next chapter. the uncertainty in velocity would be t » ^ ( 0 =^ =7^(1+^4) (5. 0) = 1/4 Find y(x.1. Exercise 5.17) 1 A K(0) ^ A— 4 I11//22 m A 2 * 10"14 cm/sec It would be over 300. the fact that the initial spread A V(0) is unavoidable (given that we wish to specify the position to an accuracy A) is a purely quantum mechanical feature. and we wished to fix its initial position to within a proton width.14) with p0 = 0. For a free particle this becomes U(t) = exp 1 (n2l fi Xlm dy dx2) 2n .18) above and compare with Eq. t) using Eq. t' = 0 : V(x. (2) The width of the packet grows as follows: 155 SIMPLE PROBLEMS IN ONE DIMENSION The increasing uncertainty in position is a reflection of the fact that any uncertainty in the initial velocity (that is to say.3 (Another Way to Do the Gaussian Problem).1. The theorem will be proved in the next chapter.1. the momentum) will be reflected with passing time as a growing uncertainty in position. the uncertainty in X grows approximately as &X~fit/2x/2m& which agrees with Eq.„=0 1 1 n\\2m)  dxn (5. In the present case. namely. since A V(0) = AP(0)/ m=ft/2 I/2 mA.18) Consider the initial state in Eq. If the particle in question were macroscopic.000 years before the uncertainty A(t) grew to 1 millimeter! We may therefore treat a macroscopic particle classically for any reasonable length of time. U(t) = e~'Ht/n. We have seen that there exists another formula for V(t).1. 0) as a power series: 00 f _ 1 VY2" n — 0 «!(2) . say of mass 1 g.1. which is approximately 10" 13 cm. (5.
the function wii definitely leak out given enough time. Notice that the convergence of at operator power series depends not just on the operator but also on the operand. so that there seems t o be a paradox. (4) Look for the following series expansion in the coefficient of x2": \ m} \m J 2! \m J (5) Juggle around till y o u get the answer.. while if it falls off continuously. on this power series. Exercise 5. fnx\ f ( x . = 0. (3) Collect terms with the same power of x.156 CHAPTER 4 (2) Find the action of a few terms ]_(itti c f ) 1. thf derivatives are singular at the boundary.1. These obey „ = w 2m(Eh 2 V) y/ where each prime denotes a spatial derivative. 2ml dx 2 2! \2m dx2! etc. n o matter how rapid the falloff. Consequently the action of U(t) = exp fi \2m/ dx2. 0) = sin [ — I. Let us ask what the continuity V(x) implies. Let us start at some point x 0 where y/ and y/' have the values f(x and y/'(x0). So there« n o paradox: if the function dies abruptly as above. Consider the wave function .4: A Famous Counterexample. .] Some General Features of Energy Eigenfunctions Consider now the energy eigenfunctions in some potential V(x). x<L/2 \x\>L/2 It is clear that when this function is differentiated any number of times we get another functi confined to the interval x < L/2. W e have here an example where the (exponential) operator power series doesn't converge. What about the spreading of the wave packet? [Answer: Consider the derivatives at the boundary. If we pretend that x is a time variable and that y/ is a particle coordinat the problem of finding y/ everywhere else is like finding the trajectory of a partic (for all times past and future) given its position and velocity at some time and il acceleration as a function of its position and time. It is clear that if we integraj . on this function is to give a function confined to x <L/2.
respectively.1. albeit a rather artificial one: V(x) = 0. 5. Consider first region III. What if the change in V is infinitely large? It means that f" is also infinitely large. in which V= oo. for the area under a function is continuous even if the function jumps a bit. y/n. But whether or not y/' is continuous. for the area under yf" can be finite over an infinitesimal region ^ i t surrounds this point. will be continuous.1a). = oo. The solution y/ is called y/j. and y/m in regions I.2. The Particle in a Box We now consider our first problem with a potential.1a) is called the box since there is an infinite potential barrier in the way of a particle that tries to leave the region jx < L j 2 .! Let us turn our attention to some specific cases. F these equations we will get continuous y/'(x) and y/(x). T h i s will b e true even in the artificial potentials we will encounter. II. however. (a) T h e b o x potential. for mathematical simplicity. Now } We are assuming that the j u m p in \]/' is finite. But can y o u think of a potential for which this is n o t true? (Think delta. and III (Fig.1) iis potential (Fig. The eigenvalue equation in the X basis (which is the only viable choice) is d2w 2m — t + ^T dx2 n2 (E V)y/ = 0 (5. we consider potentials that change abruptly at some point. II. 5. \x\ <L/2 \x\>L/2 (5. y/' will still be continuous. There are. However. some problems where. which is the area der it.a) oo oo 157 b) SIMPLE ' PROBLEMS IN v OONEDIMENSION in L/2 0 L/2 L/2 0 L/2 Figure 5.2.) .2. This means that y/" jumps abruptly there. This is the typical situation. and III. 5. This in turn means that y/' can change abruptly as we cross this point. It is convenient to first consider the case where V is not infinite but equal to some V0 which is greater than E.2) k begin by partitioning space into three regions I. (b) T h e first t w o levels and wave functions in the b o x . y/.
2.2] (5. for y/n(x)= y/ only in region II and not in alii space.10) .2 (5. § Although A and B are arbitrary coefficients from a mathematical standpoint.2. jumps to infinity there. we must set B = 0 on physical grounds since Be KX blows up exponentially as x»oo and such functions are not members of our Hiibert space. This is not so. (5.2.) These constraints applied to Eq.w see that yin = 0 It can similarly be shown that exactly those of a free particle: = In region II. + B exp[ k= i(2mE/fi2y/2x] (5. since V=0. If we now let V>cc.2. We must require that y/ u goes continuously into its counterparts y/i and as we cross over to regions I and III. In other words we require tha Wi( ~ L/2) ~ t^n( ~ L/2) = 0 Vm( + L/2) = y/n( + L/2) = 0 (5.2J (5.2) becomes d2y dx2 m 2m(V0E) ft1 Vm = 0 (5.2. the solutions Yu = A exp[i(2mE/fi2y/2x] = A eikx + B e~ikx.4 where K = [2m(V0~E)/fi2y/2. respectively.: which is solved by ym=Ae KX + Be* (5. AeikLI2\Be~ikL/2 or in matrix form ~e~ikL/2 gikL/2 =0 eikL/2 ~A~ eikL/2 _B_ "0" _0_ (5. (5.158 CHAPTER 5 Eq.: (2mEjh2y/2 It therefore appears that the energy eigenvalues are once again continuous as in i freeparticle case.1 (We make no such continuity demands on y ' at the walls of the box since.6) take the form Ae~ikL/2 + BeikL/2 = 0 (5.2.2.
. and since eigenfunctions differing by an overall factor are not considered distinct.12) e~*LeikL=2ism(kL) =0 (5.12). (5.17a) v xi' Z and from Eqs.2.13) Multiplying by e'K"/2. .— \L \ L j n odd (5. .15) '2\1/2 (nnx\ — cos . we study just Eq. . Further. L (5.2.2.11) 159 SIMPLE PROBLEMS IN ONE DIMENSION « = 0.17b) that . which says Ae~m7t/2 + Beinn/2 = 0 (5. 5. To find the corresponding eigenfunctions.17a) and (5.2. Eq. Since only one of them is independent. (5.Such an equation has nontrivial solutions only if the determinant vanishes: 41.} (5.2. we get A= efn'B (5. In summary. we may restrict ourselves to positive nonzero n.2. (5.17c) [Itis tacitly understood in Eqs.2. 4 . (5.2.9a) and (5. only if nn k =— .2. ± 1 . since ijtn = y/n for n odd and y/n= y/n for n even.17b) fi2kl __ H27t2n2 2 ml} \x\<L/2. (5.16) Notice that the case n = 0 is uninteresting since yo^O. " = 1. . we have /2\"2 ¥ innx" c o "=\l! ^2\1/2 \T". 6 . (5.6) generates two families of solutions (normalized to unity): „„(*) = g ) sin(^] n even (5.2.2. we go to Eqs.14) Since einK = (1)".3.2. En 2m .2.2. that is.9a).9b). (5.2.6) and (5. sm (nnx\ n = 2. ± 2 . 7.2.2. .
2.) Only in the left. namely. can blow up. Let us divide space into tiny intervals such that in each interval V(x) is essentially constant. As x>±oo. quantization resulted j f r o m the requirement that v^u completed an integral number of halfcycles within.e. i. by a potential that. and the cosine. (5. Thus if say. (See Exercise' 5. We argue once again that we have one more constraint than we have parameters. X and P. observe that every extra interval brings with it two frea parameters and one new interface. These solutions are also example of bound states. JC >Lj2) y/ is a sum of rising and falling exponentials (as jc »oo) and that we must choose the coefficient of the rising exponential! to be zero to get an admissible solution. it doesn't blow up within the finite intervals.2. has only three free parameters. since by assumption K h a s a definite limit as x . In the classically allowed region (x < L/2) y/ is a sum of a sine and cosine. the growing exponential. This is the origin of energy quantization here. The right. Thus we impose four conditions on y/. Thus as we go froni . leads to the quantization of energy. since. Consider next a particle bound by a finite well. (It may seem that there are four—the coefficients of the two falling exponentials. y/' does not satisfy the continuity condition at x = Lf 2. these1 being linear in y/ and y/'. (The rising exponential is not disallowed. We have already seen [Eq. Let us gain some insigha into how this happens. had a continuous spectrum of eigenvalues from — oo to +00.6 for details). which. the sine. i.We have here our first encounter with the quantization of a dynamical variably Both the variables considered so far. Bound states are thus characterized by i y/(x) •0 1 Bound states appear in quantum mechanics exactly where we expect them classically? namely. which coincided with the allowed values in classical mechanics. an overall rescaling of y/ and y/' will not help. N o w in all the finite intervals.and rightmost intervalsdoes y/ have just one parameter.4)] that in the classically forbidden region ( E < V0.) Clearly. Consider now a general potential V(x) which tends to limits V± as x>±cc and which binds a particle of energy E (less than b o t h F±). the overall scale of y/ is irrelevant both in the eigenvalue equation and the continuity conditions. jumps from 0 to V0 at \x\=L/2. we demand that] y/ and y/' be continuous at x = ±L/2.. the box so that it smoothly joined its counterparts y/i and y/m which vanished! identically. as follows. for in these infinite intervals. The particle in the box is the simplest example of a situation that will be encountered again and again.and leftmost intervals can be made infinitely wide. the continuity conditions cannot be fulfilled except possibly at certain special energies. these intervals can be made longer and longer since V is stabilizing at its asymptotic values V±. each interface between adjacent regions.. states in which a potential prevents a particle from escaping to infinity. wherein Schrodinger's equation. y/ has two parameters: these will be the coefficients of the sine/cosine if E>Vorj growing/falling exponential if E< V.» ± o o . combined with appropriate boundarj conditions.e. Since F i s everywhere finite. T o see that we have one more constraint than we have parameters. in situations where V(± oo) is greater than E. In fact. All these parameters are constrained by the continuity of y/ and f' at. two new constraints. j The energy levels of bound states are always quantized. so did the spectrum of the Hamiltonian in the freeparticle case. In the problem of the particle in a box. However.
for y/j also contains a growing exponential in general. When we cross the well.] We are now out of the fix as x>oo. Consider the f search for acceptable energy eigenfunctions. V=kmco2x2. For a nice numerical analysis of this problem see the book by Eisberg and Resnick. » [Later we will study the oscillator potential. however. which grows without • p i t as x — oo. Quantum Physics of Atoms. the particle is' so strongly inhibited from going to. Resnick.] These are the allowed energies and the corresponding functions are the allowed eigenflinctions. of the well. the constraints are always one more than the free parameters. it blows up as x+oo (and vice versa). that this implies > 0. i.o o . this can happen. y/ oscillates there and so we have two more parameters. Thus there will be no acceptable solution at some randomly chosen energy. starting with Eq.§ It is clear how these arguments generalize to a particle bound by some arbitrary potential: if we try to keep yr exponentially damped as x . happen that for certain values of energy. and so two solutions (normalizable to 8(0)) at any energy. t We are not assuming E is quantized. we can even say what the desired value of y/'(xo) is: it is  given by y/'(x0) = Ky/(x0). Molecules. y/ will be exponentially damped in both regions I and III. it shouldn't matter if we flatten out the potential to some constant at distances much greater than x 0 . If ' we start with some arbitrary values y/(xo) and y/(xo).2. the allowed levels and eigeni functions must be the same in the two potentials which differ only in a region that . How do we understand energy quantization here? Clearly. so that the particle is "even more forbidden than before" from escaping to infinity. The y/ we get integrating from region III will not generally have this feature. Wiley. New York (1974). y/ will again start to grow exponentially. See Section 5. jwevary y/'(x0) until the growing exponential is killed. one from each end (why?). Eisberg and R. 161 SIMPLE PROBLEMS IN ONE DIMENSION . we can understand energy quantiza tion as we did before. [Since we can solve the problem f analytically in region III. Verify. at some point x 0 to the right .e. instead of being a constant. Nuclei and Particles. At special energies.intervals in the finite well to the infinite number of intervals in the arbitrary tential. we can integrate Schrodinger's equation numerically. any > mlowed y/ will vanish even more rapidly than before as x +oo. we can choose yf(x0) such that they are normalized to unity. Solids.! we expect y/ to "fall off rapidly as we cross the classical turning points x 0 = ± (2E/mco2)h>2. Having found them. particle in such a state.4). but we are committed to whatever comes out as we integrate to the left of x 0 . [At any point x'0 in region I. grows quadratically.7 and Appendix F. We will find that y/ grows exponentially till we reach the well. If E is an allowed energy.. whereupon it will oscillate. there is a ratio y/'(x'0)/y/(x'0) for which only the damped exponential survives. To a . Since y/(xo) merely fixes the overall scale. (5. It is also clear why there is no quantization of energy for unbound states: since the particle is classically allowed at infinity. It can.] [ Let us restate the origin of energy quantization in another way. (Recall the analogy I with the problem of finding the trajectory of a particle given its initial position and ! velocity and the force on it. §R. taking the finite well as an example.) As we integrate out to x+oo. except at some special energies.+ . Thus only at special energies can we expect an allowed solution. since V—E. y/ will surely blow up i since contains a growing exponential. however. Since the flattenedout potential has the asymptotic behavior we discussed earlier.
203 The actual groundstate energy E\ happens to be n 2 times as large as the lower $ We are suppressing the infinite potential due to the walls of the box. which prevents the particle. <P> is time independent. Consequently (H)>ti2/2mL2 In an energy eigenstate. Since a bound state is stationary state.162 CHAPTER 5 Let us now return to the problem of the particle in a box and discuss the fac that the lowest energy is not zero (as it would be classically. corresponding to th particle at rest inside the well) but fi2n2 /2ml}.2. (H} = E so that E>fi2/2mL2 (5. If this the particle must (in thi average sense) drift either to the right or to the left and eventually escape to infinity which cannot happen in a bound state. cannot exceed L/2.2. The reason behind it is the uncertainty principle. We begin witht P2 2m so that <P2> = 2m (5. Consequently we may rewrite Eq.18 Now <P> = 0 in any bound state for the following reason. Instead we will restrict x to thd range  x\ < L/2.19 (5.2.2.19) as <7/) _<(^<^»2>_(AP)2 2m 2m  If we now use the uncertainty relation AP • we find AX>h/2 <H)> 8m(AXy its standard deviation ft2 Since the variable x is constrained by —L/2<x<L/2. J . whose position (and hence AX) is bounded by \x\<L/2. which we derive as follows. (5. from having a welldefined momentum of zero. This in turn leads to lower bound on the energy.
The desired result follows from the application of the theorem proved above. if we define V(co) = 0. and keep only the solution that has the right behavior at infinity and is continuous at x = 0.2.2.22) Jnlike in the freeparticle case. Draw the wave function and see how there is a cusp. where E0 is the I lowestenergy eigenvalue. * ' ) = I «=i ¥n(x)y/*(x') exp fXlmL2 J (5.2.1. * A particle is in the ground state of a box of length L. The uncertainty principle is often used in this fashion to provide a quick orderofmagnitude estimate for the groundstate energy.) ' (b) Prove the following theorem: Every attractive potential in one dimension has at least ^one bound state. Exercise 5.2. 2m dx V(x)\ Show that E(a) can be made negative by a suitable choice of a. consider a . Exercise 5.bound.J L i L . ' . Are there any other bound states? Hint: Solve Schrodinger's equation outside the potential for £ < 0 . < y ) i /  ^ > > £ o . it follows that V(x) = lF(x) for all x.3.2* (a) Show that for any normalized y>.1/4 Va(x) = MDr? » eax2/2 id calculate n . Show that it admits a bound state of energy M ~ma2v£/2fi2. If we denote by w> the abstract ket corresponding to y/„(x). I? Exercise 5.2. Suddenly the box spands (symmetrically) to twice its size. leaving the wave function undisturbed. Show that lie probability of finding the particle in the ground state of the new box is (8/3/r) 2 . (Hint: Expand  y> in the eigenbasis of H. or a discontinuous change of slope at x = 0.21) ie matrix elements of U{t) in the X basis are then <xjt/(0l*'> = (/(x. To show that there exists a bound state with £ < 0 . we can write the propagator as m j l h W \ ' 163 SIMPLE PROBLEMS IN ONE DIMENSION U(t)= £ \n}(n\ exp n=1 hXlmL2) (5. there exists no simple closed expression for this sum. Calculate the change in slope and equate it to ss • + « d7j/ dx dx2 (where e is infinitesimal) determined from Schrodinger's equation. Hint: Since Kis attractive. . * Consider V(x) = — aV0S(x).
J .2.2. even. (2) Equations (5.2. (5. (5.6* Square Well Potential. will have E<V0. the momentum transfer per collision.25) Verify that as V<> tends to oo. Find the force F= —dE/dL encountered when the walls are slowly pushed in. all the bound states (which alone we are interested in). Consider a particle of mass m in the state n> of a box of length L. oo and En — ti27i2n2/2mL2.2. What is E when V0 just meets this requirement? N o t e that the general result from Exercise 5. Exercise 5. imagine a circle that obeys Eq. assuming the particle remains in the nth state of the box as its size changes.2. The Continuity Equation for Probability We interrupt our discussion of onedimensional problems to get acquainted with two concepts that will be used in the subsequent discussions.23) and (5. etc. the frequency of collision on a given wall.25).2. .3.24j i where k and IK are the real and complex wave numbers inside and outside the well. we have a box.) (3) Show that there is always one even solution and that there is n o odd solution unit Vo >fi2ft2/8ma2.2. . The eigenfunctions will still be even.5 * If the box extends from x = 0 to L (instead of —L/2 to L / 2 ) show thai fn(x) = ( 2 / L ) I / 2 sin(nJix/L). the wave functions o f the lowlying levels will look like those of the particle in a box. In the ( a = k a . namely. n —1.24) must be solved graphically. (Remember a and j8 arc positive. we regain the levels in the box. Exercise 5.2b holds.164 CHAPTER 5 Exercise 5. Compare it to — dE/dL computed above. we discuss here a particle in three dimensions.2. First of all. Second. respectively] N o t e that k and K are related by k2 + K2 = 2mVQ/ft2 (5. /3 = xa] plane. Find its velocity. 5. Consider a particle in a square well potential: x <a \x\>a j Since when V0>oo.23) .4. . Consider a classical particle of energy E„ in this box.2. those of the probability current density and the continuity equation it satisfies. .2. Since the probability current concept will also be used in threedimensional problems. (1) Show that the even solutions have energies that satisfy the transcendental equation k tan ka = K while the odd ones will have energies that satisfy kcotka = K (5. let us guess what the lowering of the walls does to the states. with the obvious difference that y/ will not vanish at the walls but instead spill out with an exponential tail. and hence the average force. odd.2. The bound states are then given by thd intersection of the curve a tan a = or a cot a with the circle.
zl v(t)> dx dy dzt (y.1) 165 SIMPLE PROBLEMS IN ONE DIMENSION This is an example of a global conservation law. In quantum mechanics the quantity that is globally conserved is the total probability for finding the particle anywhere in the universe. 2><x. that is to say. t)y/(r. for it refers to the total charge in the universe. that is Q(t) = const. We get this result by expressing the invariance of the norm in the coordinate basis: since <V(0I v(t)> = < V ( 0 )  lf{t)U(t)I r ( 0 ) > = <1K0) yr(0)> then = const=<y/(OI V ( 0 > < v(0l*.2) where p and j are the charge and current densities.3. But charge is also conserved locally. such as the sudden disappearance of charge f r o m one region of space and its immediate reappearance in another. The continuity equation forbids certain processes that obey global conservation.(t)\rXr\v(t)>d'r y/*(r. We know in this case that the total charge in the universe is a constant.As a prelude to our study of the continuity equation in q u a n t u m mechanics. y. independent of time t (5. d_ dt p(r.3.3. let us recall the analogous equation f r o m electromagnetism. charge is not created or destroyed in any volume.3) This equation states that any decrease in charge in the volume V is accounted for by the flow of charge out of it. upon invoking Gauss's law. t) d3r (5.4) t The range of integration will frequently be suppressed when obvious. dt t) Vj (5. . a fact usually expressed in the form of the continuity equation dp(r. t) d3r P(r.3. y. respectively. t) d3r = — V yds (5. By integrating this equation over a volume V bounded by a surface Sv we get.
166 CHAPTER s
This global'conservation law is the analog of Eq. (5.3.1). T o get the analog of Eq. (5.3.2), we turn to the Schrodinger equation
dt and its conjugate
2m
(5.3.5)
dt
2m
(5.3.61
Note that V has to be real if H is to be Hermitian. Multiplying the first of theq equations by y/*, the second by y/, and taking the difference, we get
i*  ( ¥ ' V ) = ~
dt 2m
(V*VV~ V V V )
dt
dP Imi =  V  j dt
(5.3.
where ft j=—(y/*\y/~y/Vy/*) 2mi
(5.3.:
is the probability current density, that is to say, the probability flow per unit time per unit area perpendicular to j. To regain the global conservation law, we integrate Eq. (5.3.7) over all space: d_ dt
P(r, t) d3r=

ydS
(5.3.!
where S,rj is the sphere at infinity. For (typical) wave functions which are normalizable to unity, r3/2y/+ 0 as r>oo in order that j y/* y/r2 dr dQ is bounded, and thej surface integral of j on S ^ vanishes. The case of momentum eigenfunctions that donot vanish on S ^ is considered in one of the following exercises. Exercise 5.3.1. Consider the case where V= Vr  i V{, where the imaginary part V, is constant. Is the Hamiltonian Hermitian? Go through the derivation of the continuity equati^ and show that the total probability for finding the particle decreases exponentially e~2y',/fi. Such complex potentials are used to describe processes in which particles are absorb by a sink.
V
167
' O SIMPLE PROBLEMS IN ONE DIMENSION
Figure 5.2. The singlestep potential. The dotted line shows a more realistic potential idealized by the step, which is mathematically convenient. The total energy E and potential energy V are measured along the y axis.
v
/
0
Exercise 5.3.2. Convince yourself that if \(/ = c\jf, where c is constant (real or complex)
and f is real, the corresponding j vanishes.
Exercise 5.3.3. Consider
Find j and P and compare the relation between them to the electromagnetic equation j = pv, v being the velocity. Since p and j are constant, note that the continuity Eq. (5.3.7) is trivially satisfied. Consider y/ = A eipx/* + B e~ipx/n in one dimension. Show that j = (A 5 )p/m. The absence of cross terms between the right and leftmoving pieces in y/ allows us to associate the two parts of j with corresponding parts of y/.
Exercise 5.3.4*
2 2
Ensemble Interpretation of j
Recall that j  ^ S is the rate at which probability flows past the area d S . If we consider an ensemble of N particles all in some state y/(r, t), then N\dS particles will trigger a particle detector of area dS per second, assuming that N tends to infinity and that j is the current associated with y/(r, t).
5.4. The SingleStep Potential: A Problem in Scattering
Consider the step potential (Fig. 5.2) F(x) = 0 = V0 x<0 x>0 (region I) (region II) (5.4.1)
Such an abrupt change in potential is rather unrealistic but mathematically convenient. A more realistic transition is shown by dotted lines in the figure. Imagine now that a classical particle of energy E is shot in f r o m the left (region I) toward the step. One expects that if E> F0, the particle would climb the barrier and travel on to region II, while if E< Vo, it would get reflected. We now compare this classical situation with its q u a n t u m counterpart.
t This rather difficult section may be postponed till the reader has gone through Chapter 7 and gained more experience with the subject. It is for the reader or the instructor to decide which way to go.
168
CHAPTER 5 t =0
V o
ft
*=0 t>»a/(p0/m) Figure 5.3. A schematic description the wave function long before and lol after it hits the step. The area und IvM2 is unity. The areas under and j i//r2, respectively, are the pri abilities for reflection and transn sion.
Vo
First of all, we must consider an initial state that is compatible with quantu principles. We replace the incident particle possessing a welldefined trajectory a wave packet.! Though the detailed wave function will be seen to be irreleva in the limit we will consider, we start with a Gaussian, which is easy to han<j analytically§: y/j(x, 0) ¥l
( x ) = (KA2)'/4
eik°(x+a)
e~(x+a)2/2A2
(5.4
This packet has a mean m o m e n t u m p0 = fik0, a mean position <X') = ~a (which take to be far away f r o m the step), with uncertainties a 2
1/2
AX = 2 . / 2 '
AP =
A
We shall be interested in the case of large A, where the particle has essentially w^ defined m o m e n t u m fik0 and energy E0~ft2kl/2m. We first consider the case E{]>_ After a time t~a[pa/m]'\ the packet will hit the step and in general break into two packets: y/x, the reflected packet, and y/ T , the transmitted packet (Fig. 5.3). The area under  y/R\2 at large t is the probability of finding the particle in region I in the distant future, that is to say, the probability of reflection. Likewise the area under  y/j]2 at large t is the probability of transmission. Our problem is to calculate the reflection coefficient y/R\2 dx,
R =
•oo
(5.4.3)
and transmission
coefficient y/j\2dx,
T=
t+ co
(5.<
Generally R and Twill depend on the detailed shape of the initial wave func If, however, we go to the limit in which the initial m o m e n t u m is well defined (2
t A wave packet is any wave function with reasonably welldefined position and momentum. § This is just the wave packet in Eq. (5.1.14), displaced by an amount a.
when the Gaussian in x space has infinite width), we expect the answer to depend only on the initial energy, it being the only characteristic of the state. In the following analysis we will assume that AX = A/2 , / 2 is large and that the wave function in k space is very sharply peaked near k{). We follow the standard procedure for finding the fate of the incident wave packet, yr. Step 1: Solve for the normalized eigenfunction of the step potential Hamiltonian,
YE(X).
169
SIMPLE PROBLEMS IN ONE DIMENSION
Step 2: Find the projection a(E) = (yE\ i j / f ) . Step 3: Append to each coefficient a(E) a time dependence iE!rh and get y/(x, t) at any future time. Step 4: Identify ij/ R and y/>in y(x, t*oo) and determine R and Tusing Eqs. (5.4.3) and (5.4.4). Step 1. In region I, as V= 0, the (unnormalized) solution is the familiar one:
1/2
2mE
y/E(x) = Ae '
ik x
+ Be
ik\x
t
J
\
(5.4.5)
In region II, we simply replace E by E— V{) [see Eq. (5.2.2)], 2m(EH1
y/E(x)^Ce
ik2X
+ De'
ik2X
,
k2 =
V0)
1/2
(5.4.6)
(We consider only E> V0; the eigenfunction with E< V0 will be orthogonal to 1//7 as will be shown on the next two pages.) Of interest to us are eigenfunctions with D = 0, since we want only a transmitted (rightgoing) wave in region II, and incident plus reflected waves in region I. If we now impose the continuity of 1ff and its privative at x = 0; we get A + B^C ikx(AB) = ik2C (5.4.7) (5.4.8)
anticipation of future use, we solve these equations to express B and C in terms 'of A:
B
\kx+kJA
\EX^2+{E~
VO)1/2J
(5.4.9)
2 E 1/2 ( 2k 1 \A = \ x/2 A C= I \k\ + k2 £ + (EVo)l/2,
(5.4.10)
170
CHAPTER 4
Note that if F o = 0, 5 = 0 and C=A as expected. The solution with energy E is I B eiklX + e A C , e(x)+e,k2Xe(x)
yE(x) = A
(5.4J
where 0(x) = 1 =0 if x > 0 if x < 0
Since to each E there is a unique kx — + ( 2 m E / f i 2 ) x / 2 , we can label the eigenstatesj k\ instead of E. Eliminating k2 in favor of kx, we get
1
y/kl (x) = A
B \ exp(iki x) +— exp( — ikx x) J6( — x)
C +  exp[*(&i  2m VJfi2)' A
/2
x] 0(x)
(5.4.!
Although the overall scale factor A is generally arbitrary (and the physics depec only on B/A and C/A), here we must choose A = (27t)~ x/2 because \f/k has to ; properly normalized in the fourstep procedure outlined above. We shall ver shortly that A = (2n)~l/2 is the correct normalization factor. Step 2. Consider next
1 (2*)
1/2
f
e"lK'x + \  \ e"
*
0(—x)y//(x)
dx
I S / * O
J
e~ik2X0(x)\i/I(x)dx]
(5.4.
The second integral vanishes (to an excellent approximation) since y//(x) is nonva ishing far to the left of x = 0, while 6>(x) is nonvanishing only for x > 0 . Similar the second piece of the first integral also vanishes since y// in k space is peake around k— +k0 and is orthogonal to (leftgoing) negative momentum states. [We can ignore the 0(~x) factor in Eq. (5.4.13) since it equals 1 where i / 7 ( x ) # 0 . ] Soj
1/2
a(k 0
(2J
1/4
e*,JV/(jc) A2/2 eikxa
dx
\n
(5.4.1^
is just the Fourier transform of y//. Notice that for large A, a(k\) is very sharply peaked at kx = k0. This justifies our neglect of eigenfunctions with E<V0, for these correspond to kx not near k0. Step 3. The wave function at any future time t is
171
SIMPLE PROBLEMS IN ONE DIMENSION
y(x,t)=
a(k{)e~iE(kl)'/kii/k](x)dk,
(5.4.15)
2 \ 1/4 UTT3
exp
'ihk2^
Iexp \ 2m j
e~*,x0{x)
fakoYA
2 A 2"
exp(/'^ia)
x je'M0(x) + ^
(5.4.16)
You can convince yourself that if we set t = 0 above we regain y/i(x), which corroborates our choice A = (2n)~l/2. Step 4. Consider the first of the three terms. If 9{—x) were absent, we would be propagating the original Gaussian. After replacing x by x + a in Eq. (5.1.15), and inserting the 9(x) factor, the first term of y/(x, t) is
1/2
0(JC)JT , / 4 (A + \ m exp ik0\ x + a
iht
exp Hk0t\ 2m
—(x + a — tik0t/mf L 2A 2 (1 + ifttfmA2) = 9(x)G(~a, J k0,t) (5.4.17)
Since the Gaussian G(a, k2, t) is centered at x= —a + hk0t/m ~ fik0t/m as and 0(~x) vanishes for x > 0 , the product OG vanishes. Thus the initial packet has disappeared and in its place are the reflected and transmitted packets given by the next two terms. In the middle term if we replace B/A, which is a function of kx, by its value {B/A)0 at kx =k0 (because a{k\) is very sharply peaked at kx=k0) and pull it out of the integral, changing the dummy variable from k\ to — kx, it is easy to see that apart from the factor (B/A)09(x) u p front, the middle term represents the free propagation of a normalized Gaussian packet that was originally peaked at x = +a and began drifting to the left with mean momentum  hk{]. Thus (5.4.18)
VrR=0(x)G(a,
~k0,t)(B/A)0
172
CHAPTER 4
As t+ oo, we can set 0(—x) equal to 1, since G is centered at x = a—hk^lmK.  fikGt/m. Since the Gaussian G has unit norm, we get from Eqs. (5.4.3) and (5.4.9' B
2
R=
 y/R\ dx =
2
Et/2(E0VQ) E^2 + (E0V0)
1/2
1/2
A0
where K2kl
E0
=
2m
(5.4.191
This formula is exact only when the incident packet has a welldefined energy Eoj that is to say, when the width of the incident Gaussian tends to infinity. But it is; excellent approximation for any wave packet that is narrowly peaked in momenti space. To find T, we can try to evaluate the third piece. But there is no need to do since we know that R+T= 1 (5.4.20
which follows from the global conservation of probability. It then follows that 4Eq/2{Eq— V0)l/2
(EoVoy'2
Eo
1/2
17\V2i2 [.E0l/2+(E0V0yn
(5.4.21)
By inspecting Eqs. (5.4.19) and (5.4.21) we see that both R and T are readily expressed in terms of the ratios (B/A)0 and (C/A)0 and a kinematical factor,1 (E0— VQy/2/E^2. Is there some way by which we can directly get to Eqs. (5.4.19)j and (5.4.21), which describe the dynamic phenomenon of scattering, from Eqs.] (5.4.9) and (5.4.10), which describe the static solution to Schrodinger's equation^ Yes. : Consider the unnormalized eigenstate Vk0(x) = [A0 exp(ik0x) + C 0 exp + B0 e\p(ik0x)]9(x)
1/2
H2
)
0(x)
(5.4.2
The incoming plane wave A elkoX has a probability current associated with it equi to
jl=\M
 I
A
^
^
m
(5.4.21
while the currents associated with the reflected and transmitted pieces are *ko jR = \ B o \ * —
,2 m '
173
SIMPLE PROBLEMS IN v OONEDIMENSION
(5.4.24)
and
/
H
r
f
M m
^
(5.4.25)
(Recall Exercise 5.3.4, which provides the justification for viewing the two parts of the;'in region I as being due to the incident and reflected wave functions.) In terms of these currents
RJ*
= So A0 J'r
(5.4.26)
Co
ii A0
.2*1/2 (kg — 2m V0/ ft )
Co (EoVo) A0 Eo
1/2
1/2
k0
(5.4.27)
Let us now enquire as to why it is that R and T are calculable in these two ways. Recall that R and T were exact only for the incident packet whose momentum was well defined and equal to fik0. From Eq. (5.4.2) we see that this involves taking the width of the Gaussian to infinity. As the incident Gaussian gets wider and wider (we ignore now the A" 1 / 2 factor up front and the normalization) the following things happen: (1) It becomes impossible to say when it hits the step, for it has spread out to be a rightgoing plane wave in region I. (2) The reflected packet also gets infinitely wide and coexists with the incident one, as a leftgoing plane wave. (3) The transmitted packet becomes a plane wave with wave number (kl2mV0/ti2)l/2 in region II. In other words, the dynamic picture of an incident packet hitting the step and disintegrating into two becomes the steadystate process described by the eigenfunction Eq. (5.4.22). We cannot, however, find R and T by calculating areas under \\f/T\2 and \ y/R\2 since all the areas are infinite, the wave packets having been transformed into plane waves. We find instead that the ratios of the probability currents associated with the incident, reflected, and transmitted waves give us R and T. The equivalence between the wave packet and static descriptions that we were able to demonstrate in this simple case happens to be valid for any potential. When we come to scattering in three dimensions, we will assume that the equivalence of the two approaches holds.
Exercise 5.4.1 (Quite Hard). Evaluate the third piece in Eq. (5.4.16) and compare! resulting Twith Eq. (5.4.21). [Hint: Expand the factor (&?2mF 0 //? 2 ) 1/2 neark, ~kQ, keep just the first derivative in the Taylor series.] Before we go on to examine some of the novel features of the reflection ai transmission coefficients, let us ask how they are used in practice. Consider a gene problem with some V(x), which tends to constants V+ and V_ as x  > ± o o . simplicity we take F ± = 0. Imagine an accelerator located to the far left (x>d which shoots out a beam of nearly monoenergetic particles with ( P y — hko towa the potential. The question one asks in practice is what fraction of the particles w get transmitted and what fraction will get reflected to x = — oo, respectively. In g( eral, the question cannot be answered because we know only the mean momenta the particles and not their individual wave functions. But the preceding analyi shows that as long as the wave packets are localized sharply in momentum space, reflection and transmission probabilities (R and T) depend only on the mean momenti and not the detailed shape of the wave functions. So the answer to the question rais above is that a fraction R(k0) will get reflected and a fraction T(k0) = 1  /?(&0) w get transmitted. To find R and T we solve for the timeindependent eigenfunctio of H= T+ V with energy eigenvalue EG = fi2kl/2m, and asymptotic behavior ikox , d iko* + Be
"Ae^ Vko(x) I
X —*
CO
Jkox  Ce"
and obtain from it R=\B/A\2 and T=\C/A\2. Solutions with this asymptol behavior (namely, freeparticle behavior) will always exist provided V vanishes ra idly enough as x>oo. [Later we will see that this means  x F ( x )   > 0 as x»oo.J The general solution will also contain a piece D exp(—ik 0 x) as x»oo, but we sd D = 0 here, for if a e x p ( i k 0 x ) is to be identified with the incident wave, it must only produce a rightmoving transmitted wave Ce* 0 * as x—*oo. Let us turn to Eqs. (5.4.19) and (5.4.21) for R and T. These contain many nonclassical features. First of all we find that an incident particle with E0 > V0 gets reflected some of the time. It can also be shown that a particle with E0 > V0 incident from the right will also get reflected some of the time, contrary to classical expectations. Consider next the case E0<V0. Classically one expects the particle to be reflected at x = 0, and never to get to region II. This is not so quantum mechanically. In region II, the solution to
dx with E0 < Vo is
n
\
h2
(5.4.28
] Note that E is not restricted to be 5. in which V= V0 only between x = 0 and L (region II) and is once again zero beyond x=L (region III).v.4.29) The fact that the particle can penetrate into the classically forbidden region leads to an interesting quantum phenomenon called tunneling. where <j/ is real. no steady flow of probability current into region II.4. Standard wave theory (which we can borrow from classical electromagnetism) tells us what happens in region II: the two slits act as sources of radially outgoing waves of the same wavelength.p'. Exercise 5. (a)* Calculate R and T for scattering of a potential V(x) = V0aS(x).3. it is free once more and described by a plane wave. The impermeable screen we treat as a region with V= oo.V0 is negative. Solve for the propagator in momentum space and get U(p.) Thus there is a finite probability for finding the particle in the region where its kinetic energy Eq . Once a particle gets to region III. and hence the region of vanishing y/. This is also corroborated by the fact the reflection coefficient in this case is (E0)l/2(E0CE0) +(E0i/2 p 175 SIMPLE PROBLEMS IN ONE DIMENSION R Vo).4. we approximate by a plane wave of wave number k=p/fi. The incident particle. which must really be represented by a wave packet. Assume that the energy is positive but less than V0./2 Vo)'/2 2 ko — iK k0 + iK = 1 (5. Exercise 5.4. however. These two waves interfere on the .31) [Hint: Normalize \ftE(p) such that <££"> : d(EE'). positive. Consider a particle subject to a constant force / i n one dimension.1).2. Every once in a while an a particle manages to penetrate the barrier and come out. The rate for this process can be calculated given V0 and L. The DoubleSlit Experiment Having learned so much quantum mechanics. 5. If now a plane wave is incident on this barrier from the left with E< V0.(The growing exponential eKX does not belong to the physical Hilbert space.2. Let us label by I and II the regions to the left and right of the screen.4. since y/ n (x) = Ciy. (b) Do the same for the case V=0 for x >a and V= V0 for x <a.5.30) m(xx')2 1 +ft(x+ 2 It /•y x')— 24m (5. Consider a modification of Fig. there is an exponentially small probability for the particle to get to region III. 3. There is. it now behooves us to go back and understand the doubleslit experiment (Fig. 0) = 8{pp'ft) Transform back to coordinate space and obtain 1/2 e1 p^/bmhf (5. An example of tunneling is that of a particles trapped in the nuclei by a barrier.
Proof. dx2 d2\f/2 dx2 + Vyfi = Eyfi (5. d\j/2 dx dy/x\ dx J 0 =c (5. where y/\ and i f / 2 vanish. 5. \ i Multiply the first by y/ 2 . to get d2y/2 d2y/x . So — dyfi ~ — dyf2 Vl V2 log y/1 = log ij/ 2 + d Vi = e d y 2 (d is a constant) (5 . We now return to quantum mechanic and interpret the intensity  y/ 2 as the probability density f o r finding the particle.6 t The theorem holds even if y/ vanishes at either +oo or . Let y/i and y2 be two solutions with the same eigenvalue E: h2 2m h2 2m d2y.! It follows that c = 0." ' U s ' or d_( dx \ so that dy/2 V\— dx dyx dx . . V\y2 = Ey/2 (5. In a bound state it vanishes at bother But one can think of situations where the potential confines the wave function at one end but not other. the second by y/i and subtract.6 To find the constant c. go to x >oo.6.c o . Some Theorems Theorem 15. since they descri bound states by assumption. There is no degeneracy in onedimensional bound states.6.176 CHAPTER 5 line AB and produce the interference pattern.
Q. however.7).3.D.3). If the problem involves a magnetic field. (5.D. c. The eigenfunctions of H can always be chosen pure real in the coordinate basis. which in addition to being Hermitian. (4. What about the freeparticle case. due to nondegeneracy of bound states. If h2 d2 Ill SIMPLE PROBLEMS IN ONE DIMENSION 2m dx' then by conjugation •+V{x) y/n = Eny/n 2m dx2 V(x) y/*=En y/* Thus y/„ and yf* are eigenfunctions with the same eigenvalue. Note. the Hamiltonian is no longer real in the coordinate basis.6. as is clear from Eq.Thus the two eigenfunctions differ only by a scale factor and represent the same state. This question will be explored further at the end of Chapter 11. that while Hermiticity is preserved under a unitary change of basis. It follows that the real and imaginary parts of y/n. where to every energy there are two degenerate solutions with p = ±(2mE/fi2)1/2? The theorem doesn't apply here since y/p(x) does not vanish at spatial infinity. Proof.E.] Theorem 16. a constant . Q. y + y/* = n— —  y / r and Wr 2i are also eigenfunctions with energy E. are real. reality is not. The theorem holds in higher dimensions as well for Hamiltonians of the above form. Returning to one dimension. In this case the eigenfunctions cannot be generally chosen real. we must have y/i = cyfr. [Calculate c in Eq.E.
work with real eigeri functions with no loss of generality. except foj the h a r m o n i c oscillator. we can ignore it.. This brings us to the end of o u r study of onedimensional problems. y/= y/r+iy/i=(\ +ic)\j/r = c\j/r fl 8 Since the overall scale c is irrelevant. . i. which is the subject of Chapter 7.e.Consequently.
work in a basis and convince yourself that this step is correct.The Classical Limit It is intuitively clear that when quantum mechanics is applied to a macroscopic system it should reproduce the results of classical mechanics. We will therefore drop the third term <y/Qj y/). Qualitatively we know that the deterministic world of classical mechanics does not exist. we get and from its adjoint. when applied to slowly moving (v/c« 1) objects. When we speak of regaining classical mechanics. very much the way that relativistic dynamics. n I If you are uncomfortable differentiating bras and kets. In this chapter we examine how classical mechanics is regained from quantum mechanics in the appropriate domain. We commence by examining the time evolution of the expectation values.1) In what follows we will assume that Q has no explicit time dependence. . reproduces Newtonian dynamics. we refer to the numerical aspects. From the Schrodinger equation. our loss of innocence is permanent. We find dt = dt y> + <irt« y) + <y6 V)t (6. Once we have bitten the quantum apple.
for simplicity. />2] = P[X. Notice the structural similarity between this equation and the correspondij one from classical mechanics: do (6. (6. P]P = 2ihP so that /2m]> [from Eq. = m\¥> T which is called Ehrenfesfs theorem.5. Now [X. discuss a particle in one dimension. (1. If we consider Q = X we ge —i (6.Feeding these into Eq. F\ + [X.1) we get the relation ! < « > = ( ~ ) < v I [n. us.' If we assume H= — 2m then +V(X) <X> UJ<[X.5) .10)] m (6.: We continue our investigation to see how exactly the two mechanics are related.
5) to a more suggestive form by writing 181 THE CLASSICAL LIMIT * P_dH m dP where dH/dP is a formal derivative of H with respect to P. V(X)] we go to the X basis. (6.6) and (6. (6.5).8) le similarity between Eqs. as in this case. Since dV/dX= a/r clf/cX. dV dx ft—. We now get.6) Consider next in in To find [P.ifi — y/(x) h conclude that in the abstract. .V(X)]=ih dV — (6. in which P^iH — dx and V(X)+V(x) id for any y/(x). in the place of Eq. as long as the function being differentiated has a power series. We can convert Eq. we get (6. We would like to see how the quantum equations reduce to Hamilton's equations when applied to a macroscopic particle (of mass 1 g. (6. [P.The relation x—pjm of classical mechanics now appears as a relation among the mean values. P.8) and Hamilton's equations is rather striking.7) inhere dV/dX is again a formal derivative. The rule for finding such derivatives is just as in calculus. and X are just c numbers. calculated by pretending that H. say). dx v(x) y/(x)= . (6.
Here it 1 convenient to work in the coordinate basis where V(X)= V(x). which is the size of a proton. and F) by the functions of the mean. AP~ 10" 14 g cm/sec. namely. p 0 ) = o)(xo. we need t f approximate (8H/8X) = (dV/dX) = (V'(X)') by V'(X=xQ).. states with welldefined position and momentum Although simultaneous eigenstates of X and P do not exist. We establish this result as follows. The series is 1 V'(x) = V'(x0) + ( x . we will find in the same approximation j <C2(X. (6. Consider Eqs. and Po» since the uncertainties (fluctuations) around these values are truly negligible. To see when this is 1 good approximation.ppo) = d^(Xo. P ) } ^ ( x o .po) (6. (6. an uncertainty far below the experimentally detectable range. it is clear that we must consider an initial state that resembles th states of classical mechanics.8) which govern the evolution of <X> = x 0 and (P) = Po.10) and (6. with uncertainties AX=A and AP^fijk.) Take as a concrete example Eqs.e. There is no approximation involved in th$ first equation since < 3 H / 3 P } is just < P / m y = p o / m .po) dpo (6.10) 8H ^ (X = xo.12) Thus we regain classical physics as a good approximation whenever it is a good approximation to replace the mean of the functions 8H/3P. This in turn requires that the fluctuations about the mean have to be small. (The result is exact if there are no fluctuations. both of which are small in the macroscopic scale. such a state can be said to have welldefined values for X and P. In the second one.P=po) _dJf(xo. In the classical scale. A concrete example of such a stati is = ^ 0)2/2A2 (6. i. say.6) and (6.11). there do exist states which we can think of as approximate eigenstates of both X and P. let us expand V in a Taylor series around x0. this implies A F ~ 10" 14 cm/sec. If we let such a state evolve with time. the mean values xo(/) and p0(t) will follow Hamilton's equations.182 CHAPTER 6 First of all. In these states.xo) V(xo) + kx ~ *o)2 V"'(xo) + • . These would reduce to Hamilton's equations if we could replace the mean values of the functions on the righthand side by the functions of the mean values: • /vn / ^ ( X .po) dx0 If we consider some function of X and P. —8H/8X. P ) \ x0^<X) = ( — \ 8P / and . For a particle of mass 1 g. If we choose A ~ 1 0 ~ 1 3 c m . labeled JC0/>OAX and ( J p ) = ^ O . once again with negligible deviations. /8h\ 8H\ dP (xxo.
There is one apparent problem: although we may start the system out in a state with A~10~ 1 3 cm. a typical macroscopic particle. THE 183 CLASSICAL LIMIT .) At an intuitive level. which are corrections to the classical approximation. The classical physicist talks about making exact position measurements. is in reality an approximate eigenstate x0A)A>. each photon would carry a momentum of approximately 1 g c m / s e c and one would see macroscopic objects recoiling under their impact. we may say that these terms are negligible if the force varies very little over the "size" of the wave packet. one would soon find that the momentum of the macroscopic particle is affected by the act of measuring its position. The succeeding terms. The first term on the righthand side.) Each of these terms is a product of two factors.) In the present case. (There is no unique definition of "size. But in the present case (A F ~ 1(T 14 cm/sec) it would take 300. In summary then. the time evolution of xo and p0 will be in accordance with Hamilton's equations.000 years before the packet is even a millimeter across! (It is here that we invoke the fact that the particle is macroscopic: but for this. and thus reproduces Newton's second law. incidentally. If one really went towards the classical ideal and used photons of decreasing wavelength. F o r example. There remains yet another question. described classically as possessing a welldefined value of x and p. since the perfect measurement of velocity is merely an idealization. We see above that the uncertainty squared has to be much smaller than the inverse of the second derivative of the force.Let us now take the mean of both sides. by the time one gets to a wavelength of 10~27 cm. and besides. it exists in classical mechanics as well. represent the fact that unlike the classical particle. For example light in the visible spectrum has a wavelength of A ^ 10~5 cm and thus the minimum AX is ~ 10~5 cm. (Note. We saw that for a macroscopic particle prepared in a state . But such states do not occur in practice. But recall the arguments of Section 5. that these terms are zero if the potential is at the most quadratic in the variable x. however small. but never does so in practice. which alone we keep in our approximation.co/?0A>. one of which measures the size or nonlocality of the wave packet and the other. are these the only states one encounters in classical mechanics? What if the initial position of the macroscopic particle is fixed to an accuracy of 10~27 cm? Doesn't its velocity now have uncertainties that are classically detectable? Yes. which responds only to the force F— — V at the quantum particle responds to the force at neighboring points as well. This is clear from the fact that he uses light of a finite frequency to locate the particle's positions. corresponds to the classical force at x0. we know that with passing time the wave packet will spread. (See the third term for example. Question: While it is true that a particle in such a conveniently prepared state obeys classical mechanics. The uncertainty in the particle's position will inevitably become macroscopic. since the mean of x does. where the size of the packet is of the order of 10 13 cm. a small AP would not imply a small A F ) The problem is thus of academic interest only. while only light of infinite frequency has perfect resolution.1. the variation of the force with x. will eventually manifest itself as a giant uncertainty in position. True. We saw that the spreading of the wave packet can be attributed to the fact that any initial uncertainty in velocity." The uncertainty is one measure. it is clear that the classical approximation is good for any potential that varies appreciably only over macroscopic scales. The second vanishes in all cases. which is certainly a very small uncertainty.
the expectation value of some dependent variable P) need not. Ehrenfest's theorem does not tell us that. up to truly negligible uncertainties. I Example 6. For Hamiltonians that are at the most quadratic in X and P. In the general case. The quantum equations for the time evolution of these approximate eigenvalues and p0 reduce to Hamilton's equations. For them to obey Hamilton's equations. this replacement can be done with no error for all wave functions. (Xy = xo and </*) =p Q do not obey Hamilton's equations in all problems. o)(xQtpo). In particular. for the difference between the twtf is<X2><X>2 = (AZ)2#0. Even in those case* where xo and p^ obey classical equations. the expectation values of quantum operators evolve as do their classical counterparts. Consider where £2=X 2 . J . We conclude this chapter by repeating an earlier observation to underscore its importance.Q « X > ) ? No. in general.where A is at least 10~5 cm if visible light is used to locate the particle. The same goes for any other dynamical variable dependent on x and p. we must be able to replace the mean values (expectation values) of the functions dH/dP and dH/dX of X and P by the corresponding functions of the mean values ( X ) = x 0 and (P}=pQ. in a state given by ^(x)= A exp[ ~(x~ af/2A2]. Is <Q(X)> . unless we can replace P)} by £2(<X>.1. such a replacement is a poor approximation unless the fluctuations about the means x 0 and po are small.
quadratic in the coordinate and momentum. static equilibrium. The Hamiltonian for this system is I je=T+v= + mo) 2m 2 1 2 x2 (7. in addressing the problem of the oscillator we are actually confronting the general problem of small oscillations near equilibrium of an arbitrary system.1.1) where co = {k/m)x/2 is the classical frequency of oscillation. any system fluctuating by small amounts near a configuration of stable equilibrium may be described either by an oscillator or by a collection of decoupled harmonic oscillators. If the particle is placed at one of its minima x 0 .7 The Harmonic Oscillator 7. Consider a particle moving in a potential F(x). Since the dynamics of a collection of noninteracting oscillators is no more complicated than that of a single oscillator (apart from the obvious jVfold increase in degrees of freedom). Not only is it a system that can be exactly solved (in classical and quantum theory) and a superb pedagogical tool (which will be repeatedly exploited in this text).1. Any Hamiltonian of the above form. will be called the harmonic oscillator Hamiltonian. For small deformations x. will not interest us here. (A maximum. F= —kx. Now. the massspring system is just one among the following family of systems described by the oscillator Hamiltonian. As will be shown below. the spring will exert the force given by Hooke's law. but it is also a system of great physical relevance. A concrete example of a single harmonic oscillator is a mass m coupled to a spring of force constant k. (k being its force constant) and produce a potential V= \kx2.) Consider now the dynamics of this particle as it fluctuates by small amounts near x = The potential it experiences may be expanded in a Taylor series: V(x) = V(x0) + dV dx (xx0) + 1 d2V 2! dx (xx0 f + (7.2) . which is a point of unstable static equilibrium. Why Study the Harmonic Oscillator? In this section I will put the harmonic oscillator in its place—on a pedestal. it will remain there in a state of stable.1.
m(o2x2+— + . (It might help to refresh yoi^ memory by going back and reviewing this problem.1.1.1.) The Hamiltonian for this syste is 2 2m 2 1 2 (7. vanishes. [In other words. x is small if the neglected terms in the Taylor series are small compare to the leading term. j 2  2 ^ Jf = J f i + J f n = ~ + .186 CHAPTER 7 Now. In terms of the normal coordinates (and th corresponding momenta).x 2 ) term. which alone is retained. . since at a point of static equilibrium.] The second term in the series also vanishes since Xo is! minimum of V(x). the constant piece F(x 0 ) is of no physical consequence and may dropped.6.5b By differentiating these equations with respect to time.ma>2x2n 2m 2 2m 2 (7. since the oscillators corresponding to J#\ and J f 2 (associated with the coordinates xx and x2) are coupled by the ( x i . and hence the momenta.1.) As an example of a system described by a collection of independent oscillator^ consider the coupledmass system from Example 1. .2) readl 22 2! dx 0 2 x + 1 d3V 3! dx x3 + • • • (7.1. d2V/dx2 being identified with k=m(o (By definition.8.6 ^ Thus the problem of the two coupled masses reduces to that of two uncouple oscillators of frequencies o)\ = at = (k/m)i/2 and <yn = 3 t / 2 « = (3&/m) 1/2 . we may neglect all but the leading term and arrive at tJ potential (or Hamiltonian) in Eq. The trick is of course the introduction o normal coordinates. In the case of the massspring systen x is small as long as Hooke's law is a good approximation.1 2m = + Jt2+^mo)2(xix2)2 Now this is not of the promised form. F o r small oscillations. But we already know of an alternate description of this system in which it cai be viewed as two decoupled oscillators. (7. or equivalently. If we now shift our origin of coordinates to xo Eq. (7. the forMj —dV/dx. we get similar ones for th velocities. We exchange x\ and X2 for Xi + X2 1 \ and Xu : X[ — x2 )I/2 (7.1). we may choose V(x()) as the arbitrary reference point fd measuring the potential.
are elements of a real symmetric (Hermitian) matrix V with the following values: t \\ \=V22 = 2mo)z Vl2 = V2\ = ~ma' (7.1.1.) From the mathematical theory of Chapter 1. the expansion of V. Near an equilibrium point (chosen as the origin). This section concludes with a brief description of two important systems which are described by a collection of independent oscillators. in analogy with Eq. Thus the general problem of small fluctuations near equilibrium of an arbitrary system reduces to the study of a single harmonic oscillator. .Let us rewrite Eq. the Hamiltonian is N N dS ^ m D 1 N N (7.11) are the elements of a Hermitian matrix V.4) as 187 T H E HARMONIC z i im . . with a potential energy function F ( x j . xx). is 1 " 2 * d2V X. we are going to a basis that diagonalizes V and reduces the potential energy to a sum of decoupled terms. The first is a crystal (in three dimensions).e. dx.. d2V dxj dx/ = K' (7..Xi + . . A crystal with N() atoms (assumed to be point particles) has 3Ao degrees of freedom. xN. . Consider a system with N Cartesian degrees of freedom x . . . . the atoms in which jiggle about their mean positions on the lattice. . these being the displacements from . The second is the electromagnetic field in free space.= i J=i z £ £ (7. .1.9) For small oscillations. Now. .1. The kinetic energy piece remains decoupled in both bases. one for each normal mode.3).7) 0SCILLAT0R where V. just as the massspring system was just a representative element of a family of systems described by the oscillator Hamiltonian. a new set of coordinates ^ . .1. (7. (We are assuming for simplicity that the masses associated with all N degrees of freedom are equal. dxj (7. we know that there exists a new basis (i.1. the coupledmass system is also a special case of a family that can be described by a collection of coupled harmonic oscillators. one for each normal mode. ) which will diagonalize V and reduce to a sum of N decoupled oscillator Hamiltonians.10) j=\ i=l j= 1 ^ i= I where d V Vijdxi cx. j=.8) In switching to the normal coordinates Xi and xn (andp x a n d p { i ) .1. (7.
Hence there will exist 3JV0 normal coordinates and their conjugate momenta. the Hamiltonian will bl quadratic in the coordinates (and of course the momenta). in terms of which :W. F o r small oscillations. (7. f r o m Eq. A).2. we will have exactly 3N 0 normal modes. a(k. but the polarization has to be transverse. What are the corresponding normal modes? Recall that in the case of two coupled masses. 3). A) is like xj or x n and ©(k. and exactly out of step in the other. «(k. The coefficients of the expansion.4 t To draw a parallel with the twomass system. we arrive at the familiar equation X+oj2X = 0 with the solution x(t)~A cos cot+ B sin 0)t = xo co$(o)t + <j>) (7. t).) The modes labeled (k. . the normal modes corresponded to collective motions of the entire system. [A(r.1. A). 2. the normal coordinates.—mo) x 2 (12. (This must of course be so.2. 7. The quanting theory of the field will be discussed at length in Chapter 18. F o r a given wave vector k. the coordinate is the potential A(r.2. A) is like I or II. with the two masses in step in one case. and corresponds to plane waves travelin across the lattice. The conserved energ associated with the oscillator is 1 E=T+V=2mx2+imG)2x2=im<Q2x I (7. /) is the "velocity" corresponding to the coordinate A(r.4 By eliminating p. Review of the Classical Oscillator The equations of motion for the oscillator are.3! where x0 is the amplitude and <p the phase of oscillator.I In the case of the electromagnetic field. A). where A is the polarization index (A= 1. f) at each point in space. =£ m p = —— dx . for witl three polarizations at each k. *=^ c p ' (7. The normal frequencies are labeled <w(k. form a complete basil for expanding any state of the system.188 CHAPTER 7 equilibrium points on the lattice. in the present case.\ The normal modes are once again plane waves but with two differences: there is no restriction on k.' will be a decoupled sum over oscillator Hamiltonians. I is like {kfm)m or (3k/m) l / z . (k. Most books on solid state physics will tell yoi why there are only NQ possible values for k.1). Likewise.2. the atoms can vibrate parallel to ] (longitudinalpolarization) or in any one of the two independent directions perpendio ular to k (transverse polarization). tfo motion is collective in the normal modes.
For any  y>.Since x0 is a continuous variable. The solution in the P basis. we get the desired result. = — (Py\P\j/) 2m . . that is to say. trivially related to the solution in the X basis in this case.©V)W2 = co(xlx2)i/2 (7.5) 189 THE HARMONIC OSCILLATOR which says that the particle starts from rest at a turning point (x = ± x 0 ) . In this section and the next. picks up speed till it reaches the origin. Quantization of the Oscillator (Coordinate Basis) We now consider the quantum oscillator. With an eye on what is to follow. which in turn may be expressed in terms of the eigenvectors and eigenvalues of H.mco2X2 2m 2 ' As observed repeatedly in the past.y/> 2m 2 1 .3. (7.4) we obtain x = ( 2 E / m .2. we will solve the eigenvalue problem in the X basis and the H basis. The lowest value for E is zero.2. and corresponds to the particle remaining at rest at the origin. 1 +2 mco2{X\)/\X\ffy>0 since the norms of the states  Py/} and \Xy) cannot be negative. and slows down to rest by the time it reaches the other turning point. so that you may readily compare and contrast them with their quantum counterparts.mco\y/\XTX\v t . You are reminded of these classical results.y + l 2 mwX¥\X2\ y> ' • >t . <#> 2m 1 (y/\P2\ y. If we now set  y/') equal to any eigenstate of H.5 the passage from the H basis to the X basis will be discussed. In Section 7.„. let us first establish that the eigenvalues of H cannot be negative. 7. By solving for x in terms of E and x from Eq. a particle whose state vector  <//•> obeys the Schrodinger equation lal^/Jlvo dt with H=J?(x^X>p^P) F V P2 1 =— + . so is the energy of the classical oscillator. respectively. will be discussed in an exercise. the complete dynamics is contained in the propagator U(t).x 1 (y/\PJP\ Dv> +.
(7. ~ ~2 + ~ t \2E — / w o ) 2 * 2 W = 0 dx h\ 2 ) (7.3.3) We wish to find all solutions to this equation that lie in the physical Hilbert space (of functions normalizable to unity or the Dirac delta function).f y ^ r Tz ¥ fi ft (7.4 where b is a scale factor with units of length.Armed with the above result. W < look for a new variable y which is dimensionless and related to x by x = by (7.) We can rearrange this equation to the form d2w 2m ( 1 .3. the idea is to choos the natural length scale generated by the equation itself. we arrive at d2w dy 2 mEb2 m2a2b4 y . We begin by projecting the eigenvalue equation. Although any length b (say the radiu of the solar system) will generate a dimensionless variable y.3). (7.: .1).^ r r .3.2) (The argument of y/ and the subscript E are implicit. By feeding Eq.3.3.3. Follow the approach. ( ~ + ma}2X2)\E) \2m 2 onto the X basis.3. (7. and obtain ft2 d2 \ 1 2 2\ '•iwoVy = £y 2m dx 2 (7.4) int<j Eq. The first step is to write Eq.3. using the usual substitutions X*x dx = E\E} (7. closely—it will be invoked often in the future.3) in terms of dimensionless variables. we are now ready to attack the problem in the X basis.
we may neglect the 2sif/ term and obtain y"y2v=0 t h e solution to this equation in the same limit is K \j/ = Aym e±y2/2 (7. (7.7) g= (We may equally well choose e = 2mEb2/h2. In the present case. „ (7.3.3. Constants of order unity are not uniquely suggested by the equation.3. we will also have no feeling for the size of quantities in the relevant scale.9) k L 2 " = Jvm+2 y/" = Ay .8) at limiting values of y to learn about the solution in these limits.5) becomes y" + (2e — y2)y = 0 (7. If we insist on using the same units for all problems ranging from the atomic physics to cosmology. we will not only be dealing with extremely large or extremely small numbers.o y i \± 2m + 1 m(m — 1) y _ = y2 y y y — oc • Ay m + 2 „±y2/2 .3. In the limit ^>00.3. Not only do dimensionless variables lead to a more compact equation. Eq. which are scales generated intrinsically by the parameters entering the problem. (A distance of 1(T20 parsecs. By measuring x and E in units of i^jmcof12 and fico. they also provide the natural scales for the problem.) In terms of the dimensionless variables. we develop a feeling for what the words "small" and "large" mean: for example the displacement of the oscillator is large if y is large. is enormous if one is dealing with an atomic system.6) Let us also define a dimensionless variable e corresponding to E: mEb2 fi E =— fico . (7. our choice of e is in anticipation of the results. small on the cosmic scale.8) where the prime denotes differentiation with respect to y.3.The last terms suggests that we choose 191 THE HARMONIC OSCILLATOR (7.) The next step is to examine Eq.
8 and obtain u" — 2yur + (2e ~ l)w = 0 (7. consisted demands that we expand the cosine and sine and d r o p terms of order y2 and beyon^ We then get V > A + y—*0 cy+0()f) where c is a new constant [=J?(2e) l / 2 ].3. (7.3.. Consider next the limit. i.11) we find I C„[n(»l)/'2a/+(2el)/] = 0 n =0 Consider the first of three pieces in the above series: £ «=0 Cnn{n\)y. Equation (7.1) that indicat( that a powerseries solution is possible.3. upon dropping t h e / f term.3. for the other possibility is not a part of the physical Hiibert space since it grows exponentially as y>oo.n1 (7. and nJ some negative n. if we assume O O u(y)= I Qy" n=0 (7. [The series begins with n = 0. we pick yF e~yl/2. We therefore infer that y/ is of the form (7. Of the two possibilities ym e±y2/2.} Feeding this serif into Eq.8) becomes. T o determine u(y) completely. y/" + 2e\ff = 0 which has the solution y/ = A cosIv'Jej] + B sin[V2e_y] Since we have dropped the y2 term in the equation as being too small.e.3.11 This equation has the desired features (to be discussed in Exercise 7. we feed the above ansatz into Eq.K where u approaches A + cy (plus higher powers) as and ym (plus lower powers as 7+00.3.192 CHAPTER 7 where we have dropped all but the leading power in y as y*co. since we know that as u^A + cy+ 0{y2).3. (7.12 the equation will determine the coefficients.3.1 .
is that u{y) does not behave like yT as >>*oo (as deduced at the outset) since it contains arbitrarily high powers of y. (7. and Cn + 2 (7.Due to the « ( « .3. Feeding this equivalent series back into Eq. (2+l2g)/ (1 + 2)(1 + 1) _ (7. this series also equals n2 n = 2 193 THE HARMONIC OSCILLATOR In terms of a new variable m = n . C 7 . We have seen that as y+ oo. .2//)] .2 e ) (n + 2)(n + l) . since we saw at the outset that the oscillator eigenvalues are nonnegative.3.3. C 6 . C3.14) Since the functions y are linearly independent (you cannot express yn as a linear combination of other powers of y) each coefficient in the linear relation above must vanish. The first sign of sickness in our solution. . There is only one explanation. .2 the series becomes o o X oo Cm+2(m + 2)(m+l)y"^ S CB+2(» + 2 ) ( » + 1 ) / « =0 n=0 since m is a dummy variable.1 . the recursion relation above generates C2.15) Thus for any C0 and C\. The function w(j>) is given by n u(y) = C0 ft (12C)/ (0 + 2 ) ( 0 + l ) t (1 ~ 2s) (4+l2e) 4  (0 + 2 ) ( 0 + l ) (2 + 2 ) ( 2 + l ) ' (2+l2g) (6 + 1 —2s) 5+ + C. But we know something is wrong. (7. C3.13) we get ^ y"[Cn+ 2(n + 2)(h + 1) + C„(2e .0 n=0 (7. since s has not been constrained in any way. . .3.! ) factor.16). Eq. We thus find (2« + l . It appears as if the energy of the quantum oscillator is arbitrary. then the two possibilities for u{y) are >f or u{y) fe y—»oo .3. there are just two possibilities ¥(y)—*ytcc yme±y2/1 If we write y(y) = u(y) e y2/2 . .16) ( 1 + 2 ) ( 1 + 1) (3 + 2 ) ( 3 + l ) ^ where C0 and C\ are arbitrary. .
which is not bounded by any finite power of y as J^GO. We may explicitly verify this as follows.20) .15)] C„+ 2 C„ "00 Compare this to the growth of ym eyl. Consider the power series for u(y) as y*co.194 and CHAPTER 9 Clearly u(y) in Eq.3.. (7. Cn — coefficient of y" = 1 jk\.3. 1.3. The growth of the series is characterized by the ratio [see Eq..2k+m 2 ^ y ym ey = I n (7. with n = 2k + m or k = (n — m)/2. (7.3. we are now led to conclude that no e is allowed.18) the coefficient Cn + 2 (and others dependent on it) vanish. (7.18) tells us that energy is quantized: the only allowed values for E= eha (i.w ) / 2 ] ! Likewise Cn + 2 — so Cn + 2 r [(nm)/2]! [(n + 2m)/2]\ (nm 1 + 2)/2 n In other words.3.16). so that \i/(y)czym ey2 e~yl/2z ym e+y / 2 .16) grows as ym e^. values that yield solutions in the physical Hilbert space) are E„ = (n+^)ha). which is the rejected solution raising its ugly head! Our predicament is now reversed: from finding that every s is allowed. 1 [(»+ 2 . 2 .. If we choose C  = 0 when n is even (or C 0 = 0 when n is odd) we have a finite polynomial of order n which satisfies the differential equation and behaves as y" as >>>oo : lC1y+C3y3 + C5y5 + ' •• + Cny" J Equation (7. (7. (7. it is governed by its coefficients C„_oo as y* oo. If e is one of the special values 2n+ 1 « = 0. Since 00 . .e. « = 0. Just as the series is controlled by C 0 (the coefficient of the lowest power of y) as >>~>0.1. u{y) in Eq. .3.2. corresponds to the latter case.3. Fortunately there is a way out.
x'.27) x / f „ ( x ' ) exp[—/(«+ \/2)m(t — t')] .For each value of n. t')= £ A„ exp „=0 \ '2 2n (7.3.15) determines the corresponding polynomials of nth order. We can now express the propagator as mm 2 Itr / \ A _/ x \Hn(x)An exp 2fi / M C 0 U(x. (7.3. (7.21) 195 THE HARMONIC OSCILLATOR Hi(y) = . called Hermite polynomials.3.24) (7.3.3.7ifi2 (n\) = Vn(x) \ 1/4 E X / / P max R ^ R R ( mm R . The normalized solutions are then yE(x) = y/in+]/2)>i mm . The following recursion relations among Hermite polynomials are very useful: H'„(y) = 2nHni Hn+](y) as is the integral Snn{x'/22"n!) = 2yH„2nHni (7.25) Hn(y)H.3.3.22) The derivation of the normalization constant mm lnK22n(n\Y] »/4 An = (7.3.\ 2 ( y ~ j / ) HA(y) = \2(\Ay2 + b4) The arbitrary initial coefficients C 0 and Ci in Hn are chosen according to a standard convention. H„(y): H»{y) = \ Hx(y) = 2y H2(y) =2(12y2) (7.(y) dy = (7.23) is rather tedious and will not be discussed here in view of a shortcut to be discussed in the next section.26) which is just the orthonormality condition of the eigenfunctions ys„(x) and y/„ (x) written in terms of y = (mm/fi)]/2x. t. Eq.
C„. show that \f/(x)<j>(x) dx = 0 " —0 0 Use this to show that i/2(x) and i//i(x) are orthogonal. (7.25).3. 2yu'.3. Write y/ as a product of the asymptotic form and an unknown function The function u will usually be easier to find than <//.3. Using the values of Gaussian integrals in Appendix A.2 verify that y2(x) and yo(*) are orthogonal. Step 3. Step 4. Verify that H^y) and HA(y) tj>(x) obey the recursion relation. 0 = exp imoj (x2 + x'2) cos Q) T— 2xx' H 2 sin coT (7.3. of the three pieces u.8).3 show that <«Pn> = 0 and thus that <^2> = (AA:)2 and </ .11) but not Eq.: where T=tt'. t.3.3.3. Step 1. If y/(x) is even and •0 0 is odd under x>x. Eq.1.3.3. * Using the symmetry arguments from Exercise 7.5. Show th <1X2 1) = Vajlmco and { \ \ P 2 \ \ ) = lmcoH.3.11) on the other hand. Exercise 7. behavior of y/. Using Eqs.2.1 )u. (7. and ( 2 e . (7. while the piece has two more powers of y. Introduce dimensionless variables natural to the problem. (7. the last two have the same powers of y. from which the solution does not follow so readily. devoted to path integral formalism. show that J/2 \2m(o) mojti 1/2 <n'\P\ny = \ Exercise 7.15).2 > = (AP)2 in these states. By feeding in a series into the latter.196 and CHAPTER 9 Evaluation of this sum is a highly formidable task. Extract the asymptotic (y>oo. . Inifi sin o Tj 1/2 U{x. Exercise 7. Exercise 7. and C„_2 obtains.3. Show that y/a(x) saturates the uncertainty bou AXAP>:f\/2. x'.* Consider the question why we tried a powerseries solution for Eq.15). Exercise 7. In Eq. Try a power series to see if it will yield a recursion relation of the form Eq. (7.3. The result happens to be mm .4.23)(7.3. Step 2.3. We will not attempt it here sii we will find an extremely simple way for calculating U in Chapter 8. verify that a threeterm recursion relation between C„+2. This concludes the solution of the eigenvalue problem.3. Before analyzing oj results let us recapitulate our strategy. (7. The problem is that y/" has two powers of y less than ley/.
) (2) The levels are spaced uniformly.6. We have here a special case of the correspondence principle.Exercise 7. (We know vaguely that when a system is big. . which states that as the quantum number tends to infinity. for example. The fact that the oscillator energy levels go up in steps of fia allows one to construct the following picture. We pretend that associated with an oscillator of classical frequency © there exist fictitious particles called quanta each endowed with energy fica. with an amplitude of 1 cm.20) as the energy of n such quanta. We now discuss the eigenvalues and eigenfunctions of the oscillator. Note that the quantization emerges only after we supplement Schrodinger's equation with the requirement that y/ be an element of the physical Hiibert space.2 7 erg apart. The correspondence principle tells us that the quantum number is a good measure of bigness. oscillating at a frequency of 1 rad/sec. which is what obtained for all but the special values of E\. (7. In this case it meant the imposition of the boundary condition y/(x>oo)*0 [as opposed to ^(x >oo)»oo. we regain the classical picture. we say that An quanta have been created (or destroyed). (7.3. the quantum number associated with this oscillator is while the difference in n between adjacent levels is unity. * Consider a particle in a potential F(x) = i m a V . The following are the main features: (1) The energy is quantized. we forget about the mass and spring and think in terms of the quanta. We view the nUa piece in the energy formula Eq.3. the quantum oscillator has a discrete set of levels given by Eq. a mass of 2 g.20). Consider.3. In other words. it may be described classically. Why does the classical oscillator seem to have a continuum of energy values? The answer has to do with the relative sizes of the energy gap and the total energy of the classical oscillator. = 00. Its energy is E= \ma)2xl= 1 erg Compare this to the gap between allowed energies: AE= tw3 ~ 10^27 erg At the macroscopic level. Stated differently. it is practically impossible to distinguish between a system whose energy is continuous and one whose allowed energy levels are spaced 10 . When the quantum number n goes up (or down) by An. In contrast to the classical oscillator whose energy is continuous. X>0 x<0 197 THE HARMONIC OSCILLATOR What are the boundary conditions on the wave functions now? Find the eigenvalues and eigenfunctions.
thinking of the oscillator in terms o f these quanta has proven very useful. respectively.31T . since all these problems are mathematically identical and differ only in what the coordinate and its conjugate momentum represent. labeled by the 3N0 values o f (k. X).p = 0 ) is precluded by the canonical commutation relation [X. X). say to take into account nonleading terms in the Taylor expansion of the potential. This result is common to all oscillators. In the case of the crystal. X) even when phonons are absent. (3) The lowest possible energy is fito/2 and not 0. We start with a normalized trial state  y/> and consider m I mco\X2) 2 (7. P] = ih. What we need then is a compromise <//min(x) that minimizes the total mean energy without violating the uncertainty principle. or the interaction between the crystal and some external probe such as an electron shot at it. X). which is discussed in Chapter 18. If we choose a wave function y/(x) that is sharply peaked near x = 0 to minimize the mean potential energy <5m(o 2 X 2 >. If. which will be discussed in Chapter 18. For a crystal whose Hamiltonian is exactly given by a sum of oscillator pieces.3. called phonons.Although it seems like a matter of semantics. X) in each mode (k. J S 2m Now (A P)2=(P2y(P)2 + (7. the wave function in P space spreads out and the mean kinetic energy (P2/2m) grows.) In the following discussion let us restrict ourselves to the mechanical oscillator and examine more closely the zeropoint energy. Such a state. (The zeropoint fluctuation of the field has measurable consequences. is forbidden by the uncertainty principle. there are 3JV0 oscillators.3 (AX)2=(X2)(X)2 (7. a crystal has an energy {ha>(k. is a reflection of the fact that the simultaneous eigenstate \ x = 0.3.3. with M= A/ > =0. Unlike the classical oscillator. the phonon concept proves very useful. Thus.) Similarly. the interaction of the electromagnetic field with matter may be viewed as the interaction between light quanta or photons and matter. This energy. however. The quantum state of the crystal is specified by giving the number of quanta. which can be in a state of zero energy (with x=p = 0) the quantum oscillator has a minimum energy of ha/2. (The two effects mentioned above may be seen as phonon phonon interactions and phononelectron interactions. the introduction of the phonon concept is indeed a matter of semantics. Let us therefore try to find a state that is quantum mechanically allowed and comes as close as possible (in terms of its energy) to the classical state x=p = 0. and the electromagnetic field has an energy X) in each mode of frequency a even when photons are absent. The converse happens if we pick a momentum space wave function sharply peaked near p = 0. with frequencies to(k. at each (k. Let us now begin our quest for y m m(x). whether they describe a mechanical system or a normal mode of the electromagnetic field. we consider deviations from this. We saw that it is the absence of the state jc = 0. called the zeropoint energy. p = 0) that is responsible for this energy.
37) . (7. Finally we choose.3.34) where the equality sign holds only for a Gaussian.) For these states (from which we must pick the winner) <//> 2m + . as will be shown in Section 9. for which <//>Gaussian = — o«7(AA) ( 2 7 . (Since <X> and <P> are independent of each other and of ( A ^ ) 2 and (AP) 2 . the one with the AX that minimizes <//>Gaussian • By requiring ^K^Ocaussian d(AX) we obtain {AXf=h/2mco and (H)min = ho)/2 (7.mco2(AX)2 2 (7. such a choice is always possible.3. 3 .33) Now we use the uncertainty relation AXAP>h/2 (7.3.3.3.3. the Gaussian has the lowest energy.32) OSCILLATOR The first obvious step in minimizing <i/> is to restrict ourselves to states with <X> = <?) = 0. 3 6 ) What we have found is that the mean energy associated with the trial wave function is sensitive only to the corresponding AX and that.so that (A P ) + ( P y 2 2 199 THE HARMONIC < = —yLL 2m + _ m(02[(AX 2 1 )2 + <X> 2 ] . We get • W ^ g ism(AX) + 2 t7'3'35) We minimize < # ) by choosing a Gaussian wave function. from the family of Gaussians.39) (7. of all functions with the same AX.3.38) 2 =A = 0 ~% 8m(AXy 1 + mco 2 2 /TO itv (7.
we have found that the folio normalized function has the lowest mean energy: (mcu V/4 r . If we apply the above result (VmJHly/^) to  v ) = <(y/\H\y) (for all  y > ) I Vo) = groundstate vector.0\H\y0)<(y.3.^ j f mcox2\ fid) / rrN <*>.2 We have thus managed to find the oscillator groundstate energy and state ve without solving the Schrodinger equation.=T (7.3. \ Also. In more typical instances. As for the wave function. we can only get an estimate for AX. Our ability here to obtain all the information about the ground state using the uncertainty relation is a consequence of the special form of the oscillator Hamiltonian [which allowed us to write < / f ) in terms of (AX) 2 and (AP) 2 ] and the fact that its groundstate wave function is a Gaussian (which has a privileged role with respect to the uncertainty relation). the use of the uncertainty relation will have to be accompanied by some handwaving [before <//) can be approximated by a function of (AX) 2 and (AP) 2 ] and then too will yield only an estimate for the groundstate energy. .200 and CHAPTER 9 Thus.3 forally> £o=<Vol#!Vo> = < Vminl H\ W i n ) = ~ (7. It would be a serious pedagogical omission if it were not emphasized at juncture that the uncertainty relation has been unusually successful in the above context. it follows I V0> = I Vmin) (7.42) that y/min) (7. since there was only one state.3. with energy fico/2. we get <Vmin# Vmin><<>ol#l Vo) = E0 (7. (7.2: EQ=(y.4! Now compare this with the result of Exercise 5.41) and (7.\H\¥) If we set  y ) =  y/ min ) we get E0 = iy/0\H\y/0}<<ty/min\H\ It follows from Eq. by systematically hunting in Hilbert space.2.3.3.  Vmin). the spread associated with it.M .
] Notice.2) the excursions outside the turning points are small compared to the classical amplitude. 2 stand for the classical turning l/1 points.1. The small arrows at \y[ = {2n + \ ) t .  y/(x)\2 seems to go just the other way (Fig. n even n odd In Chapter 11 on symmetries it will be shown that the eigenfunctions had to have this property. however.2. The position of a given classical oscillator is of course exactly known. that when n is large (Fig. 7. Consequently the eigenfunctions are even or odd: y/„(—x) = y/„(x).1.4. The correspondence principle. 7. where am I likely to catch it? If the velocity of the oscillator at a point x is the time it spends near the x. however. for quantum mechanics is expected to differ from classical mechanics.3. depending on whether n is even or odd. There is no contradiction here.1) Y/„(X) contain only even or odd powers of JC. Normalized eigenfunctions for n = 0. but dies out exponentially as [Verify that the classical turning points are given I/2 by)>o=±(2w + l) . 4 Figure 7.1). (6) The probability distribution P(x) is very different from the classical case. This exponentially damped amplitude in the classically forbidden region was previously encountered in Chapter 5 when we studied tunneling. for the ground state in particular. varies inversely with V(JC) : Pci(x)cc 1 t>(x) 1 — x . and hence the probability of our catching it there during a random spot check. 4 t 1 2 A0 —II / I1 \ / 1 2 i t V . r (4) The solutions (Fig. (5) The wave function does not vanish beyond the classical turning points. Recall that y = (ma)/fi) x. But we could ask the following probabilistic question: if I suddenly walk into a room containing the oscillator. and 3. tells us that for large n .45) which is peaked near ±JC0 and has a minimum at the origin. In the quantum case. = ¥n{x).201 THE HARMONIC OSCILLATOR i 4 2 —i—r* ) 2 *V .2\l/2 ) (7. 7.
* The Oscillator in Momentum Space.4.202 and lf„(y)l CHAPTER 9 4 .2. we can see how the classical limit is reached: the quantum distribution P(x) =  wiggles so rapidly (in a scale set by the classical amplitude) that only its mean can be detected at these scales. Note however that what I am calling nonexistent is an oscillator that actually has the properties attributed to it in classical mechanics. (7.1/4 —jpjlmfi o > There are several other pairs. which shows the situations at n = l l . which are related by the substitution mco 1 /ma. 1 Vo(/>) = mnfi(o. can shoot past the "classical" turning points.) 7.2 0 Figure 7. and have a zeropoint energy of k h a even while they play dead. whose mean is measured in practice and agrees with the classical probability curve. (Refer back to Exercise 7. Exercise 7. Thus. The Oscillator in the Energy Basis Let us orient ourselves by recalling how the eigenvalue equation p 2 J + ma2X2 2m 2 J )\E> = E\E> . From Fig. mco^>\ /mco. You may wish to watch out for them.5. A remark that was made in more general terms in Chapter 6: the classical oscillator that we often refer to. for example. and this agrees with Pci(x). and not one that seems to have them when examined at the macroscopic level.7. In other words. .3. show that the momentum space eigenfunctions may be obtained from the ones in coordinate space through the substitution x~*p.2.2).3. all oscillators. is a figment lodged in our imagination and doesn't exist.3. Probability density in the state n=j The broken curve gives the classical probabi] distribution in a state with the same energy. By setting up an eigenvalue equation for the oscillator in the P basis and comparing it to Eq. the two must become indistinguishable. and thus have discrete energies. 7. including the 2g mass and spring system. are ultimately governed by the laws of quantum mechanics. such as AX and AP in the state n>. We are reminded here of the doubleslit experiment performed with macroscopic particles: there is a dense interference pattern.
which allows us to work in the energy basis without having to know ahead of time the operators X and P in this basis.P] = ifiI=ift (7. The next few steps will seem rather mysterious and will not fit into any of the familiar schemes discussed so far. we first compute the X and P operators in this basis. Now suppose that we want to work in the energy basis.4. Let us first introduce the operator a J ^ T x J . F o r instance. To solve the problem in the momentum basis. i.2) which follows from X~+x. P^—iH d/dx. given their form in the coordinate basis.4. But there is a clever way due to Dirac. (7. will then become a differential equation that we will proceed to solve. (2) We solved for the components <x  E) = y/E(x) and the eigenvalues.l \ 2n \2mcon.4) (Note that m a ^ l / m a as X <*• P. We must first find the eigenfunctions of H. All we will need is the commutation relation I [X. You must be patient till they begin to pay off. P (7. so that we can carry out the change of basis.1). but is basis independent.e. P) in this basis.4. P+ ifi d/dx.ip'x/fi xS(xx') (2>r£)1/2 (given) ift8'(pp') <x'p) dxdx' ipx'JH (2 jtH)1/2 = We then find P and H(X.) They satisfy the commutation relation (which you should verify) [a. there is not much point in setting up the problem in the E basis.was solved in the coordinate basis: (1) We made the assignments X~*x.4. <*is>.4./ ..M P V2man (7. 203 T H £ HARMONIC OSCILLATOR <P'\X\P> = <p'\x} <x\X\x') . J] = 1 (7.5) . The eigenvalue equation.3) and its adjoint / flt= \1/2 ^ \ 2fi / / 1 \1/2 X . Once we have done this. But finding < x \ E } = y/E(x) amounts to solving the full eigenvalue problem in the coordinate basis.
: no) 1/2) (7. We wish to solve the eigenva equation for H : H\ ey = e\ ey (7.4. a]a] =a (1. 1 . whose eigenvajues e measure energy in units of fico. H] = .] Let us next define an operator \/2)ft(o (7. a a+ 1/2] = [a.1 The utility of a and af stems from the fact that given an eigenstate of H. H] = [a.a + (74.4.1 .4.4 and [a1.204 CHAPTER 7 Note next that the Hermitian operator afa is simply related to H: + mco 2. = = H])e> (aHa)\ey (e\)a\ey (7. i a*a = X + P2 + — [X.4.6) comes from the noncommutative nati H. Two relations we will use shortly i A t t [a. Consider Ha\£y^(aH[a. where s is the energy measured in units of fico. P] 2H Imcafi 2 ft = 1 2 ~Hca so that H=(afa+ [This method is often called expressing H — P2+X2 (ignoring (XiP) = The extra fico/2 in Eq. f the "method of factorization" since we a constants) as a product of ( X + i P ) ~ a a (7. they genera others. of X and P.
4.7)] £ q ~ 2 (7. They are also called destruction and creation operators since they destroy or create quanta of energy fico.11) that ae> is an eigenstate with eigenvalue e— 1. .4.14) One refers to a and a as lowering and raising operators for obvious reasons. a\e> = Cs\€\> where Ce is a constant.15) [from Eq.4.4. . £ . So. .e. it must be that the downward chain breaks at some point: there must be a state ] e 0 ) that cannot be lowered further: a\ £o> = 0 Operating with a . 6 + 2 .oo. .16) fe are using the fact that there is no degeneracy in one dimension. We are thus led to conclude that if e is an eigenvalue of H. . so are g+1.13) (7. and . we get a a\ £o) = 0 or ( H . /7])e> = (atH+at)\e) (7. e + 3. (7.We infer from Eq. .4. (7. .12) 205 THE HARMONIC OSCILLATOR = (f+l)a+le> so that at\e) = C£ + l\s+\) (7. and \ e— 1) and \s) are normalized eigenkets..1 /2) £0> = 0 or H\£O> = \\£O) (7.4.t Similarly we see that Ha*\ e ) = (a^H— [</. £ + oo. i. The latter conclusion is in conflict with the result that the eigenvalues of H are nonnegative. .4.
14).. (7.12) (7. Co quently it follows from Eqs.4. Are these the only levels? If there were another family. We want to detei j j r the constant C„ appearing in the equation a\n} = C„\n~ 1) Consider the adjoint of this equation <«!</= <« — 1 j C* By combining these equations we arrive at {n\afa\n} = (n — 1  « .4.1.1 (7. it too would have to have ground state  £o> such that a\ £o> = 0 or or (7.206 CHAPTER 7 We may.2.4.4.4. Since e — n+ 1/2.13 C„ = (n)1/2 ei4f> is arbitrary) .18) that je 0 ) and £o> represent same state. n = 0..2.4.1' n = 0. (7. We now calculate the constants Cs and CE+i appearing in Eqs.. or En= («+ \/2)ft(Q.' (7.I)C*C„ i\n) = C*C„ (n\n\ny =  C„2 (since \rt— 1) is normalized) (since #  « > = ( « + 1 / 2 )  n > ) (7.4. let us label the kets by the integer n..1.4. The same goes for the families built from  s 0 ) and  eo) by the re action of a*.. (7. however. raise the state 1 e 0 ) indefinitely by the repeated application < We thus find that the oscillator has a sequence of levels given by £„ = ( « + 1 / 2 ) .16) and (7. lj But we know that there is no degeneracy in one dimension (Theorem 15).
4. we invert Eqs.21) and (7. So we have a\ri) = nl/2\n— similarly be shown (by you) that aV> = («+i)I/2i«+i> (7.4.4.] By combining these two equations we find afa\ n> = « V / 2  n .4) to obtain / X= ^ \l/2 (a + a (7.26) (7.26) and (7.25) (7.It is conventional to choose 0 as zero.28) —\2ma / (mafC\/2.4.4.21) 207 THE HARMONIC OSCILLATOR [Note that in Eqs.24) = n\n) (7.n To find the matrix elements of X and P.27) !> = («+ 1 )i/28„. The details are left as an exercise.4.22) are very important.4.4. (7.4.30) 0 .4. „ t (7.4.4. They allow us to compute 'the matrix elements of all operators in the  n ) basis.3) and (7.4.4. The two basic matrices in this energy basis are n = 0n=1 n=0 «= 1 a n=2 0 1 1/2 n=2 0 0 0 0 0 0 2 1/2 0 3 1/2 (7. First consider a and af themselves: (n'\a\n) = nx/2(ri\n <«'aV> = (n+ l)]/\n'\n+ ~ 1> = nx/2dn>.29) and then use Eqs.1 > = n1/2ni/2\u> terms of N = afa led the number operator (since it counts the quanta) H~N+ \ (7.22) the larger of the ris labeling the two kets appears under the square root.21) and (7. (7.27).4.23) Equations (7.22) 1) (7.4. (7.4. +l (7.4.
(7.4. Consider. of the ground state 0>: i \ n i i\ ri'1  a ( « .4. <3.28) and (7.29): 0 1 1 / 2 0 2 1 / 2 0 0 31/2 0 X *) * Jl/2 \ 1 / 2 0 2 0 0 l/2 3 0 . for example.2mca 7 (7.4.InV2 ) l«2> 1/2 0> (7. .27) or Eq (7.21) and (7.22) combined with our mnemonic involving images of the tran formed vectors a\ri) and a\n>.4./2 3 1 / 2 (7.4.4.2 The a and af operators greatly facilitate the calculation of the matrix of elements i other operators between oscillator eigenstates.22) also allows us to express all normalized eigenvectors w) in te./2 .4.3< Equation (7.4.Y32).208 and and its adjoint !V2 0 CHAPTER 9 0 0 0 2 l / 2 0 0 0 0 3 0 i/2 Both matrices can be constructed either from Eqs.3! 0 0 jl/2 1 1 / 2 0 —21/2 0 0 'rri ( mcofi V / 2 0 0 0 2 l/2 0 3 0 . (7.26) and (7.3! 0 The Hamiltonian is of course diagonal in its own basis: 1/2 H fia 0 0 0 3/2 0 0 0 5/2 0 0 (7. We get the matrices representingX and P by turnii to Eqs.4.4.4.
32)] and evaluate the inner product <xw>.4. P] = in.1/2/ . and a ad. but is true in all bases. Consider a remarkable feature of the above solution to the eigenvalue problem of H. namely. P*—iH d/dx). the P basis recommends itself. and then we use the Fouriertransformed version of Postulate II.2ma [2(3 1/2 ) + 4(3 1 / 2 ) + 3(3 i / 2 )] What if we want not some matrix element of X. the only nonzero contribution comes from dda. which tells us to find the eigenvectors x> of the matrix X [Eq. which gives the action of X and P in the X basis (X+x. P+p. for to do so we first had to solve for the energy eigenfunctions in the X basis. Usually we work in the X basis and set up the eigenvalue problem (as a differential equation) by invoking Postulate II. which was begging the question. . add. which follows from Postulate II. Instead we used just the commutation relation [X. In the present case we could not transform this operator assignment to the energy eigenbasis. V2 / 209 THE HARMONIC OSCILLATOR Mir in \Ho / max ma) 1/2 x exp max ma 1/2  I X dx whereas in the n> basis 3/2 <3X 2> =  3 2 ma 3/2 o\(a+ay\2> . (7. in particular the energy basis. Since we obtained the complete . In some cases (the linear potential problem). \ 1/2 /.t„t < 31 (a* + aV + ad a + ad a \ 2 maj + daa + o W + dda + d d d ) j 2) Since a lowers n by one unit and d raises it by one unit and we want to go up b 7 one unit from « = 2 to n = 3. Now dda\iy ««V2> = 2 1 / 2 a V  1 > = 2 1 / 2 2 1 / V  2> = 2 1 / 2 2 1 / 2 3 1 / 2 ) 3> = 3 1 / W  3 > = 3 1 / 2 4 1 / 2 a4> = 3 1/2 4 1/2 4 1/2 13> « W [ 2 > = 3 1 / V a  3 > = 3 1 / 2 iV!3>3 1 / 2 33> so that 3/2 <3*32>' . X*in d/dp. but the probability of finding the particle in n> at position x l We can of course fall back on Postulate III. A more practical way will be developed in the next section. 3 3! 22 2!.the X basis one would have to carry out the following integral: .co f 1 1 \ n f i j .
4. we go to the X basis. we may trade ( present Postulate II for a more general version: Postulate II. F o r example if we project the eigenvalue equation P\P>=P\P> onto the X basis. W h a t we do measure are probabilities <G)^> 2 for obtaining some resu cd when Q is measured. we meet this requirement and arrive at Postulate II as stated earlier.3d is equally satisfactory.210 CHAPTER 7 solution given just this information. This in fact is the case. The independent variables x and p of classical mechanics n become Hermitian operators X and P defined by the canonical commuta [X. (7. the present version of Postulate II allows us some latitude in the choice of P.3i • . or the eigenvah spectrum of operators such as the Hamiltonian. Now.4.P] = iH of the q u a n t u m operators and the Poisson brackets (PB) of their class counterparts {x.1 dx \f/p(x) =pyp(x) (7. squares of matrix elements  < ^ I  Q  I / 2 )  2 . How can physics be the same as before? The answer is that the wave function is never measure directly. Clearly in its own basis We must then pick P such that [ X . Dependent variables co(x. it would appear that the essence of Postulate is just the commutator. P] = ifi.36j Dirac emphasized the close connection between the commutation rule [X./ « — + / ( * ) X basis dx d (7. we now get (7. it is not at all obvious that in every problem (and 1 just the harmonic oscillator) the same physics will obtain if we make this our start] point.4. P] = iH. If we m a k e the conventional choice P—ifi d/dx.3 f r o m which it follows that yp(x) is no longer a plane wave cceipx/fi.p) are given by operators Q: T o regain our old version.4. In one of the exercises that follow you will be guided toward the proof that these measurable quantities are in fact le invariant under the change to the nontraditional operator assignment Eq.4./>} = ! . we can add to —ifi d/dx any function of x without altering the commutator: assignment XX basis x (7. In other words.
for the same reason as in the N = 1 case.10).p)} = Y(x. P] = iH{x.4.4..which allows us to write the defining relation of the quantum operators as [X.PN obeying the commutation rules [X. Eq.. is a choice but not the choice satisfying the canonical commutation rules. 0 (7. Once the differential equations are obtained.p+P)=Q.2. .. Xi • x. (7.. . Xj] = iH{xi. Px. A F ] = A[Q..39). it turns out that if {fi>(x. XN.. A(x. see Exercise (7. pj} = iHdij [Xi...p) = ifi (7. the ones among dependent operators follow from the repeated use of the relations [ « . X basis I Pi X basis •/fi^OX. Given the commutation relations between X and P.iti{xi. we may abandon Cartesian coordinates in looking for the solutions.p) . The Cartesian coordinates xx.. p N of the classical description of a system with N degrees of freedom now become Hermitian operators X\. .0 [Pi.Pj] = iH{Pl.pj}= Similarly (o(x. .38) 211 THE HARMONIC OSCILLATOR The virtue of this viewpoint is that its generalization to the "quantization" of a system of N degrees of freedom is apparent: Postulate II (For N Degrees of Freedom). .4. Pj] .p)...39) [We restrict ourselves to Cartesian coordinates to avoid certain subtleties associated with the quantization of nonCartesian but canonical coordinates. namely. T] + [Q.•} .T]A Since PB obey similar rules (Exercise 2.1) except for the lack of emphasis on ordering of the classical variables. A]T and [QA. *.T] = Q[A.p)^o)(x+X.] It is evident that the generalization provided towards the end of Section 4. r] + [Q.xN and momenta p \ .7.4.
* Compute the matrix elements of X and P in the  n ) basis and comj with the result from Exercise 7.4. ( Exercise 7. that] to say for "quantizing. hence the formal similaii between classical and quantum mechanics.3. Using = results from the previous exercise show that for the oscillator (k . <*(0>. {2)Q=X2.2* Find <X>. <X2>.* (Virial Theorem). P)] = iHT(X. P) OM except for differences arising from ordering ambiguities. Show that c(k) = k/2 by considering a circular orbit.5* At r = 0 a particle starts out in y(0)>= 1/2 1/2 (0> +11».6* Show that <a(0> = e~ia" (a( 0)> and that <flT(/)) = e'a"<«t(0)>. (3) find (X(t)} and (P(t)] using Ehrenfest's theorem and solve for (X{t)') and <P(/)) and compare with part (2).AP in the state \n). (7. Although the new form of postulate II provides a general. </>(0)>.40) for the case (1)Q=Z. Exercise 7. Exercise 7.7. given just the commutation relations (and a little help fro Dirac) is atypical.4.212 CHAPTER 7 then [Q(X. </>2>.4. P). Verify Eq.4." in practice one typically works in the X basis and ignores the latitude in the choice of Pi and sticks to the traditional one.4.2) <r>=<r> in the quantum state n>. The solution to oscillator problem. Exercise 7.4. Exercise 7. Exercise 7.4.1. the average (over the orbit) kinetic and pot energies are related by T=c(k)V when c(k) depends only on k. Exercise 7. A~X2 + P2 A = P2 The second case illustrates the ordering ambiguity. The virial theorem in classical mechanics states for a particle bound by a potential V(r) = ark. </>>. first encountered in Chapter 6.4. Show that (n\X4\n} = (ti/2mco)2[3 + 6n(n+ 1)]. AX .3. (2) find ( x m = wmx\¥m. .4. which leads to the simplest differential equations. basisindepend^ specification of the quantum operators corresponding to classical variables.4. (1) Fill I v(0>. A(X. P \ —ifid/dxj. <P(0>.
Ly.9 (Important). Consider the unconventional (but fully acceptable) operator choice d p^m—+f(x) dx in the X basis. and note that there are no ordering ambiguities.rje'>=jt5(jtjc') new X basis Next verify that (x\p\x')= ii\~+f(x) S(xx') m . and Lz. Ly] = ifiLz. (3) Verify that [Lx.4.3) for the definition of the PB].7. (1) Verify that the canonical commutation relation is satisfied. (2.Exercise 7. Exercise 7.8 * Consider the three angular momentum variables in classical mechanics : lx=ypzzply = Zpx~Xp. the quantum counterparts. l y }=h [see Eq. (2) Verify that {lx. (2) It is possible to interpret the change in the operator assignment as a result of a unitary change of the X basis: <g(X)?f> I y\ Jgix)/* X> where g(x)= f{x')dx' First verify that <jc. 213 THE HARMONIC OSCILLATOR h= xpyypx (1) Construct Lx.4.
pN. Since the problem has rotational symmetry we use polar coordinates p = (x2+y2y/2..2).e. since they bofl represent the same abstract operator Cl(X. In this section we revisit this problem in order to explain some of the subtleties arising in the direct quantization of nonCartesian coordinates without the use of Cartesian coordinates in intermediate stages.pN to operators obeying the canonical comma tation rules. Since d presence of / is related to a change of basis.pj+—ifi d/dqt) does not generate the correct Hamiltonian H.. ..4.. . ^ Section 4. . —ifi d/dx) and (o(x. given a basis x). .10* Recall that we always quantize a system by promoting the Cartesi® coordinates X\.4. —ifi d/dx+f) can have the san eigenvalues and a onetoone correspondence between their eigenfunctions. J4?(qi+qf. we first set up the different® equation in Cartesian coordinates and then change to spherical Coordinates (Section 4. If nonCartesian coordinates seem more natural in some cases.41) basis Since p and <fi are not mixed up as x and y are [in the (x2 + y1)17 2 term] the polar version can be more readily solved. xN\ and momenta p\. P +ifift new X basis dx +f { x ) This exercise teaches us that the "X basis" is not unique. • Exercise 7. the invariance of the physics under a change • / ( f r o m zero to nonzero) follows. such as til eigenvalue problem of a Hamiltonian with spherical symmetry. by multiplying by a phase factor which changes neither the norm nor the orthogonality The matrix elements of P change with f the standard choice corresponding t o / = 0 . What is novel here is that we are changing from one X baa to another X basis rather than to some other Q basis. in terms of which (j) = tan ~l(y/x) (7. Another lesson to remember is thl two different differential operators co(x.. . P). .214 CHAPTER 7 i. .. (1) Consider a particle in two dimensions with 2m which leads to in the coordinate basis.2 it was pointed out that if & is written in terms of nonCartesian but canonical coordinates q\ . . we can get anothfl x>. even though the operator assignment satisfies the canonical commutation rules.qN\ px .
+ We know Eq.41) and (7.42) which disagrees with Eq. . and hence the H constructed with it.4. Now this in itself is not serious.41).f The question we address is the following: why not start with 34? expressed in terms of polar coordinates and the conjugate momenta Pp = dpm P : xpx+yp> (y+r)1'2 215 THE HARMONIC OSCILLATOR (where e p is the unit vector in the radial direction). The problem comes from the fact that p dp d<f> and not dp d<j) is the measure for integration. we end up with H h2 coordinate basis 2m dtf> 82 \ d p p 1 82 \ d<p (7.pp. If we choose operators dp P^iti that obey the commutation rules. lonical commutations rules? Let's do it and see what happens. (7.42) do not have the same eigenvalues. since the quantization procedure in terms of Cartesian coordinates has empirical support.41) is the correct one. What do we do now? (2) A way out is suggested by the fact that although the choice Pp*—ifi d/dp leads to the correct commutation rule. <> and p^ to quantum operators obeying /.4. for as seen in the last exercise the same physics may be hidden in two different equations. the Hamiltonians in Eqs. also called lz) 2m + Imp + (verify this) and directly promote all classical variables p. = xpy ~ypx (the angular momentum. (7.4. it is not Hermitian! Verify that C® / dp oc (*2n / \* ^J J (r^ilpj¥2pdpd(^ = <PpVlV2> (You may assume pyty2~*0 as p>0 or oo. and pj.) t What we will see is that Pp= iti d/dp.4. (7. In the present case this isn't true: as we will see.4. are nonHermitian.
4. and yet do not get the right quantum Hamiltonian. There are ways of constructing H given Jf (the path integ formulation suggests one) such that the substitution Pp*—ifi(d/dp + l/2p) leads to (7.4. 7.32)] and then take the inner product <x«>.216 CHAPTER 7 Show. (3) In the Cartesian case we saw that adding an arbitrary f(x) to —ifi d/dx didn't haw any physical effect. chosen Her tian operators. however.dp2 p dp 4p2 which still disagrees with Eq.4. We have satisfied the commutation rules. — (7. (7. We will not get into a discussion of these prescriptions for generating since they finally reproduce results more readily available in the approach we adopting. as it stands.5.41). (7.ff. While this ambiguity is present even in the Cartesian ca it is resolved by symmetrization in all interesting cases. 0+2®).4.41). I .4> is indeed Hermitian and also satisfies the canonical commutation rule. Notice that the additional term is indeed of nonzero order in fi. With nonCartesian coordinates ambiguity is more severe. The angular momentum P^—ift dfd<j) is Hermitian. the amplitude for finding the particle in a state  « ) at the point x could not be readily computed: it seemed as if one had to find the eigenkets (x) of the operators X [Eq. Passage from the Energy Basis to the X Basis It was remarked in the last section that although the n> basis was ideally suiti for evaluating the matrix elements of operators between oscillator eigenstates. that (7. The key to the myster lies in the fact that Jt doesn't determine H uniquely since terms of order fi (or higher) maj be present in H but absent in . on singlevalued functions: y(p. <j)) = \)/(p.pp*tfi — + 1 2 p. But there is a more direct way to get ^„(x) = <x«>. whereas here the addition of a function of p to —ifi d/dp seems important Why? [Is f(x) completely arbitrary? Mustn't it be real? Why? Is the same true for the ifi/I piece?] (4) Feed in the new momentum operator Pp and show that H tf 2m <L+IJL coordinate basis . In the present case the quantum Hamiltonian corresponding to 2m is given by 8 dp 2m p H coordinate basis *Jt \ p*p.
5.k We start by projecting the equation defining the ground state of the oscillator a0) = 0 (7.5. (7. 1 / d In the X basis Eq. For later use we also note that (since d/dy is an/i'Hermitian).1) 217 THE H A R M O N I C OSCILLATOR 01 the X basis: 0>+<x0>=^0(*) /ma a l I'if I 1 \ p ~\2h) J n m \ \2 n / 2 x \2 m a h ) J \2 matj dx (7.5.5) Wo(y) or y0(y) = A0e or ( = A 0 exp I max2 — y2/2 .1) then becomes .2) r In terms of y = (mo)/ti)l/2x.5. or J +^U(>0 =O dyj (7.
3./2y and (a + a^)\ny = n/2\n!> + (» +1).y2/2 il/2V (7.5.1./2«+1> 2nHni(y) . you should have a good grasp of how q u a n t u m mechanics works. Starting with a + af = 2. Exercise 7. We now conclude our rather lengthy discussion of the oscillator. dy. (7. Exercise 7. Project Eq.5.3. If you understan this chapter thoroughly.22).5.1) on the P basis and obtain Yo{p).7.5. (7.3. Exercise 7. Project the relation a\n}=n'/2\nl) on the X basis and derive the recursion relation H'n(y) = using Eq.? By projecting the equation h>=T^0> onto the X basis. \jcnj A comparison of the above result with Eq.218 CHAPTER 7 or (upon normalizing) ' \l/4 / 2v ma \ nil exp ( max in (7. (7.5.5.5.2. we get the normalized eigenfunctions (ma <x n)=\j/n \ 1/4 \ma y J (»!) 1/2 .22) shows that y2/2 (7.
p and and 2„2 2m 2 Show that I and that Note that EcX is independent of m and co. k=\A* 1CP16 ergs/° K. being Boltzman's constant.3. The index i is now continuous and corresponds to the variables x and p describing the state of the oscillator. derive the relation 219 2yHn(y)2nH„X(y) THE HARMONIC OSCILLATOR 9 Hn+i(y) = Exercise 7. (The "probability" referred to above is in relation to a classical ensemble of similar systems and has nothing to do with quantum mechanics. The Boltzman formula P(i) = where epEii)/Z Z = ^ep£(i) gives the probability of finding a system in a state i with energy E(i).and Eq. (7. when it is in thermal equilibrium with a reservoir of absolute temperature T= 1 /file. . i>x.) (1) Show that the thermal average of the system's energy is E=^E(i)P(i) = —\nZ dp (2) Let the system be a classical oscillator.22).5..4* Thermodynamics of Oscillators.e. i.
X This agrees with experiment at high temperatures but not as Empirically. for crystals whose atoms behave as point particles with no internal degrees of freedom .oe/T T j (e°*/T\ )2 where 0£= fico/k is called the Einstein temperature and varies from crystal to crystal. Show that if the oscillators are treated classicaly. The mean thermal energy of the crystal Crystal is £ cl or Equ summed over all the normal modes. • 3k Although C(T) C q u ( r ) ^ 0 as T*0. Show that Zqu = and ef"u°/2(\e/"u°rl I. This discrepancy arises from assuming that the frequencies of al J More precisely. Show that Cqu(T) = 3k \ e> . the specific heat per atom is Cc](T) = . C(T)*2k {T large) (T+0) Following Einstein. She that Cqu(T) Cqu(T) t»Oe 3k ) T<. treat the oscillators quantum mechanically. asuming for simplicity th they all have the same frequency co. which. is equivalent to 3jV0 decoupled oscillators. for small oscillations. (5) Consider a crystal with N0 atoms.d E ^ = 3 k ' N0 dT Ik 1 I which is independent of T and the parameters of the oscillators and hence the same for i crystals.<0E V T .220 CHAPTER 7 (3) For the quantum oscillator the quantum number n plays the role of the index i. (4) It is intuitively clear that as the temperature T increases (and (3 = 1 jkT decreases) i oscillator will get more and more excited and eventually (from the correspondence principle? qu T—*ao Ed Verify that this is indeed true and show that "large T" means T»fi(o/k. the exponential falloff disagrees with the observi T 3 behavior.
Figure 7.6 . o. {0 E is chosen to be 1320 K. .3.8 i 1.0 normal modes are equal.] This discrepancy was eliminated by Debye.) i 0 ^ i 0. . 7. Comparison of experiment with Einstein's theory for the specific heat in the case of diamond. But Einstein's simple picture by itself is remarkably successful (see Fig. which is of course not generally true.4 T/0 E . [Recall that in the case of two coupled masses we get tui = (k/m)V2 and con = (3k/my/2.2 .3). . . o. • 0.
8
The Path Integral Formulation of Quantum Theory
We consider here an alternate formulation of quantum mechanics invented by Feynman in the forties.t In contrast to the Schrodinger formulation, which stems from Hamiltonian mechanics, the Feynman formulation is tied to the Lagrangian formulation of mechanics. Although we are committed to the former approach, we discuss in this chapter Feynman's alternative, not only because of its aesthetic value, but also because it can, in a class of problems, give the full propagator with tremendous ease and also give valuable insight into the relation between classical and quantum mechanics.
8.1. The Path Integral Recipe
L We have already seen that the quantum problem is fully solved once the propag a t o r is known. Thus far our practice has been to first find the eigenvalues and " eigenfunctions of H, and then express the propagator U(t) in terms of these. In the path integral approach one computes U(t) directly. For a single particle in one dimension, the procedure is the following. To find U(x, t;x', f ) : (1) Draw all paths in the xt plane connecting (x', (') and (x, t) (see Fig. 8.1). (2) Find the action S[x(0] for each path x(t). (3) U(x,t;x', f)=A t efiWol/(i (8.1.1)
all paths
where A is an overall normalization factor.
J The nineteen forties that is, and in his twenties. An interesting account of how he was influenced by Dirac's work in the same direction may be found in his Nobel lectures. See, Nobel Lectures—Physics, Vol. IH, Elsevier Publication, New York (1972).
I
223
224
(x,t) and
CHAPTER 9
Figure 8.1. Some of the paths that contribute to the propagator, contribution from the path x(i) is Z = exp{iS[x(t)]/h}.
8.2. Analysis of the Recipe
Let us analyze the above recipe, postponing for a while the proof that it reprl duces conventional quantum mechanics. The most surprising thing about it is tfl fact that every path, including the classical path, xci (7), gets the same weight, that is to say, a number of unit modulus. How are we going to regain classical mechanics in the appropriate limit if the classical path does not seem favored in any way? To understand this we must perform the sum in Eq. (8.1.1). Now, the correct way to sum over all the paths, that is to say, path integration, is quite complicated and we will discuss it later. F o r the present let us take the heuristic approach. Let us first pretend that the continuum of paths linking the end points is actually a discrete set. A few paths in the set are shown in Fig. 8.1. We have to add the contributions Za = eiSlXa(')yn from each path xa{t). This summation is done schematically in Fig. 8.2. Since each path has a different action, it contributes with a different phase, and the contributions from the paths essentially cancel each other, until we come near the classical path. Since 5 is stationary here, the Z ' s add constructively and produce a large sum. As we move away from xcl(t), destructive interference sets in once again. It is clear from the figure that U(t) is dominated by the paths near x cl (t). Thus the classical path is important, not because it contributes a lot by itself, but because in its vicinity the paths contribute coherently. How far must we deviate from x d before destructive interference sets in? One may say crudely that coherence is lost once the phase differs f r o m the stationary value S[x0](t)]/fi = Sc\/fi by about it. This in turn means that the action for the coherence paths must be within fin of Sci. For a macroscopic particle this means a ;ery tight constraint on its path, since Sc\ is typically ~ 1 erg s e c ^ 1027/z, while for an electron there is quite a bit of latitude. Consider the following example. A free particle leaves the origin at f = 0 and arrives at x= 1 cm at t = 1 second. The classical path is (8.2.
Figure 8.2. Schematic representation of the sum Paths near xcl {/) contribute coherently since S is stati ary there, while others cancel each other out and I be ignored in the first approximation when we calcu
ReU
U(t).
225
THE PATH INTEGRAL FORMULATION OF QUANTUM THEORY
Figure 8.3. Two possible paths connecting (0, 0) and (I, 1). The action on the classical path x = t is m/2, while on the other, it is 2m/3.
Consider another path x=r (8.2.2)
which also links the two spacetime points (Fig. 8.3.) For a classical particle, of mass, say 1 g, the action changes by roughly 1.6 x \ 0 2 % and the phase by roughly 1.6 x 1026 rad as we move from the classical path x=t to the nonclassical path x=t2. We may therefore completely ignore the nonclassical path. On the other hand, for an electron whose mass is ^ 10  2 7 g, 8S~ fi/6 and the phase change is just around a sixth of a radian, which is well within the coherence range 5S/1i<n. It is in such cases that assuming that the particle moves along a welldefined trajectory, xc] (/), leads to conflict with experiment. *
8.3. An Approximation to U(t) for a Free Particle
Our previous discussions have indicated that, to an excellent approximation, we may ignore all but the classical path and its neighbors in calculating U(t). Assuming  that each of these paths contributes the same amount exp(i'S c i/fi), since S is stationi ary, we get U(t) = A'e'tSd/n (8.3.I)
where A' is some normalizing factor which "measures" the number of paths in the coherent range. Let us find U(t) for a free particle in this approximation and compare the result with the exact result, Eq. (5.1.10). The classical path for a free particle is just a straight line in the xt plane:
(8.3.2)
corresponding to motion with uniform velocity v= (xx')/(t1'). mv2j2 is a constant,
Since S£ =
J,
2
tt'
226
and
SO that
CHAPTER 9
U(x, t; x', t')—A' exp
im(x — x')7 _ 2n(tt')_
(8.3.3)
To find A', we use the fact that as t — t' tends to 0, U must tend to Comparing Eq. (8.3.3) to the representation of the delta function encountered in Section 1.10 (see footnote on page 61), 1 —jv') = lim exp A+0 (7lA2)]/2 (valid even if A is imaginary) we get m 2
,1/2
(xx'f
A' =
Khi{tt')_
so that 1/2 U{x, t; x', 0)=U(x, t\: r'} = f———) exp x') \2 nfiit} im(x~xry 2fit
(8.3.4)
which is the exact answer! We have managed to get the exact answer by just compul ing the classical action! However, we will see in Section 8.6 that only for potentiia l of the form V~ a + bx + cx2 + dx + exx is it true that U(t) = A{t) eiS*f*. Furthermore, we can't generally find A(t) using U{x, 0; x') = <5(jcx') since A can contain an arbitrary dimensionless f u n c t i o n / s u c h t h a t /  » 1 as t>0. H e r e / = 1 because we canj construct a nontrivial dimensionless / u s i n g just m, fi, and t (check this).
I
8.4. Path Integral Evaluation of the FreeParticle Propagator
Although our heuristic analysis yielded the exact freeparticle propagator, will now repeat the calculation without any approximation to illustrate pa integration. Consider U(xN, tN; x0, t0). The peculiar labeling of the end points will be ji ified later. Our problem is to perform the path integral r^N
Xo
iS[x(t)]/h ®[x{t)}
(8.4J
where *XN ®[x(t)]
227
THE PATH INTEGRAL FORMULATION OF QUANTUM THEORY
The discrete approximation to a path x(t). Each path is specified by T  1 numbers V x(fi),..., x(r,vi). To sum over paths we must integrate each x, from  oo to +00. Once all integrations are done, we can take the limit N*<x>.
Figure 8.4.
is a symbolic way of saying "integrate over all paths connecting x0 and xN (in the interval t0 and Now, a path x(t) is fully specified by an infinity of numbers x(f0), • • •, x(t),..., x(tN), namely, the values of the function x(i) at every point t in the interval t0 to tN. To sum over all paths we must integrate over all possible values of these infinite variables, except of course x(t 0 ) and x(tN), which will be kept fixed at x 0 and xN, respectively. To tackle this problem, we follow the idea that was used in Section 1.10: we trade the function x(t) for a discrete approximation which agrees with x(t) at the N+ 1 points t„ = t0 + ns, n = 0,.. . , N, where e=(tN—t0)/N. In this approximation each path is specified by N+ 1 numbers x(t0), x(t\),. . ., x(tN). The gaps in the discrete function are interpolated by straight lines. One such path is shown in Fig. 8.4. We hope that if we take the limit AT> 00 at the end we will get a result that is insensitive to these approximations.t N o w that the paths have been discretized, we must also do the same to the action integral. We replace the continuous path definition
'In
fr.
&(t)dt
= to
1
mx2 dt
by
*
,=0 2 \
(8.4.2) £ J
where x, = x(f,). We wish to calculate • xN U{xn, tm', Xo, to) — xo 1* 00 C 00 = lim A
oo £.0 X
*
J co
v
v
—00 \
i m N ,_ 1 (x/+1 XjY exp in 2 h e —O O (8.4.3)
00
dxx • • •dxN
I We expect that the abrupt changes in velocity at the points t0 + rts that arise due to our approximation will not matter because does not depend on the acceleration or higher derivatives.
228
and
CHAPTER 9
It is implicit in the above that x0 and xN have the values we have chosen at the outset. The factor A in the front is to be chosen at the end such that we get the correct scale for U when the limit oo is taken. Let us first switch to the variables
m
1/2
* We then want
=
W J
*
lim A'
N^ co
exp
oo oo
dyx • • dyNx (=o i
(8.4.4)
where (N\)/2
Although the multiple integral looks formidable, it is not. Let us begin by doing the yx integration. Considering just the part of the integrand that involves y\, we get
(8.4.:
Consider next the integration over y i . Bringing in the part of the integrand involvil y2 and combining it with the result above we compute next
/.
171 \
\
1
/2
p oo
dy2 f . \ V2 I7t\
A1/2 (.2yi + yl)/2i
Jyo + Zyi)2/^
(•in)2
1/2
(ys~yo)2/3i
(8.4.1
By comparing this result to the one from the y{ integration, we deduce the patter if we carry out this process N— 1 times so as to evaluate the integral in Eq. (8.4.4 it will become fiVk^'V2 yUL> p(yN ytf/Ni NV2
or {in) (N — 1 ) / 2 1/2 m(xN — xo} /2tteMi
2
229
THE PATH INTEGRAL FORMULATION O F QUANTUM THEORY
N Bringing in the factor A(2He/m)
(N — 1 ) / 2 from up front, we get m IjrftiNe)
1/2
I f
U=A(
\ N/2 (27i1iei\ I \ m J
exp
im(x N — XoY ITLNE
If we now let N ^ c o ,
0, N s ^ t N — t Q , we get the right answer provided Infiei m
N/2
A=
= BN
(8.4.7)
It is conventional to associate a factor 1 / B with each of the N— 1 integrations and the remaining factor 1 / B with the overall process. In other words, we have just learnt that the precise meaning of the statement "integrate over all paths" is
I
where
1 @[x{t)] = lim B N—*co
dx\ B
dx 2 B
dx N 1 B
(2kTI£I\
\
A >/2
(8.4.8)
m
/
8.5. Equivalence to the Schrodinger Equation
The relation between the Schrodinger and Feynman formalisms is quite similar to that between the Newtonian and the least action formalisms of mechanics, in that the former approach is local in time and deals with time evolution over infinitesimal periods while the latter is global and deals directly with propagation over finite times. In the Schrodinger formalism, the change in the state vector  y/) over an infinitesimal time £ is
which becomes in the X basis y/(x, £)  yr(x, 0) : l£ h n2 e 2 2m dx' + V{x, 0) y/(x, 0)
i
— l£ \¥(£))\y,m n
H\¥( 0)>
(8.5.1)
(8.5.2)
230
and
CHAPTER 9
to first order in s. To compare this result with the path integral prediction to the same order in e, we begin with
y/(x, s) =
U(x, e; x')y/(x',
0) dx'
(8.5.3)
The calculation of U(£) is simplified by the fact that there is no need to do any integrations over intermediate x's since there is just one slice of time s between the start and finish. So m(xx') t\2 2s 'x + x' .0.
U(x, s; x') = ——— \2nfiis
exp<i
sV
fi}
(8.5.4)
where the ( m / 2 n f i i s ) x / 2 factor up front is just the 1 / 5 factor from Eq. (8.4.8). We take the time argument of V to be zero since there is already a factor of e before it and any variation of V with time in the interval 0 to s will produce an effect of second order in e. So 1/2 y/(x, s) im(x — x') 2 sfi
 ( — ) \2 nfdej
x y/(x', 0) dx'
exp
exp
Ity^
0
(8.5.5)
Consider the factor exp[/m(x — x') 2 /2efi]. It oscillates very rapidly as ( x  x ' ) varies since s is infinitesimal and fi is so small. When such a rapidly oscillating function multiplies a smooth function like y/(x', 0), the integral vanishes for the most part due to the random phase of the exponential. Just as in the case of the path integration, the only substantial contribution comes from the region where the phase is stationary. In this case the only stationary point is x = x, where the phase has the minimum value of zero. In terms of r] = x' — x, the region of coherence is, as before, mrj 2 sfi or (2sfiirV/2 MS I ,
\ m J
(8.5
Consider now m \2 nfiis x y(x+
Qxp(imr]2/2fis)~ — uu
exp (8.5.7)
rj, 0) drj
We will work to first order in s and therefore to second order in r] [see Eq. (8.5.6) above]. We expand du/ ox n2 d2w —~ + • • 2 ox
231
THE PATH INTEGRAL FORMULATION OF QUANTUM THEORY
y/{x+t1,0)
= y(x, 0) + 7 7
+
exp
iHH
IS  1   V(x, 0) + ' n
since terms of order r j s are to be neglected. Equation (8.5.7) now becomes m 1/2 fimrj is yf(x, 0 )   V(x, 0)y/(x, 0) \2fie, fi
yr(x, s) = I 2nHis) + 7 7 dy dx 1 n2 2
d2y  drj dx
Consulting the list of Gaussian integrals in Appendix A.2, we get m Inftir.) is H\ or is 32  r2 d 2m dx (2kMs\ m J
1/2
1/2
y/(x, s) = 
^ \ m }
fis (infUs^^ lim \ m
1/2 J
d2\jf dx2
V(x, 0 ) y f ( x , 0)
V(x, s)  1y(x, 0)
V(x, 0) y/(x, 0)
(8.5.8)
which agrees with the Schrodinger prediction, Eq. (8.5.1).
8.6. Potentials of the Form V= a + bx + cx1 + dx + exxt
We wish to compute
(* X
U(x, t;x') =
« eiS[x(t")Vm
x(oi
(8.6.1)
t This section may be omitted without loss of continuity.
1) becomes U(x. i c ) ) integrates to give S^Xd ] = Sc].x') = Qxp<~S[xcl(n+y(t'')]^[y(t'')] (8. we have for intermediate integration variables Xi = x(t'/) = xcl (t'!) + y ( t f ) = xa ( t f ) Since x d (/") is just some constant at t". dxt = dy'j and fx 2[x{f)\ = 1*0 my(t")] (8.! so that Eq. line in y and y. (8. xci) + y^ V dx 1 (d2& 2 \ dx2 Xc\ dse y+ dx Xd y yz + 2 d2^ dx dx d2S£ yy+ ~ .' .6. if i recall = \mx2 — a~bx — cx2 — dx — exx (8.2 (8. t. The second piece.2 Sx *cl y dt" / 1 (8i The series terminates here since ££ is a quadratic polynominal.xcl + y) dt" &(xci. >'(0) = j ( 0 = When we slice u p the into N parts. vanishes due to the classical equation of motion. Since all the paths agree at the end points.6.232 and Let us write every path as x(f)=xd(0+XO It follows that i ( 0 = *d (t") + y(t") CHAPTER 9 (8.6. In the last piece. The first piece i f (x c ] .i The next step is to expand the functional S in a Taylor series about x c l : S[xc]+y] &(xcl +y.
we would get Eq. x') = exp x iSc]\ fi 9[y(tl] rt exp 1 2 2 \ .13) Since the coefficient b does not figure in Eq.6. t) is not known.9) m (8.6.tx') = eiSc. In this case we know that [see Eq. it can only depend on t. we can extract all the probabilistic information at time t.11) with c = e = 0. F o r the harmonic oscillator. c = j m © 2 .10) U(x.6. An . N o t e that even if the factor A(t) in y/(x. The path integral method may be extended to three dimensions without any major qualitative differences./nA(t) (8.co y ) dt"$[y(t")] (8. t. N o w if we were doing the freeparticle problem.6.11). (8. . (8. (8.14) The evaluation of this integral is discussed in the book by Feynman and Hibbs referred to at the end of this section.233 1 d2<e _ 2 dx2 " d2£ dx dx 1 d2se 2 dx2 Consequently Eq.6.4)] * A(t) =  m 2nfiit) 1/2 (8.12) is valid for potentials that are at most quadratic in the coordinates and the velocities.12) where A(t) is some unknown function of t. Notice the ease with which the Feynman formalism yields the full propagator in these cases. So U(x. Consider in particular the horrendous alternative of finding the eigenfunctions of the Hamiltonian and constructing f r o m them the harmonic oscillator propagator. (8.6.my —cy— eyy j dt" (8.3.6. and we have to do the integral ' ] A(t) = exp i / f i J m(y .6.6. it follows that the same value of A(t) corresponds to the linear potential V=a + bx as well.11) Since the path integral has no memory of x c ) . the form of U in Eq. In particular.8) _ THE PATH INTEGRAL FORMULATION O F QUANTUM THEORY —e (8.5) becomes f —c (8. (8.6.6.
2xxf] where A(t) is an unknown function. Show that for the harmonic oscillator with if = { mx2 —  mo)2y^ U(x. Now consider the extraction of the eigenfunctions. as per Eq. A(t) = \2nnitJ / 1/2 agrees with the exact result.x') = A(t) exp j m0) [(x 2 +x' 2 ) cos at . P. 9h(o/2.6. Hibbs.5).6. els What happened to the levels in between? 1 (2) (Optional).8) starting from Eq. Exercise 8. (8.4). (8.31). for the cal of the oscillator.. .1 * Verify that \ rn \ U(x.f x .6. 0 = 1 yn(x)v!(x')ef£»(''v* « (8. (8. All this was thanks to the factor e multiplying V in Eq.4) and the fact that r/^e l / 2 . (5.interesting problem in this class is that of a particle in a uniform magnetic field.5. (8.4. We know that given the eigenfunctions and the eigenvalues we can co struct the propagator: U(x.5. it did not matter very much: any choice x+arj. R.. Ffl further details on the subject of path integral quantum mechanics. and  ^i(x) 2 by expanding in powers of a =exp(icot)..x'. Let x = x' and f'f Find E0. x'. (where rj = x' — x) for 0 < a < 1 would have given the same result since the difference between the choices is of order Tje~e3/2.6. McGrawHill (1965).6. B] expanding both sides of Eq.\ ^o(x) 2 . t.15] j Consider the reverse process (since the path integral approach gives U directly). fi •i «• l/2 (1) Set =f = Assume that A(t) = (mco/2nifi sin (ot) for the oscillator. Hint: Start with xc](/")= x0 + v0t"+ \{f/m)t"2 and find the constants xQ and v0 from the requirement that xci (0)=. you should find that E=h(o/2.. Feynma and A. 5h(o/2. . see R. ai^ Chapter 21.6. Path Integrals and Quantum Mechanics.5.* Recall the derivation of the Schrodinger equation (8.2.15).6. t. Note that although we chose the argument of Vto be the midpoint x+^/2. t.8.) Exercise 8. Ei. 0) = A(t) exp(iSci/h).4. J t Exercise 8.3. (Recall Exercise 2. Exercise 8.7. for V(x) = . and xc] (t) — x. Eq.
This point will come up in Chapter 21. I . Along the way you will see that changing a makes an order s difference to y/(x. but will also make the answer sensitive to the argument of A in the linear term.e. use the midpoint prescription. in contrast to the scalar potential case.) Note that £ now gets canceled. but the onedimensional version will suffice for making the point here.3. Thus.. (We should really be using vectors for position and the vector potential.Consider now the case of a vector potential which will bring in a factor 235 THE PATH INTEGRAL FORMULATION O F QUANTUM THEORY to the propagator for one time slice. Choose a — 1/2 and verify that you get the onedimensional version of Eq. (4. This will not only bring in an A2 term. e) so that we have no choice but to use a = 1/2.7). going to order e to derive the Schrodinger equation means going to order if in expanding the exponential. i.
given a state  y > .v [Q. AQ • AA. The derivation will make clear the conditions under which such a relation will exist.^ 0 ). In the next section we will derive the Heisenberg uncertainty relations.p).2. Generally the lower bound will depend not only on the operators but also on the state.<Q> ) 2  y> ] 1 / 2 (9. If we consider two Hermitian operators Q and A.1. po) has associated with it welldefined values for any dynamical variable a>(x. A] = iT (9.2) There are. namely.1. states for which AQ —0.In quantum theory. and these are the eigenstates ©> of Q.2. Introduction In classical mechanics a particle in a state (x 0 .9 The Heisenberg Uncertainty Relations 9. however.1) and an uncertainty about this mean: ( A Q ) = [<i// (Q . o n e c a n o n l y give the probabilities P(co) for the possible outcomes of a measurement of £2. Of interest to us are those cases in which the lower bound is independent of the state. Derivation of the Uncertainty Relations Let Q and A be two Hermitian operators. 9. The probability distribution will be characterized by a mean or expectation value ^ <Q> = <yQy/> (9. o)(x 0 . with a commutator .1) 237 . which will provide a lower bound on the product of uncertainties. they will generally have some uncertainties AQ and A A in an arbitrary state.1.
j (9. In terms of H and A (AQ) 2 ( A A) 2 = < y/\Ci< W  A:l\W > (9. we know that the commutator has to enter the picture somewhere.! (9 23 which has the same commutator as Q and A (verify this).2. we get from Eq.1 = A] + + ^ . = <£Vl*V><AyAv> since and A2=ArA If we apply the Schwartz inequality \V]l2\V2\2>KVi\V2>\2 (9. (AQ^(AA)2><^VAy/>z (9. Thisj arrange through the following identity: &A+Aq 2 _ QA = +  CiAAQ 2 (9.238 CHAPTER 9 You may readily verify that T is also Hermitian.<A> ) 2  y/> where <H) = (y/\Q\ v> and <A> = < v  A  y/}.5 (where the equality sign holds only if  V. Let us next define the pair D = Q<Q> ~ A = A<A> (9. where c is a constant) to states )0y/> and  A y > .2 .4). > = V2).: Now.2.<Q> ) 2  y > < W \ ( A .\ ( Q . (9.2. A ] .2. Let us start with the uncertain product in a normalized state \y/}: i (AQ ) 2 (AA) 2 = (y.2.1 Let us now use the fact that (£ly/\Ay/) = < vl^Al y/} = <y^A y/> to rewrite the above inequality as (AQ) 2 (AA) 2 > I <yrGA y/}\2 (9.
the expectation value of the commutator is pure imaginary. A ]  y >  z (9.9). A]+y> = 0 (9. According to Eq.2.2.13) Since the first term is positive definite. for which T=fi.1) .2.14) which is the celebrated uncertainty relation.10) into the inequality 239 T H E HEISENBERG UNCERTAINTY RELATIONS We next use the fact that (1) since [ f t A] = /T.2.2. where T is Hermitian.2.3. we get (AQ ) 2 (AA) 2 > >kv\ [ f t A] +  v> + K W \ r  w> 12 [ f t A] + ! Wy+ 5 < Y\T\ V/)2 (9. Recalling that \a + ib\2 = a2 + b2.15) 9.2.< X > )  y > (9. In this case (AQ)2(AA)2>i<^[ftA]+ 4 +y 4 (9. we may assert that for any 1 y/} (AQ) 2 (AA) 2 >£ 2 /4 or AQAA>V2 (9. the expectation value of the anticommutator is real.12) I ( A Q ) z ( A A ) z >  < y ^ [ f t A] + + i[Q.11) This is the general uncertainty relation between any two Hermitian operators and is evidently state dependent. Let us note that the above inequality becomes an equality only if (1) Dy> = cAy> and (2) <y[Q.15) such a state is characterized by (i>~</>>)y> = c ( X . (9. (9. (9. The Minimum Uncertainty Packet In this section we will find the wave function y/(x) which saturates the lower bound of the uncertainty relation for X and P.where [ f t A]+ is called the anticommutator. (2) since [ f t A]+ is Hermitian.2. Consider now canonically conjugate operators.3. we get Feeding Eq.
(9.3. implicitly defined by these equatio the X basis.3. Eq.\(P(P})(X(Xy) + (X<X»(P(P))\Y)=0 ( where <P> and ( X ) refer to the state 11//>. whatever < X ) may be.240 CHAPTER 9 and (y. which in this frame reads (y\(P(P))X+X(P(P)) y> = 0 If we now exploit Eq.4) now becomes W(x) (9 = Y(0)el<p>x/* e~lclx2/2* In terms of A2 = ft/\c\?x/ y/(0)e'^ "e iiPyxjft (9 x2/2A2 y/(x) = 1 . (9.1) and its adjoint. j (C+C*KY\x2\wy=o \ . Eq.1) becomes ~in£~ <p>) VM = c(x >)yr(x) or ~ = ^[<P> + c(x~(X))]dx \ff(x) n ( Now. Eq.3. (9. In this frame. (9.2).3) ha solution y/(x) = y / ( 0 ) eKP>x/n eicx2/2* (< Let us next consider the constraint. a f r o m which it follows that c is pure imaginary: c = i\c\ O u r solution. it is always possible to shift our origin (to x = < so that in the new frame of reference <T> = 0.3.3. we find W\c*X2 + cX2\y/} =0 . Eq. (9.
Applications of the Uncertainty Principle I now illustrate the use of the uncertainty principle by estimating the size of the groundstate energy and the spread in the groundstate wave function. may be written entirely in terms of the electron's variable as p2 + p2 . It should be clear from this example that the success we had with the oscillator was rather atypical. for the special case ( X } = (P) = 0.7) Thus the minimum uncertainty wave function is a Gaussian of arbitrary width and center. to simplify the notation. \(X +Y 2 2 \ + Zy 2 /2 (9A2) the first step in minimizing ( H } is to work only with states for which (/*. If the origin were not chosen to make (X)> zero.1)J Let us begin by mimicking the analysis we employed for the oscillator. like is arbitrary.) = 0.where A . was used in the quest for the state that minimized the expectation value of the oscillator Hamiltonian.4. We evaluate ( H ) in a normalized state  y } : 2m \(X2+Y2 + Z2)1/1/ <#+<#>+<#>j 2m Since <p2y=<APxy+iPxy2 etc. We choose as our system the hydrogen atom. we would have instead ¥(x)= 241 T H E HEISENBERG UNCERTAINTY RELATIONS y«jr» ei<p>(x~<x>)/K e^< x >)2^2 (9. This result. assuming the proton is a spectator whose only role is to provide a Coulomb potential for the electron. The Hamiltonian for this system. For such states 2m \(X2+Y2 + ZZ)1/2/ t The operator (X2+ Y2 + Z2) 1/2 is just 1/r in the coordinate basis.3. We will occasionally denote it by 1 /r even while referring to it in the abstract. 9. .4. pi z " 2m 2 ! + ^ (Xz+Y2 Z2)l/z (9.
F o r such states APl + APi + APl 2m [(AX)2 + (AY)2 + (AZ)Z)1/2 For a problem such as this. and Z) and the functions of the] mean ( ( X ) .' where the ^ symbol means that the two sides of Eq. Now the handwa\ begins..4) are not strictly equ but of the same order of magnitude.)2 + {APzf 2m From the relations ( X z ) = (AXy + ( X y etc. Y. So we write (APx)2 + (APy)2 + (APzy 2m Once again.(x + 2 Y +ZZ)1/2/ z <(x2+ y2+z2)1/2> (9. and < Z » are of the same order of magnitude. < K>.4. we argue that <(x2 + y 2 + z 2 ) 1 / 2 <(x2+r2+z2) 1 / 2 > > «x 2 > + < y2> + < z 2 > )1/2 and gett (APX)2 + (AP.We cannot exploit the uncertainty relations APxAX>fi/2. yet since <//> is not a function of AX and A P. We argue that (see Exercise 9. 1 . .4. etc. with spherical symmetry.4. and AZ. it is intuitively clear that configuration of least energy will have (AXf = (AY)2 = (AZ) I We are basically arguing that the mean of the functions (of X. (9. «^ 2 >+<y 2 > + <z2» 1 / 2 it follows that we may confine ourselves to states for which <X> = < y > = <Z>=0 in looking for the state with the lowest mean energy.2). The problem is <. A Y.(X2+ Y2 + Z2yl/2y is not simply related to AX. They are in fact equal if there are noj fluctuations around the mean and approximately equal if the fluctuations are small (recall the discussioii toward the end of Chapter 6).
7). / 2 (AX) := 0 4 me Finally.and (APX)2 = (APy)2 = (APzf so that 3(APx)2 2m Now we use APxAX>fi/2 to get W 8 m(AX)2 3 1 / 2 AX e2 3 1 / 2 AX (9. the true groundstate wave function is not a Gaussian but an exponential y/(x. where a0= ft2/me2 .4.7) What prevents us from concluding (as we did in the case of the oscillator). Likewise.4. 'U\>~2me* me (9. z) = cexp[(x2+y2+z2)1/2/ao].7) is an approximate inequality. y.4.2 m e 4 / 9 n 2 or that the groundstate wave function is a Gaussian [of width 3(3 1/2 )# 2 /4me 2 ] is the fact that Eq. Eq.5) 243 THE HEISENBERG UNCERTAINTY RELATIONS <H>Z We now differentiate the righthand side with respect to AX to find its minimum: en2 &m(AXy or (9.6) e2 3 . the exact groundstate energy Eg=me4/in2 (9. (9. that the groundstate energy is .4. only by a factor ^ 2 .8) differs from our estimate.4. However.4. (9.
244 CHAPTER 9 is called the Bohr radius.2. the AX associated with this wave function is AX=h2/me2 (9.4.R2 = X2+ Y2 + Z2) 2m (R ) 2 l/2 corresponds to a onedimensional problem.4.4.2. Exercise 9. but not always as much as in the case of the oscillator. <f>)r2drd(cos 6) d<f> = 4K I \if*{r)y{r)r2 dr= 1 for a function of just r.1. Exercise 9. (3) Show that < l / r > ~ l / < r > ~ w g 2 / ^ 2 in this state. However. In conclusion. * Compute AT. Note that in three dimensio the normalization condition is <vl V>= V*(r> 4>)v(r. Ignore the fact that the hydrogen atom is a threedimensional system pretend that (P2 = P2 + P2 + P2. Exercise 9. (9. the uncertainty principle gives us a lot of information about ground state.4 which also is within a factor of 2 of the estimated AX in (9. 0.9).3. verify that r/ao r=(x2+y2 + z2y/2 is a normalized wave function (of the ground state of hydrogen).4. where T=P2/2m. * Consider the oscillator in the state \n= 1) and verify that ' — \j ^ 1 1 2 2 \x / (x ) may n Exercise 9. (2) Calculate (AX)2 in this state [argue that (AX)2 = 5</2>] and regain the result quo in Eq.4.4. (1) By referring to the table of integrals in Appendix A.AX.6).4. Why is this relation not so famo . Assuming APAR>h/2 estimate the groundstate energy.
since time t is not a dynamical variable but a parameter.8 Figure 9. The EnergyTime Uncertainty Relation There exists an uncertainty relation AEAt>fi/2 (9. Example 9. [The output of the source is not just e b u t rather 0(t) e~'°". i. recall that eigenstates of energy have a time dependence e~ . who catches it at the point x2..e. It will however be seen that initially the atoms make transitions to several levels not obeying this constraint. and skating toward each other on trajectories that are parallel but separated by some perpendicular distance (Fig. When skater A reaches some point x.] Similarly when the excited atoms get deexcited and drop to the ground state. 9. where AE and At are related by (9. At time / = 0..5. At the point x\.12). The following example should clarify the meaning of this statement. 9. does not emit photons of definite energy) if it has been in operation only for a finite time. even if the dial is set at a definite frequency co during this time. co = E/ft. Another way to describe this uncertainty relation is to say that violations in the classical energy conservation law by AE are possible over times At~fi/AE.5. . (Range of the Nuclear Force. we expect transitions to take place only to a level (if it exists) fico above the ground state. the deviation AE from the expected finalstate energy will decrease according to A E ~ H / t . they do not emit photons of a definite energy E=Ee — Eg (the subscripts e and g stand for "excited" and "ground") but rather with a spread A E ~ ft/At. a system that has been in existence for infinite time) has a welldefined frequency.£t/fi. is not associated with a pure frequency co — E/h or definite energy E. we turn on light of frequency co on an ensemble of hydrogen atoms all in their ground state. (9. Consider the following example. The content of this equation is quite different from the others involving just dynamical variables.) Imagine two ice skaters each equipped with several snowballs. Since the light is supposed to consist of photons of energy fico. skater A throws the snowball towards skater B. whose transform is not a delta function peaked at co.2. We interpret this result by saying that the light source is not associated with a definite frequency (i.1) which does not follow from Eq. only a wave train that is infinitely long in time (that is to say. At being the duration for which they were in the excited state.5. even if its time dependence goes as e~tEl/f> during this period. To see how this comes about.1. a definite energy is associated with a definite frequency. as t increases. The rough meaning of this inequality is that the energy of a system that has been in existence only for a finite time At has a spread (or uncertainty) of at least AE. Now. [The time dependence of the atomic wave function is not e~iEe'/ft but rather 0(t)0(Tt) e~iEe'/f> assuming it abruptly got excited to this state at ? = 0 and abruptly got deexcited at t = 7\] We shall return to this point when we discuss the interaction of atoms with radiation in a later chapter.5.1).1. Thus a system that has been in existence only for a finite time.e. However. Only as /^oo do we have a rigid law of conservation of energy in the classical sense.1).
He (^4) will then recoil away from B and start movi along a new straight line. which have a mass p and energy fic2. This is roughly how elementary particles interact with each other: if they throw photc at each other the force is called the electromagnetic force and the ability to throw and cat photons is called "electric charge. He too will recoil as a result. . unlike the two skaters endowed with snowballs. For example. the lightest object that can be exchanged and hence responsible for the nuclear for of the longest range. neither is the pion the only particle that can be "exchanged between nucleons nor is the number of exchanges limited to one per encounter. We would like to estimate the range of the nuclear force using the uncertainty principj Now. (The pion however. the observer would conclude that the range of the force is 10 meaning A and B will not affect each other if the perpendicular distance between them exce< 10 ft. he would conclude that there is a repulsive force between A and B.) Also our analogy with snowballs does not explain any attractive inU action between particles. If A (or B) c throw the ball at most 10 ft..1 246 CHAPTER 9 let him throw a snowball toward B. If this whole process were seen by someone who could not see the sm balls. which is called the Compton wavelength of the pion ai is a measure of the range of nuclear force." If the projectiles are pions the force is called the nuclc force. produce a pion from nowhere (violating the classical law of eneri conservation by ~//c 2 ) provided it is caught by the other nucleon within a time At such tli At~fi/AE=ti/pc2. nucleoid in the nucleus do not have a ready supply of pions. Even if the pion travels toward the receiver at the speed of light. Let B now catch the snowball. The value of r is approximately 1 Fermi = 10~n cj The picture of nuclear force given here is rather simpleminded and should be taken wi a grain of salt.e. shown in the figure. however. itci only cover a distance r = c At = fi/pc. the protons and neutrons (i. nucleon can.
The rule for quantizing this system [Postulate II.1.lc) w. (7. we have restricted our attention (apart from minor digressions) to a system with one degree of freedom.1. In this section we consider N particles in one dimension. and start with the case N—2.Xj] = iH{xlixj}= [Pi.p2). usually the coordinate basis.4. a single particle in one dimension. Eq. In practice one works in a basis.10 Systems with N Degrees pof Freedom 10.pj}=0 (10.1a) [Xt. number of spatial dimensions. N Particles in One Dimension So far.P\) and (X2. We now consider the quantum mechanics of systems with N degrees of freedom.1.Pj] = 0 (10. namely.1.39)] is to promote these variables to quantum operators {Xi . The increase in degrees of freedom may be due to an increase in the number of particles. This basis consists of the kets which are . Pi) obeying the canonical commutation relations: (i= 1. It might be occasionally possible (as it was in the case of the oscillator) to extract all the physics given just the canonical commutators.j (10. 2) [X„ Pj] = iH{xhPj} ift5. or both. The TwoParticle Hilbert Space ^ Consider two particles described classically by (xupi) and (x2.1b) V & * ifi{pi.
1.2x2) In this basis \y/y^(xix2\y/} X^xt Pi^iii ^ e dxj = y(xx. the momentum basis. we define operators X\ and P\ obeying [XuPi) = ifiI (10.5) as the absolute probability density for catching particle 1 near Xi and particle 2 near x 2 .p2). Consider a system of two particles described classically by (xi. § Note that any function of X\ and P. P2) to define the Q basis. <32 As a Direct Product Space I • There is another way to arrive at the space V i ^ . There is.1.1. We denote by V l g .3) We may interpret P(xi. If we want the quantum theory of just particle 1./>i) and (x2.6) l = <VlV> = There are other bases possible besides xix 2 ).More generally. provided we normalize 1i//> to unity ^  <xix2 y>2 dx\ dx2 = P(x 1.1. for example. x 2 ) dx 1 dx2 (10.x2) = Kx1x2\¥y\2 (10.1. i . Pi) and 02(X2.4) L ff (10. and that is to build it out of two oneparticle spaces. consisting of the simultaneous eigenkets \p\p2} of P\ and P2.x2) (10.7) The eigenvectors  x j ) of X\ form a complete (coordinate) basis for the Hiibert space I Note that we denote the bra corresponding to \x[x'2y as (x[x'2\.2) ^ 2 ! * 1 * 2 > = * 2 XIX 2 > and are normalized a s  <xix^xix 2 > = S(x[ xl)S(x.1.248 CHAPTER 10 simultaneous eigenkets of the commuting operators X{ and X2 (10. commutes with any function of X2 and Pj. we can use the simultaneous eigenkets ct)ict>2> of two commuting operators? Q](Xi. f V. 2 the twoparticle Hiibert space spanned by any of these bases.
Now a position measurement will yield a pair of numbers (xi. Thus Eq.P[1)] = ihIw (10. A similar picture holds for particle 2. not every element of V]<S)V2 is a direct product. This should be compared to the direct sum (Section 1. there is one and only one basis vector xi><g)x2> of V i ® V 2 . Q j .8a) 249 SYSTEMS WITH N DEGREES OF FREEDOM where / ( 1 ) is the identity operator on V i . The product lxi>®x 2 >.1. JCI)(H)X2). [X?\Pp] = iftI(2) (10. (10. The coordinate basis. let us denote the corresponding ket by xi>®x 2 >: .Vi of particle 1.) of H ^ X j . or. „ .9) v l*i><g>l*2>~r Note that *i>(8>x2> is a new object. (particle 1 at X] (particle 2 at x2 .©2=V1@V2 in which case the dimensionalities of Vi and V 2 add (assuming the vectors of Vi are linearly independent of those of V 2 ). I**?. Other bases. (10.1. Since the operators X\. What will be the coordinate basis for this system? Previously we assigned to every possible outcome jti of a position measurement a vector jxi> in V] and likewise for particle 2.7) reads [X?). etc. we can use the momentum basis \ p i y ® \ p 2 y . Pi.1.1. is just one possibility. is the product of vectors from two different spaces. act on V].1. the statement makes heuristic sense: to each basis vector xi> of V] and x 2 > of V 2 .4): V.8b) Let us now turn our attention to the twoparticle system. quite unlike the inner product <vMV2> or the outer product I v ' i X ' H both of which involve two vectors from the same space. The dimensionality (number of possible basis vectors) of Vi<S>V2 is the product of the dimensionality of Vj and the dimensionality of V 2 . more generally. The direct product is a linear operation: 'H (a\xiy + a'\x\y)®(P\x2>) = a P \ x O ® \ x 2 > + a'P\x\y®\x2^ (10. P i ) are also possible.. Although all the dimensionalities are infinite here. let us append a superscript (1) to all of them. and refer to as the direct product of the spaces Vi and V 2 . called the direct product. x2). and in particular. such as />i> of Px or in general. Since after the measurement particle 1 will be in state Jxj > and particle 2 in \x 2 }. e>i)®[6) 2 ). For instance  V 0 = I*'1>®X2> +  X 1 > ®  X 2 > .10) The set of all vectors of the form  x i > ® X2> forms the basis for a space which we call Vi ® V 2 . Although these vectors span V 1 ®V 2 .
It is intuitively clear that when two particles are amalgamated to form a single system.13) Let us define a direct product of two operators. p ( 2 l ) ® a \ as ^(1)8X2)^(1)^2) ( l 0 1 1 6 } The following properties of direct products of operators may be verified (say by acting on the basis vectors xi><g)x 2 »: .250 CHAPTER 10 cannot be written as  y > = IVi>®!V2> where and \\j/ 2 } are elements of Vi and V 2 .1. Ti (l) and A^2) (denoted by r.1. this defines the inner product between any two vectors in V i ® V 2 . Let us denote by X]1)<8)<2) the counterpart of X{i\ and refer to it also as the " X operator of particle 1. it suffices to define its action on these.1.11) Since any vector in Vi(g)V 2 can be expressed in terms of the xi>(g)x 2 > basis. (10.e.13). whose action on a direct product ket ©i>(8>®2> is ( r f 1 ) ® A ? ) )  f f l i > ®  a ) 2 > = r 1 (1) G) 1 ><8)A^fi) 2 > In this notation. Pi(1) and J f . P22\ which acted o n Vi and V 2 . Thus C it must be an eigenket of with eigenvalue x i : ^I)®(2)Xl>®l^2>=X1x1>g)X2> (10. N o w the ket xi>(g)x 2 > denotes a state in which particle 1 is at J x . the position and momentum operators of each particle. Since the vectors xi>(g)x2> span the space. In other words X\l)®(2)\x.. from V 2 to V. i.(1)® A 2 2 ) ).lJ Note that x i 1 ' ® ^ 5 does not really care about the second ket x 2 >. respectively. <g) V 2 by defining the momentum operator for particle 2.14) We can similarly promote Pj 2 ) . The inner product of txi>(g)x 2 > and xj>(g))x2> is = 8(x'ixl)8(x'2x2) (10." Let us define its action on V i ® V 2 .y®}x2) = \X\l)xO®)I{2)x2> (10. say. AT(il). as X(l)®(2)=X(l){2)/(2) (10i115) (10. we may write in view of Eq. it acts trivially (as the identity) on  x 2 ) and acts o n xi> just as A^1* did.1. respectively. must have counterparts in Vi(g)V2 and have the same interpretation.
) etc.1* Show the following: (1) [Q.17b) f then I [Q. (2) (n. A{1)®t2)] = r.8 and ( a 2 '2 ) . It is recommended that you reread the preceding discussion after working on the exercise.1.1. respectively (the ± signs label the basis vectors.17d) * The notion of direct products of vectors and operators is no doubt a difficult one. with no simple analogs in elementary vector analysis.+ b 0 — o  ® (2) = Cri . ® V 2 is spanned by four vectors  + > ® ] + > .17a) SYSTEMS WITH N DEGREES OF FREEDOM (operators of particle 1 commute with those of particle 2).1.1.) The space V . Thus 6 = < +  c r r >  .Show (using the method of images or otherwise) that ++ ++ a (1) { ji .) 251 (1) ®/ \ / (2 (I) ® A £ ] = 0 for any Q. +><g)>.17c) and similarly with 1 >2.  ~ > ®  + X ! ~ > ®  _ > . * Imagine a fictitious world in which the singleparticle Hiibert space is twodimensional..) ® (2) + a2 ( l ) ® ( 2 ) ) 2 = ( « i 2 ) ( .1. Let us denote the basis vectors by  + > and  .2.(11®(2). Exercise 10.> ®/ (2) =+. The following exercise should give you some valuable experience. Let + + b d 4tf .(1)<8i/(2) (10. ( 1 > ®a 2 ( 2 > (10. 2) and A^ 2> (10. (. + / h be operators in V.Exercise 10. (l> ®n 2> )(0 ( 1 l> ®A? ) ) = (Q0)l°<g> (rA)£ (2) (10.+ 0 a 0 b 0 c 0 d 0 c d .1. ) ® / ( 2 > + / l l ) ® (Q2) ( 2 ) + 2Q. (4) (Q 1 (. +0 . and V 2 .> .
Ppm2] have the same connotation and commutation rules [Eqs.19). .x 2 plane.1. Notice first that both spaces have the same dimensionality: the vectors xix 2 ) and xi>(g)x 2 > are both in onetoone correspondence with points in the x i . p j 1 ^ ] = m < v ( 1 ) ® / ( 2 ) = m <v(1)®(2> 2 zj»®<2>]= [ p ^ \ p/1)®(2)]=o i. We will.1. . .2 (1 L18) °' N o w we are ready to assert something that may have been apparent all along: the space V i ® V 2 is just Vi® 2 » xi><S>l*2> is just xix 2 >. . . Although the succinct notation suppresses the label ( 1 ® 2 ) of the space on . we shall use the more compact symbols occurring on the righthand side of Eqs. Since X and P are defined by their commutators we can make the identification I We can also identify the simultaneous eigenkets of the position operators (since they are nondegenerate) : *l>(g)X2>=XiX2> In the future.1.17c) it follows that the commutation relations between the position and momentum operators on V i ® V 2 are [ArP®(2). however. return to the concept of direct products of vectors and operators on and off and occasionally use the symbols on the lefthand side.1) and (10. etc.252 CHAPTER 10 (Recall that <a®</? is the bra corresponding to a>®/?>. (10. by taking the matrix product of cVt®(2> and C2 )®(2) and by directly computing the matrix elements of M 1 '® <722).18)]. (10.1. . P2 and X(2)(8>(2). Notice next that the two sets of operators Xx. and A ^ 0 ® ^ is just X\.1.17a) and (10. (10.j= 1.) ~e f 0 0 " g h 0 0 0 0 e / _0 0 (2) g (3) ~ae af be bf ag ah bg bh ( W ^ W ^ ® ^ ce cf de df Sg ch dg dh_ Do part (3) in two ways.. F r o m Eqs. .
) Consider next the operators.1.) (10. ( x ) V 2 . We now expand the function C ^ x i ) in the basis «i(jci): I I C0)2(x\)= X 0.which the operators act.20) Notice that the coefficients of the expansion depend on the value of x i . (10.x2). the CM kinetic energy operator of the twoparticle system: „ 1 CM which really means 2 M T ( ^ ( 1 ) = (P. which represents the abstract ket \\j/} from V ^ .1. If we choose Q = X .1. for \y/) belongs to V l g l 2 and tt> 1 )®o> 2 ) spans Vi(S)V2. p(0®(2) = (i>.22a) What does this expansion of an arbitrary T//(JCJ. it should be clear f r o m the context. which used to be ihd/dxi now becomes —ifi d/dx\.(l)®(2) 253 SYSTEMS WITH N DEGREES PIM 2(ml+m2) " ~ __Pcm_{PI+PI) _PI 2M — 2 2M — + Pj + 2PlP2 2M OF FREEDOM . Similarly let co2(x2) form a basis for V 2 . ) ( g > i f >)2 + 2P\l)®Pp The Direct Product Revisited Since the notion of a direct product space is so important. where the partial derivative symbol tells us it .22b) with these basis vectors we regain Eq.1. (10. If we keep x j fixed at some value. The momentum operator on V i .22b) P t»l 0)2 which means V l t S 2 = V . 2 )^>®< 2 > + ( p 2 ) ( l ) ® ( 2 ) + 2i. we get the familiar basis  x i ) ®  x 2 ) .1 (*. for example.22a). we revisit the formation of V1(g)2 as a direct product of V] and V2. Let Q/ 1 3 be an operator on Vi whose nondegenerate eigenfunctions y/aj^Xi) = «i(Xi) form a complete basis.1. say xu then y/ becomes a function of x 2 alone and may be expanded as V(*i>*2) = I Ca2(xi)a2(x2) 0)2 (10. Consider now a function y/(xi. (In the coordinate basis. By dotting both sides of Eq.1. ( 1 ) (g>/ ( 2 ) ) 2 + ( 7 ( .22a) is the coordinate space version of the abstract result I V> = 1 I C aM> „ 2 fli I >®©2> (10.1. the direct product of the kets coi) and \co2} becomes just the ordinary product of the corresponding wave functions. x2) in terms of ®I(XI) X (a2(x2) imply? Equation (10.21) Feeding this back to the first expansion and dropping the bar on xx we get I = X Z C au(02 o} x (xi)o3 2 (x 2 ) <H[ a>2 (10. Consider. but this time in the coordinate basis instead of in the abstract.
1. This is the coordinate space version p ( i ) ® ( 2 ) _ p ( i ) 0 j ( 2 ) y o u a r e encouraged to pursue this analysis further.e. Px) + H2(X2. We now examine these two classes. Class B: H is not separable. i. \y/(t)} = \E)eiEt/n Eq..254 CHAPTER 10 operates on x t as before and leaves x2 alone. It evolves in time accord™ to the equation P\ mm 2m \ .23) becomes [Hx(Xx.1. P 2m { P2 2m? Vx(Xx) + + V2(X2) = Hx + H2 (10.2: Class A corresponds to two particles interacting with external potentials V\ and V: but not with each other.1.1. (10.e.1. theii energies are separately conserved and the total energy E is E\ + E2. the decomposition means that the two particles evolve independently of each other. while in class B there is no such restriction. Let us see these results reappear in quantum theory.U There are two classes of problems. i. Class A : Separable Hamiltonians. Classically.. X2) \ ¥ } = H\¥) (10.2^ (10. P2)]\E) = E\E) (10. Pi 2m2 V(Xx. . Class A : H is separable. In particular. Evolution of the TwoParticle State Vector The state vector of the system is an element of Vt<g)2. V(XuX2)*Vx(Xx)+V2(X2) and H^Hx+H2 (10. F o r a stationary state.
1. By projecting the eigenvalue Eq. (10.4).28a) 255 SYSTEMS WITH N DEGREES OF FREEDOM It should be clear that the state l / s j ) ® ! / ^ ) corresponds to particle 1 being in the energy eigenstate isi> and particle 2 being in the energy eigenstate \E 2 ).1.1.31) The subscripts E} and E2 have no specific interpretation yet and merely serve as labels.29) in the coordinate basis to illustrate a useful technique that you will find in other textbooks.1.1. say.  > and H?)\E2) = E2\E2) (10.28b) (10. we obtain fi1 d2 K2 d2 V2(X2) 2m i dx2 where ViM 2m2 dx2 WE(Xj .1.1. Feeding this ansatz into Eq. Eq. feeding \E) — l ^ ) ® \ E 2 ) .17a)] we can find their simultaneous eigenstates. Clearly H}E) = (Hl+H2)\El)®\E2) so that E=EX+E2 (10. where j £ i > and \E2) are solutions to i / H £. we took the energy eigenvalues from each space.1.1. (10. and making the usual operator substitutions. (10. the position eigenvectors..30) We solve the equation by the method of separation of variables. (10.1.28) and (10.29) It is worth rederiving Eqs.30) and then dividing both sides by . which are none other than  £ ] ) < g  £ 2 ) = \E\E2).X2) = (X1X2\E) (10. x 2 ) = y/ E Xxi)y/ E2 (x 2 ) (10.1. (10.Since [Hx.1.) Finally. x2) = E\j/E(xj.27) on this basis. E~E\ + E2 into Eq.1. x2) ¥E(XI. instead of.> = £.28c) = (Ei+E2)\El)®\E2) = (Ei+E2)\E) (The basis Z?i>(g)J2i2> is what we would get if in forming basis vectors of the direct product V(x)V 2 .26) we get \il/(t)) = \El) e£Eit/*®\E2) e'iE"/R (10. H2]=0 [Eq. We assume I We(X. (10.
29) on  x i x 2 ) = xi>(g>x 2 >.1. Since xi and x 2 . Case B : Two Interacting Particles.34) to be the projections of Eqs.28) and (10.1. F(X.1.1.) VE.33) and (10. If. We will call these constants Ei and E2. x 2 ) = V(xi ./n e'iE^¥E2(x2) e~iE>!/* (10.34) where y/E} and y/ El are eigenfunctions of the oneparticle Schrodinger equation with eigenvalues E\ and E2i respectively.V(xx. (10. and hence the two functions. equals a constant E. Consider next the more general problem of two interacting particles with Jf= A 4. x 2 ) f? = ¥E](XI) iE.(X\) ft2 82 VE2(XI) 2m2 dx2 + F2(*2) This equation says that a function of xi alone. however.32) breaks down into three equations: 1 WEX* I) 1 n2 2mx 82 82 2 Vi{xi) WEAx\) = E\ ~h2 2m2 ox V2(X2) yE2(X2) = E2 Ei+E2 = E (10.36) .T) = VE(XI . plus one of x2 alone.x 2 ) (10.1.X2.x2)*Vi(xi)+V(x2) Generally this cannot be reduced to two independent singleparticle problems.1.256 CHAPTER 10 WEI(X1)\I/E2(X2) w e get 2 a2 1 . (10.35) where V(xi.1. We recognize Eqs. Consequently WE(X 1 . (10.(JC.r e _2m\ dxi +• 1 + F. it follows that each function separately equals a constant. x 2 ) 2m 1 2 m2 A 4 (10. may be varied independently. Thus Eq.1.
1. x) JPCM ' XCM/F> VEJ?) 1/2 (2 nh) M (10.x2. In the quantum theory.1.P) Pc 2M ""b p2 2p + V(x) (10. In this case ECM = .1. the quantization condition is just [ ^ C M j PCM ] — (10. moving x. which is free.1.39a) (10. one can always.1.40) VE(XCM . with the reduced mass p=m}m2/(m\ under the influence of V(x): .37a) and the relative coordinate X — Xj x2 (10.which describes two particles responding to each other but nothing external.1.6) and Cartesian.P2)^Jt(xCM.41) h — — + £ r el J7_PC „ 2M The real dynamics is contained in ¥ E t J X ) which is the energy eigenfunction for a particle of mass p in a potential V(x). has mass M=m{ +m2 and momentum * pCMMxCM = miXi+m2x2 and another.5. the C M .38) which is just the result from Exercise 2. Since the new variables are also canonical (Exercise 2. p2 P2 ^ + f L + H = v ( x ) 2M and the eigenfunctions of H factorize: 2p (10.pI'. one usually chooses to study the problem in the C M frame. Since the C M drifts along as a free particle. momentum p = px. P] = m and all other commutators zero.37b) reduce the problem to that of two independent fictitious particles: one.1.39b) [X.4 modified to one dimension. by employing the C M coordinate 777 ]X] + 7772*2 257 SYSTEMS WITH N DEGREES OF FREEDOM XcM : m\ + m2 (10.PCM\ '^relative +m2).7.
As mentioned earlier in the book. Uj= I.. x2. by promoting xx. promoting these variables to operators obeying [X„Pj] = iK5vttc. (We are assuming that all the variables are Cartesian. Verify that the normal coordinate also canonical.258 CHAPTER 10 PcM/2M drops out of the energy. d/dxu) in the differentia* equation.. In such cases the oscillators become independent and their energies add both in the classical and quantum cases. II Now quantize the system. the most far ones being Hamiltonians quadratic in the coordinates and momenta which may be reduced to a sum over oscillator Hamiltonians by the use of normal coordinates. by using I and relative coordinates (or other sets of coordinates) reduce the problem independent oneparticle problems.3* Consider the Hamiltonian of the coupled mass system: jf = £L + £L +1 m(02[x\ + jc + (x. In short. . Now change) from xu x2 (and of course c/cx\.x2)2] 2m 2m 2 We know from Example 1. II Write the eigenvalue equation for H in the simultaneous eigenbasis of Xx and X\\.U : X) i Xi >1/2 and the corresponding momenta _P\±P2 PiM=~ >1/2 (1) Rewrite 2tf in terms of normal coordinates.p\. This result (with respect to the quantum oscillators] was assumed in the discussion on specific heats in Chapter 7. d/dx2) to x\. xu (and d/dxt. and X2. i.etc. one can change coordinates and then quantize or first quantize and then change variables in the differential equation. Write the eigenvalue equation for H in the simultaneous eigenbasis of X. and the plane wave factor in y/ representing i motion becomes a constant. one generally cannot.1. N Particles in One Dimension All the results but one generalize from N=2 to arbitrary N. if the change of coordinates is canonical. In general.e. There are a few exceptions.. and p2 to quantum operators..6 that 3fC can be decoupled if we use normal coordinates JCi. You should end up with the result from part (1). Exercise 10. The only excej is the result f r o m the last subsection: for N>2. one can forget all about the C M in the quan^ theory just as in the classical theory. if one warns .} = £. (2) Quantize the system directly./?.8.. that {*.
. We will also write a position eigenket as r> and the orthonormality relation (xyz\x'y'z') = 3 5(xx')5(yy')5(z — z') as <r[r'> = £ (r—r'). (2) Write down the corresponding wave functions in terms of single oscillator wave functions. Exercise 10. Recall that a particle in a onedimensional box extending from x = 0 to L is confined to the region 0 < x < L. convenient to use a different notation in the two cases. and Py. and the x. are positive integers. and so on. We will denote the two Cartesian coordinates of the single particle by x and y rather than X. Consider now a particle confined in a threedimensional cubic box of volume L3. it is best to first quantize the Cartesian coordinates and then change variables in the differential equation. N are involved. however. and z axes along the three edges meeting there. . Choosing as the origin one of its corners.mo>yy2 2 2 i 2m (1) Show that the allowed energies are E=(nx+\/2)ncox+(ny+\/2)ncoy.5). \pxpy). ..2.) 259 SYSTEMS WITH N DEGREES OF FREEDOM 10. its wave function vanishes at the edges x = 0 and L and beyond (Exercise 5.2. . y ^ ~ y and that the parity depends only on n = nx + ny. * Quantize the twodimensional oscillator for which Jf + . Likewise the momenta will be denoted by px and py.2. and x2. It is. The generalization to three dimensions is obvious. When several particles labeled by numbers 1. nx.ny = 0 . 2 .2. their common eigenkets \xy). Verify that they have definite parity (even/odd) number x>—x. Thus PiP2> will represent a twoparticle state in which particle 1 has momentum pi and particle 2 has momentum p2 and so on. More Particles in More Dimensions Mathematically. . the problem of a single particle in two dimensions (in terms of Cartesian coordinates) is equivalent to that of two particles in one dimension. show that the normalized energy eigenfunctions are where E= n2nl 2 ML' and n. and Px. 1 . this extra label will also be used.to employ nonCartesian coordinates. The quantum operators will be called X and Y. y.2. . i ^ (ni + 4 + ni) 2 Exercise 10.mcoW + . . The same goes for the momentum eigenkets p> also.1* (Particle in a ThreeDimensional Box). respectively.
We shall say two particles are identical if they are exact replicas of each other in every respect—there should be no experiment that detects any intrinsict difference between them. Near holes 1 and 2 rest two identical billiard balls. Imagine a billiard table with four holes. Exercise 10. 10. Two identical billiard balls start near holes 1 and] end up in holes 3 and 4. Let us call these balls 1 and 2. The Classical Case  • Let us first orient ourselves by recapitulating the situation in classical physics. Although the definition of identical particles is the same classically and quantum mechanicallv the implications are different in the two cases. numbered 1 through 4 (Fig. such as its charge or mass and not its locatior. The diction of P2. normalized eigenfunctions of the first three states (that is. nx. the holes near which they find themselves. . 10. Show that the degeneracy of a level with f E=(n + \)fia> is (w + 1). . leads to some very surprising results. even though the two final configurations would be in tinguishable to an observer who walks in at t = T. . when properly applied to a system containing identical particles. for the cases « = 0 and 1). or momentum.ny. 2 . Reexpress the first four states in terms of spherical coordinates. that they would end up in holes 4 and 3.3. (2) Write the corresponding eigenfunctions in terms of singleoscillator wave functions and verify that the parity of the level with a given n is (—1)".* Quantize the threedimensional isotropic oscillator for which ^=PI+P2y+P'+1I M P V + / + *2) 2m 2 (1) Show that E=:(n + 3/2)fia). Identical Particles The formalism developed above. Reexpress your results in terms of polar coordinates p and <r (for later use). respectively is wrong. (3) Consider next the isotropic oscillator (a) x = coy). namely. . Show that the degeneracy of a level with energy £= (n + 3/2)fao is («+l)(« + 2)/2.1).1. as predicted by P\. respectively.3. t By intrinsic I mean properties inherent to the particle. nz = 0. .2. Write explicit. 1.t=0 Figure 10. The difference between the labels reflects not any intrinsic difference in the balls (for they are identical) but rather a difference in their environments.
there must exist some experiments in which these two configurations are inequivalent. We declare that Px is correct and P2 is wrong. the resulting configuration would appear exactly the same. the two balls which seemed identical to us would be distinguishable to someone who had been following them from an earlier period. To summarize. Although both balls appear identical to the newcomer. Consequently two configurations related by exchanging the identical particles are physically nonequivalent. Now. the distinction exists. we are able to trace the ball in hole 3 back to the vicinity of hole 1 and the one in hole 4 back to hole 2. it is possible in classical mechanics to distinguish between identical particles by following their nonidentical trajectories (without disturbing them in anyway). Similarly at t = 0. We will now discuss one such experiment. Now of course it is not really necessary that either we or any other observer be actually present in order for this distinction to exist. and one that will play a dominant role in what follows. two players propel the balls toward the center of the table. that the newcomer does not? The answer of course is—their histories. In order for this distinction to be meaningful. What do we know about the balls (that allows us to make a distinction between them and hence the two outcomes). Imagine that at time t = 0. TwoParticle Systems—Symmetric and Antisymmetric States Suppose we have a system of two distinguishable particles 1 and 2 and a position measurement on the system shows particle 1 to be at x = a and particle 2 to be at . if he can make the distinction. that if these two balls are exchanged. Nonetheless these two configurations are treated as distinct in classical physics. which completely outlaws the notion of continuous trajectories for the particles.Now it follows from the definition of identity. there exists no physical basis for distinguishing between identical particles. Consequently two configurations related by the exchange of identical particles must be treated as one and the same configuration and described by the same state vector. One imagines in classical physics the fictitious observer who sees everything and disturbs nothing. is that in quantum theory. If seen in isolation they would appear identical: an observer who walks in just at t—T and is given the predictions of Pi and P2 will conclude that both are right. At once two physicists P\ and P2 take the initialvalue data and make the following predictions: 261 SYSTEMS WITH N DEGREES OF FREEDOM Say at time T we find that ball 1 ends up in hole 3 and ball 2 in hole 4. An immediate consequence of the above reasoning. We now proceed to deduce the consequences of this restriction. the configurations predicted by them for t=T differ only by the exchange of two identical particles.
upon equating the coefficients of  a b ) and \ba) that 0 = ay. the state vector just after the measurement. t We are assuming a arise.y\ba) = a[P\ba) 4. We write the state just after measurement as I V> = *i = «. are physically distinct) neither is acceptable. The problem is that our position measurement yields not an ordered pair of numbers (as in the distinguishable particle case) but just a pair of numbers: to assign them to the particles in a definite way is to go beyond what is physically meaningful in quantum theory.3.3.y\ab)] we find. but the choice we are agonizing over does 9 fl .2): (S\ab) 4. given an unordered pair of numbers a and b we can still define a unique sum (but not difference). Since  y ) and a\ if/') are physically equivalent. the state is acceptable. If we impose the constraint Eq. satisfy the constraint mL \y(a. 9 yap iH If a = b.a)) (10. neither. Since the particles are distinguishable. Since under the exchange \ab) «> \ ba) and the two vectors are not multiples of each o t h e r j (i. (10. x2 = b) = \ab) (10. and the state vector lies somewhere in the twodimensional degenerate (with respect to X\ +X2) eigenspace spanned by them. Suppose we repeat the experiment with two identical particles and catch one at x~a and the other at x~b.262 CHAPTER 10 x = b. Now. we require that  y/(a. We have seen that in quantum theory two configurations related by the exchange of identical particles must be viewed as one and the same and be described by the same state vector. there are just two product vectors. In other words. Is the state vector just after measurement  a b ) or ba)1 The answer is. b)). Let \\}/(a. What our measurement does permit us to conclude is that the state vector is an eigenstate of X} +X2 with eigenvalue a+b. b)) = P\ab) + y\ba) be the allowed vector..e. It is given by and corresponds to having found particle 1 at b and particle 2 at a.1) where we are adopting the convention that the state of particle 1 is described by the first label (a) and that of particle 2 by the second label (b). ab) and \ba) with this eigenvalue.b)) = a\¥(b.® mI fl where a is any complex number. the sum of the eigenvalue being insensitive to how the values a and b are assigned to the particles. the state obtained by exchanging them is distinguishable from the above.
They are \ab. photon. S ) or ft)ifi)2. and particles such as the electron. x2 = b} + \x\=b.3. let co represent the eigenvalues of a complete set of commuting operators.t Although we have made a lot of progress in nailing down the state vector corresponding to the measurement. So we rule out this possibility.3) 263 SYSTEMS WITH N DEGREES OF FREEDOM It is now easy to construct the allowed state vectors.3. A}. A) = jab) ~\ba) (10. . If not.) More generally.5) called the antisymmetric state vector (a = — 1).  Thus if we catch two identical bosons. (These are unnormalized vectors. proton. F \\(/} = \x]=a.so that = ±1 (10. Their normalization will be taken u p shortly. if some variable Q is measured and the values coi and CD 2 are obtained. the state vector immediately following the measurement is either  ca \ca2. Bosons and Fermions t Although both S and A states seem physically acceptable (in that they respect the indistinguishability of the particles) we can go a step further and make the fallowing assertion: [ i A given species of particles must choose once and for all between S and A states. 1: Nature seems to respect the constraints we have deduced. one at xa and the other at x = b.3. Particles such as the •ion. S} t We are assuming Q is nondegenerate. Then the space also contains linear combinations feuch as i Bv Pnich are neither symmetric nor antisymmetric. ' Suppose the contrary were true. we have still to find a way to choose between these two alternatives. x2 = a) =\aby + \bay = \ab.S) = \ab) + \ba) (10. and neutron are always found in Intisymmetric states and are called fermions.4) called the symmetric state vector (a = 1) and \ab. and graviton are always found in symmetric states and are called IOJO/W. the l ^ t e vector immediately following the measurement is B. and the Hiibert space of two identical particles ©ntained both S and A vectors.
we do not particles in any particular way. which relates the spin of the particle toil "statistics"—which is the term physicists use to refer to its bosonic or fermionic nature. (This should be clear classically. Before going on to this second method. astrfl physics. Howev* this connection. twofermion This is the celebrated Pauli exclusion principle: Two identical fermions cannot b&J the same quantum state.A> Note that although we still use the labels x\ and x2. 2ft. let us address a question that may have plagued you: our analysis has only told us that a given type of particle. where it is not possible to define spin or any form of angular momentum. as will be clear by going through the arguments in one dimension. A} = \WCO)\Q}CO} = 0 (10. With this important derivation out of our way. The only difference will be the increase in the number of labels. the position . has to be either a boson or a fermion. The first is by further cerebration. A}=\(0i(02} Let us now set to i = w2 = co.. 3ft/2. Recall that the spin of the particle is its internal angular momentum. let us note that the requirement that the state vector of two identical particles be symmetric or antisymmetric (under the exchange of the quantum numbers labeling them) applies in three dimensions as well. ft. This principle has profound consequences—in statist® mechanics. the state vector after the measurement would ] been  y} = =a. x2 = b} — JCI =b. proven in three dimensions. We will have occasion to return to it often. withk the framework of quantum field theory. to be specific. The magnitude of spin happens to be an invariant for a particle (and thus serves d a label.. to the same ip> for bosons. does not apply to one dimension.264 CHAPTER 10 Had the particles been fermions. The spin statistics theorem. For example. and those with spin equal to an odd multiple of ft/2 are fermions. and to state vectors differing only for fermions.) Thus the only way to find if a particle in one dimension is a boson or fermion is to determine the symmetry of the wave function experimentally. provable in quantum field theon asserts that particles with (magnitude of spin) equal to an even multiple of ft/2 • bosons. Since the relevant arguments are beyond the scope of this text I merely quote the results here. = a and x2 = b or vice versa. We find \cow. in understanding chemical properties of atoms. There are two ways to the answer. We are now in a position to deduce a fundamental property results from the antisymmetry of their state vectors. in nuclear theory. like its mass or charge) and can have only one of the following values:! fi/2. This is the second method.36) — \(02(Dx) attach them to I at x = a and i = j Either choice lea by an overall sign of fermions.. but does not say which one. Thus having caught the bosons we need not agonize over whether x.. etc. Consider a  d)\C02. to be discussed in a moment. say a pion. x2 = a} = \ab} — \ba} = \ab.
e. (1)2 = b). (o2 = a) there is one (unnormalized) bosonic vector \coi=a. call them + and . v. As a prelude to the discussion of such an experiment. We now turn to the second way of finding the statistics of a given species of particles.9) I Since spin has no classical counterpart. i.3. Let us call the Hilbert space of symmetric bosonic vectors V 5 and the Hilbert space of the antisymmetric fermionic vectors V ^ . There is no corresponding fermionic vector (the Pauli principle). co2 = a) and one fermionic vector \coi = a. because it appeals to a simple experiment which determines whether the twoparticle state vector is symmetric or antisymmetric for the given species. the method that works in one or three dimensions. co2 = b} + \a}]—b. (i)2 = by — \(Oi—a. under the exchange of all the labels.3. We first examine the relation between these two spaces on the one hand and the direct product space V ^ on the other.eigenket of a spinzero boson will be labeled by three numbers x. 002S2) ~~  (02$2 j (10. and z. Thus two electrons can be in the same orbital state if their spin orientations are different.3. We express this relation as V1(g)2 = V 5 ® V ^ (10. A} = \W\Si. the operator representing it is not a function of the coordinate and momentum operators and it commutes with any orbital operator Q. The space V l g ) 2 consists of all vectors of the form £0]£02> = l®i)<B>l®2>. If we denote by a all the orbital labels and by s the spin label. (o2 = a) is already symmetric and we may take it to be the bosonic vector. let us study in some detail the Hilbert space of bosons and fermions. the states will be labeled by the orientation of the spin as well as the orbital labels that describe spinless bosons. will be of the form 265 SYSTEMS WITH N DEGREES OF FREEDOM £02^2. but we mean by a quantum state a state of definite to and s.. Thus V1(g)2 has just enough basis vectors to form one bosonic Hilbert space and one fermionic Hilbert space. . Thus spin may be specified simultaneously with the orbital variables.or spin up and down (the meaning of these terms will be clear later). J We shall consider just spini particles. If a = b. the vector \wx = a. for which this label can take only two values. Q)2 — b} and \a)\=b. the state vector of the fermion that is antisymmetric under the exchange of the particles. which have spin at least equal to fif 2. Bosonic and Fermionic Hilbert Spaces We have seen that two identical bosons will always have symmetric state vectors and two identical fermions will always have antisymmetric state vectors.To each pair of vectors \coi=a.8) Thus we find once again that two fermions cannot be in the same quantum state. For fermions.7) We see that the state vector vanishes if a)i = Q)2 and S] = S2 (10.
You may readily check that ict»2. S ) = jcoco) (10.J dist where £ d j s t denotes a sum over all physically distinct states.1 Any vector  \f/s> in V s may be expanded in terms of this f i basis. If co\=co2 = co the product ket 1coco) itself both symmetric and normalized and we choose coco.3. S\ ysy\' (10. i. S} corresponding to a variable Q with discrete eigenvalues.3.n (10. S} when measurement is made on a system in state  \f/s}. S ) and \w2(Oi. a2) may be written as 1 =<VslVs>= £ dist (10. Thus in studying \ identical particles we need only consider V s or V A. the normalization facta is just 2" 1 / 2 .3. Another way is to count them both and then divide by 2. It is however convenient. As usual ! identify Ps(co\. to view V s and VA as subspaces of V® 2 and the eleme of V s or V A as elements also of Vi® 2 . J Since every element of V. . cq2) = \((0\(02.is perpendicular to every element of V^ (you should check this) the dimensionality of V ^ equals the sum of the dimensionalities of V s and V^.3. the state of two bosons is an element of V s i that of two fermions an element of VA. It can also be shown that a system starts out in V 5 (V^) remains in V^(V^) (see Exercise 10. \(ox(o2y + !<»2<»I> Since <»i<02> and o 2 <»i) are orthonormal states in V l g ) 2 ..e. Let us now consider the normalization of the vectors in V 5 . S). W\(D2. Consider first I eigenkets \a)1a)2.12b) In this manner we avoid counting both <»i<02. unnormalized state vector is \(ox(o2. bookkeeping purposes.5).3. then (02 dist 1 = 1 I 0)2 = Wmm a>t=a)m.1 is a normalized eigenvector.The normalization condition^ j \j/s) and Ps(coi. 2 4 Our analy has shown that at any given time. which are physically equivalent.266 CHAPTER 10 with V s getting slightly more than half the dimensionality of V 1 ( g .s. S) = The preceding discussion assumes cox ^co2.3. S\<*>\(02.S') = 2 1/2 [  « I « 2 > + £0 2 £0 1 >] (10.111 as the absolute probability of finding the particles in state \cojco2. If coi and co2 take va between comin and co max .
X2)\2 dx\ dx2 (10.3.51 \f/S)\ (10. . S\\f/s) (10. x) are all given by the limits x t *x 2 ^x of Ps{xi.x2) 2 Vs) + (x2XI I \f/S> ] = <*1*2 Vs) (10.17) However. 5 ) . in this case Ps(.13) to obtain Ps(xu * 2 ) = l<*i*2 S S\ ys}\: 267 SYSTEMS WITH N DEGREES OF FREEDOM (10. x 2 ).19) where we have exploited the fact that  y/s) is symmetrized between the particles and has the same inner product with <(x\x2\ and (x2x}\. 51 y/s> =  2[ y/s(.s(X.Xi.3.x.v(. We do not get into these since the points on the line x\ =x2.x\. note that 1 =Y f 1 <*1*2 . the normalization J The points J J = x2 = x pose some subtle questions both with respect to the factor 1 / 2 and the normalizaC tion of the kets I**. dx i dx 1 = Ps(x. x2) 2 x2) and  y s ) is 2 dx\ dx2 <JClJC2. etc. x2)= due to the rescaling.What if we want the absolute probability density for some continuous variable such as J ? In this case we must take the projection of  \ff S ) on the normalized position eigenket: (10. x 2 ). In the following discussion you may assume that quantities such as Ps(x. Now.15) where the factor 1 / 2 makes u p for the double counting done by the dxi dx2 integration^ In this case it is convenient to define the wave function as ys(xi..3.3. x 2 )l' (10.3. ys(xi.x2) so that the normalization of y/ s is =2 1/2 <*ix2. y/. Consequently.3.18) .x make only an infinitesimal contribution to the integration in the xxx2 plane (of any smooth function). x).14) The normalization condition for Ps(xi.3.16) 1 = <//.
A) = 2~1/2[\(o]Q}2)\(o2o)i)] (10. The normalized state vector just after measurement is then 3. (10.22) . 4 ) + <*2*I4.3. is related to it by jt scale factor of 2 1 / 2 .4) + 4.* 2 4.3) = Tyi I ( 10 . x2) = 2 1/2 (xix2. thatj kets enter the definition of the wave function Eq. only as bookkeeping devices.3 .20) in obvious notation. Note. and the norma tion integral above. (^) (10.21a) MM S i n = <x. however. as is an element of V l l g > 2 as well. We measure the energy of two noninteracting bosons in a box extending f r o m x = 0 to x = L and find them to be in the quantum states n = 3 and n = 4.21b) These considerations apply with obvious modifications to the fermionic space V A • The basis vectors are of the form \Q}. 3 ) + <x 2 *I3.17) becomes l = <Vs\ Vs) =  ys2 dx i dx2 = (Vski^Xxi^i Ys) dx 1 dx2 which makes sense.Q}2.3.19). were it not for the: that the quantity that is of physical interest < x x x 2 . S\\f/s). They are not elements of the inner product (. (10.3.268 CHAPTER 10 condition Eq.3.3.x2\ Ws} where in all of the above. Let us now consider a concrete example. 3>] 2(2 ' ) + ^4(^2)^3(^1)] = 2~]/2[\f/i(x])\f/4(x2)+W4(xx)\j/3(x2)] (10. The wave function is Vs(xi. S\ \ f / s ) = ZyZ ) 4 ) + <x.x\x 2 \y/) would be of no interest to us.
(10. 3 .4>4.27) l/2 V3O1) y4(x 1) V4(*2) Vl(x2) Had we been considering the state \(D\(o2.X2) The normalization condition is dx i dx 2 Pa(xi.3.3. 3 ) . x2) = 2~1/2(x^x2 = <*l*2l Va) 269 SYSTEMS WITH N DEGREES OF FREEDOM .X2) =2 (10. The corresponding wave function may be written in the form of a determinant: Va(xi.J J1/2 Va(xi. whether it is a boson or fermion. A\\f/A) (10. JC2)r (10.3. let us say we have two identical noninteracting pions and wish to find out if they are bosons or fermions.y4(x 1)^3(^2)] (10.28) (XL) Determination of Particle Statistics We are finally ready to answer the old question: how does one determine empirically the statistics of a given species.e..co2.24) 1 = Va(xI . x2) = l\ \)/A{x 1.3.22)]. without turning to the spin statistics theorem? For concreteness.26) (We may equally well choose K.3> \Va> = >1/2 (10. x2)\2 dxx dx2 (10.) The wave function is once again yA(xi.(The case ©] = (02 does not arise here.3. i. A) [Eq.3. if we had obtained the values n=3 and n = 4.v.3. . x2) = <. = x 2 or coi .23) and as in the bosonic case Pa{xi.25) Returning to our example of two particles in a box.xix2\ yA) = 2 = 2 1/2 [Vs(xi) \i/4(x2) . 4 ) I ¥A> which makes no physical difference). then the state just after measurement would have been 3. J The determinantal form of y/A makes it clear that \j/A vanishes if .
whose labels are disregarded in the position measurement. *2> . The basic idea works just as well in three dimensions. the interference terms arise. the probability for: one at Xi and the other at x 2 . One refers to the tendency of fermions to avoid each other (i. the probability density for two dis label carrying (but otherwise identical) particles. we are not allowed now to assign definite labels to thej particles.P 0 (xi>x. because in q^ turn theory. (The subscri]: denotes distinguishable. Say we find one in the state n = 3 and the other in the state n. remind^ that there is more to identical particles in quantum theory than just their iden^ characteristics: they have no separate identities. The differ between the two cases is most dramatic as X]—•x 2 —: PA(XI»X. )  2  V4(X2)\2 ) ^3(^2)112 + \ MX. The sum of these two terms then gives PD(x\. § The label can.)2 ±[\l/*(x])\i/4(x])\f/Z{x2)V3(x2) + (xO^C^i)^3(^2)^4(^2)] (10. called interference terms. with no regard paid to their labels. but other* identical. the first term represents the probability that particle 1 isi and particle 2 is at x 2 . PS/A(XI . where the probabilities for finding a ] at a given point x on the screen with both slits open was not the sum of the proba^ ies with either slit open. probability distribution in x space would be.§ with particle 1 in state 3 and described by a probability distribt] j«//3(x)2. We put them in a onedimensional boxt and make an i measurement. x 2 >x)>0 (Pauli principle applied to state  x » (lOj whereas Psixi^x. depending on their statistics. In both cases. There is a parallel between] situation and the doubleslit experiment.2 = 2\2~U2[\f/3(xl)lf/4(x2)± = I V 3 ( * . Had they separate identities (a the classical case) and we were just indifferent to which one arrives at xi and wl one at x 2 . avoid the state x i = x 2 = x) as obeying "FermiDirac statistics" and the tendency of bosons to I We do this to simplify the argument. and particle 2 in state 4 and described by the probability distribu  \f/4(x)\2. X 2 ) = 2 \F/S/A{X\ .270 CHAPTER 10 We proceed as follows. In this case. The interference terms tell us if the pions are bosons or fermions. for example. Just as we were not allowed then to assign a definite trajectory to the paf (through slits 1 or 2).) The next two terms. we would get just the first two terms. while the second gives the probability for the excha event.3 Compare this situation with two particles carrying labels 1 and 2.4. x2^x) = 2[\ v3(*)2 ^4(*)2 +1 M*)2I Vi(X)\2] (1( which is twice as big as .. when an event can take place in two (or more) indistinguishable we add the corresponding amplitudes and not the corresponding probabilities. .e. be the electric charge. x 2 ). x 2 >x).
—) pairs. while the ( + . x2) (once again ignoring the hypercharge). but not zero. In other words. As mentioned earlier on. the nature of the interference term will betray this fact. s can have only two values. P(x. If we now make measurements on the ensemble and extract the distribution P(x\. The (+. It will then be up to us to find the nature of the hidden degree jf freedom which provides the distinction. x 2 ) helps us decide not only whether the particles are bosons )r fermions. From the above discussions. if we get a ratio less than 2. we can readily imagine ieciding (once and for all) whether pions are bosons or fermions by preparing an lsemble of systems (with particles in n = 3 and 4) and measuring P(xi. r) describing labeled particles. they are like two distinct particles and can be in the same orbital state.for the other. Since we are ignorant of spin. and n3. but also whether they are identical in the first place. (—.) pairs. Let us say that an energy measurement shows the quantum numbers of the particles to be «i. Note that P(jci . Since the particles are identical. + ) pairs. which are identical with respect to mass and charge. r) smaller than Pd{*." Let us assume we are ignorant of hypercharge. respecrely. K) pairs. call them + and —. the label we are ignoring is the spin orientation s. which describes labelcarrying particles. it is also clear that one cannot hastily conclude. The safe thing to do here is once again to work with an ensemble rather than an isolated measurement. —) pairs are identical fermions and will produce t negative interference term. our ensemble will contain ( + . upon catching two electrons in the same orbital state in three dimensions that they are not fermions. ad (+. n2." after the physicists who first plored the consequences of the antisymmetrization and symmetrization requireents on the statistical mechanics of an ensemble of fermions and bosons. some (K. two bosons. if ^particles that we think are identical differ with respect to some label that we are not aware of. Systems of N Identical Particles The case N= 2 lacks one feature that is found at larger N. x). we will find the interference term has the + sign but is not as big as it should be. x2). x) should be twice as big as PD(x. (This is the reason for referring to the bo sonic/fermionic nature of a particle its statistics.jnglomerate as "obeying BoseEinstein statistics. for example.) Given the striking difference in the two distributions. call them K and K. we will actually be including some (K. K) pairs. + ) and (—. If we assume that 5 never changes (during the course of the experiment) it can serve as a particle label that survives with time. Imagine.) pairs will not. . Thus we will find P(r. but different with respect to a quantum number called "hypercharge. . If s = + for one electron and . we know the ensemble is contaminated by labelcarrying particles which produce no interference terms. all we can conclude from this observation is that the total lergy is E=\ 271 SYSTEMS WITH N DEGREES O F FREEDOM (ft2K2 \2mL2 Itf + nl + ni) . In preparing an ensemble that we think contains N identical pairs. We illustrate it by considering the case of three identical particles in a box. In this case. If the ensemble contained only identical bosons. This will tell us it our ensemble has identical fermion pairs contaminated by pairs of distinishable particles.
Since double exchange of the same two labels is equivalent to no exchange. Bosons will always pick the S states and fermions. (10.33) «2«1«3>+«3"1«2>«3«2«1>] called the totally antisymmetric state. the wave function associated with « 1 « 2 «3 ) S/A).3. = (3!) 1 [l«lW2«3> — l«lW3«2> + «2«3«1> (10. The physical states are elements of the sixdimensional eigenspace spanned by these vectors and distinguished by the property thai under the exchange of any two particle labels. the wave function in the X basis is VS/A(Xu x2. nt=n2= n3 = n. ^^M . and M3«IM2>. the state vector changes only by a factor a . S)= /2 [HiM2"3> + "l«3"2> + "2«3"l> + I«2«l«3> + «3«2«1> + «3«1«2>] called the totally symmetric state.272 CHAPTER 10 N o w there are 3! = six product states with this energy: \nin2n3}. and 2 /ll«2W3. the A states. then the product state  nnn) is normalized and symmetric and can be used as the S state. for which a = — 1 for all three possible exchanges. for example. x 3 ) = ( 3 ! ) " 1 / 2 < x i x 2 * 3 .(*3)± + \ifH3(xi)\ifni(x2)\ffn2(x3) )\f/„Xx2)Wn3(x3) ± ynj{xi)\(/n2{x2)ynXxi)\ (10. As in the N= 2 case. is Eqs.! for which a ~ + 1 for all three possible exchanges (1 2. we conclude as before that a = ± 1. It follows that no two fermions can be in the same state.35) + The normalization factor (3!) 1/2 is correct only if all three n's are different. (10.334) and /* O O I VS/A\2 dx 1 dx2 dxj. « 2 »3fli). A similar question does not arise for the fermion state due to the Pauli principle. = 1 oo —oo —oo F o r instance. S/A\ y/s/A> = < x i x 2 * 3  Vs/a} (10. 1 3).3.34). If.33) and = (3!)":1 / 2 [ ( X i ) ^„2(x2) ^3(^3) ± Wmt*l) ^3(^2)V m (x 3 ) + ^2(xi)^„3(x2)v^. «i«3«2>.3. 2 *» 3.3. There are only two states with this property: 1 «!« 2 «3.
Consider the case N= 3.1 ? Such states do not exist. In our example. this separability is ruined (to give just one example) by the gravitational interaction between the mass and Pluto. (1 <• 3). when we study the oscillations of a mass coupled to a spring. Two questions may bother you at this point. it still needs to be formally ruled out. while the A states pick up a — 1 for all three exchanges. When Can We Ignore Symmetrization and Antisymmetrization? A basic assumption physicists make before they can make any headway is that they can single out some part of the universe (the system) and study it in isolation from the rest. the direct product space for /V>3 is bigger (in dimensionality) than VS®VAQuestion II. i. What if two pions always pick the S state while three always pick the A state? While intuition revolts at such a possibility.3. Classically.. All these results generalize directly to any higher N. While no system is truly isolated. Since a general proof for this case and all N involves group theory. we will not discuss it here. Note that since we get only two acceptable vectors for every /V! product vectors. and (2 3).The fermion wave function may once again be written as a determinant: Vn2(x l) (3!) 1/2 V«i(*2) V« 2 (*2) 273 SYSTEMS WITH N DEGREES OF FREEDOM V«3(*l) (10. We do so at the end of the next subsection. There are three possible exchanges here: (1 <>2). For instance.3. the isolation of the system is expressed by the separability of the Hamiltonian of the universe: ^universe ~ J^sys ^rest ( 10. What about states for which some of the a ' s are +1 and the others . one easily sees that y/ vanishes if two of the x's or n's coincide. Question I. one can often get close to this ideal. It follows that the time evolution of the system's p's and q's are independent of what is going on in the rest of the universe.36) VsC^) V«i(*3> Vmfa) Using the properties of the determinant. we ignore the gravitational pull of Pluto. .e. You may verify this by exhaustion: take the 3! product vectors and try to form such a linear combination. so will three or more. which depends on their relative separation.37) where J%ys is a function of the system coordinates and momenta alone. The S states pick up a factor a = + 1 for all three exchanges. we have extended our definition of bosons and fermions from N= 2 to all N. We have tacitly assumed that if two identical particles of a given species always pick the S (or A) state. If we neglect this absurdly small effect (and other such effects) we obtain separability to an excellent approximation.
esi) P(x s) P (xrtsi) (10..38) where i//sys is a function only of system coordinates. the uncertainty in the position of each pion is compensated by a separation that much larger. If we catch a pi{ somewhere on earth at time t. xM) = Ge(Xe)GM{XM) (10. X r e s t) d x trest = 1 Vsys(* s )( 2 = I Vsys(^)t2 rest)  d xrest (10. N o w it seems reasonable that at least in certain cases it should be possible! get away with the product state and ignore the symmetrization or antisymmetrizatij conditions. Suppose. the band of uncertainty about each trajectory merely to be much thinner than the minimum separation between the particles di their encounter. we will no longer have P(xs. X[. (Even in classical mechanics. i.3. but on the moon. we can follow their wave functions: we know the first wa function is a Gaussian GE{xE) centered at a point in the lab on earth and that second is a Gaussian GM(xM) centered at a point on the moon. and i not care about the rest.3. Even if there is no interaction between the system and the rest.3.e. it is not necessary to know the trajector exactly to follow the particles.. collectively referred to as xK Thus if we want the probability that the system has a certain coordinate JC^. the Ha tonian is separable. It seems reasonable that we can give the particles the lat "earth pion" and "moon pion. the systems will not be statistically independent). for example. Although we ca follow their trajectories.) We therefore believe that if we assumed \f/(xE." which will survive with time. product states are not allowed and only S or A states must I used.4 . We could have obtained this result by simply ignoring ^ r e s t from the outset. Once the state vector fails to factorize.274 ruiiyrcD m Quantum mechanically. the theory tells us that it is almost certainly the " pion" and that the chances of its being the "moon pion" are absurdly small. we find one pion in the ground state of! oscillator potential centered around a point on earth and another pion in the sar state. that at t = 0. we find (symbolically) P(Xs) = Vuniverse(Xj.3. separability of H leads to the factorization of the wave function of the universe: Vuniverse V s y s ' Vrest (10.e. and we can not integrate' P(xTest) and regain P(xs).4 (i. Things get complicated when the system and the "rest" contain identical par cles.
e.. the person on earth can compute the probability for finding the earth pion at some x by integrating out the moon pion: I GM(xm )  2 dx\t = \GE(XE)\: 275 SYSTEMS WITH N DEGREES OF FREEDOM (10.\2=  G ^ 1 )  2  ^ 2 )  2 + + I ^ O l ' t ^ W i 2 G%(Xx)Gm{Xx)GUX2)Ge{X2) (10. Jti)]. and one near x2 is P(xx.46) The first term is what we would get if we begin with a product wave function Eq. x2) = 2 y.3.3. X2) = 2~X/2[GE(XI)GM(X2) + GM{XI)G^X2)] (10. This is given by setting either one of the variables (say *i) equal to xE and integrating out the other [since P(xi. sums over) the earth pion will obtain * P{xM) = \GM(xM)i (10. where xM is a point on the moon. Similarly if we asked for P(xM).41) and integrate out xM. we will again get \ GM(XM)\2 . the probability (density) of finding one particle near JC. We get \GE(x2)\2dx2 I GM(X2)  2 dx2 + 1 GM(XE)\: + GE(XE)GM(XE) I GUX2)GE(X2) dx2 + GUXe)Ge(XE) G%(X2)GM(X2) dx2 (10. Let us start with Vs(xi. (10.3.43) Let us now verify that if we took a properly symmetrized wave function it leads to essentially the same predictions (with negligible differences).45) + GUxi)GE(Xl)G%(x2)GM(x2) Let us ask for the probability of finding one particle near some point xE on the earth. Now. Given this product form. x 2 ) = P(x2. The other three terms are negligible since \jm IS peaked on the moon and is utterly negligible at a point xE on the earth. who does not care about (i.3.3.44) We use the labels xi and x2 rather than xE and xM to emphasize that the pions are indeed being treated as indistinguishable. with no regard to the other. There is no need to divide by 2 in doing this integration (why?).we should be making an error that is as negligible as is the chance of finding the earth pion on the moon and vice versa.42) Likewise the person on the moon.3.
by assumption. . XE2 . Sucl X It is being assumed that the particles are free. we must symmetrize between them. But i labeling is quite useless.276 CHAPTER 10 The labels "earth pion" and "moon pion" were useful only because the Gaussians remained well separated for all times (being stationary states). 0) 2 (the "fate" of the defl function)! or by noting that AP= oo for a delta function (the particle has all possilj velocities from 0 to oo) and which. and were wave pack drifting toward each other. XE2) ' W(XM) (10.4 The extension of this result to more particles and to fermions is obvious. point is that at the start of any experiment. W(xei . If thei Gaussians had not been bound by the oscillating wells. one at x = a and the other at x — b. These labels acquire a physical significance only if they survive for some tin Labels like "a particle of mass m and charge + 1 " survive forever. Let two of the three pions be on earth and the third one < the moon. therefore.1 (10. XM) — yA(XE\ » XE2) ¥ ( X M ) If we integrate over the moon pion we get P(xE.. and the total function will be. We are thus led to conclude that two pions on earth will have a probability distribution corresponding to two fermions if there is a third pion on the moon and t distribution expected to two bosons if there is not a third one on the moon. since after the passage of an infinitesimal period of til the delta functions spread out completely: the probability distributions become< stants. a. All these considerations apply with no modification to two fermions: the cases differ in the sign of the interference term. t. one can always assign the particles soij labels. the labeling (and the factorized wave function) woij have become invalid when the Gaussians begin to have a significant overlap. We may describe the initial state by a product wave function. A dramatic illustration of this point is provided by the following example.49. You may verify this by examining \ V(x.3. XE2)\: (10. Then. which is irrelevant to considerations. XE2. t — 0 we catch two pions. spreads out in no time. while the longev of a label like "earth pion" is controlled by whether or not some other pion isj the vicinity.3. We can give them labels a and b since the two delta functions do not overlap even if a and b are in 1 same room. in obvious notation. Suppose threepion systems picked the A state while twopic systems picked the S state. At this point the answer to Question II raised at the end of the last subsect becomes apparent. Wfiat if there are three pions. XM) — WS{Xei . the following function should provide an excelle approximation: . xEl) = 2 y/A(xE].3. two on earth and one on the moon? Since two on the earth (assuming that their wave functions appreciably overlap) can < confused with each other.
so that we end up with a symmetrized wave function in p space.50) 277 SYSTEMS WITH N DEGREES OF FREEDOM The formal reason is that for any choice of the arguments x t and x2. A word of caution before we conclude this long discussion.3. Thus when a momentum measurement (which says nothing about the positions) yields two numbers. 3. say the P basis. there are no grounds in P space for distinguishing between them. The physical reason for this is that the two pions have the same momentum distributions—with (P) = 0 and identical Gaussian fluctuations about this mean—since the momentum content of the oscillator is independent of its location. Show that the total number of allowed. however. and 4. we wish to switch to another basis. b. we may describe the system by a product wave function in this space •fusing labels like "fast" and "slow" instead of "earth" and " m o o n " to distinguish Jletween them). It should be clear that these arguments apply not just to X or P but to any arbitrary variable Q.3. which in our example was . and c available to them. Consequently.absurd conclusions are averted only if the statistics depend on the species and not the number of particles. distinct configuraSons for this system is (1) 27 if they are labeled (2) 10 if they are bosons (3) 1 if they are fermions .3* Imagine a situation in which there are three particles and only three ites a. If. (For example. but not in another space where the distinction between them is absent. we cannot assign them to the pions in a unique way. Write down a symmetrized. if Xi is on the earth and x 2 is on the moon. Exercise 10. If two particles have nonoverlapping wave functions in x space. if there are two particles with nonoverlapping wave functions in p space. only the first piece is important. By the same token. normalized state or. and  y/>. symmetrization is important because the pspace wave functions of the pions overlap strongly and there exist values for the two momenta (both ~0) for which both terms in the symmetric wave function are significant. then it is only in x space that a product wave function provides a good approximation to the exact symmetrized wave function.) Physically it is because the chance of finding one pion in the territory of the other is negligible and interference effects can be ignored. the n values obtained were 3.1* Two identical bosons are found to be in states swn the normalized state vector describing the system when (//) #0.3. Write '* Exercise 10. Exercise 10. we must consider the Fourier transform of the symmetric function and not the product. Formally. x2) = 2~l/2iG^xOGM(x2) + GM(xx)GE{x2)] (10. * When an energy measurement is made on a system of three bosons in 'ox. only one or the other of the two terms in the righthand side is important.2.3.
(There are two possible vi in this case. Repeat for £ sys = n2 jlmL2. Energy measurement of the system yields the value Esys = h27t2/mL2. Exercise 10.) You are not told if they are bosons or fermions. assume they are given by power series and consider any in the series.6. You may assume that the < degrees of freedom are orbital. peek into the discussion leading to Eq.. (It is Hermitian and unitary.22). show that any eigenstate oH continues to remain an eigenstate with the same eigenvalue as time passes.X2) = \X2.4* Two identical particles of mass m are in a onedimensional box length L.) (2) Show that its action on the basis ket ©i. element VS/A never leave the symmetric or antisymmetric subspaces they start in. (3) Show that Pi2X]P{2 =X2. (11. * Consider a composite object such as the hydrogen atom.2. i. Then show 1 F 1 2 D(Xi.. P2.. and hence that VS/A are its eigenspaces with eigenvalues ± 1.278 CHAPTER 10 Exercise 10. Exercise 10. [Consider the action on x.Given this. If you need help. Write do] the state vector of the system.5 * Consider the exchange operator P\2 whose action on the X basis ij P\1\X\.> (1) Show that P\ 2 has eigenvalues ±1. . P\2X2Pi2~X.3.3.3. X}. for the functions of X and P.] (4) Show that the Hamiltonian and propagator for two identical particles are unaffected under H+P\2HP\x and U~>Pl2UPU. Will it bell as a boson or fermion? Argue in general that objects containing an even/odd numt fermions will behave as bosons/fermions. © 2 ) is also to exchange the labels 1 2. x2> or \pl.P2)Pn=&{X2. and similarly for Pi and P2.e. X. Pt).p1). Px\X2.
Our previous experience.11 Symmetries and Their Consequences 11.p).1. we explored the consequences of the symmetries of the Hamiltonian. Here we address the corresponding results in quantum mechanics.1a) I <P>^(P> (11. Equivalently.2.2. Overview In Chapter 2.1b) 279 t It may be worth refreshing your memory by going through Sections 2.8.2.! 11. suggests that in the quantum formulation the expectation values should play the role of the classical variables. . we cannot define translational invariance to be the invariance of the energy under an infinitesimal shift in the particle position. we still have two equivalent ways to interpret the transformations: (11. however. We saw the following: (1) If J f is invariant under the infinitesimal canonical transformation generated by a variable g(q. Having agreed to formulate the problem in terms of expectation values.1. We therefore make the correspondence shown in Table 11. (2) Any canonical transformation that leaves J f invariant maps solutions to the equations of motion into other solutions. How shall we define translational invariance? Since a particle in an arbitrary state has neither a welldefined position nor a welldefined energy. then g is conserved. an experiment and its transformed version will give the same result if the transformation is canonical and leaves Jf invariant.7 and 2. Translational Invariance in Quantum Theory Consider a single particle in one dimension.
j y/E) such that (11. (11.2.1.7) and corresponds to physically displacing the particle to the right bj The second point of view is to say that nothing happens to the state vector is the operators X and P that get modified by T(£) as follows: X^T\£)XT{£) P^T\£)PT(£) such that T\E)XT(£)=X+ T\e)PT(£) =P £l (H. Physically.4 = \¥ey (11. . sources of external field if any. Correspondence between Classical and Quantum Mechanical Concepts Related to Translational Invariance Concept Translation P+P Classical mechanics e Quantum mechanics OPWP) <P> = 0 (anticipated) Translational invariance Conservation law The first is to say that under the infinitesimal translation.2.280 CHAPTER 10 Table 11. the equivalence of the active and passive pictures is due to the fact that moving the particle one way is equivalent to moving the environment the other way by an equal amount.) to the1 left by £.1 This point of view is called the active transformation picture (in the terminolog Section 1.: (Ve\P\yE) = (\if\P\y) (11. the translation operator.2. This is called the passive transformation picture. Physically it corresponds to mov the environment (the coordinate system. (11. which translates the state (and which' be constructed explicitly in a while) n£)\¥y Eq.: In terms of T(e).2. etc. each state y/> modified into a translated state.2) becomes (v\T\e)XT(£)\v} = <_y/\X\v)+ e (11.2.3 <¥\Tf(£)PT(£)\¥)=(¥\P\¥) (11.
(11.4).2.2.5) between <y/i and \y/)>.\T\s)PT(s)P We now reason as follows: \y. it is this point of view that best exposes many formal relations between classical and quantum mechanics. and gives another. (11. y/> dx £\y/} dx' (x' = x+£) (11. (2) Since  y > is arbitrary. We will find that it is possible to construct T(s) given either of Eqs. In what follows. .} = 0 (1) The operators being sandwiched are Hermitian (verify).o o <x< oo. (11. we show the equivalence as follows.0 281 SYMMETRIES AND THEIR CONSEQUENCES <y.2. equivalently. Once the action of T(S) on a complete basis is known. and of course that the two yield the same result. . its action on any ket  y/) follows:  ye) = T{E)\yy = T(E) !* co j*r><j*. The passive transformation picture is nice because the response of the quantum operators X and P to a translation is formally similar to that of their classical counterparts.2. The answer appears obvious if we work with kets of definite position. the action of the Hilbert space operator T(s).5). (3) The operators themselves vanish. p). we will examine both pictures. implying Eq. In this case it is clear that r(£)*> = x+£> (11.7) In other words if y/(x) t As we shall see. if the particle is originally at JC. we first rewrite Eq. it must end up at x + e. we can choose it to be any of the eigenvectors of these operators. y/) dx £><J*. Notice that T(E) is unitary: it acts on an orthonormal basis jf). (11. . The active transformation picture is nice in that we work with the quantum state which now plays the role of the classical state (x.! We begin by discussing translations in terms of active transformations. \x).2. we get Eq.2. Let us examine how the ket  y £ > is related to  y/)> or. It follows that all the eigenvalues vanish.4) as (y/\ t\E)XT{E)XSL\ y/} .5). \x+e)>. To go the other way.2.4) or (11.o o < x + S< co.Mathematically.6) In other words.2. If we sandwich the operator equation (11.
6) but rather T(E)\x) = eiesUy*\x+E) (11.10) .2. The reason is that in our derivation we have assumed more than what was explicitly stated. Thus the wave function \ye(x) is obtained by translating (without distortion) the wave function y/(x) by an amount E to the right.) In ignoring g(x).1 (Note that as s>0.2. Let us see how.2.8 F o r example.1b)? Itii automatically satisfied: I* 00 <Ye\P\ y£> = * ij/*(x)\ — ifi — j \f/e(x) dx \ dx) ' OO ip*(x— E) \ — ifi — j y/(x— E) dx V dx) oo ¥*{x')\—i1i~\¥{x')dx' (x' = x—s) (11. Y o u may verify that the action of T(E) defined by Eq. (11. Classically.2. (11. translation is specified by i w independent relations X^X+S I while in the quantum version we seem to find that in enforcing the former (on position eigenkets).2. that T(e)x:) =  x + e > While our intuition was correct. the latter automatically follows.2 — liu = <V\P\V> N o w there is something odd here. if \(/(x)~e~x2 is a Gaussian peaked at the origin {x s)2 y(xE) ~e~~ ~ is an identical Gaussian peaked at x = E.2. If we start with Eq. as it should. (11. As seen in chapter 7. (11. it follows. (11.282 CHAPTER 11 then 1 <x\T(E)\y>=y(xE) ' (11. We reasoned earlier. and the general result consistent with our intuition is not Eq.8] satisfies the condition Eq. that since a particle initially located at x must end up at x + E. H o w about the condition Eq.1a). we had essentially assumed the quantum analog of p+p.2. our implementation was not. the X basis is not unique. on physical grounds.
2.2* Using T\s)T(e) = 1 to order s. we find that <xy —. deduce that Gt = G.11b) Note that there was nothing wrong with our initial choice Tx> = £>—it fas too restrictive given just the requirement <JT>»<X> + e. Exercise 11. We find G by turning to Eq. but not so if we also jnsidered <P>^<j p ). Having defined translations. Exercise 11. (11. we ay expand T( s) to order e as is fi T(s) = I G (11.2.I instead of Eq.13) The operator G.2.i/ti) is introduced in anticipation of what is to follow. In all those cases we will make the analog of the naive choice T ( f )  x ) = 1* + s) to shorten the derivations.2. we must first [instruct the operator T(s) explicitly.2. called the generator of translations.2 for the proof) and is to be determined. let us now define translational invariance in the same spirit. Verify Eq. we find n dx .2.12) To derive the conservation law that goes with the above equation.11b) iere f=g'.Demanding now that <Jp>></>>. We define it by the requirement (11.8): spanding both sides to order s. (11. is Hermitian (see Exercise 11.6).1.11a) 283 SYMMETRIES A N D THEIR CONSEQUENCES T(s) (11.2. we eliminate / and reduce g t o a less constant (which can be chosen to be 0).2. (11.2. The constant ( . <x> + £ (11. Since s=0 corresponds to no translation. This situation reappears when we go to two or three dimensions and when we consider rotations.2.
n H]\Y. The momentum conservation law now follows from translational invariand Eq.\H\T(6)Y. We see that exactly as in classical mechanics.2. we get.H]\Y) = 0 ^ (11.12). (11. G = P and T(E) = I~P n (11.14): = <T(E)Y. but it measures positid from a new origin.1 (11.2.2.2. <¥\[P.^]> = 0^<P> = 0 I (11.) = IW\H\Y> + Y <VLIP. given that it acts as follows on X and P: T\E)XT{E)=X+EI T\E)PT(E) = P (11.2.K Translation in Terms of Passive Transformations Let us rederive T(S). if we combine it with Eq.}+ 0(E2) so that. the momentum is the generator 1 (infinitesimal) translations.Y~<Y. shifted to the left by e: This is the meaning of Eq.2. (11.2.1 The operator T^{E)XT(E) is also a position operator.2.lj It now follows from Ehrenfest's theorem that <[P.\T\S)HT(E)\Y. upon equating the coefficient of E to zero. (11.284 CHAPTER 11 so that ax Clearly G is the momentum operator.17a .
2.17a) (using the fact that Gf = G) v n ) \ n or is .G]=mi (11.2. (11.Equation (11.20a) or [P.17b) states that under the shift in the origin.2. (11.[ X . .18a) I This allows us to conclude that \x. the m o m e n t u m is unaffected.17b) we find P+f(X) (11.2.2.19) fi (11.2.2. We define translational invariance by the requirement ^ T\E)HT{S)=H (11.20b) T(s) = I T isP Having derived the translation operator in the passive transformation picture. which we choose to be zero. G ] = EI fi (11.G]=0 which eliminates f{X).2. it reduces / t o a c number which commutes with X and P.. let us reexamine the notion of translational invariance.t So once again (11. Writing once again isG 285 SYMMETRIES A N D THEIR CONSEQUENCES T(e) = I we find from Eq.2.21) f For the purists.18b) G= If we now turn to Eq.
(11. and both sets of variables undergo identical changes in a translation. P) (11. and for any unitary operator U. U1Q(X. Applying this result to the case UT(e) variable to a translation: fi(JT. A Digression on the Analogy with Classical Mechanics^ The passive transformation picture has the virtue that it bears a close formal resemblance to classical mechanics. the reader may skip this digression. consider a typical term in the series such as PX2P. with operators Q in place of the classical variables $ In a less advanced course. using UfPX2PU= tfputfxutfxutfpu Q. because of the Ehrenfest's theorem.2^ which implies that <P> = 0. P) (11. (11. We have.H (11. P)r= we get the response of any dynamical Q(tXT.E. TfPT) = Q(JT 4. I . The reason is simply that H is the same function of X and P as J f is of x: and p.2.2.H= (/+ isP/H)H{I~ ~~ n [H.22) Thus the transformed f2 is found by replacing X by X+ si and P by P.P)U=Q(WXU. IfPU) For the proof. If we now apply this to Eq. The conservation of momentum follows if we write T{e) in Eq.21) we get the following definition of translation invariance: H(X + el.286 CHAPTER 10 We can rewrite Eq.2. P) = H(X.el.2.21) in terms of P and expand things out to first order in e: 0= T\e)HT(e) .21) in a form that is closer to the classical definition of translational invariance.D.23) N o t only does this condition have the same form as its classical counterpart but it is also satisfied whenever the classical counterpart is.2. (11. P)rfQ(X. P] ieP/fi) . But first we need the following result: for any P\ that can be expanded in a power series.
P} ft combining which we obtain 8X=~^[X. P) . the infinitesimal unitary transforma tion T(e) generated by P is the q u a n t u m image of the infinitesimal canonical trans formation generated by p : if we define the changes 8X and 8P by 8X=T\E)XT{E)~X I I 8X={I+ 8P=T\E)PT(E)~P we get.a [Eqs. on the one hand. we obtain SQ = —ft [Q. P] = 0+<P> = 0 by Ehrenfest's theorem . Eq. P] = 0 More generally.23)].22). f r o m T= I—izP/fi ieP/fi)X(I isP/K) X = —ft [X.2. If the problem is translationally invariant.p} = CO(X+ £.17). p) ~ £ 8p = s{p. (11.17). In fact.n(X. we have 8H= — f\ [H. P) These are the analogs of the canonical transformation generated by p : 8X—E{X.p}= 0 p) 8M = £{a>.22) and T—ieP/fi. (11.p)CO(x. P] = Q(X+ eI. (11.2. EIX= EI and on the other. upon combining. P] 8P=(I+i£P/ft)P(I~i£p/ft)P=—[P. (11.P] ft 8P = = eI n [P. f r o m Eq. 8X=X+ 8P=PP=0 (working to first order in e).2.2.2. (11.
. it preserves the PB between the x's and the p's. ^ $ More generally if [£}. ifTU. we have seen that the transformation generated by any g(x. P) insvi The correspondence holds for finite transformations as well. fl if then u p o n premultiplying by the unitary operator UG(S) and postmultiplying UG(£)..g} has as its image in q u a n t u m theory the infinitesimal unitary transformation Uaie)* J— isG/fi in response to which — IF 5Q = 1i [Q. quantities preserved are the commutation relations between the X's and the Ps. G] Now. then a similar relation holds between the transformed operators U*BU.288 CHAPTER 10 while classically The correspondence is achieved through the substitution rules already familiar us: Q <»• m [ft. p) is canonic i.e. ufPj u] = This completes the proof of the correspondence infinitesimal canonical { transformation generated by g(x. A] ^ {co. for these may be viewed as a sequence of infinitesimal transformations. 1} In general. we find that the transformed operators o b e y ! [u'Xi u. 9] = T. p). In the q u a n t u m theory. This is the quantum version of the result that PB are invariant under canonical transformation.p) [ infinitesimal unitary transformation generated by G(X. 5(0 = £{(0. the infinitesimal canonical transformation generated by g(x.
for example. J Note that a single particle whose H is translationally invariant is necessarily free. In the quantum theory there will exist a unitary operator such that. For instance. the state is U{t)\p)> and we find pu{t)\py=u(t)p\py = u{t)p\py=pu{t)\py (i 1. (11.2. we cannot find a unitary transformation in the quantum theory. its momentum eigenvalue remains constant. x2. [) If a system starts out in such an eigenstate. and pu.25) since the propagator is a function of just H * Suppose at / = 0 we have a system in an eigenstate of P: P\P>=P\P> (H. H] = 0>[P.2.27) In other words.2. p2 to Xi. xn. the state at time t is also an eigenstate of P with the same eigenvalue. in the coupled oscillator problem.1. §End of digression. p\.3. 6 / ( 0 1 = 0 (11.26) After time t. X If the transformation is not regular. + J 2 ) / 2 = ^ I and so on. for example. the end effect will be the same. We divide the interval a into N parts of size a/N. For such states with welldefined momentum. the conservation law < P ) = 1 reduces to the classical form. p = 0. or first quantize and then perform the unitary transformation—since the quantum operators respond to the unitary transformation as do their classical counterparts to the canonical transformation. if X\ U~ ( J .J We can see why we can either perform the canonical transformation at the classical level and then quantize. we performed a canonical transformation from Xi.P\. since unitary transformations preserve the eigenvalue spectrum.2.24) tells us that we can find the simultaneous eigenbasis of P and H. * When H is time independent. the result is true if P commutes with H{t) for all t. first note that 289 SYMMETRIES A N D THEIR CONSEQUENCES I  [P.§ Let us now return to the problem of translational invariance. Exercise 10. If H= H(t). (This agrees with our result f r o m Chapter 5. Notice that in a problem with translational invariance. we know U(t) = exp( — iHt/h). Eq.The correspondence with unitary transformations also holds for regular canonical transformations which have no infinitesimal versions. To prove this. Xj = (X[ +x2)/l. Finite Translations What is the operator T(a) corresponding to a finite translation a! We find it by the following trick. that the energy eigenstates of a free particle could be chosen to be momentum eigenstates as well. (Why?) . where. As N^co.
^ ) " JV.e. i.33) A Digression on Finite Canonical and Unitary Transformations^ Though it is clear that the correspondence between canonical and unitary transformations. A Consistency Check.2. established for the infinitesimal case in earlier discussions. oo V N We may apply this formula. must carry I Optional. (11. to the present problem.32) =T(a + b) (11. iFis" merely a statement about how translations combine in space.a ) about the point x. which is the full Taylor series for y / ( x . since the only operator in the picture and commutes with everything in sight.. e~ax= l i m ( l . true for c numbers. Now. which are supposed to translate quantum states. and the line above it] in the vefj act of deriving the formula for T(a). A translation by a followed by a translation by b eqq a translation by a + b. it is necessary that the law of combination of the translation operators coincide with the law of combination of the translations they represent.2.^T — • • • ^ + — X basis • (Tad. although we presumed this [see Eq. (11. bel like a c number.iaP/fi) satisfies T(a)T(b)=T(a We find that this is indeed so: T(a) T(b) = e~iaP/* • e~ibPin = +b)P/n + b)? (11. This result has nothing to do with quantum mechanics is true whether you are talking about a quantum system or a sack of potatoes.29).290 CHAPTER 11 a/N becomes infinitesimal and we know T\a/N) =I ~ P UN Since a translation by a equals N translations by a/N. For this interpretation to be consistent. T{a) = N>oo lim [T(a/N)]N=e~iaP/* by virtue of the formula (11. Since T{a) we find <x\T(a)\y.2. let us verify that our result T(a) = exp( .dx (UJ dw dx d2w a2 dx 2! ¥ (11. . Now.y = ( x ) ~— a + . we have just built operators T.
In the quantum theory we have Q^T\a)nT(a) Using the identity 1 1 291 SYMMETRIES A N D THEIR CONSEQUENCES = e'aP/*£leiaP/ti e~AB e+A=B+ [B.p}. ~={(o.over to the finite case. under an infinitesimal displacement 5a. etc. (Recall the series for sin 0.2. applying the above result to the variable dco/da. P].p) = {{co. 5a) = Sa{a>.f l •+ a I [Q. A].p}+~ {{(o.2. if we set £l=X2 we get X2^>(X + al)2. etc. In the classical case.) . Verify that if._ 1 21 ' Q .p} da . we get da (do)/da) = d co/da = {dco/da. A] + — 2! 3! we find ~ J\.2.3.34) if we make the usual jbstitutions. A] + — [[B.F} + (11.2. for definiteness. The response of co to the finite translation is given by the Taylor series >out the point a = 0: a a>+o) + a{a).35) vhich we see is in correspondence with Eq. you the same result. P] + . you relate lie transformed coordinates x and y to x and y by the infinite string of Poisson brackets. x = x cos 6 — y sin 9.p} +• • (11.p). (11.34) For example. * Recall that we found the finite rotation transformation from the infinitelimal one. instead.8). the case of translations.a1! — I [[Q.p) id so on.p} or dco . by solving differential equations (Section 2. let us nonetheless go through the details. Consider. Exercise 11.
(e)y> = y ( x i . that is < X i . < x u . P) T( e) = H(X + e/. . .~ Y * ti P^l'P * where P is the total momentum operator. (11. Implications of Translational Invariances Consider a system with translational invariance. XFFJL— n P\\f/) = il • • •. There are some profound consequences of translational invariance besid momentum conservation.e . we get [7Xfl).2.2. Starting with the analog of Eq. (11.2^ from which it follows that T(e) = I . on expanding both sides to order e. Any system whose parts interact with each other. ..8). We take these up next.40) Whereas in the singleparticle cases this implied the particle was free. Premultiplying both sides Eq. .2J = Pi> Translational invariance means in this case (suppressing indices). .2.#] = 0 .292 CHAPTER 10 System of Particles We will not belabor the extension of the preceding ideas to a system of particles. will have this property. P) (11. .xw7. but nothing external.21) with T and using its unitarity. . X N dw dXi ) ~ i= i X (11. . x * . .. P) = Tf( e)H(X. H(X. You may verify that T\e)XiT(s) T\e)PiT(e) = Xi+er (11. here it merely requires that H (or rather V) be a function of the coordinate differences.e ) we find.
In other words. But this does not correspond to repeating the same experiment and getting a different result. which differed only by a translation at t = 0. + f.2. respectively. R 2 . then T(a)\ v(0)> is the state vector of the system prepared by B. in that every experiment. At t = 0 two observers A and B prepare identical systems at x = 0 and x = a. N o w this problem has no translation invariance. differ only by the same translation at future times. since the condenser.42) where the subscripts 1 and 2 refer to the electron and the proton and F(R)$ to the potential due to the plates. P. P 2 ) ^ / / ( R I . Therefore the two systems. which clarifies the meaning of this remark.41) The consequence of this relation is illustrated by the following example (Fig. electromagnetic. A symbolic depiction of translational invariance.1.e. which is just the translated version of ^ ' s system at time t. The Hamiltonian is 2mx 2m2 RiR2 +elV(Rl) + e2V(R2) (11. We have already seen this result in the classical framework. the time evolution of each system appears the same to the observer who prepared it. weak... and strong (e. <7(01=0 or T(a)U(t) = U(t)T(a) (11.41) the latter may be rewritten as T(a)U(t)  y/(0)>.1). Using Eq. will give the same result. Consider the following illustrative example. (11. After time t. P 2 ) which in turn means that if the atom alone is translated (away from the condenser) it will behave differently. R 2 + £.2. if repeated at a new site.2. the state vectors evolve into U(t)[ v(0)> and U(t)T(a)\ y/(0)>. Now it turns out that every known interaction—gravitational. We pursue it further now.g. i. nuclear)—is translationally invariant. 11.. P!. The states are represented schematically by wave functions. .Figure 11. which materially affects the I Remember that R is the operator corresponding to the classical variable r. # ( R . If  y(0)> is the state vector of the system prepared by A. A hydrogen atom is placed between the plates of a charged condenser. The two systems look identical to the observers who prepared them. Translational invariance of H implies that the same experiment repeated at two different places will give the same result (as seen by the local observers). IWO)>/ T(o)t(0)> / It follows that [T(a).
we redefine our system to include the (N) charges on the condenser and write (11. P) which implies that if the atom and the condenser are moved to a new site. between what constitutes the system and what is irrelevant (for practical purposes) to its evolution. The content of the assertion made above is that every known interaction translational invariance at the fundamental level—if we expand our system to inclu< all degrees of freedom that affect the outcome of an experiment (so that there are not external fields. Nov it is true that (dropping indices).2.294 CHAPTER 11 dynamics. the behavior of the composite system will be unaffected. The real utility of the concepts of translational invariance and momentum conservation lies in these approximate situations. not via the external field which breaks translational invariance. The translational invariance of natural laws reflects the uniformity or homogeneity of space. . no system is truly "isolated" except the whole universe (and only its momentum is exactly conserved). It is translational invariance that allows experimentalists in different parts of the earth to claim they all did the "same" experiment. and complement each other. Who cares if the universe as a whole is translationally invariant and its momentum is conserved? What matters to me is that I can take my equipment to another town and get the same results and that the momentum of my system is conserved (to a good accuracy). But in practice one draws a line somewhere. 11. represents the fact that one part of free space is as good' as another. = We will examine further consequences of translational invariance toward the end of the next section. To incorporate it in what is translated. but through the Coulomb interaction. This is why we apply momentum conservation to evei problem whatever be the underlying interaction. which does not. This result should be vie not as obvious or selfevident. correct. is left behind. P)=//(R. Time Translational Invariance JH.'rj'i r Just as the homogeneity of space ensures that the same experiment performed at two different places gives the same result. I use the term "isolated" in this practical sense. //(R + £.3.43) N o w the charges on the condenser enter H. and to corroborate. The fact that the dynamics of an isolated! system (the condenser plus atom in our example) depends only on where the parts of the system are relative to each other and not on where the system is as a whole. homogeneity in time ensures that the t To be exact. but rather as a profound statement about the Couloi interaction. It is the invariance of the natural laws under translations j that allows us to describe a hydrogen atom in some distant star as we do one on j earth and to apply to its dynamics the quantum mechanical laws deduced on earth. only interactions between parts of the system) the total H translationally invariant.
will be 295 SYMMETRIES A N D THEIR CONSEQUENCES I + <?)> = I—Hit. Now Ehrenfest's theorem for an operator Q that has no time dependence% is Applying it to Q = H in a problem with time translational invariance.3.5) Thus time translational invariance requires that H have no t dependence.2) The outcome will be the same in both cases if 0 = \y/(t2 + s))~\\ff(h + £)) =1 . Let us prepare at time tx a system in state  y/0> and let it evolve for an infinitesimal time e.same experiment repeated at two different times gives the same result./dt} on the righthand side. it follows that H is timeindependent: dH dt (11. we find <£> = 0 (11.3) Since  y/0) is arbitrary. i .6) which is the law of conservation of energy.3. beginning with the same initial state. it follows that H(t2) = H(h) Since t2 and are arbitrary.3.3. The state at time ^ + e.3. [H(t2)H(h)]\yo} (11. Let us see what feature of the Hamiltonian ensures this and what conservation law follows. to first order in e.4) = 0 (11. t there will be an extra piece ift(dQ.3.) n IVo> (11.1) If we repeat the experiment at time t2. the state at time t2+ £ will be l£ I~H{t2) n IV('2+£)> IVo> (11.
By repeating the natural laws over and over t h r o u g h all of spacetil nature gives tiny earthlings. for example.296 CHAPTER 10 A n important simplification that arises if dH/dt = 0 is one we have repea exploited in the p a s t : Schrodinger's equation m ^ H w y dt admits solutions of the f o r m iEi/n \V(t)> = \E)e where the timeindependent ket  E ) satisfies H\E> = E\E> T h e entire dynamics. however.e. all k n o w n interactions—from gravitational t o strong—are translational invariant. before and after we learned . In particular. (11. But this is not to say that the Newtonj scheme worked till Einstein came along. In other words.2. boils do1 the solution of the timeindependent Schrodinger equation (11. If the system includes just the electron and the p r o t o n .3.42). i. This simply means repeating the experiment without recharging the condenser. t).9). the determination of the p r o p a g a t o r U{t).. which has no t dependence. Consider. w h o p r o b e just a miniscule region of space for a fleet m o m e n t (in the cosmic scale).43). (11.2. The considerations t h a t applied to space translation invariance apply h < well. which tell us that what little ittlej k n o w is universal and eternal. we enlarge the system to include the N charges on the cond we end u p with the H in Eq. If. just a sequence of h a p h a z a r d events with no rhyl or reason. If nature were not to follow the same rules over spacetit there would be n o rules to find. H will depea time—it will have the f o r m of Eq. For example. which dl not change with time. a chance of comprehending the universe at lai Should we at times be despondent over the fact that we k n o w so few of nat laturi laws. a hydrogen a t o m between the plates of a discharging denser. if we suitably define the system (to inclui sources of external fields that affect the experiment) the total / / w i l l be independi t. Consequently.t % The invariance of the laws of nature is not to be confused with our awareness of them. Einstein showed that Newtonian mechanics and gravitation I approximations to relativistic mechanics and gravitation. the relation of Newton's scheme to Einstel (as a good approximation in a certain limit) has always been the same. The very cycle of physics—of deducing laws from s<J p h e n o m e n a studied at some time a n d place a n d then applying them to other phent ena at a different time and place—rests o n the assumption that natural laws spacetime invariant. will lead to a diffi result. The spacetime invariance of natural laws has a p r o f o u n d impact on our qi for understanding nature. w i t h F = F(R. let us find solace in these symmetry principles.
its action on an arbitrary ket follows: ny> = n  O O \xXx\yr>dx r oo  — x ) < x  i y ) dx (where x' = — x) (11. parity is a discrete transformation. (which we will study in the next chapter).4.3) follows. . and rotations. it will be readily found that 11/J) =  />>.* > in analogy with the classical case.4. Given the action of n on a complete (X) basis. as you will see in a moment.4.5) — The function * ) *s the mirror image of y/(x) about the origin. (11. we define the action of the parity operator on the X basis as follows ni*> = i . Classically. Given this.4. Applying Eq.1) —i parity In quantum theory. Parity Invariance Unlike spacetime translations. (11.4) x'Xjc'yOdx' It follows that if < x  y > = V'(x) (11.11.4.2) nLp>=!/>> (11.5) to a momentum eigenstate.4.4. the parity operation corresponds to reflecting the state of the particle through the origin 297 SYMMETRIES A N D THEIR CONSEQUENCES X parity ^ X (11.
where evenparity vectors have even wave functions and oddparity vectors have odd' functions. (4) o r n _ 1 = n + = n . In particular.x > n2*> =  (*)>Hx> Since this is true for an entire basis.X. P) = P (11. (3) IT is Hermitian and unitary. A moment's "reflection" will proveJ Since njc> = i . P) is parity invariant if nfH(X. The eigenvectors with eigenvalue ± 1 are said to have even/odd parity.1 (11 and a common eigenbasis of n and H can be found. i Rather than define II in terms of its action on the kets.298 CHAPTER 10 The eigenvalues of IT are just ± 1 . IV = I (ilj Please note that (1) n ^ r r 1 (2) The eigenvalues of IT are ± 1. In this case P)Tl = H{ . if we conskk just bound states in one dimension (which we saw are nondegenerate). every eigenva tor of H is necessarily an eigenvector of II. In basis. ¥(CD) need not be even or odd even if  <//> is a eigenstate (check this). The same goes for the P basis since In an arbitrary basis. F o r example. . we may also define! through its action on the operators: nfxn=x UiPU We say H(X. the oscillator Hamiltonil .P) = H(X.
satisfies Eq.) If H is parity invariant. This means formally that the Hamiltonian cannot be made parity invariant by any redefinition of the system if weak interactions are involved. since the initial S was down.% Consider the following concrete example of a P decay: 60 Co> 6 0 Ni + e . In the other real experiment (B). what I see in the mirror (experiment M) does not correspond to what can happen in real life. then at a later time the state of his system will be related to mine by the parity transformation. Let another observer set up another system which is just the mirror image of mine (experiment B). and vanishes outside. (11. (When x>x they vanish. Now it turns out that the electron likes to come flying out in a direction opposite to the spin of 6 0 Co—and this implies parity noninvariance.4. is not a solution to the equations of motion. Of course the mirror also shows an electron coming down the z axis. My electron comes out down the z axis.8) and its eigenfunctions have definite parity equal to ( .4 for a discussion of why the parity transformation is essentially a mirror reflection in three dimensions.2) (experiment A). which is what I see in a mirror in front of me. since y/„ is given by the sine function only between 0 and L. which are responsible for nuclear ft decay (among other things).l ) " + l .e.1 ) " . then UU(t)=U(t)U (11. the final states U{t) y(0)> and C/(/)n i^(0)> will not be mirror images of each other (since TMJJ1 UYl). Let the /? decay take place.4.+ v i where e~ is an electron and v is an antineutrino. In this case the eigenfunctions have parity ( . Notice how the spin S gets reversed under a mirror reflection. Let us see how.4. Although you are not yet familiar with spin. 11. These are the laws of weak interactions.9) This means that if at t = 0 I start with a system in a state  y ( 0 ) X and someone else starts with a system in the parity operated state I I  ^ ( 0 ) > . At f = 0 I prepare a system that consists of a ^Co nucleus with its spin up along the z axis (Fig. . Physically this means that if two observers prepare initial states  y ( 0 ) > and IT^(0)> which are mirror images of each other. J See Exercise 11. Let M denote a fictitious experiment. i. The particle in a box has a parityinvariant Hamiltonian if the box extends from L/2 to L/2. V(x) is not parity invariant and the eigenfunctions 299 SYMMETRIES A N D THEIR CONSEQUENCES I s i n g have no definite parity. Whereas all natural laws are invariant under spacetime translations (and rotations) some are not invariant under parity.. Consequently. n being the quantum number of the state. n being the quantum number. you may pretend here that ^ C o is spinning in the sense shown. If the box extends from 0 to L. Thus B starts out as the mirror image of A but ends up different. the dynamics forces the electron to come up the z axis.
Hayward R. it has no asso conservation law in classical mechanics. If a certain phenomenon can happ< translated or rotated version can also happen. * You are told that in a certain reaction. S. it need not end in a st.. . same parity if weak interactions are at work. but not its mirrorreflected ve if the phenomenon involves weak interactions. Yang. In higher dimensions when we use a plane mirror (say lying on the t T.. (Note that since parity is a discrete operation. C. Lee and C. But in the actual experiment B.. R. 104. 1413 (1957). isolated system starts out in a state of definite parity. which I perform. B starts out as the mirror ima After the decay. * Prove that if [II. where m is an integer. Phys. The possibility that weak interai could be parity noninvariant was discussed in detail by Lee and Yang in 195i confirmed shortly thereafter by the experiments of C. 254 (1956). Argue that parity is violated.) > Exercise 11.2. S.4. i. W. it points down (experiment M). 1 . An example of parity noninvariance.  Exercise 11. Hudson. where x ^ — x may be viewed as the effect of reflecting through a (p mirror at the origin. the spin is chosen to be down.e. the spit nucleus points up the 2 axis. is down the 2 axis. / / ] = 0. E.* A particle is in a potential V(x) = Vo sin(2nxfa) which is invariant under the translations x*x + ma. In its actual mirror image. a system that starts out in a state of evet parity maintains its parity. P. In exp B. Rev. This is certainly tr one dimension. antiparallel to the initial nuclear spin S. Is moma conserved? Why not? • Exercise 11. I This then is the big difference between parity and other transformation! as spacetime translations and rotations. Rev. the momentum of my electron. the electron comes out wi spin always parallel to its momentum.4. Wu and collaborator Exercise 11.1.A M B Initial States S mirror *t Final States mirror Figure 11.. N.4* We have treated parity as a mirror reflection.2. p e . Phys.e. In experiment A.4. the dynamics forces the elec come up the 2 axis. which is a real experiment. Ambler.4. Wu. The mirror image of cou shows the electron coming down. 105. D.3. In terms of conservation laws. i.
reverses its speed. The backward motion of the planet will bring it back to the timereversal initial state at t=2T. What we see in the movie can really happen.5. Here are the details. p(T) at 0 = jr/2 after doing a quarter of a revolution. we will now prove it with the help of Newton's Second Law using the fact that it is invariant under t . the orbit is running counterclockwise.(l 1. Just for this discussion let us use a new clock that has its zero at the . indeed.1) The above equation defines timereversal invariance (TRI). noninvariance under mirror reflection implies noninvariance under parity. only one (z) coordinate gets reversed. What will it do? We know it will retrace its path and at time 27" end u p in the timereversed state of the initial state: f x ( 2 F ) = x(0) p(2T) = p( 0) (11. Consider a planet that is on a circular orbit around the sun. Let us call the initial position and momentum x(0). While the correctness of Eq. On the other hand. when a cartoon character falling under gravity suddenly starts clawing his way upwards in thin air using sheer will power./>(0). it is perfectly possible for a set of people and cars to execute this motion.1) is intuitively clear. At t=T. Since rotational invariance holds for weak interactions.t : the acceleration is even in time and the potential or force has no reference to t. any quantity like position or kinetic energy. In general. m 301 SYMMETRIES AND THEIR CONSEQUENCES 11. Verify that reflection on a mirror in the x—y plane is the same as parity followed by 180° rotation about the z axis. which involves an even power of t in its definition is left invariant and any quantity like momentum or angular momentum is reversed in sign under the timereversal operation.5. We can describe T R I more graphically as follows. Suppose we take a movie of the planet from f = 0 to t=T.^ . At t = 0 it starts at G = 0 and has a velocity in the direction of increasing 6. whereas the parity transformation reverses all three coordinates. TimeReversal Symmetry This is a discrete symmetry like parity. but ignore this fact for this discussion.) We now define the timereversed state as one in which the position is the same but the momentum is reversed: I  xr(t) = x(t) pr(t)=p(t). Say that after time T the planet has come to a final state x(T). and lets it go. In some movies they get a big laugh out of the audience by showing cars and people zooming in reverse. it was shown how Superman could make it happen even as you are watching the movie. In other words.5. you may laugh since this is a gross violation of Newton's laws. (We should really be using vectors. Now Superman (for reasons best known to him) stops it dead in its tracks. As a serious physics student you should not laugh when you see this since these motions obey Newton's laws. More generally. Let us first understand what it means in classical physics. we start playing the film backward.plane). if you see a movie of some planetary motion you will have no way of knowing if the projector is running forwards or backward. In other words.
timereverse its state. We blame the magnetic interaction for this failure of TRI: the force now involves the velocity which is odd under timereversal. t) = H(x)\ff(x. The ideas will be illustrated in the simplest context. Let us say it is released at t = 0 just like the planet. 5 seconds after the reversal.t ) ) = F(x r (t)) Not all problems are timereversal invariant. given that jt(7) does? We find it does: d2xr(t) dt2 d2x(t) dt2 d2x(~t) d(~tf m m m : F ( x ( . Let us wait till it has gone around by it j 2 and at this time. it is readily seen that starting from t= T it will once again go on a counterclockwise circular orbit tangential to the first at the point of reversal. This is clear from the fact that the detailed probability distribution in x is unaffected by this change. so that t = 0 defines the point when the motion is timerever When the movie is run backward we see the trajectory In other words. j . the object is where it was 5 se before the reversal. Let us consider a particle in one dimension with a timeindependent Hamiltonian H. Consider a positively charged particle in the x y plane moving under a magnetic field down the zaxis. dt m t) Let us first note that performs timereversal. Will it return to the timereserved initial state at t = 2Tl No. In the ^representation the wave equation is d f j x . The reversal of velocities follows from this: dt d{t) and does not have to be additionally encoded. with its velocity in the direction of increasing 6. it is clear from looking at plane waves (or the momentum operator — ih(d/dx)) that —p under complex conjugation. We now ask how all this appears in quantum mechanics. Due to the v x B force it will go in a counterclockwise orbit. On the other hand. t = T. The question is this: Does this obey Newton's Second Law d2xr(t) „ m—^=F{xr) .302 CHAPTER 10 point of timereversal.
if we have a magnetic field. we have seen at the end of Chapter 6 that every pgenfunction implies a degenerate one which is its complex conjugate. 0). For H= P2/2m+ V(x) this is the case. (11. complex conjugate it. 0) * It is clear that in order for the end result.e. require that H(x) = H*(x) (11. that the Hamiltonian be real. If we had included in our system the currents producing the gnetic field.2) 0) i. i. 2T).If the system has TRI. and you could have seen the charges in the wire. to obey y/(x. and the charge we chose to focus on would have followed the timereversed trajectory. which is i j / ( x .iH*(x)T/1i 303 SYMMETRIES A N D THEIR CONSEQUENCES iff * (x.e. We find the following happens at each stage: iH(x)T/n . even in ler dimensions. Indeed.*(x. Notice that the failure of T R I in the presence of a magnetic field does not resent any fundamental asymmetry under timereversal in electrodynamics. we must find the analog of Eq. I . is real.. The ws of electrodynamics are invariant under t+—t. There will be a minor twist when the problem involves Spin which has no classical counterpart. On the other hand. If H has TRI. the field would have been reversed at t=T.5. let that evolve for another time T a n d see if we end up with the complex conjugate of the initial state.2T)=y. For most problems where the coordinate basis is the natural choice the fcbove discussion will do. So let us prepare a state y/(x. P enters linearly d H(x)^H*(x). let it evolve for time T.1). iH (x) T/ft iH *(x) T/ft W*{x. There are ways to formulate T R I in a basisindependent way but we will not !o so here. This can be handled by treating spin as we brould treat orbital angular momentum. certain experiments together with general arguments from [uantum field theory suggest that there exist interactions in this universe which do iot have this symmetry at the fundamental level. the entire system would have followed the lereversed trajectory. you would have found them running backward.. if you had taken a movie of the experiment and layed it back.5. and reversed them also. On the other hand. The asymmetry in our example ose due to our treating the magnetic field as external to the system and hence not be timereversed.
let us quickly review translations in two dimensions.1.2.1.3) Px and Py are the dot products of P with the unit vector (i or j) in the direction of the translation.2.2) coordinate basis In terms of the vector operator P . rotational invariance of H leaves enough room for interesting dynamics.14) f r o m Eq. (11. 12. rotational invariance was not discussed. respectively. Since there is nothing special about these two directions. we conclude .13). The reason is that most of the problems we discuss involve a single particle (which may be the reduced mass) in an external potential. By a straightforward extension of the arguments that led to Eq. (11.1) (12. Px coordinate basis (12. P = Pxl + Py\ (12. We first consider two dimensions and then move on to three. Translations in Two Dimensions Although we are concerned mainly with rotations.12 Rotational Invariance and Angular Momentum In the last chapter on symmetries. and whereas translational invariance of H implies that the particle is free.1. which represents momentum. but because it is all too important and deserves a chapter on its own. we may deduce that the generators of infinitesimal translations along the x and y directions are.1. not because it is unimportant.
(12.(Py} sin 0o ( P y ' ) R = ( P x ) sin 0 o + <P y > cos 0o where <X)R = and <X> = < ^  X  V r > .5a) (12.2.2. (12. (12.2). we define the action of i7[/?] on position eigenkets: £/[/?] y) = jc cos 0 o ~ y sin 0 O .2. e t c .7)  We will suppress the rotation angle w h e n it is either irrelevant or o b v i o u s .6) (¥R\X\WR) (12.2.2. We will take the simpler route of dropping it f r o m the outset and proving at the end that < P x ) and < P y ) transform according to Eq.2. which specifies how ( X ) and < T ) t r a n s f o r m : in omitting a possible phase factor g(x. It is represented by the 2 x 2 matrix in Eqs.2.3) 307 ROTATION INVARIANCE AND ANGULAR MOMENTUM The rotated state  y/ R ) must be such that <X>* = < X > c o s 0 O ~ < Y } sin 0 O < 7 > * = < X > sin 0 o + < 7 > c o s 0 O ( P x ) R = ( P x ) cos 0 O .4).2.4a) (12.1) and (12. One way to show this is to keep the phase factor and use Eqs. In the active transformation picture! \yr*>=U[R]\yr> (12. let £/[/?(0ok)] be the operator associated with the rotation R(<f>0k). (12. we are also ensuring that < P x ) and transform as in Eq. (12.2.5).Let us denote the operator that rotates these twodimensional vectors by i ? ( 0 o k ) .4b) (12.2. In analogy with the translation problem.2.5). y).2.5b) to eliminate it.5a) and (12.5b) As in the case of translations. Explicit Construction of U\R] Let us now construct i/[i?]. Consider first an infinitesimal rotation case we set In this t J/[*(erk)]=7 ie7 Lz (12. x sin <f)o + y cos 0o> (12.2. Just as T(a) is the operator in Hilbert space associated with the translation a. this equation is guided by more than just Eq.2. .
e. it can be shown that (x. is to be determined. y) + — (yez) + — ft dx dy <X. and as a result <Px~) and (Py} transform just as do. y\Lz\ V> = Y(X.2. i.py) momentum space. (12. y} = \xysz.] Expanding both sides to order £z is 3 0 <x.2. the generator of infinitesimal rotations. (11..9) t f Exercise 12. (11. [Hint: Re derivation of Eq.2.y) .y\Il^^\y/}=y/(x xez + y> (1 + y£z.2.2. (12.8) from Eq. in and (Y) — ft (Px.2.6).6). which becomes to first order in e z U[R]\x.Py\Lz\ Thus I~i£zLz/ft rotates the momentum space wave function ¥(px. Starting witi Eq. \ dy) \ dxj So coordinate basis * x\ —ift — J —v( — ifi — \ dy) \ dxj ( V ) or in the abstract Lz=XPyYPx (12) Let us verify that < P x ) and ( P y } transform according to Eq. in accordance with Eq. y\I\ (xe2) y\L2\ y/) = jc( .2. (12.9).yx£z) (12.8) to Eq.5).2.308 CHAPTER 10 where L:.2.2. .1.5).12) m °ba"s [ifi ^pyifi ^ dP* d px Py > it is clear that dw dpx dw (Py£z)+—{px£z) dpy by e.* Provide the steps linking Eq.ifi — 1 — v( — ifi — if/(x. (12. Since (12. (12.
2. by changing to polar coordinates.17a) (12.2. (12. [Y.16a) (12.2.2.16) and (12.15b) AND 309 ROTATION INVARIANCE ANGULAR MOMENTUM = Pxsz + Py By feeding Eq.2.Lz]=mpx These commutation relations suffice to fix Lz as XPy  Exercise 12. (12.2. Using these commutation relations (and your keen hindsight) derive l2=XPy— YPX.14a) (12. So let us consider an alternative form for Lz.2.7) into the above we can deduce that [X.16b) (12.2.Lz] = iftX [Px. It can be shown. that Lz • .2.ifi — dx) it is hard to see that e~l4>oL::/tl indeed rotates the state by the angle F o r one thing.18) coordinate basis * x\ .2.We could have also derived Eq.11) for Lz by starting with the passive transformation equations for an infinitesimal rotation: Ut[R]XU[R]=XYez U\R\YU{R]=Xez U\R]PXU[R] U\R]PyU[R] = + Y PXPysz (12.2.2. [py.2. expanding the exponential is complicated by the fact that x(iftd/dy) and y(ifid/dx) do not commute.14b) (12.15a) (12. At least show that Eqs.Lz] = ihP.19) .ifi — coordinate basis ^A.2. The finite rotation operator £/[/?(0 o k)] is i (f> o U[R(<f>0k)j=lim \I——LZ\ = exp( — i(f>0Lz/fi) n^od \ n N J Given N (12.17b) YPX.2.17) are consistent with L z = XPyYPX.i f i — ]y( \ vy) . t (12. (12.Lz] = ihY (12.2.
(2) There exists a common basis for L z and H.310 CHAPTER 10 This result can also be derived more directly by starting with the require that under an infinitesimal rotation £ z k.22) also implies the following: (1) An experiment and its rotated version will give the same result if is rotationally invariant. N o w it is obvious that exp( . the combined effect of translations and rotations is nothing very simple. Besides the conservation of <L Z ).y/{p. Y. </>) becomes y/(p.<j>0 coordinate basis \ v (12. and Py respond to the rotation as do their classical counterparts [Eqs.14) and (12. H is rotationally invariant whenever J f is.) The Consistency Check.2. Py) (12.</>0) by Taylor's theorem. (12. Px. In contrast to pure translations or rotations. for exp( . 0) .211 >f (12.2. Y.H]= 0 Since X. which have a simple law of composition.2. We identify Lz as the angular momentum operator. Y. Exercise 12. (12.(f>0d/d(j)) \f/(p. I i > conserved in a problem with rotational invariance: if U\R]H{X. Let us now verify that our rotation and translation operators combine as they should. Px.23 Qfhj T' rotates the state by an angle about the z axis. and also by the direct method mentioned above.ifoLz/ft) — > exp . Thus the rotation operators have the right law of combination Physical Interpretation of Lz. 0 . It is also obvious that U[R(<f)ok)]U[R(<f>o^)}= U[R((<f>o+ 0o)k)]. \y(x. and then verifying that the product of the corresponding quantum operators equals the .3 * Derive Eq.15)] and H is the same function of these operators as M is of the corresponding classical variables. (We will spend a lot of time discussing this basis as we go along.2. We seem to be facing the prospect of considering every possible combination of rotations and translations.19) by doing a coordinate transformation on E q (12.2. Py) U[R] = H(X.2. since (i) it is obtained f r o m lz = xpy—ypx by the usual substitution rule (Postulate II).22i it follows (by choosing an infinitesimal rotation) that [LZ. finding their net effect.10). and (ii) it is the generator of infinitesimal rotations about the z axis. Px. Eq.2.2. y) = i f / ( p . (12.
By subjecting a point in the x y plane to these four operations we find x X+ 8X Ly+£yJ — • (x+ R(EZk) _(x+ ex)(y+ €y)e.operator corresponding to the result of all the transformations. Consider the following product of four infinitesimal operations: r w f at. the f o u r f o l d product w o u l d equal I.24) I+{£XPX+£yPy) \ i fi £y£zPX i .23) E i. however. s Zf and beyond. £X£ZPy ft IexLz fi J I{exPx+eyPy) ft (12. The deviation from this result of / is a measure o f n o n c o m m u t a t i v i t y . as can be seen by rearranging the factors so that the t w o o p p o s i t e rotations and the t w o o p p o s i t e translations cancel each other.2. Given two symmetry operations that d o not c o m m u t e .yj £x)ez + (y + > _y+£x£z.2. which will prove to be a giant step toward our goal.X In the above. the fourfold product provides a nice characterization o f their n o n c o m mutavity. £y). this characterization is complete. Let us take one small step in this direction.£y£zi + sxez\. A s we shall see.23) imposes the following restriction on the quantum operators: V[R( £ z k)]r(£)C/[R(£ z k)]T(£) = n  £y£zi + exez\) (12.£) U[R(et k)] T(E) where £= sxi + £y\. unaffected by the secondorder terms that we have ignored. (12. H o w about the coefficients of terms such as £X£2Z. Note that although these are secondorder terms. . X — (y+ £y)£Z _(x+ £x)£z+y_ (12. We do.e. . But many other tests are possible. we have ignored terms involving £ x .25) By matching coefficients (you should do this) we can deduce the following constraints: [Px.L:]=ihPy [Py . how about finite t Note that if rotations and translations c o m m u t e d ..17)]. Equation (12. they are fully determined in our approximation.ezk)] T{ . So our operators have passed this test. that the net effect is a translation by . or more generally. Lz] = iftPx which are indeed satisfied by the generators [Eq. i.2.e. retain the £ x £ z and £yez terms since they contain the first germ of noncommutativity. U[R{ .2.2.
If we convert the law of combination X — • X+ £x — • X+ £x Eyl * X y+ey • X  y £yj (12. (12. Eq. We have just finished testing the first two. we did verify that the arguments of the T operators combined according to the laws of vector analysis.2. and Lz.Py] =0 =I (12. let tis do it explicitly again.2 into the operator constraint T(£y\)T(~£x\)T{£y\)T{exY) we deduce that [Px. Py. Py. (2) This relation will be automatically satisfied if the generators pass the tests we have finished conducting. . Equation (12. (12. [Although earlier on. / Exercise 12. (12.2.2 A f l A + f t A 2 + A 2 n = [A. I. and [Px.2.23) keeping terms of order ex (You may assume £^ = 0. the possible commutators are [P v .] When I say that there are no further tests to be conducted. [A. Lz].24) to order exe\. it is not hard to explain when it is time to stop testing.. but not really. {Py. in view of the identity . Py]. The following exercise should illustrate this point. and Lz.2. verify that the new constraint coming from the e x el term is satisfied given the commutation relations between Px. We can stop the tests when all possible commutators between the generators have been considered.4 * Rederive the equivalent of Eq. I mean the following: (1) Every consistency test will reduce to just another relation between the commutators of the generators. we did not consider the fourfold product.2.26) is just a special case which brings out the commutativity of Px and P:.) Use this information to rewrite Eq.27). equating coefficients of this term deduce the constraint .312 CHAPTER 10 rotations? H o w about tests other than the fourfold product. Although the third was tested implicitly in the past. such as one involving 14 translations and six rotations interlaced? There is a single answer to all these equations: there is n o need to conduct an> further tests.23 which of course is satisfied by the generators Px and Py.].2Lz Px Lz + PXL2Z + iJ Px = h2Px This seems to conflict with statement (1) made above. fi ]] Using the identify. In the present case. given the generators Px. Although it is beyond the scope of this b o o k to explain why this is so.
is a 3 x 3 matrix. Examples of V are ?=PX\ + Py\ and R = X i + Y\ [see Eqs. its components are operators in Hilbert space. it transforms as a vector in V (R).2) (12. In order to exploit this fact we must first find the eigenfunctions of L:. and on the other. The condition OI£«IV2> = < V r 2 Z. But first note that 4 seems to be arbitrary: it can even be complex since 0 goes only from 0 to 2K.Vector Operators We call V = Vxi + Vy\ a vector operator if Vx and Vy transform as components of a vector under a passive transformation generated by U[R]: 313 ROTATION INVARIANCE AND ANGULAR MOMENTUM where R0 is the 2 x 2 rotation matrix appearing in Eq.3. H and Lz share a common basis. with the obvious difference that Ri.3) where R(p) is an arbitrary function normalizable with respect to p dp. The Eigenvalue Problem of Lz We have seen that in a rotationally invariant problem.3.2. The same definition of a vector operator holds in three dimensions as well. where we could argue that p had to be real to keep  y  bounded as x>ao.3. (12. Let us impose it.14) and (12. (Compare this to the eigenfunctions eipx/f> of linear momentum.2.3. We begin by writing L:\hy = h\l z y in the coordinate basis: I ift — dq> = / z Vt(P>0) (12.>* J This will ensure that y/ is normalizable with respect to (12. r  V r.1) The solution to this equation is Vi£p> 0 ) = K(p) eil:<t>/n (12. 12.3.2.) The fact that complex eigenvalues enter the answer.15)].% We shall have more to say about R(p) in a moment. (12. signals that we are overlooking the Hermiticity constraint.4) /•xi f* 2K dxdy= * pdpd(j> n * r > .1). Note the twofold character of a vector operator such as P : on the one hand.
We shall see in a moment that the Hamiltonian in a rotationally invariant problem does just this..3. Consider a superposition of two allowed 4 eigenstates: J4>Qf> Hp. Eq.5) to Eq. but also to be an integral multiple of ft: L = mh. m = 0. show that the>e values are either .2. h. 2h. Let us try to deduce the restriction on lz from another angle. If this requirement is to be satisfied for all y/i and \y 2 . . (12.6).8) I One calls m the magnetic quantum number. one can show (upon integ by parts) that it is enough if each y/ obeys 0 ) = y(p. The natural thing to do at this point is to introduce some operator that commutes with Lz and whose simultaneous eigenfunctions with Lz pick out a unique basis in each Vm. where m is an integer. . 3h/2. Notice that lz = mfi implies that y is a singlevalued function of 0.. By arguing on the ground of symmetry that the allowed values of lz must be symmetric about zero. (12... ± 1 . <j>) = A(p) eup J + B(p) e By demanding that upon a I n rotation we get the same physical state (not necessarily the same state vector). (12. It is m I possible to restrict lz any further this way. .3. but it can be specified by its energy and angular momentum in a rotationally invariant problem. h/2.3 ^ This forces L not merely to be real... Physically this means that a state is not uniquely specified by just its angular momentum (which only fixes the angular part of the wave function). . for the space of all normalizable functions R is infinite dimensional. Let us now return to the arbitrary function R(p) that accompanies the eigenfunctions of Lz. ~3h/2. Provide the steps linking Eq.3.3. show that lz — l'z = mfi.2.3. we find l=e _ 2itilz/fi (12. (However.2 h. h. . (12. h/2.) Exercise 12.3)... .3.. Exercise 12.3. Its presence implies that the eigenvalue lz = mh does not nail down a unique state in Hilbert space but only a subspace V m . 0.3^ (12.. ± 2 .1. The dimensionality of this space is clearly infinite.3. or .. 2k) If we impose this constraint on the Lz eigenfunctions. .314 CHAPTER 12 becomes in the coordinate basis " oo C 2n 'o JQ 1*2 it 0 •'o m 34>)¥{pdpd$ (12. see Exercise 12.
we seek simultaneous eigenfunctions of H and Lz.* A particle is described by a wave function g.3.4>) = Exi/E(p.) Exercise 12.11) (We shall use p to denote the mass. The eigenvalue equation for H is . * A particle is described by a wave function V f ( p . 0 ) = V(p).9) INVARIANCE AND ANGULAR MOMENTUM which would have been nondegenerate eigenfunctions of Lz if the p coordinate had not existed.LZ] = 0 in this problem.(l)) (12. = .3. y/(p.3. We have seen that the most general eigenfunction of Lz with .<t>) = Aep2/2Alcos2<l> Show (by expressing cos 2 <> in terms of O m ) that f I P(L = 0) = 2/3 P<Jx = 2h) = 1 / 6 /»(/.2 « ) = 1/6 (Hint: Argue that the radial part p2/2A2 e is irrelevant here. These obey the orthonormality condition In d(f> = 8n (12. since m will denote the angular momentum quantum number.n 2 f d 2 1 d 1 82\ _ 2p \dp p dp p dcj> ¥E(p.10) It will be seen that these functions play an important role in problems with rotational invariance.3.4.3.3. Exercise 12.) Since [H.It proves convenient to introduce the functions <M0) = ( I n T ^ t 1/2 im<t> 315 ROTATION (12. <I>) = A e~pl/2A2^j cos <f> + sin Show that P(lz = ti) = P(lz = n)=i Solutions to Rotationally Invariant Problems Consider a problem where V(p.
316
CHAPTER 10
eigenvalue mfi is of the form y a p , 0) = R(p){27tyx/2eim* = tf (p)® w (0)
where R(p) is undetermined. In the present case R is determined by the requ that
EM
(13
be an eigenfunction of H as well, with eigenvalue E, i.e., that y/Em satis^ (12.3.11). Feeding the above form into Eq. (12.3.11), we get the radial equation\ determines REm{p) and the allowed values for E: fj2(d 2p \dp
2 2 i
1 d p dp pJ
2

"
REm(p)=EREm(p)
(1^
As we change the potential, only the radial part of the wave function, R, chal the angular part O,,, is unchanged. Thus the functions <I>,„(0), which were obtained by pretending p does not exist, provide the angular part of the wave function in I eigenvalue problem of any rotationally invariant Hamiltonian.
Exercise 12.3.5*. Note that the angular momentum seems to generate a repulsive] tial in Eq. (12.3.13). Calculate its gradient and identify it as the centrifugal force. 12.3.6. Consider a particle of mass n constrained to move on a circle ofj l2/2pw2. Solve the eigenvalue problem of H and interpret the degeneracy"
Exercise a.
Show that
Exercise
12.3.7.*
(The Isotropic
Oscillator).
Consider the Hamiltonian 2/v2 , v2\
2 2 f j = fPx + Py i 1 2p 2
(1) Convince yourself [H, Lz] = 0 and reduce the eigenvalue problem of H to the radial differential equation for REm(p). (2) Examine the equation as and show that R U P ) — T0 l m l p~> P (3) Show likewise that up to powers of p REm{P) So assume that JR£ffl(p) = p m e ' ^ ^ U ^ p ) . •e^'2''2*
(4) Switch to dimensionless variables e = E/fi(o, y(fico/fi)l/2p. (5) Convert the equation for R into an equation for U. (I suggest proceeding in two stages: R=ylmlfJ=e~ y2,2U.) You should end up with
317
ROTATION INVARIANCE AND ANGULAR MOMENTUM
U" +
2w[+r y >  2 y U'
+(2e~2\m\2)U=0
(6) Argue that a power series for U of the form
U(y)=
£
r  0
Crf
will lead to a twoterm recursion relation. (7) Find the relation between Cr+2 and Cr. Argue that the series must terminate at some finite r if the y>oo behavior of the solution is to be acceptable. Show e = r + m + 1 leads to termination after r terms. Now argue that r is necessarily even—i.e., r = 2k. (Show that if r is odd, the behavior of R as p*0 is not p1"1.) So finally you must end up with E— (2k + \m\ + \)ft(o, Define n = 2k + \m\, so that k = 0,1,2,...
(8) For a given n, what are the allowed values of m? Given this information show that for a given n, the degeneracy is n +1. Compare this to what you found in Cartesian coordinates (Exercise 10.2.2). (9) Write down all the normalized eigenfunctions corresponding to n = 0, 1. (10) Argue that the « = 0 function must equal the corresponding one found in Cartesian coordinates. Show that the two n = 2 solutions are linear combinations of their counterparts in Cartesian coordinates. Verify that the parity of the states is (—1)" as you found in Cartesian coordinates. Exercise 12.3.8. * Consider a particle of charge q in a vector potential
t
E„ = (n + \)fi(o
(1) Show that the magnetic field is B = 5k. (2) Show that a classical particle in this potential will move in circles at an angular frequency co0 = qB/pc. (3) Consider the Hamiltonian for the corresponding quantum problem:
[Px + qYB/2c]2
t
[PyqXB/2cf
2n
2n
Show that Q = (cPx + qYB/2)/qB and P=(PyqXB/2c) are canonical. Write H in terms of P and Q and show that allowed levels are E=(n + \ /2)H(o0.
318
CHAPTER 10
(4) Expand H out in terms of the original variables and show
where H(co0/2, n) is the Hamiltonian for an isotropic twodimensional harmonic osc" of mass fx and frequency oj 0 /2. Argue that the same basis that diagonalized H(a>0/2, n) diagonalize H. By thinking in terms of this basis, show that the allowed levels for H E=(k + sm — \ m+ 5)ficoQ, where k is any integer and m is the angular momentum, vince yourself that you get the same levels from this formula as from the earlier [E={n + 1/2)/j©O]. We shall return to this problem in Chapter 21.
12.4. Angular Momentum in Three Dimensions
It is evident that as we pass f r o m two to three dimensions, the operator Lz up two companions Lx and Ly which generate infinitesimal rotations about t and y axes, respectively. So we have Lx — YPZ — ZPy Ly = ZPX — XPZ Lz=XPy~YPx (12.4. (12.4.1 (124.li
As usual, we subject these to the consistency test. It may be verified, (Exercise 12. that if we take a point in threedimensional space and subject it to the folio rotations: R(exi), R(eyj), ^ (  e ^ i ) and lastly R(£y\), it ends up rotated  e ^ k . In other words R(  £y] )R(  £XI)R(£y\)R(£X i) = R(£x£yk) (12.
It follows that the q u a n t u m operators U[R] must satisfy £/[*(£y\)]U[R(£xi)]U[R(ey\)]U[R(£xi)] = U[R(e^k)] (12.
If we write each U to order £ and match coefficients of £ x £ y , we will find [Lx,Ly] = ifiLz (12.
By considering two similar tests involving £ y £ z and £ z £ x y we can deduce constraints [Ly,Lz] = ifiLx (12... (12.4.
[Lz, Lx] = ifiLy
You may verify that the operators in Eq. (12.4.1) satisfy these constraints. So they are guaranteed to generate finite rotation operators that obey the right laws of combination. The three relations above may be expressed compactly as one vector equation
319
ROTATION INVARIANCE AND ANGULAR MOMENTUM
L x L=
Yet another way to write the commutation relations is {Lj,Lj} = ifi X £ijkU
(12.4.5)
(12.4.6)
In this equation, i and j run from 1 to 3, Lx, L2, and L 3 stand for Lxy Ly, and Lz, respectively^ and are the components of an antisymmetric tensor of rank 3, with the following properties: (1) They change sign when any two indices are exchanged, Consequently no two indices can be equal. (2) fii23= 1. This fixes all other components. For example, ei32=l, e3i2 = (  l ) (  l ) = + l (12.4.7)
and so on. In short, Eijk is +1 for any cyclic permutation of the indices in £123 and 1 for the others. (The relation
c = axb
between three vectors from V 3 (i?) may be written in component form as 3
Ci= X
(12.4.8)
3
Z Zijkdjbk
(12.4.9)
Of course a x a is zero if a is a vector whose components are numbers, but not zero if it is an operator such as L.)
Exercise 12.4.1 * (1) Verify that Eqs. (12.4.9) and Eq. (12.4.8) are equivalent, given the definition of eiyA. (2) Let U\, U2, and U3 be three energy eigenfunctions of a single particle in some potential. Construct the wave function ^ { x i , x 2 , x 3 ) of three fermions in this potential, one of which is in V1, one in U2, and one in U3, using the eok tensor. Exercise 12.4.2* (1) Verify Eq. (12.4.2) by first constructing the 3 x 3 matrices corresponding to R(exi) and /?(£ v j), to order s. (2) Provide the steps connecting Eqs. (12.4.3) and (12.4.4a).
t We will frequently let the indices run over 1, 2, and 3 instead o f x, y, and r.
320
CHAPTER 10
(3) Verify that Lx and Ly defined in Eq. (12.4.1) satisfy Eq. (12.4.4a). The proof f w other commutators follows by cyclic permutation.
We next define the total angular momentum operator squared
L /jjf I I il^
aS»
It may be verified (by you) that [L2,Li] = 0, i=x,y,orz
(I
Finite Rotation Operators. Rotations about a given axis commute. So rotation may be viewed as a sequence of infinitesimal rotations about the sac What is the operator that rotates by angle 9, i.e., by an amount 6 about parallel to 0? If 8 = then clearly
L U[R(ex JM — e i)] = „i9x x/fi
The same goes for 8 along the unit vectors j and k. What if 8 has some art direction? We conjecture that = L (where 0 = 8 / 0 ) is the generator of itesimal rotations about that axis and that i e
ft N
U[R(Q)] = lim
N^oo V
/
0L]
/
=ei90L/t> _ el9 L/fi (i2.4.i:
Our conjecture is verified in the following exercise.
Exercise 12.4.3* We would like to show that 9 L generates rotations about the parallel to 9. Let 50 be an infinitesimal rotation parallel to 0, (1) Show that when a vector r is rotated by an angle 50, it changes to r+(58xr might help to start with r±<S0 and then generalize.) (2) We therefore demand that (to first order, as usual)
y(r)
> y ( r  < 5 0 x r ) = ^(r)(<S0 x r) • Vy/ {/[/{(58)]
58/fi)L§, show that L§=$ L.
Comparing to U[R(5Q)] = I(i Exercise 12.4.4*
fornM Recall that V is a vector operator if its components V, transform!
Ut{R]VlU[R] = XRiyVJ (12.M
(1) For an infinitesimal rotation 59, show, on the basis of the previous exercise, that
321
ROTATION INVARIANCE AND ANGULAR MOMENTUM
s
£ RvVj= K+ (50 x V ) , =
j
Vt+ZZswiSO^V*
J k
(2) Feed in U[R} = \(i/ti)5QL
into the lefthand side of Eq. (12.4.13) and deduce
that
\ V u L f \ = mY,HkV k
k
(12.4.14)
This is as good a definition of a vector operator as Eq. (12.4.13). By setting V = L, we can obtain the commutation rules among the L's.
If the Hamiltonian is invariant under arbitrary rotations, U\R]HU[R]=H (12.4.15)
it follows (upon considering infinitesimal rotations around the x, y, and z axes) that [//,L,] = 0 and from it [H, L2] = 0 (12.4.17) (12.4.16)
Thus L2 and all three components of L are conserved. It does not, however, follow that there exists a basis common to H and all three L's. This is because the L's do not commute with each other. So the best one can do is find a basis common to H, L2, and one of the L's, usually chosen to be L z . We now examine the eigenvalue problem of the commuting operators L 2 and Lz. When this is solved, we will turn to the eigenvalue problem of H, L 2 , and L z .
12.5. The Eigenvalue P r o b l e m of L2 and L. There is a close parallel between our approach to this problem and that of the harmonic oscillator. Recall that in that case we (1) solved the eigenvalue problem o f / / i n the coordinate basis; (2) solved the problem in the energy basis directly, using the a and a' operators, the commutation rules, and the positivity of H; (3) obtained the coordinate wave function yn(y) given the results of part (2), by the following trick. We wrote «j0> = 0 in the coordinate basis as
322
CHAPTER 12
which immediately gave us y/o(y) ~e~y2/2, up to a normalization that could determined. Given the normalized eigenfunction y/o(y), we got y/„(>0 by the applicalfcc: the (differential) operator d/dy)n/(2nn\)l/1. In the present case we omit part (1), which involves just one more boutwtli differential equations and is not particularly enlightening. Let us now consider part (2). It too has many similarities with part (2) nft; oscillator p r o b l e m . ! We begin by assuming that there exists a basis j a, /?> cdfco to L2 and Lz: L2 \apy = a\ap}
Ls a/3> = ^  a j 3 > We now define raising and lower operators Z/f which satisfy L±] = ± fiL± and of course (since L2 commutes with Lx and Ly) [L2,L±} =0 (1 Lx i iLy
(fl
Equations (12.5.4) and (12.5.5) imply that L± raise/lower the eigenvalue o f Ls fi, while leaving the eigenvalue of L2 alone. For example, L7(L+\af}y) = (L+Lz + = (L+p + tiL+)\a(}y hL+)\apy
(13
and L2L+\ a py = L+L2\apy = aL+\afiy
F r o m Eqs. (12.5.6) and (12.5.7) it is clear that L+\ apy is proportional to the nor ized eigenket j a, p + fiy: L+\apy — C+{a, P)\a, p + fiy
(12
J If y o u have forgotten the latter, y o u are urged to refresh y o u r m e m o r y at this point.
It can similarly be shown that
323
ROTATION
L\af$y = C(a, P)\a,
P~fi)
(12.5.8b)
INVARIANCE AND ANGULAR MOMENTUM
The existence of L± implies that given an eigenstate }ap} there also exist eigenstates \a, /? + #>, a, P + 2 # > , . . . ; and  a , / ?  # > ,  a , / J  2 # > , . . . . This clearly signals trouble, for classical intuition tells us that the z component of angular momentum cannot take arbitrarily large positive or negative values for a given value of the square of the total angular momentum; in fact classically \lz \ < ( / 2 ) 1 / 2 . Quantum mechanically we have (aP\L2L2z\aP) which implies = (af3\L2x + L2y\aP) (12.5.9)
(since L2x+L2y is positive definite) or a>P2 (12.5.10)
V
a — p2>0
Since ft2 is bounded by a , it follows that there must exist a state a/? max > such that it cannot be raised: L + a/? m a x > = 0 Operating with L  and using LL+ =L2L2Z (L2Ll — fiLz, we get a/? max > = 0 (12.5.11)
(a  p 2 m ^  ^j8max)l aj8 max > = 0 a = PmAPrmx + ft) (12.5.12)
Starting with a/? max > let us operate k times with L  , till we reach a state a/? m i n > that cannot be lowered further without violating the inequality (12.5.10): La/? m i n > = 0 L+ L~a/J m i n > = 0 (L2 — L2Z + fiLz)\a pm\n) = 0
a = /?min(j3min (12.5.13)
A comparison of Eqs. (12.5.12) and (12.5.13) shows (as is to be expected) Pmi« = Pmax (12.5.14)
324
CHAPTER 10
T a b l e 12.1. S o m e L o w  A n g u l a r  M o m e n t u m S t a t e s (Angular momentum)
k/ 2
0 1/2
Pmux
0
a
0 (l/2)(3/2)«2
\ap> 0, 0>
(3/4)fi2, fi/2> (3/4)fi2,fi/2>
fi/2 n
1
(D(2)82
2fi 2 , « > 2fi2,0> 2fi 2 ,  « >
3/2
Since we got to ajS m i n ) from a/J m a x > in k steps of ft each, it follows that
Pmax /?min 2/? max ftk
ftk ^max = ~>
k = 0,
1,2,
. . .
(12.5.!
k\
a = (j8max)(jSmax+«) =
k
^ (  J   + l
(12.5.15
We shall refer to (k/2) = (Pmax/K) as the angular momentum of the state. Notiq that unlike in classical physics, /?max is less than a , the square of the magnitudej angular momentum, except when a = jSmax = 0, i.e., in a state of zero anj momentum. Let us now take a look at a few of the lowangularmomentum states list Table 12.1. At this point the astute reader raises the following objection. A.R.: I am disturbed by your results for odd k. You seem to find that have halfintegral eigenvalues (in units of ft). But you just convinced us in 12.3 that Lz has only integral eigenvalues m (in units of ft). Where did you go R.S.: Nowhere, but your point is well taken. The extra (halfintegral) eige ues arise because we have solved a more general problem than that of Lx, Ly( and L2 (although we didn't intend to). Notice that nowhere in the derivation use the explicit expressions for the Us [Eq. (12.4.1)] and in particular Lz>\ d<t>. (Had we done so, we would have gotten only integral eigenvalues as you exj We relied instead on just the commutation relations, L x L = iftL. Now, these coi tation relations reflect the law of combinations of infinitesimal rotations in dimensions and must be satisfied by the three generators of rotations whateve nature of the wave functions they rotate. We have so far considered just scalar' functions y/(x, y, z), which assign a complex number (scalar) to each point. Now, there can be particles in nature for which the wave function is more complicated, say a vector field ¥(,*, y, z) = ^ ( x , y, z)i + y/y(x, y, z)j + yz(x, y, x)k. The response of such a wave function to rotations is more involved. Whereas in the scalar case the effect of rotation by 59 is to take the number assigned to each point (x, y.:)
which followed from just the commutation relations. as the orbital angular m o m e n t u m . But the old maxim—if something can happen. who know very little about spin. our analysis doesn't imply that there must exist particles with spin integral or half integral—but merely reveals the possible variety in wave functions. and J. applies to the problem of arbitrary J and not just L.5. as the spin angular momentum (or simply spin).5. So our result. / . the generators of infinitesimal rotations will be of the f o r m J.1 it follows that under an infinitesimal rotation £k. (A simple example from two dimensions is given in Fig. y) = VX{x + y Ez . as the total angular momentum. and Sz which shuffle the components y/x. which has to be done by 3 x 3 matrices Sx.yy(x + ¥x(x+yez.325 y Figure 12.y . y/z of In such cases..1. Of course. yez. that S2 can have only integral or halfintegral eigenvalues if the commutation relations are to be satisfied. or S t look like in these general cases. We will study them in Chapter 14 on spin. y') ( 2 ) and then rotate the vector itself by the infinitesimal angle. but we do k n o w this: the 7. and L z will only do part (ii) but not part (i).1 for a concrete example). We do not yet know what J. Sy. 12. whatever be the nature of wave function they rotate. One refers to L. From Fig. x and reassign it to the rotated point (V.^Li + S.16) with L as a special case when the wave function is a scalar. Ly.xez) . Jz is not restricted to have integral eigenvalues. y. z'). T h e differential o p e r a t o r Lz does the first part while a 2 x 2 spin matrix S. z) (i) must itself be rotated by 59 and (ii) then reassigned to z'). y . does part (1) (see Exercise 12. it will—is true here and nature does provide us with particles that possess spin—i.'s must obey the same commutation rules as the Ls. in the vector case the vector at (x. Exercise 12.1* Consider a vector field *F(x. y/y. But our analysis tells us.) The differential operators Lx.xe2)ez + yy(x +ysz. So in general we have JxJ=/#J (12. i Yx vUx.1. for the commutation rules reflect the law of combination of rotations and must be obeyed by any triplet of generators (the consistency condition). where Lt does part (2) and 5. 12.5.e.yxe7)ez xez) ys'y(x. particles whose wave functions are more complicated than scalars. d o e s the second. Thus the answer to the question raised earlier is that unlike Lz. The effect o f the infinitesimal rotations by Ez ROTATION INVARIANCE AND ANGULAR MOMENTUM on a vector y in t w o d i m e n s i o n s is to ( 1 ) first reassign it to the rotated point (*'. y) in two dimensions. y) = y .
.17): we simply ignore the states with halfintegral m arie j. 7 = 0. An example of halfintegral eig will be provided when we consider spin in a later chapter. m)\j. iez ~LZ _iez "o jft in Vx n n o . They obey L 2 /m> = / ( / + l ) f i 2  / m > . .5./ Our problem has not been fully solved: we have only found the eigenvaluesthe eigenvectors aren't fully determined yet.17b m=j. We will determine them now.5. 1.5.2 . Here is a summary of what we have foul eigenvectors of the operators J2 and Jz are given by J2\jm>=jU+ Jz\jm)=mfi\jm).) Let us return to our main discussion.5. y) of y). we get J±\jm> = C±(j. Note that in the above m can be an integer or halfinteger depending on j . m± 1 > where C±(j.j We shall call j the angular momentum of the state. a.. 1/2. (12. (As in the oscillator problem. / (12. The results for the restricted problem J = L that we originally set out to sc are contained in Eq. 3 / 2 . /=0..1.j (12. I Vy so that Jx = L™<8>I(2> + = LZ + SZ I0)®S™ where I ( 2 ) is a 2 x 2 identity matrix with respect to the vector components. we will denote these states by Irn). . This example only illustrates the fa.191 .j. finding the eigenvectors means finding the matrices corresponding to the basic operators whose commutation relations are given. 7 (1) is the identii. To emphasize the generality of the] we have found. (12. we will express them in terms of J ' s rather than L's and also] to a more common notation. .. . operator with respect to the argument (x. . that JZ = LZ + Sz if the wave function is not a scalar.2. (In the present example. .5. To remind us in these cases that we are dealing with J = L. (12.5. M>. If we rewrite Eq. . Lz\lm) = mh\lm}.17a (12. and m (instead of L±. m) are yet to be determined.) Let us continue our analysis in terms of the J ' s .326 CHAPTER 12 Show that (to order ez) y'x J V'y I "1 o" 1. and /?).." eigenvalues ±ti.. .8) in terms of J±. . 1 J .1?" (t^Sb m = l. 1.
Equation (12. W e c h o o s e it to b e unity according to standard convention. m)  2 = j { j + W = 2 ~ m2%2  mfi2 n2(Jm)(j+m+l) or C+0'. m)<y. — ( j f i ) for J:.~ CX{j.w ) 0 + rn + 1)] It can likewise be shown that 1/2 so that finally 1/2 J±\ftn) = fi[0'T w)0'± w + l)]1/2li.If we take the adjoint of J+\j™> = C+{j.5. . m + 1 > 327 ROTATION INVARIANCE AND ANGULAR MOMENTUM we get <Jm\ J. m+l\j\m+\) <Jm\ J2 — j l — fijz\jm)> =  CM m)\2 or C+O". we obtain m)20".l ) f i .m _1 = . [0 .w)0+w (12. . • . so that each family with a given angular momentum j has only 2j+l states with eigenvalues j'H. ( j .21a) J There can be an overall phase factor in front o f C+. for we can write down the matrix elements of Jx and Jy in this basis: <j'ni\Jx\jm> = <J'mt\ I jm) + 1)]172 + 5jr bm<. ±/> they kill the state. m + 11 Equating the inner product of the objects on the lefthand side to the product of the objects on the righthand side.{5jr +.5. <Jm\ JJ+\jm> = j C+(j. m)\j. m ± 1 > (12. .20) brings us to the end of our calculation. w) = n[(J .5.20) Notice that when act on y.
5.23) ft/2 1/2 0 ' 0 0 0 A 0 0 0 0 0 ifi/2 0 0 0 0 m/2 0 0 0 0 0 0 0 0 ih/21/2 0 0 0 0 ifi/21/2 0 ifi/21/2 0 0 0 0 ift/2l/2 0 I (12.0) 0 0 0 0 (1.1) 0 0 in 2 (1. In what follows. Jx.5. . This is J T h e q u a n t u m n u m b e r s j and m d o not fully label a state. they are block diagonal : they have no matrix elements between one value of j and another.328 CHAPTER 12 0 ' m '  Jy\jm> = <j'm'l jm> 2/ (1 Using these (or our mnemonic based on images) we can write down the matnce. Jz.0) <U) (0. a state is labeled by  ajm).1) 0 0 0 0 0 2H (11122] 0 0 0 m 2 0 0 0 0 0 0 0 0 0 is also diagonal with elements mft. and Jy in the  jni) basis!: jm (0.0) 0 0 0 0 2fi 0 2 (1.24 Notice that although Jx and Jy are not diagonal in the \jm ) basis.0) (1. where a represents the remaining labels. 0 0 0 0 0 0 0 0 8/2 0 0 0 0 h/2 0 0 0 0 0 0 0 0 8/2 0 l/2 0 0 0 ft/21/2 0 fi/2" 2 0 0 0 0 (12. corresponding to J2. w e suppress a but a s s u m e it is the same throughout.1) 0 0 (U) 0 in 2 « (1.
(9. But this is easier said than done. More generally. (12.jy0)]=mjzu). . if has components only in V 0 . are infinite dimensional and exponentiating them is not practically possible. so is the linear combination 0 • J.9)]. corresponding to a certain . The matrices J. V j .5. . Jy] = iftJz.. j=o.. .m2] (use symmetry arguments to relate ( J 2 ) to (J2}). .. But the situation is not as bleak as it sounds for the following reason./. we just need the matrix Du).. (2) Show that in these states < Tl> = <Jy}=]2 h2\j(j+1) . . the blocks do not mix when we multiply them. Since the J's are all block diagonal.<]). all rotation operators £/[/?] will be represented by block diagonal matrices. (1) Verify that the 2 x 2 matrices J^/2).25) and Jz0/2) obey the com 329 ROTATION INVARIANCE AND ANGULAR MOMENTUM Exercise 12. Consider the series representing D(J)\ t The material f r o m here to the end o f Exercise 12. The block diagonal form of the rotation matrices implies (recall the mnemonic of images) that any vector  in the subspace V. . in practice. it will be satisfied within each block.3* (1) Show that <7JC> = </ V > = 0 in a state \jm>./+ ^dimensional block at a given j is denoted by Z>0)[/?]. Let us see why. then we have [jxU). What makes the situation hopeful is that it is possible. are block diagonal.. (3) Check that AJ x • AJy from part (2) satisfies the inequality imposed by the uncertainty principle [Eq.2. First note that since J.5. mutation ruk[J^/2\J^2)] = ihJ<1/2\ (2) Do the same for the 3 x 3 matrices J. ±j).2.i. to evaluate these if j is small. \ j—j} goes into another element  (//j) of V. . V 2 .Jy[/2).7 m a y be s k i m m e d over in a less advanced course. and so is its exponential. we need just the first ( J + 1) matrices Du). . Finite Rotations! Now that we have explicit matrices for the generators of rotations. spanned by the (2j+1) vectors \ j j ) . (4) Show that the uncertainty bound is saturated in the state  j.because J± (out of which they are built) do not change j when they act on \jm}. In particular when we consider a commutation relation such as [Jx. Thus to rotate \ y/j}.. JXi Jy. Consequently. and Jz. If we denote the (2/+ 1) x (2j+ 1) block in J. The (2'.5. . by j j ^ . V. we can construct the matrices representing £/[/?] by exponentiating ( —/0 • J / ft). (3) Construct the 4 x 4 matrices and verify that Exercise 12.5.
up to a normalize. tion factor.5.is assumed to be invariant under any rotation. Jz\jj}=jfi\jf>. Consequently the series representing may be reduced to D^=zf„(0)(eo j f y It is possible.4) that (0 • J 0 ) )" for n > 2/ can be written as i^bt combination of the first 2j powers of 9. Let  be an element of Vy. = UJt U\ The primed operators have the same commutation rules and hence eigenvalues as the Jt. . with ]. and a rotation cannot change this. This means that Vj itself does not contain invariant subspaces. we find that \j. they are called invariant subspaces.5). . we get + 59~J+ + 259z Jz)]\jj> and J\jj> = *(2j)l/2\jJ 0ijS9z)ljj}\i(2jy/2d9+\j\jl> Since V. The eigenvectors are just \jm}' = U\jm)>. we can easily establish that the ( 2 j + 1) orthonormal vectors. In the following analysis we drop all primes./ > generated by the operators J?. F o r example. which also belongs to V . Consider any unitary transformation U which converts \jj} into \y/} and a different triplet of operators T H defined by J'. .)}'=  y/}. . • •. Since we haven't chosen a basis yet. from  y/'). to find closed expressions for f„(0) in terms of trigonometric functions. By considering more of such rotations. \jj}.J ° ' . Formally it is because 2 [ J . such that it is the basis vector \jj}. also belongs ti Subtracting (1 .. which is irrelevant in what follows. The invariant subspaces have another feature: they are irreducible. . of V .j) all belong to Vy. . in practice.jalso belongs to V. .330 CHAPTER 10 It can be shown (Exercise 12. for modest values of j (see Exercise 12. (What if we had already chosen a basis  j j ) ..5. .  j.{i/2K)(59+Jwhere S6± = Since J+\jj) = 0. D(X/1\R] =c o s 0 / 2 ) sin(£ « Let us return to the subspaces V .) Let us apply an infinitesimal rotation 50 to \y/}.ij59z)\jj"). The physics behind the invariance is simple: each subspace contains states of a definite magnitude of angular momentum squared j(J+\)fi2. I j. This gives = [/(//weJ)M> = [I. . is as big as the latter.. Since they go into themselves under arbitrary rotations. £/[i?]] = 0 and so £/[/?] cannot change the eigenvalue of J2. [Y'} = (89x±i59y) 1>. . any invariant subspace V . We prove this by showing thai. and furthermore. let us choose one such that  y ) is one of the! basis vectors.
Since \H. For the set of all rotation operators. (For example. We are concerned here with subspaces of smaller dimensionality.. Such vectors form a ^dimensional subspace. further block diagonalize all the D°\ We show that if this were not true. Let it be possible to block diagonalize all the D°\ say.. . . then a contradiction would arise. (In a technical sense. J ] = 0. etc. where the boxed regions represent the blocks. the elements of which do not generally commute with each other. this irreducible form is the closest one can come to simultaneous diagonalization. Z)(0>. The block diagonal matrices representing the rotation operators U[R] are said to provide an irreducible (matrix) representation of these operators. Dm 0 D(1/2) ljm> basis 0 Dn) Consider next the matrix representing a rotationally invariant Hamiltonian in this basis. respectively.means that we cannot. It follows that V. any vector with just the first d\ components nonzero will get rotated into another such vector. has the same dimensionality as Vy . Z? (1) . Vj is its own subspace and is invariant. which also commutes with .) We have seen this is impossible. AH this is summarized schematically in the sketch below. contains two invariant subspaces of dimensionalities dx and d2.Thus V. H has the same form as J2. The unboxed regions contain zeros. by a change of basis within Vy. Thus Vy has no invariant subspaces. as follows: 2j+ x 331 ROTATION INVARIANCE AND ANGULAR MOMENTUM U'm) basis 2j + l t 0 new basis dt I I i 0 (The boxed regions are generally nonzero).) The irreducibility of V.
the relation between degeneracy and rotational inva is different in the two cases.5..5. (Hint: In the case J=JZ what happens when both sides are applied arbitrary eigenket ym)? What about an arbitrary superpositions of such kets?) (3) It follows from (2) that J2J+1 is a linear combination of J°. since [H. rotational invariance is the cause of degeneracy in both cases. Arg the same goes for J2j+k.. J]. Exercise 12.j } ..c o s ( 0 / 2 ) / ( l / 2 ) .5.J 0 ) .. as is the case for classical states with these properties. given two elements of Vy. namely. if we are given two states with the^ magnitude of angular momentum but different orientation. it is not always true thai may be rotated into each other (Exercise 12. k= 1 .. Consider the family of states  j j ) . .7). Q u a n t u m mechanically. Onei them as states of the same magnitude but different orientation of angular momentum.e. Classically. in the classical sense.. namely. since [H.5.5. we argue as folk (1) One may be reached from the other (in general) by the combined action] and U[R].id • J { i / 2 ) /ft) . but] degenerate states are not always rotated versions of each other in the quanti (Exercises 12. all to of a given j are degenerate in a rotationally invariant problem. However.332 CHAPTER 10 all the generators. (2) Show that 1 (jjn)[jum[J{jT)b\ •• (J+jti)=0 where J=0. Using results from the previous exercise and Eq. Conside . J2] = 0. 2 . . (12. i takes this remark literally.6). J2j. jfi. = 0.4.. is an eigenspace of H with eigenvalue £}. (1) H is diagonal. . we argue that they^ degenerate because (1) One may be rotated into the other.. (j—\)fi. . Although the result is true classically. \j.5.(2i/£)sin(0/2)<? • J ( i/ 2 ) = (cosOx\){—) isin (2) D0)[R] = wp(iOxJ(])/ft) 0x('^A + Ia) Exercise 12.6 and 12. . (2) These operators commute with H. i. (jfi). In short..... Generalize the result to 0• J ( j ) . J±] = 0. (2) Within each block.5. .5. Exercise 12..23J that (1) Z)(1/2)[i?] = exp( . one is led to believe that one mayl these into each other.. (Hard)..e.6. * (1) Argue that the eigenvalues of Jf and Jy * are the same as I J? '. [ H . . \jm}. . H has the same eigenvalue Ej. It follows from (2) that V. i. (2) This rotation does not change the energy.
1> for any 0X 333 ROTATION INVARIANCE AND ANGULAR MOMENTUM Theerror stems from the fact that classical reasoning should be applied to <J>.) (2) Let it act on 1.istance. s . which depends on only three parameters. 1> into just 1.1).) that can't be satisfied by varying R.) Exercise 12. 1> is * [ . and not direcly to \jm}. and 7 can one rotate 1.5. P. the construction of the eigenfunctions of L2 and Lz in the coordinate basis. m) into a linear combination that involves l. 0> >[i?(M)]l.. the family 1. 11. 0 X .5 with Jx*Jy.1 ) .s i n 6»J + cos It is not too hard to see why we can't always satisfy  jm'} = D^[R]\jm} or more generally. . (4) Show that one can always rotate any 1.i. and a called Euler angles. that the state with zero igular momentum along the z axis.0>. Rather than parametrize an arbitrary rotation by the angle 8. (Use the result from Exercise 12. p. verify that a 180° rotation about the > axis applied to ]1. but it will not be a rotation matrix corresponding to £/[/?]. (5) To see that one can occasionally rotate  /m) into \jm'). that takes /m) into \ jm'y or  y/.7: Euler Angles. y and any m. /?. 0). which describes a single rotation by 6 about an axis parallel to 0. which is a vector in Hilbert space. y. which define three successive rotations: U[R(a. Verify that <J> responds to rotations like its classical counterpart. y)] explicitly as a product of three 3 x 3 matrices. complex relations between the components of  y]) and 11//.1).y)] =e e~ip^/n e (1) Construct D(1)[i?(cr.1> and show that <J> in the resulting state is ( J ) = #(sin P cos ai + sin /? sin aj + cos pk) (3) Show that for no value of a. 1. namely.5. we may parametrize it by three angles.} into  y/j). (Of course one can find a unitary matrix in V. 0 y . m'. which responds to rotations like an ordinary vector. These abstract equations imply (2/ + 1) linear. 0>. > Angular Momentum Eigenfunctions in the Coordinate Basis We now turn to step (3) outlined at the beginning of this section. p.1> turns it into 11. 1. and Qz. satisfy by any choice of R.m'). for example.1> by some utable (5 nl) angle about the * axis. for two normalized kets  y/j} and  y/j). Using £> (l) [i?(0 A i)] from part (2) in the last exercise how that l 11. for some a. may be obtained by rotating 1. by showing that (J) in the state £> ( 1 ) [£(0 x i)]l. ft. It may seem. . given the information on the kets \lm).e.
0) = R(r)(sm where R(r) is an arbitrary (normalizable) function of r. we will see that H will nail d o w n R if we seek simultaneous eigenfunctions of H. as we did in the study of Lz in two dimensions. The " t o p m o s t " state j//) sa L+//> = 0 If we write the operator L± = LX± iLy in spherical coordinates we find f 8 „ 8 (12.5. But first let us introduce. <$) the eigenfunction corresponding to  / / ) . 6.334 CHAPTER 10 Consider the states corresponding to a given /.<f>) = 0 \8G 8(j>)Y * Since is an eigenfunction of Lz with eigenvalue Ifi.5. 6. (12. L2.8 (Optional). When we address the eigenvalue problem of rotationally invariant Hamiltonians. the function that would i . and Lz.5. or 1 cotG^HO J d( sin 9) sin 6 (12. we let W\{r.30) I 0)1 U\{r. we find satisfies ( —•+ i c o t 0 — )y/'(r.1 coordinate basis ± n e ^ i — ± i cot e \89 (12 d(j)J Exercise 12.29) \86 dU'i V1. \ 1 v p:Q Y ^ d(j>) ^ coordinate basis "hST" • ifi — cos d>~—I.sin 6 cot d — ^ 60 5<j>) If we denote by y/'i(r. Verify that J • ± d L in sin© $ 1cos 0 cot „0 — \. 4>)=Ufa and find that 6) e1 1/4.2!) (12. 6.5.
which are regular between 0 = 0 and k. I. for m > 0. = 2 r j 2 < / ( c o s 0)d<t> = 1 • (12. the others ensure that * l*' c \Y'. .1 / reflects our convention. . (12. "(2/+l)f An ll—m d{cos 0) _ 1/2 TO = 1 (l+m)! 1/2 27! . The result is.32) 335 ROTATION INVARIANCE AND ANGULAR MOMENTUM Whereas the phase factor ( .5. <py 1 (1) (2/)1/2 ft L \89 Yi 84>J J We can keep going in this manner until we reach 7/ _1 .36) d<j> If we seek common eigenfunctions of the f o r m ! / ( 9 ) e'"1*. Another route to the Yf is the direct solution of the L 2 . Lz eigenvalue problem in the coordinate basis where 1 d 89 1 a2 sin 9 89 and of course L7~* — ift sin7 9 8ft (12.5. nondegenerate solution in the absence of the radial coordinate : (2 / + ! ) ! 4k nl/2 j— (sin 0)1 eihp 27! (12.5.5. . 2 . 1. (f>)dn=sltsr.5.(21) 1(1—m)!_ 21 (12. These functions are called spherical harmonics and satisfy the orthonormality condition YT (e.1= 0. .33) J 1 J _ 1 J< .35) (sin 0) For m < 0.\ dQ. Ylr\e.<t>)Y?'(Q.40).5. Since l\y = h(21)l/1\l.l JQ We may obtain by using the lowering operator. see Eq. f We neglect the function R(r) that c a n tag a l o n g as a spectator. we will find that L 1 has eigenvalues of the form I (/ + \)ft2.1> (12.34) L\lI> = h[(I+l)(\)f/2\l. .have been the unique.
we uf readily see (assuming y/ is normalized to unity) that (* OO P(L2 = l(l+l)ft\ Lz = mft) = C/"(r) V dr (12. 6. 6.336 CHAPTER 10 where / > \m\.5. under the action of t/[i?].9. distance r with angular m o m e n t u m (/. which are Hermitian on singlevalued funi of 6 and 4>. Exercise 12. The Yf functions are mutually orthogonal because they are n.5. i. . —I. (j)) in terms of Y/m(6>.5..t The expansion Eq. m)..5 ^ If we compute (y/\L2\y/} and interpret the result as a weighted average. will getmi\ec with each other by . . which is insensitive to 6 and is Hermitian with respect tol integration..37a) tells how to rotate any yr(r. 4>)yr(r.1. t N o t e that r is just the eigenvalue o f the operator (Xz+ Yz + Z2)'/2 which c o m m u t e s with L1 and I . the coefficients CJ"(r). (10.. (j)) using rdep coefficients [consult Eq. (12.: y(r. I In practice. 0 /= 0 m = — / where <700 = r r (0. We may expand any y/(rf 9.20) for a similar expansion]: O O / (12.^i It is clear f r o m the above that C? is the amplitude to find the particle at a radi.5.e. Show that L2 above is Hermitian in the sense yf(L2y2)dQ. (2) Each block will rotate the Cfm into linear combination of each other.= The same goes for L z . m\. one can explicitly carry out these steps only if y/ contains onh YT's with small /.M)=I I CT(r) 5 7 ( 0 . 4>) (12. A concrete example will be provided in one of the exercises. l—\. <jj) by an angle 0 (in principle): (1) We construct the block diagonal matrices. erate eigenfunctions of L2 and Lz.
5.5.40) Closely related to the spherical harmonics are the associated Legendre PT (with 0 < m < / ) defined by 1/2 polynomials (2/+ \ ) ( I . „ d a mz — sin 0 12— 2 ^ W ) = o 80 n sin 0 . For example. Polar plots of these functions may be found in many textbooks. is almost entirely confined to the x . We already know p = mH from the analysis of —iti(d/d<f)).5.36). The Shape of the Yf Functions.I T S * PT(cos 6) (12. Likewise. for large /. <£) = Y7m={\)m(YT)* (12. / 2 sin Q e±i(p cos 6 (12.5.  Yj\ ocsin' 61. as one would expect of a classical particle with all its angular momentum pointing along the z axis.f Here are the first few Y? functions: 337 ROTATION INVARIANCE AND ANGULAR Y?l= T ( 3 / 8 ^ ) .10.5. So assume that the simultaneous eigenfunctions have the form and show that satisfies the equation 1 \sin 0 50 8 . almost entirely confined to the z axis.39) MOMENTUM Y0{=0/An)yl y2±2 = (15/32rc) 1 / 2 sin 2 0 e±2i<t> Yf=T(15/$7i)1/2 sin 0 c o s 0 e±i<p ^2 = ( 5 / 1 6 ^ ) 1 / 2 ( 3 COS2 6 — 1) Note that f 3T(0. Write the differential equation corresponding to L2\afi) = a\apy iri the coordinate basis. For large /. Pf(cos 9)=Pj(cos 0) is called a Legendre polynomial.41) If m=0. Exercise 12. using the 1 } operator given in Eq. (12.m ) \ 4 7i(l + m)\ ( . the functions  Y™ exhibit many classical features.  is.y plane.
Thus the aj diverges when u*l (0>O or n). and r.5. By rewriting the equatio terms of w = cos 0. 2 .(12. Derive Y{ starting from Eq.] Lower it to get Y° and Y\X and compj with Eq. (12.5. / = 0. z. Prove the i for Y\.5. Show that if a/fi2 = (/)(/ + 1 ) .5 = / ( / + l ) . ./2r (12. commute with II. that 3 \ \ 1/2 x±iy 2. />(/z = 0)=x 1 P(lz = +fi)= \/6 = P(lz = —fi). . P\.) Exercise 12.13 * Consider a particle in a state described by y/ = N(x + y + 2z)e~ar where N is a normalization factor.1. Verify that L_ does not alter the parity.5. . 1. thereby proving the result for all Yf. (12. theseriei terminate and be either an even or odd function of u. Exercise 12. Exercise 12. show that for a particle described by y above. <j>^<p + n under parity... and P2 and compare (ignoring overall scales) with the Y/ functions. The funcf Pa(u) — P?a+1)fi2(«) = P?(u) = Pi (u) are just the Legendre polynomials up to a scale fa^ Determine P0.32). y. show that P„ satisfies d2P« dPi (a) 2 Convince yourself that a power series solution Pa= I M= 0 CnvT I will lead to a twoterm recursion relation.5. they should share a basis wi Verify that YTu(~\)lY?.. (2) \m\ <1 We will consider only part (1) and that too for the case m = 0. .338 CHAPTER 10 We need to show that a (1) . (1) Show. n /=0.. (First show that O^nO. by rewriting the functions in terms of .5.39).28) and normalize it youi [Remember the ( . Show that (C n + 2/C„)»l as n^oo.11.1 o YV J ± ) 'An) ! (2) Using this result.12* Since L2 and L.2.x. .1 ) ' factor from Eq.
0)=RElm{r) Yne.6.5. L] = 0 for a spherically symmetric potential. <p) <12. and bearing in mind that the angular part of V2 is just the L2 operator in the coordinate basis [up to a factor ( . . The Schrodinger equation in spherical coordinates becomes Tti2( 1 8 2 8 1 d .14.5.36)].5. y cos 9X . Consider a rotation 6Li.5. we seek simultaneous eigenfunctions of H. L2.6."1. 9. z) ' yR = y/(x.$) = EWE(r. [Hint: (1) y?. J Check your result against that anticipated above.z sin dx> z cos 9x+y sin dx) Let us verify this prediction for a special case § which must go into v^Aze^ \jfR = A(z cos 9xy sin 9X) e~r2/al (1) Expand y in terms of Y{. (2) Use the matrix e~lf>xLx/n to find the fate of i// under this rotation. F.42).e. and Lz : V e . we get t See Exercise 12.4>) (12. Solution of Rotationally Invariant Problems We now consider a class of problems of great practical interest: problems where V(r. (12. (12.6. Y?.fi2r2)~\ see Eq. which corresponds to "0" 1 (2) Use Eq.5. n d 1 82\ 1 T t — r — + 2 sin 0 — + 2 . 5 + V(r) _ 2fi V 8r 8r r sin 0 89 80 r sin 2 9 8<fi) xWE{rfQ.2) Feeding in this form.<f>) = V(r). Under this 339 ROTATION INVARIANCE AND ANGULAR MOMENTUM x^x y ^ y cos 0X — z sin 0X z*z cos 9t + y sin 0X Therefore we must have y.Exercise 12.] 12.1) Since [H.
=0 Exercise 12.1. At this point it becomes fruitful to introduce an auxiliary function UEi d4 as follows: Rsi= and which obeys the equation l(l+\)ft2 U£l/r (13 Ldr1 h2 E V(r) UE.3) and Eq.) (3) Having found E.6. oo.(r)UEI = EUEl (12. in all but the 1=0 states.* A particle is described by the wave function y/E(r. there is the repulsive centrifi barrier. as anticipated earlier. We find. (12. the (2/+ l)fold dege of H.2* Provide the steps connecting Eq. Exercise 12.c o t (2) In addition to the actual potential V(r). find E.6.6.( .5) as an eigenvalue equation /(/+i)fi2 2 pr2 J UEl = D.5).340 CHAPTER 10 the radial equation 1 8 2 8 ~2 V r 8r 8r 1(1+1) + V(r)\ REl = ERei 2 r (12. The equation is the same as the onedimensional Schrodinger equation for the following differences: (1) The independent variable (r) goes from 0 to oo and not from . (3) The boundary conditions on U are different from the onedimensional c We find these by rewriting Eq. (12.QJ Notice that the subscript m has been dropped: neither the energy nor the) function depends on it. consider finite r and find V(r). (12. 0. 1(1+ l)hz/2pr2.6.6.4>) = A e~r/a0 (a0 = const) (1) What is the angular momentum content of the state? (2) Assuming \i/E is an eigenstate in a potential that vanishes as leading terms in Schrodinger's equation.
6. . (12.10.7b) Exercise 12.6.6. the expression in the brackets in Eq.9) Now this condition is satisfied if V * c.7b) vanishes at the upper limitt and the Hermiticity of D. Show that Eq. U*(D. then we demand that 341 ROTATION INVARIANCE AND ANGULAR U?(D. (12.3. (12. c = const (12.6.6. In other words.8a) or U El > ikr (12.6.7a).10) i— t For the oscillating case. In either case.7a) MOMENTUM This reduces to the requirement 'vt^VSH? dr dr j = 0 (12. if U\ and U2 are two such functions.U\) dr = (DjUifUidr (12. is Hermitian with respect to them.and demanding that the functions UE! be such that D. hinges on whether or not dU2 TT dUf Uf——U2 — dr dr = 0 (12.6.6.7b) follows from Eq.6. a necessary condition for REI\Vdr I UEI\2 dr to be normalizable to unity or the Dirac delta function is that U El >0 (12. we must use the limiting scheme described in Section 1.8b) the first corresponding to b o u n d states and the second to u n b o u n d states.U2)dr = LJ. Now.6.
342 CHAPTER 10 If c is nonzero. Wei try to extract further information on UEi by analyzing the equation governing? these limits. (2) Show that V 2 (l /r) = —4. without making detailed assumptions about V(r).. § O r compare this equation to Poisson's equation in electrostatics V 2 0 = ~ 4 / r p . for R is still integrable. consider sphere centered at the origin and use Gauss's law and the identity V2<£ = VV$). In this case we k n o w f r o m Coulomb's law that 0 = t ] .* (1) Show that r sin 6 (consider a test function). This is because of the (13 the proof of which is taken up in Exercise 12. Thus unless V(r) contains function at the origin (which we assume it does not) the choice <?#0 is unte Thus we deduce that V. Here p = <S3(r). j 1/2 the angular part of y/ has to be F o = (4TC)~ . The problem with c ^ O is that the corresponding total wave funct r 7u 0 does not satisfy Schrodinger's equation at the origin.6.  represents a unit point charge at the origin.§ General Properties of Um \ We have already discussed some of the properties of UE! as r+0 or co.4. To see what happens at r = 0. This in itself is not a disqualification.6.r<S3(r) (Hint: First show that V 2 (l/r) = 0 if r^O. the equation is dominated byj $ A s we will see in a moment.4. is incompatible with the requirement that ' as r»0. Consider first thel r+ 0. >0 r>0 (i: Exercise 12. Assuming V(r) is less singular than r~2. then R J\ i^w U r v — ' r diverges at the origin.
and the answer may be sensitive to the potential. Although £/0(r)>0 as r>0./ ) (regular) (irregular) (12.13) 343 ROTATION INVARIANCE AND ANGULAR MOMENTUM We have dropped the subscript E. 2. for which K(r)ocr2) and we cannot say anything in general. it will dominate the result (as in the case of the isotropic oscillator. Consider now the behavior of XJEl as r>oo. So let us consider the case where rV(r)*0 as r~*ao. £ < 0 : The particle is bound. The region r+ao is classically forbidden and we expect UE to fall exponentially there. If / = 0 .6. £ > 0 : the particle is allowed to escape to infinity classically. .) There are now two cases: 1. At large r the equation becomes d2UE dr2 2fiE ' ft2 VE (12. since Ro(r)=Uo{r)/r^O a t r = 0.=0 will also behave as r / + ' with / = 0 .15) ft (We have dropped the subscript / since the answer doesn't depend on /. note that a particle in the / = 0 state has a nonzero amplitude to be at the origin. the centrifugal barrier is absent.6. If we try a solution of the form t w we findi a(a — !) = /(/+1) a or a = 1+1 and Ir' ' r" Ur +1 or ( . t/. If V(r) does not vanish as r+oo.6.centrifugal barrier: /(/+1) (12. In the problems we will consider. We expect UE to oscillate a s r*co.14) We reject the irregular solution since it does not meet the boundary condition 1/(0) = 0. since E becomes inconsequential in this limit. The behavior of the regular solutions near the origin is in accord with our expectation that as the angular momentum increases the particle should avoid the origin more and more. The above arguments are clearly true only if 1^0.
t A l t h o u g h A and B are arbitrary in this asymptotic form. rV(r)+ 0 as r>c can take the limit as r>oo in the integral and f ( r ) approaches a constant as r j If instead V(r) = as in the Coulomb problem. we ignore f" and find f kfi2 ifi J /(r)=/(ro)exp=F M 2 V(r') dr' } ro (12. the particle behaves as a free particle far from the origin.' r L kfi2 \rjJ ) This means that no matter how far away the particle is from the origin.! Nov might wonder why we demanded that rF(r)>0 and not simply V(r)>Q as r] To answer this question.15) are of the form UF = Aeikr+Be—ikr k=(2pE/ny/2 u that is to say.5. The solutions to Eq. which i UEi{r = 0) = 0 .6.. § W e are considering the case o f equal and o p p o s i t e charges with an eye o n the next chapter.e.ft 344 CHAPTER 12 Consider the first case. this problem only gets worse. let us write UE=f(r) e ±ikr and see if f ( r ) tends to a constant as r>oo.6. their ratio is determined by the requir that if UE is c o n t i n u e d inward t o r = is due t o the constraint it must vanish. If V(r) falls even slower than a Cou potential.§ then / ( r ) = / ( r 0 ) exp± and UE(r)~exp± ilkr + ^zln \ kfi2 r 02. it is t completely free of the Coulomb potential.6. is because Dt is nondegenerate even for £ > 0 . and n o t two.5) we find (ignoring the centrifugal barrier) f"±(2ik)f'2 n ^f=0 Since we expect f ( r ) to be slowly varying as r>oo. (12. fl . If V(r) falls faster than r~\ i. T h a t there is just o n e free parameter solution (the overall scale). Feeding in this form of UE int? (12.1 where r 0 is some constant. see Exercise 12.
Now. (If A/B were arbitrary. Show that Dt is nondegenerate in the space of functions (J that vanish as r+ 0. we could choose B=0 and get a normalizable bound state for every £ < 0 .) Note that UEl is nondegenerate even for £ > 0 . All the results from the £ > 0 case carry over with the change 1 Thus f UE >Ae~Kr + Be+Kr (12. (Recall the proof of Theorem 15.Consider now the case E < 0. 6. Section 5.17) with k ^ i K ] UE~exp( = ± I n cr rjefKr (12.6. fyr2 dr d£l = SEE 8„> S„ We will consider the case £ > 0 in a moment. obeys 0) I W*im(r. Exercise 12.5). Jo and VE!m(r. In the Coulomb case we expect [from Eq. This means that E. .6. the energy eigenfunctions are normalizable to unity. (12. When E< 0.18) is true only if rK(r)>0.6. we will find that this is indeed the case. label a state fully in three dimensions. Only for certain discrete values of E will the eKr piece be absent.5. I. <p)wE'rm (r.6. both eKr and e~Kr will be present in UE. For arbitrary £ < 0 . (12. 4>) = RB(r)rr(G. these will be the allowed bound state levels.6. we have ™O O UEl(r)UEl(r)dr=SEE. the growing exponential is disallowed. Eq.6. and m. 9. ) As before. As the operator D/(r) is nondegenerate (Exercise 12.19) ^^e2/*:*2 When we solve the problem of the hydrogen atom.18) K= (2n\E\ffi2y/2 345 ROTATION INVARIANCE AND ANGULAR MOMENTUM r> oo Again B(A is not arbitrary if we demand that UE continued inward vanish at r = 0. 0.6.
Now we premultiply both sides by d} to get dUi(dt.6.= 1+1 p dp (12. Rev. 59.20) becomes (d.6. / ( / + 1) r2 i/£/=0.dl)Ui=U.U. we end up with d2 .6.6. 0.U The variable k. This problem looks a lot like the harmonic oscillator except fi the fact that we have a potential 1 fp2 instead of p2.25. Eq. Phys. (t>) = REl(r)YT(e.) In terms of these. which has disappeared. Infeld. = d]U.23 (12.6.24 t T h e present analysis is a simplified version of the w o r k o f L. we obtain d2 ^P 2 /(/+!) P2 t I//=t/i (12..211 (Note that d/dp is antiHermitian. (12. These are d — + d.346 CHAPTER 12 The Free Particle in Spherical Coordinates^ If we begin as usual with VWr. So we define operators analogoi to the raising and lowering operators.211 i and its adjoint d dp l+l P i (12. (12. 737 (1941).6. will reappear when we rewrite the answer terms of r = p/k. = fi2 Dividing both sides by k2. and changing to p = kr.) You may verify that d) d} — di+ j d]+1 so that (12.6. . 0) and switch to UEi.. (12.
Let us begin with the equation U. .28) Now.+ l (12. If.6.27) le constants in front are chosen according to a popular convention. we are really interested in the functions Rf= U.) ( d t+ l V „: \ . we can find the others.26) where c} is a constant. Given UQ./p.6. Uo must be included.6.It follows that d]Ui=c.§ These obey (from the above) pRl+]=d](pR. We see that d) serves as a "raising operator" in the index I.! From Eq.12). for it can always be absorbed in the normalization. /r = kU. (12.20) it is clear that if 1=0 there are two independent solutions: t/£(p) = s i n p . w e will gain some insight into the origin o f such a ladder o f solutions.U. one is considering the equation in a region that excludes the origin. § Actually w e w a n t Rt = U. But the factor k m a y be absorbed in the n o r m a l i z a t i o n factors of V and R. cos p (12.+l=d]Ui (12. P dp) t In Chapter 15. We choose it to be unity./p. Consider now the tower of solutions built out of UQ and U*.6. however. (12.6.J dp d p) \ R l or R i+1 j+ J . N o w Uo is unacceptable at p = 0 since it violates Eq. p dp) p'"1 i+i Ro .
.co p \ 2/ 1 / ln\ • — cos p P V 2 / (12. 3 cos p J2(P)= — — s i n p —.30b) called spherical Neumann functions of order /.6. .—J [ — y ^ ) (12.32) nt (12.+ Here are a few of these functions: .6. the ji(l) functions are finite and in fact jt(p) • (12. \P p) P As p—*co. / 3 1\ ~ — — cosp \P pi * — sin ( p /'.29) r b = cos p These generate the functions —j called the spherical Bessel functions of order /.. sin p p cosp p .30a) n i p (12..33) t O n e also encounters spherical Hankel functions ht= in. .32) J Despite the apparent singularities as p>0.6. sinp 7o(p)= .6.6. i n s o m e problems.348 CHAPTER 10 so that finally we have = \p dp] Now there are two possibilities for Ro: — Rq (12. v n2(p)= «o(p) = cosp cosp p sin p p 3 sin p — P 1\ .6. these functions behave as j. and R? = = ( y \ . .
6.6 * (1) Verify that Eqs. 0.6. y.22) are equivalent to Eq.6.6. we would have readily obtained VfE(x.37) (2) Verify Eq.2 I These satisfy VBnUr.6. (12. Show that there are no bound states for V0<n2fi2/&prl.6. * Show that the quantization condition for / = 0 bound states in a spherical well of depth — V0 and radius r0 is k'/K = .6.3 ) .24).6. (Recall Exercise 5.38) . z) 1 (2n fi) 3/2 2 ft2k2 2p 2p (12. Find the energy levels of a particle in a spherical box of radius r0 in Exercise 12. . (12.20) =~8(kk') (12.6.21) and (12. on the other hand.6. (12.6.14). are singular niip) (21. Verify that j0 and j\ have the limits given by Eq.34) AND 349 ROTATION INVARIANCE ANGULAR p0 MOMENTUM and correspond to the irregular solutions listed in Eq.) Connection with the Solution in Cartesian Coordinates If we had attacked the freeparticle problem in Cartesian coordinates.9. (12. (12. .6.6.2.6. (12. The Neumann functions. (5)(3)(1).8* the / = 0 sector.tan k'r0 where k' is the wave number inside the well and IK is the complex wave number for the exponential tail outside. Freeparticle solutions that are regular in all space are then fi k E= 2p 2 7. Exercise 12.33). (t>) =ji(kr) YT(6.14). These are just the regular solutions listed in Eq. (12.6.where (2/+1)!! = (2/+ 1)(2/—1)(2/.7.1)!! j+1 (12.6.36) We are using here the fact that jl(kr)Mk'r)r2dr *o Exercise 12.35) 1 WEtmWE'l'mr2 drdCl = —8(k~k')8tv nk 8. Exercise 12.6. (12. 0).
describing a particle moving in direction with energy E=ft2k2f 2p.41) This relation will come in handy when we study scattering. As p*r/fi = ( p r cos 6)ti=kr we get ikr cos 6 2/2 rk 2/2 cos 6 (K It should be possible to express this solution. (12. Since we have y?(0)= 00 €fkrme= s ClMkr)pl(cos e)t C/= Co.(*) = .(cos0) 1/2 / ht+Y _ /=0 It can be show that C.(cos 0) d(cos 0) = [2/(2/+ l)]<Sfl.(2m + I)!! after projecting out Ch Hint : Consider the limit kr^O . Exercise 12. This concludes our study of the free particle. as a linear combination of the functions which have the same energy.6.6.350 CHAPTER 10 Consider now the case which corresponds to a particle moving along the z axis with m o m e n t u m p.6.10. (21+ V 4 it lV/2 P. Verify Eq. <j>) / = 0 m=l (12.* Y * .6. = i'(2/+1) so that i'(2l+ \)ji (kr) Pi (cos 6) 1—0 (12. (Optional). or equivalently. the same k : go ikr cos 6 / = 1 1 CTjiikr) YT(6.(cos 0)P. only terms with m = 0 are relevant since the lefthand side is independent i (j).41) given that (1) P.lj 2 l\ (3) d'(x2[)1 dx1 J 0f ' ( i . Physically this means that a particle moving along the z axis has no angular* m o m e n t u m in that direction.4 Now. 1 (2) />.
6. (p) /(/+ 2 pr 2 2p UE!=0 (12.43) VElm = we obtain the radial equation 1 [dr2 fi2 .46) (12. we find U~e where (12. will be discussed in detail in the next chapter.6.6.6.6.42) If we write as usual (12.49) where we have incorporated the known behavior [Eq. So we let U(y) = eyl/2v{y) and obtain the following equation for v(y): I/2 (12.48) It is clear upon inspection that a twoterm recursion relation will obtain if a powerseries solution is plugged in. (12. the hydrogen atom. A= fiat (12.47) v"2yv' + 2A1 /  y = 0. .14)] near the origin.45) /\ T ' is dimensionless.6. The most celebrated member of this class. (0.H(D\X2 +Y2 + Z2) 2 (12.6. 351 ROTATION INVARIANCE AND ANGULAR MOMENTUM le Isotropic Oscillator The isotropic oscillator is described by the Hamiltonian H= P* + p2y + P 2n K .We close this section on rotationally invariant problems with a brief study of the isotropic oscillator.6.6.44) As rtco. We set v(y)=yl+l I cHf n—0 (12.2„2 Y.
2. 0 Of particular interest to us is the fact that states of different / are degenerate.49) derive the twoterm recursion relation. (12. 0 . 3 1. for the degeneracy in / can be attributed to additional invariance properties of H.6.2 . .6. (12.1.6. we will see in Chapter 15.52) (12. the allowed / values are /=n .2. . n . This is. Derive the quantizations condition. however.2 / = 1.6. Write them as linear combinations of the n = 0 and n = 1 eigenfunctions obtained in Cartesian coordinates.53) (12.6. .3. 0 m = ± l .0 n=2 n=3 m = 0.. For this reason it is occasionally termed accidental degeneracy. a misnomer. 1 or 0 Here are the first few eigenstates: «=0 n~ /=0 m= 0 m=± (12. The degeneracy of the different I states (which are not related by rotation operators or the generators) appears mysterious. ± 1 . ± 1 . ±2. ±3. Exercise 12. and how they explain the degeneracy in I.48) and (12. 1 /= 1 /= 0. where the problem was solved in Cartesian coordinates.6..6J If we define the principal quantum number (which controls the energy) n = 2k + / we get E=(n + 3/2)fio} At each n. The degeneracy in m at each I we understand in terms of rotational invariance. Eq. Argue that C0 7^0 if U is to have the right properties near y = 0.50)..6. . . ± 2 .2k = n. £ = 0. (12. Exactly what these extra invariances or symmetries of H are. (2) Calculate the degeneracy and parity at each n and compare with Exercise 10. (3) Construct the normalized eigenfunction \j/„im for « = 0 and 1.352 CHAPTER 12 By going through the usual steps (left as an exercise) we can arrive at following quantization condition: E=(2k + l+3/2) fico..11 * (1) By combining Eqs.
At a later stage. as a result of which the relative coordinate is essentially the electron's coordinate and the reduced mass is essentially m. (13. and a proton of charge +e and mass M.1. we will see how we can easily take into account the finiteness of the proton mass.1) UEl= 0 (13.13 The Hydrogen Atom 13. &. determines the energy levels in the rest frame of the atom. .1. (f>)= 1 ^ YT{6. the Schrodinger equation <?2 E +l(l+\)n2 2 mr (13. So we try to factor out the asymptotic behavior. </>) (13. when we compare the theory with experiment.1. <t>) = REl{r)YT(G. However. of an electron of charge — e and mass m. In this case we have just the electron moving in the field of the immobile proton.1.3) It is clear upon inspection of Eq. Since the potential energy of the electron in the Coulomb potential <j) = e/r due to the proton is V— —e2/r. as well as the wave j functions!  y. By using C M and relative coordinates and working in the C M frame.2) . let us first solve the problem in the limit M~+ao.1. since m / M ~ 1/2000. The Eigenvalue Problem We have here a twobody problem. we can reduce the problem to the dynamics of a single particle whose mass fi =mM/(m+M) is the reduced mass and whose coordinate r is the relative coordinate of the two particles.2) that a power series ansatz will lead to a threeterm recursion relation.Elm(r. t It should be clear f r o m the context whether m stands for the electron m a s s or the z c o m p o n e n t of angular m o m e n t u m .
Um where W= —E is the binding energy (which is the energy it would take to liberate the electron) ana that UEI ~ r—>0 ~ exp [—(2m Wjft2)1 /2 r] (13. Taking into account the behavior near p — 0 [Eq.5) Equation (13.1.1. You may verify that if we feed in a series into Eq. (12.354 CHAPTER 13 We already know from Section 12.1.4) suggests the introduction of the dimensionless variable p= (2mW/h2y/2r (13. a twoterm recursion relation will obtain.5)] we try Vv = pl+l Z Ckpk (13.8) (13.1.6) and the auxiliary function vE) defined by UE.1. = e~p vEl The equation for v is then d v _ 2 — + e2X „ dv dp where X=(2m/f?W) dp LP /(/+!) u=0 (13.6.19)]. (13.10) and obtain the following recursion relation between successive coefficients: Ck+} Ck The Energy Levels Since 1 ktcc e2X + 2(k + (k + l+2)(k l+\) + l+ 1) —/(/+ 1) (13.9) I and the subscripts on v are suppressed. (13.1.6 that u p to (possibly fractional) powers [Eq.1.1 J (13.1 y+i r' (13.1.1.12) .8).
. 2. 3p. Is denotes the state ( « = 1.19) .15). we demand that the series terminate at some k. called a Rydberg (Ry). and 2 states at n = 3. and so on.15) fr the allowed energies are _ 4 E »= ^Ti> 2n n 1.1. No attempt is made to keep track of m. l=nkl=«l.1. . let us note that the degeneracy at each n is Y ( 2 / + l ) = «2 /=o (13. 1 . 3.1.2. It is convenient to employ a natural unit of energy.. . . 1.. (13..1. / = 0 ) . p. : /=0.1.3.1. (13.1. 4 . In this spectroscopic notation.1.0 (13. 2s and 2p the / = 0 and 1=1 states at n = 2. .. d . states. This will happen if ep 355 THE e2X = 2(k + l+\) (13.2.«2..h. g. and 3d the / = 0 .1.. .18) It is common to refer to the states with / = 0. (13. for measuring the energy levels of hydrogen: me 4 Ry = — r y 2ft2 (13. according to Eq.14) In terms of the principal quantum number n = k + l+1 (13. f .1. We discuss these later. as s. 1..p 2p is the behavior of the series pm e2p.17) That states of different I should be degenerate indicates that H contains more symmetries besides rotational invariance. For the present. 2 ...16) and at each n the allowed values of / are. (13. and would lead to U~e Pv~pme e ~pm as p>oo.9)] _ 4 In(k + l+1) k = 0..13) HYDROGEN ATOM or [from Eq. . 3s.1.
1.23) .1. This polynomial is called the associated Laguerre polynomial. r _ 2m ( me ft2 \2ft2n2 1/2 (13.1.1. T h e first few eigenstates of hydrogen. n= I troscopic n o t a t i o n .22) Recall that __ /2m W\l/2 ' " n r me Yn } Lk(x) = (\)k(dk/dxk)L%k.Figure 13.20) Figure 13.1. it is a straightforward matter to determine the wave functions and to normalize them. (13. in terms of which = n (13. The corresponding radial function is (13. L^e'idr/dxnie'**"). Consider a given n and /.21) the corresponding function vt is p!+1 times a polynomial of degree nl1. The e is measured in R y d b e r g s and the states are labelled in the I .10) terminates at k = nl1 (13. Since the series in Eq.1. The Wave Functions Given the recursion relations.1 shows some of the lowestenergy states of hydrogen.
\ Ir 2 . Exercise 13.11) and (13.26). The following are the first few normalized eigenfunctions. Exercise 13. i 1 \1/2.— I e~r/2a° a0. VE~(r)meVKnl e~Kr in a Coulomb potential V= e2/r [k = (2m W/h2)]/2].8)(13.' ^ 1 .1.1. Eq.0=! \32nalJ i a0 V/2 \647ra 3 0/ .1. L will be dominated by the highest power.1. which provides the natural distance scale for the hydrogen atom.1.26) (If l=n~\. = Wnim: 1/2 ^1.1.f—) \naQJ \na0J As r*oo.6. (12. and {r)n~l e~r/na° (13.1. r " .4).24) called the Bohr radius.1.27) e~r/2aoco&0 ^2. Exercise 13.1.18). in a Coulomb potential (see Exercise 13.0. this form is valid at all r since Lo /+1 is a constant.1.0.0 = 1 r/a0 .1. (13. Derive the degeneracy formula.25) Rni ~ (independent o f / ) (13.e~r/2a° sin e e±i4> Clo Exercise 13.19)] that as r*oo. Show that this agrees withEq.1. Recall from the last chapter [Eq.4.In terms of the length ft a0 = — 5 me 2 357 THE HYDROGEN ATOM (13.W — ) I^i.e . i U O .1. (13.2.1.1. (13.26) was anticipated in the last chapter when we considered the behavior of UEi as r*oo.3.10). (13.nal.1.14) starting from Eqs. Derive Eqs.) Equation (13.1. obtain \y2w (normalized). . (13. Starting from the recursion relation.
358 CHAPTER 13 Let us explore the statement that a0 provides a natural length scale for thS hydrogen atom. the radial function has n — l— 1 zeros and the density in r has several bumps. Thus the Bohr radius gives the most probable value of r in the ground state and this defines the "size" of the atom (to the extent one may speak of it in quantum theory). at least in the state o f / = « — l . we shall use the same symbol t o refer t o it in the abstract.oce~2r/na° r2" dr n The probability density in r reaches a maximum when — (e~~2r/na°r2n) = 0 dr or r = n2a0 (13.32) (13. (13.5) that <r> = . which states that if V= crk. Consider the state described by 171(0. we may define the size by <r>. this equals a0.28) Let us ask for the probability of finding the electron in a spherical shell of radius r and thickness dr: HL t P(r)r 2 dr dQ.1.1.1. In any eigenstate <#> = E= < r > + < V} = i^/lmy It can be shown (Exercise 13.1. . that jg (r~}„im = y [3w2 — / ( / + 1)] (13./ .<e 2 /r> (13.30) When n— 1. (j>) (13. then the averages f and U are related by f = k 2 J E v e n t h o u g h r represents the abstract operator (X2+ Y2 + Z2)]/2 v tt o n l y in the coordinate basis.33) which is just the quantum version of the classical virial theorem.1.. In this case. I f / ^ n — 1 . which indicates that the size grows as n2a{). by using properties of L 2 ' .i < K ) . so as t o keep the n o t a t i o n simple.1.1.J It can be shown. If n> 1 we see that the size grows as n2.31) Rather than go through the lengthy derivation of this formula let us consider the following argument.
2. Consider fi = R . £ = = " 2 fi2n2 2aon2 from which it follows that —me4 —e 2 359 = ke2/r> THE (13.35) 1 n Although (13. Several questions need to be answered.1.1.5. ( f l ) = 0 for any ft. however. In the present case.1. * (Virial Theorem). because. it is not true in an s state that <1 />4> ^ 1 / (r4>." Quotation marks are used once again. m} is a stationary state.36) — # <0 the two are of the same order of magnitude (see Exercise 9. Since n. the ^degeneracy can be explained in terms of other symmetries the Hamiltonian has ibesides rotational invariance. exhibits "accidental degeneracy.) This completes our analysis of the hydrogen spectrum and wave functions. since <1 /r 4 > is divergent while l/<r 4 > is not.1. Now.34) HYDROGEN ATOM (13. (One must be somewhat cautious with statements like <1 /r> ~ 1 /<>>. we have seen that the symmetries of H imply . But first let us address a question raised earlier: what is the source of the degeneracy in I at each nl Exercise 13.It follows that E=]^V) Now. The Degeneracy of the Hydrogen Spectrum The hydrogen atom. <1/ r) is well defined in all states and indeed <1 /r> and 1 /<>> are of the same order of magnitude. such as (1) What are the numerical values of En. in the state labeled by n.37) which agrees with the result Eq.1. 1. a0. For example.31). I.P and use Ehrenfest's theorem to show that <7> = (—1/2)<K> in the state \n.3. m>.1.4. 13. etc. as in the case of the oscillator.2) and we infer that (r)„~n2a 0 (13.? (2) How does one compare the energy levels and wave functions deduced here with experiment? These questions will be taken up in Section 13. (13. like the oscillator.
This is how the degeneracy in m is "explained" by rotational invariance.1. Precisely what this operator is and how it manages to raise / by one unit will be explained in Section 15. So it must be that since [N.e. This would then explain the degeneracy in I at each n. we must be able to build some operator out of the components of N. gravitational. if there is an extra symmetry (besides rotational invariance) there must be some extra conserved quantities (besides angular momentum). we have used the symmetrization rule to construct N from n. i. H ] = 0.. R 2 1/2 (X +Y +Z ) 2 v (13.L X p ] (verify this). which means that we can raise and lower the m values at a given / without changing the energy. it cannot escape to infinity.4. § Since [P. So.2. which commutes with H and which raises I by one unit. k Exercise 13.3) We have seen that the conservation of L implies that [L ± . L] ^ 0 .2) N = — [PXLLXP] 2m — 2g . Now it is well known classically that the CoulombJ potential is special (among rotationally invariant potentials) in that it conserves the RungeLenz vector pxl m e r n= r (13.> 3 [ ( P x L) + ( P X L ) f ] = j [ P X L . there will be an operator N which commutes with H: [N. (1) Express n in terms of r and p alone (get rid of I).360 C H A P T E R 13 the conservation of the generators of the symmetries. Let us see why the conservation of the RungeLenz vector ti implies closed orbits. as we follow it from some arbitrary time onward.1)? In quantum theory then.2.2. Consequently. devoted to the study of "accidental" degeneracy.2. The conservation of n implies that not only is the orbit confined to a plane perpendicular to 1 (as in any rotationally invariant problem) it is also closed (Exercise 13. . say.#]=0 and is given by§ (13. (2) Since the particle is bound. p x I . Show that e2 n = 'rmax \2E + max I V rn + Or generally any 1 jr potential. H] = 0. it must reach a point rmax where its distance from the origin stops growing.2. You will also find therein the explanation of the degeneracy of the oscillator.
361 THE HYDROGEN ® '"min 2 2 E+ e' ATOM f min Thus r max and r min are parallel to each other and to n. 1 and rmm are parallel. all three vectors r m a x . (Why?) Convince yourself that for a circular orbit.5 Mev MC2= (0.5) . (2) ask how the predictions of the theory are actually compared with experiment. Consider now an estimate of the Bohr radius a0 = fi2/me2 J A more exact value. and n are aligned with the major axis of the ellipse along which the particle moves.2) (13.e. r m i n .1) (13.3.3.3.3. We will express the rest energies of the particles in millionelectron volts or MeV : mc 2 ^ 0.. 13.511 is a more exact value) (938.3. The conservation or constancy of n implies that the maximum (minimum) separation is always reached at the same point r mai (r min ).3. (13.4) m + M M as are the relative coordinate and the electron coordinate. In fact. Numerical Estimates and Comparison with Experiment In this section we (1) obtain numerical estimates for various quantities such as the Bohr radius. etc. n must and does vanish. i.3)1 (1/1836)$ (13. the orbit is closed..3) 1000 MeV m / M ^ 1/2000 Consequently the reduced mass fi and electron mass m are almost equal: mM mM M= (13. energy levels. while n 1 and rmax are antiparallel. Numerical Estimates Consider first the particle masses.
The best way to remember e2 is through the fine a = e 2 1 / 1 fie 137 \ 137. the number of spacetime dimensions. 13.6 n E„ = eV t M a n y o f the tricks used here were learned f r o m Professor A . 8 (1973. a will be 1/137. This constant plays a fundamental role in quantum mechanical problems involving electrodynamics. . R o s e n f e l d at the University of California Berkeley.R y / « 2 We estimate me Ifi 2 Ry = mc 2 fic. e V ~ 13. e.25 x 1Q6 (137) 2 So. ene.5 x 10° (0. returning to our main problem.3 eV (13.3) (13.054 x 10~27 erg sec A more useful thing to remember for performing quick estimates 1st ftcszlOOO eV A where 1 angstrom ( A ) = 10 structure constant'.To find this we need the values of fi and e. mass. its numerical value has an absolute significance: no matter what units we use for length. Since it is a Godgiven number (independent of mortal choice of units) one tries to relate it to fundamental numbers such as 7T. etc. and time.3.6) 0.04. we can now estimate a0: * me 2 a0 (2000X137) mc 2\ A A \e 2 0.6) cm. It was mentioned earlier that ft = 1. Thus.53) Consider next the energy levels £„ = . Since it is dimensionless. using the more accurate value of Ry. Anyway. several attempts have been made to arrive at the magic figure of 1/137. although no one tries to explain why c = 3 x 1010 cm/sec.
it was pointed out that the Compton wavelength of the pion was the distance over which it could be exchanged.The electron in the ground state needs 13. we will see pair creation and we will have three (or more) particles instead of the one we started to locate. we use more and more energetic probes. as we try to locate the particle better and better. It can also be defined as the lower limit on how well a particle can be localized. the lower limit is zero. The preceding analysis shows that this is not strictly correct. we treated the electron as a localized point particle. we need a photon of momentum ? A P~± AX Since the photon is massless. since we admit position eigenkets x>. In the nonrelativistic theory we are considering. So we demand AE<lmc2 he AX' or a 2mc n mc <2mc2 AX> If we attempt to localize the particle any better. say photons to be specific. relativity allows the production of a particleantiparticle pair in the measurement process. But in reality. but it . The first h2 e2 % = —.6 eV down the infinitely deep Coulomb potential.— =—=Xe me he mc 363 THE HYDROGEN ATOM (13. To locate it to some AX.6 eV to be liberated or ionized.8) is called the Compton wavelength of the electron and is 137 times smaller than the Bohr radius. the corresponding energy is AJ? AE~ — AX n c in view of Einstein's formula E2 = c2p2 + m2c4. Let us digress to consider two length scales related to a0. One may imagine that it is 13. If this energy exceeds twice the rest energy of the particle.3. What does Xe represent? In discussing the nuclear force. In our analysis of the hydrogen atom.
Consider a particle of mass m in V(r) = ~e2/r. Let us now return to the hydrogen atom.11) . — x 137 250 103 A If we multiply Xe by a we get another length. The source of these mnemonics is the Bohr model of the hydrogen atom. called the classical radius of the electron . since the "fuzziness" or "size" of the electron is a times smaller than the size of the atom.5 x —J— A ^ . Let us now consider mnemonics that help us remember the dynamics. for they are phrased in terms not allowed in q u a n t u m theory. a would have been of order unity. The mnemonics discussed so far are concerned only with the numbers. the size of the electron and the size of its orbit would have been of the same o r a ^ and the point particle approximation would have been untenable. ft 10"5A re = ale = —= mc he mc (13. The dynamical equation is mv 2 e 2 (13. Let us note that A e = a •fl0—0.3. the Coulomb energy of the distribution (the energy it takes to assemble it) will be of the order e2/r„ where re is the radius of the sphere. we arrive at the classical radius.9) If we imagine the electron to be a spherical charge distribution. These must be used with caution. If we attribute the rest energy of the electron to this Coulomb energy. About a decade or so prior to the formulation of q u a n t u m mechanics as described in this text. moving in a circular orbit of radius r.10) or mv — — r (13. Bohr proposed a model of the atom along the following lines. a0 ft/mc do =a ~ 1 137 H a d the electric charge been 10 times as big.3.364 rwAPTFR n also shows that it is a fair approximation.3. In summary.
3.mv2 = .3. (13. a mnemonic for the mnemonic) is the equation a=P (13.mc2(v/c)2 2 2 mc2B2 = — .14) I Thus.mc2a2 2 2 . It also follows that any energy is allowed since 1 E=mv2 2 e 2 365 THE HYDROGEN e 2 r 2r = — mv 2 \ ATOM 2 (13. you can go back to this model for the formulas (though not for the physics.e.. if you ever forget the formula for a0 or E„. Given this.11) we get n2h2 m r 2 2 (13.3. since it is perched on the fence between classical and quantum mechanics.12) Bohr conjectured that the only allowed orbits were those that had integral angular momentum in units of fi: mvr = nfi Feeding this into Eq. but quantizes angular momentum and so on). it speaks of orbits..3. we get the ground state energy as Ei = .3.. The most succinct way to remember the Bohr atom (i.16) I where /? is the velocity of the electron in the ground state of hydrogen measured in units of velocity of light ( f i = v/c).13) m• e2 r or r = n2 — = n2a0 me and (13.Thus any radius is allowed if r satisfies this equation.
3.3. In principle. (Recall k = 2aH/p. . (13. which states that the binding energy is ~ ( 1 / 1 3 7 ) 2 times the rest energy of the electron. If we measure the energy at time t>T. change the Hamiltonian f r o m the Coulomb Hamiltonian.e E n If we rewrite E\ as —e2/2a0. Exercise 13. However.1. by turning on some external field (i. if we perturb it for a time T. An equivalent way (which avoids the use of velocity) is Eq. i. Let us consider first the energy levels and then the wave functions. since  nlrn) is not a stationary state of H° + Hl.3. 3 . In practice. .2. CeTus ask how these are to be compared with experiment.. to H°+H]) its state vector can start moving around in Hilbert space. . One measures the energy by detecting the photon emitted by the atom.) Comparison with Experiment Q u a n t u m theory makes very detailed predictions for the hydrogen atom.* The pion has a range of 1 Fermi = 10 force. we obtain a family of lines as we vary n. at least for several values of n'. 5 A as a mediator of nuclear Exercise 13. it will stay that way forever. The frequency of the detected p h o t o n will be comi " En En' Thus the frequency of light coming out of hydrogen will be Ry/ 1 1 ft \ri2 n J (13.366 CHAPTER 10 Given this. 2. one measures the differences in energy levels as follows. we can get the formula for a0.17). we may find it corresponds to another state with n ' ^ n . . * Estimate the de Broglie wavelength of an electron of kinetic energy 200eV. The ri = 1 family is . If we start with the atom in an eigenstate  n l m ) .19) F o r a fixed value ri = 1. how could one forget that the levels go as n 2 . These families have in fact been seen.e. one can measure the energy levels by simply weighing the atom. Estimate its rest energy.3. The equation «=j3 also justifies the use of nonrelativistic q u a n t u m mechanics.
(13. etc. or since v/c~a. Then there is the correction due to the fact that the kinetic energy of the electron is not jmv2=p2/2m in Einstein's theory. However. say the one corresponding to the transition n = 2*n' = 1: 13. in particular Eq. a correction of order 2 a relative to main piece. in the variable v j c2. . Much of these may be explained by corrections that are calculable in theory. there are tiny corrections due to quantum fluctuations of . This difference is.2) only in that m would be replaced by p.16) for the energy levels. As explained in Chapter 10. etc. which is the difference between the energy at velocity v and the energy at rest. which is 3mu 4 /8c 2 . . They will be included (in some approximation) in Chapter 17. This is a correction of order v2/c2 relative to the p2/2m piece we included. small discrepancies. . 4 . (13. l\ 10 v ^ — eV fi The_ wavelength is estimated to be co 10 ^ i A more refined estimate gives a value of 1216 A. First we must correct for the fact that the proton is not really immobile. but instead mc2[(\ ~v2/c2)~l/21].1. or in terms of the momentum. The n' = 3 family called the Paschen series.5 e V / . The Dirac equation. Let us estimate the wavelength of a typical line in the Lyman series. however. in very good agreement with experiment.1.called the Lyman series (it corresponds to transitions to the ground state f r o m the upper ones): 367 THE HYDROGEN ATOM • »T(ri) (133 20)  The n'=2 family is called the Balmer series and corresponds to transitions to the states 2/m> from n = 3 . there are. There are other corrections of the same order. . .3 / J 4 / 8 W V . which we will not solve in this book. in all cases. The jmv2 term is just the first in the power series expansion of the above. observable in practice: one sees it in the difference between the levels of hydrogen and deuterium (whose nucleus has a proton and a neutron). Equally good is the agreement for all other observed lines. In Chapter 17 we will take into account the effect of the next term. This in fact would be the only change in all the formulas that follow. it too doesn't give the full story. this is done by writing Schrodinger's equation for the relative (and not electron) coordinate and working in the C M frame. which differs from 1 by less than a tenth of a percent. However. This equation would differ from Eq. and these go by the name of finestructure corrections. This would simply rescale the entire spectrum by a factor p / m = M / ( M + m ) . takes into account the relativistic corrections to all orders in v/c. that we have here a twobody problem.
We can also estimate the surface temperature as follows.) Consider next the wave functions. Suppose we apply an external perturbation H] for a short time E. The evaluation of the integrals entails detailed knowledge of the wave functions. During this time. important to bear in mind thai all these corrections are icing on the cake. respectively. Exercise J3. The following example should give you a feeling for how this is done. the system goes f r o m n/ra> to j .V(e)> = fi  nlm} The probability of it being in a state \n'l'm'y nlm}) is (assuming  n ' l ' m ' ) is different from m \{n'l'm'\w(m2 = (n'l'm'\H'\nlmy ft Thus q u a n t u m theory can also determine for us the rate of transition to the state  « 7 ' m ' ) . The relative strength of the Balmer and I .3. The agreement between theory and experiment is spectacular. These corrections ca cu a l l b l e in theory and measurable experimentally. It is. agreement of the calculated rates with experiment is a check on the predicted wave functions. Let Tbe the surface temperature. The probabilities P{n = l) and P(n = 2) of an atom being at n = l and M = 2. The presence of these lines will tell us that there is hydrogen at the surface of the sun. are related by Boltzmann's formula P(n= 1) where the factor 4 is due to the degeneracy of the n = 2 level. We shall see a concrete example of this when we discuss the interaction of radiation with matter in Chapter 18. T o test the predictions. which i§' ] coordinate space.3. white light tries to come out and the atoms at the surface absorb what they can. will be some integral over and iy„im with H sandwiched between them. From the interior of the sun. Instead of looking at the emission spectrum. The atoms in the ground state will now absorb the Lyman series and this will lead to dark lines if we analyze the light coming from the sun. however. This rate is controlled by the matrix element <ji'l'm'\Ft\nlrn). (Much of the present speculation on what the correct theory of elementary particles is will be put to rest if one can come u p with a description of these particles that is half as good as the description of the hydrogen a t o m by Schrodinger's equation. we can also look at the absorption spectrum of hydrogen. and conversely. Say some hydrogen atoms are sitting at the surface of the sun.368 CHAPTER 13 are the electromagnetic field (which we have treated classically so far). one once again relies on perturbing the system. Now only atoms in « = 2 can produce the Balmer lines in the absorption spectrum. that the simple nonrelativistic Schrodinger equation by itself provides an excellent description of the hydrogen spectrum.
The fact that. one averages over angles to get a spherically symmetric <p. however.2He has its n = 1 shell filled. One then computes the potential <j)\(r) due to this electronic configuration^ If it coincides with </>0(/) (to some desired accuracy) one stops here and takes the configuration one got to be the ground state of the atom. Although it is possible. which is in the Is state. from which we may infer T. What do the eigenstates look like? They are still labeled by (nlm) as in hydrogen. As a result. Consider the one due to Hartree. (Recall t In this discussion electron spin is viewed as a spectator variable whose only role is to double the states. Multielectron Atoms and the Periodic Table It is not possible to treat multielectron atoms analytically even if we treat the nucleus as immobile. while states of high angular momentum.. Formally this is because the potential is no longer 1 jr and physically this is because states with lower angular momentum have a larger amplitude to be near the origin and hence sample more of the nuclear charge. The " r a d i u s " of each state grows with n. Show that for T= 6000 K. P{n = 2)/P(n = 1) is negligible and that it becomes significant only for Tm 105 K. a configuration denoted by Is2. a more practical method is to follow some approximation scheme.) 369 IIV niUKOubN ATOM 13. to treat an arbitrarily complex atom by solving the exact Schrodinger equation numerically. „ . this time starting with (p\(r).„ . see the nuclear charge shielded by the electrons in the inner orbits.4. with a slight dependence on /. R=300K. which are suppressed at the origin. States of a given I and n are said to form a subshell. and computes the allowed energy eigenstates. with opposite spins (the Pauli principle will not allow any m o r e ) ! until all the electrons have been used up. This configuration is denoted by l j 1 . If not. in a semiclassical sense. Here one assumes that each electron obeys a oneparticle Schrodinger equation wherein the potential energy V= (r) is due to the nucleus and the other electrons. it is in the configuration \s22s]. at each n the energy goes u p with /.] The degeneracy in I is. in principle.. . Lithium fLi) has its third electron in the 2s state. § If necessary.. call it (r). This is a fairly good approximation. One then fills them u p in the order of increasing energy.e. (The Boltzmann constant is k n 9 * ]0 _ s eV/K. i.e ) times the probability density associated with its wave function. And what are the wave functions? They are the eigenstates in the potential 0(r)! To break the vicious circle. each electron is assigned a charge distribution which is ( . [This is because (f> (r) is rotationally invariant. Let us now consider the electronic configurations of some low Z ( Z is the nuclear charge) atoms. with states of different m degenerate at a given n and /. putting in just two electrons in each orbital state. one goes through one more round. they may be viewed as moving on a sphere of radius equal to the most probable value of r). signals the soundness of this scheme. one begins with a reasonable guess for the potential. lost. in practice. Hydrogen (*H) has just one electron.Lyman lines will tell us P(n = 2)/P(n = 1). A mnemonic is eV at room temperature. In computing the electronic contribution to (p(r). . one soon finds a potential that reproduces itself. Helium ( 2 He) has two electrons in the Is state with opposite spins. States of a given n are thus said to form a shell (for.
This is because the growth in energy < H to a change in n from 3 to 4 is less than the growth due to change in / from 2 { f l at n = 3. Na passes on its electron to F and the system as a whole lowers its energy. in t!1P "rare earth" elements. with all similar elements in the same column. which want to get rid of two electrons. we can guess who will interact with whom. The 3s and 31 subshells are filled when we get to argon ( 18 Ar). If "Ma could get rid of this electron.1 eV compared to an n = 2 electron in Ne. Consider an element such as 10Ne. This quasiperiodic behavior was emphasized in 1869 by Mendeleev. who organized the elements into a periodic table. sodium ( 1 ] Na).. has a solitary electron in the 3s state. we see the likes of oxygen. oxygen ( s O).. Neon is in the configuration 1 s22s22p6.. i.e. called the ionic bond and form the N a F molecule. Na. since the binding energy in F is 17. while two places to the right we have elements like magnesium. this would change. eager to give up an electron. i. which occupies the 3s state. The next one. For example. consider an element like "Na. t  neon ( . Its binding energy is 5.e. Ne. we expect all elements with filled outer shells to be chemically inert. Given the maxim that happiness is a filled outer shell. for example. i. This is true. On the other hand. eager to accept an electron. For . elements with one electron to spare In the last column we see the inert elements. which has one more electron. the atoms cannot part company. 0 Ne). but there is a large gap in energy to cross. If we look one place to the left (in Z) of 10 Ne. such as 18 Ar. which are held together by electrostatic attraction. He. boron ( 5 B). has its n = 2 shellfilled. while if we move one place to the left. which want two electrons. Given the electronic configurations. in which just the 3s and 3p subshells are filled. For example. potassium (l9K) INK its 19th electron in the 4s and not 3d state. N a + and F~. we meet those that behave like Na. the 6s shell is filled before the 4 / shell. carbon ( 6 C). The origin of this inertness is the same as in the case with filled shells: a spherically symmetric electronic charge distribution and a large excitation energy. we see H. we meet the likes of F. which has a binding energy of 21.Tit next element. we see a perfect acceptor for this electron: we have here 9 F. and fluorine (9F). we can see that several elements will have similar chemical properties because they have similar outer shells. which has a closed outer shell. Since the total electronic charge is spherically symmetric (lR«/2 1 YT\2 is independent of 6 and (j>). So when " N a and 9 F get together. for they have now become charged ions. one can anticipate many of the chemics properties of the elements. in which the elements are arranged into a matrix. Having carried out the transfer.y«6shells are also inert. etc. If one of the electrons in the outer shell could be excited to a higher level. This electron sees a charge of +e when it looks inward (the nuclear charge of 11 shielded by the 10 electrons in the n = 1 and 2 shells) and is thus very loosely bound. Thus the atom is rarely excited and is chemically inert. etc. Once we grasp that the chemical behavior is dictated by what is happening in the outermost shell. nitrogen ( 7 N). whose n = 2 shell is all full except for one electron..that the s state is lower than the p state. it shields the nuclear charge very effectively and the atom has no significant electrostatic affinity for electrons in other atoms. it could reach a stable configuration with a closed n= 2 shell. Li.) We keep going this way through beryllium ( 4 Be). It is also true that some elements with filled . we see a certain chemical tendency over and over again.4 eV. It follows that as we move in Z. If we move two places to the left. This phenomenon occurs often as we move up in Z.e.6 eV. As we go down the first column. If we move one place to the right of the inert elements.
such as the formation of H 2 0 when two H atoms get together with an oxygen atom. they all behave alike. Since what happens in the interior does not affect the chemical properties. Ignore interelectron effects. deep in the interior (but of a higher energy). Consider the rare earth elements. Wiley. let us stop here. which have very similar chemical properties. 90 Th to l03 Lw. for a nice trip through the periodic table. not only can N a give its electron to F. which have a filled Is subshell and a 5/ subshell that is getting filled up. and that the nonrelativistic description holds. Assume levels fill in the order of increasing n.4.t Exercise 13. Basic Physics of Atoms and Molecules. There are many more properties of elements that follow from the configuration of the outer electrons.3. This way all three atoms get to fill their outer shells at least part of the time. * Visible light has a wavelength of approximately 5000 A . the energy levels of a multielectron atom go as Z2. Since the Coulomb potential is Zejr. why is the energy ocZ2? Exercise 13. Exercise 13.4. Balmer. it can give to CI. Why doesn't the chemical behavior change with Z in this range? The answer is that in these elements the 6s subshell is filled and the 4/subshell. Fano. Which of the series—Lyman. is being filled. If you want to know more.2* Compare (roughly) the sizes of the uranium atom and the hydrogen atom.* Show that if we ignore interelectron interactions. . Likewise F can get its electron from K as well. New York (1959). which has a lone electron in the 4s state. you must consult books devoted to the subject. Paschen—do you think was discovered first? 371 THE HYDROGEN ATOM J S J See. Since we must stop somewhere.4. which is one shy of a filled 3p subshell. Fano and L. More involved things can happen. U. Chapter 18.instance. The same goes for the actinides.1. in which each hydrogen atom shares an electron with the oxygen atom. 58 Ce through 71 Lu. forming the covalent bond.
1. The actual experiment. protons. Fortunately the problem can be solved by a shrewd mixture of classical intuition and reasoning by analogy. photons—have the spin degree of freedom.2. neutrons. It is. The reason is that these phenomena involve a quantum degree of freedom called spin. What is the Nature of Spin? The best way to characterize spin is as a form of angular momentum.14 Spin 14. Consequently. say the z direction. In this chapter we study just electron spin. that the result is .e. is measured. L y . however. however. which has no classical counterpart. for most particles—electrons. shows that this is wrong. how does it enter the Hamiltonian (dynamics of spin)? 14.. The treatment of the spins of other particles is quite similar. As the operators L x . not the angular momentum associated with the operator L. Introduction In this chapter we consider a class of quantum phenomena that cannot be handled by a straightforward application of the four postulates. with the exception of the photon. our existing formalism predicts that if the angular momentum along. The problem is very important. i. Photon spin will be discussed in Chapter 18. nor can we immediately write down the quantum Hamiltonian that governs its time evolution. neither can we obtain the spin operator by turning to Postulate II. as the following experiment shows. which moves at speed c and can't be treated nonrelativistically. In the next three sections we address the following questions: (1) What is the nature of this new spin degree of freedom? (2) How is the Hilbert space modified to take this new degree of freedom into account? What do the spin operators look like in this space (kinematics of spin)? (3) How does spin evolve with time. a result of zero will obtain. and L z will give zero when acting on it.e. in a state with a constant (spaceindependent) wave function. i.. An electron is prepared in a state of zero linear momentum.
the wave function is transformed as follows: * J / "I \ i_ is iti dfd((> 0 . and (2) the components of the wave function get transformed into linear combinations of each other.2) We identify the generator of infinitesimal rotations about the 2 axis. Also spin was first discovered on the basis of spectroscopic evidence and not from an experiment of the above type. This angular momentum is called spin. for it was imagined in the early days that if the electron has angular momentum without moving through space.1) where Sz is an n x n matrix.3.3. then it must be spinning like a top. in particular. By generalizing our findings from Exercise 12. Kinematics of Spin The discussion following the general solution to the eigenvalue problem of angular momentum (Section 12. allows us to understand the above experiment. this equation reads§ n I . for a consistent mechanical model doesn't exist. We adopt this nomenclature. 14. as the 2 component of angular momentum.. the generator of infinitesimal rotation is not just L but something more.1 to an n component wave function in three dimensions.5) suggests the way for treating particles with intrinsic angular momentum or spin. The differential operator L does part (1). not associated with its orbital motion. while an n x n matrix S is responsible for part (2). More on this later.\ It follows that the electron has "intrinsic" angular momentum. but not the mechanical model that goes with it. Fortunately one can describe spin and its dynamics without appealing to any model. we can say that under an infinitesimal rotation around the z axis.374 CHAPTER 10 ±fi/2.5. . \ >i~ is ~ 1 — sz fi ifi d/d4>_ i •  0 (14. starting with just the observed fact that it is a form of angular momentum. We see it has two parts : JZ = LZ + Ss $ In practice one measures not the angular momentum. but a related quantity called magnetic moment. Let us now develop the formalism that deals with spin and. Recall that if a particle is described by a wave function with many (n) components. In abstract form.3.J fi \¥> lv> (14. § The spin operators will be denoted by the same symbol (5") whether they are referred to in the abstract or as matrices in some basis. The reason is that under an infinitesimal rotation two things happen to the wave function: (1) the values at each spatial point are reassigned to the rotated point.
z. 3 / 2 . . y. the way to describe the electron is through a twocomponent wave function called a spinor: ¥+(x.24): 0 1 1 0 Sv =  0 i i 0 sr= 1 o o l (14. y/+ = 0. . n) they commute. 1. have an eigenstate of Sz with eigenvalue %jl\ if y/~ ^ 0 . However. (12. the Sz eigenvalue is ( ..5. . are generators of rotations. z) _ yr{x. and the commutation relations were satisfied block by block.6) Now recall that in Chapter 12 we found matrices Jx..4) Since L and S act on different parts of the wave function (the former on x. . the infinitedimensional matrices were built out of (2/+ 1) x (2/+ 1) blocks. the latter on the indices i= 1 .3.8a) v "o" _1 (14. = ifi £ SijkSk k (14.3.24)] that obey precisely these commutation relations.8b) we If y/.3. Lj] + [S.3. So which block shall we pick for the electron spin operators? The answer is given by the empirical fact that S? has only the eigenvalues ±ti/2.3. Jj] = ifi £ eijkJk k (14.7) Thus. We proceed as follows. . (14.22)(12. But these matrices were infinite dimensional. they must obey the consistency condition [J.3.3.. z) _ ¥ = (14. Sj] = ift X £ijkLk +1 £ijkSk k k Using the known commutation rules of the L h we deduce [5.5. .and more generally J =L+ S (14.5. . This singles out the 2 x 2 blocks in Eqs. 1/2. Jy. with y = 0 . and we may infer from Eq.3) Our problem is to find the number («) of components appropriate to the electron and the three spin matrices that rotate its components. Since J.5.5) (14./ z / 2 ) . (12. y.4) that [L. . y.3. and Jz [Eqs.22)(12. .= 0..
which puts it between a scalar. y.3. lead to spin. the components of the spinor are complex.10) that y/+ and y are independent of x.e. [For any particle.376 CHAPTER 10 Let us now proceed to interpret the experiment mentioned earlier. Thus the spin of the electron is always 1 / 2 (in units of fi) and serves as an invariant label of the particle. When we come to the Dirac equation.] We have deduced that the electron is to be described by a twocomponent wave function in the coordinate basis. Since we prepared a state of zero m o m e n t u m . It follows that Lz acting on y/ gives zero. However.11) and yields a value ifi2 on any state y/.vY) etc. it is easy to see that the introduction of spin has doubled the size of Hilbert space.3. and z.J Let us restate this result in Hilbert space. i. The basis vectors  xyzsz) diagonalize the mutually commuting operators X. if we demand that the equation befirstorder in time and space.9) We deduce from P  y/> = 0. like its charge or rest mass.e.m 0 •im 0 (14. which has three. However. and a vector. The state vector t We made the deduction given the empirical input from experiment. if it was oo dimensional before. The electron spinor is a rwocomponent object. i. 7. and Sz (one can also think of other bases such as pj r > or )p. if you know what I mean.). . A significant difference between spin and orbital angular m o m e n t u m is this: we can change the magnitude of orbital angular m o m e n t u m of a particle (by applying external fields) but not the magnitude of its spin. Sz doesn't: there is an amplitude y/± for obtaining ± f i / 2 .. now it is 2oo dimensional. —ifiVy/+ "o" _0_ (14. we will see that incorporating relativistic kinematics will automatically lead to a multicomponent wave function. First.. we want the operator P to give zero when acting on y/. " l o" _0 1_ (14. which has one component. The S2 operator is s2=n2 0 0 (5)(5 + l). Z. The operator P simply differentiates both components of y/: I .3. the magnitude of spin is decided by the number of components in the wave function and is an invariant.
z — — co. we denote it by a function iy(x).1 2 _0 ¥+(x. sz— = +*/2) 377 SPIN +fi/2) V(x. z — co.basis (xyzsz  if// — y/(x=—co. z. 2 from — 00 to 00. sz = h/2) W{x = 00.e. we may compactify Eq. y.y.12) Y(x. sz= +fi/2) lv> R.f i / 2 . y = —co. Sz is a 2 x 2 matrix: SzlV> • R.3..y.sz~ —fi/2) (14. y. R. . . .3. The normalization condition is l = <Vl ¥> = 1 < \j/1 xyzsz > (xyzsz \ dx dy dz (  v +  2 + l y/\2)dxdy dz (14. since it depends on discrete (.y.3.sz !//(X= GO.v) as well as continuous (x. Likewise the components below the dotted line define a function y/(x. we write it as a column vector whose components are functions. The horizontal dashed line separates the components with sz = fi/2 from those with sz = . define a continuous function \j/+(x. . y — co.S. 2). z) indices.J3) This notation blends two notations we have used so far: if the vector has components labeled by a discrete index /'(/'= 1 .12) to the form » V+(x. the component of  i/O1 will vary smoothly.y — 00. y = 00. Sz=—flj2) Clearly y/{r. but here. (14.14) In the compact notation.3. y.15a) . y.SZ basis z) z) y/(x. z— 00. z) _w~{x. ±fij2) gives the amplitude to find the electron at r with sz=±fi/2.is a 2oodimensional column vector in this basis: y/(x = co . z).3. 2 = 00. y. while if it is labeled by a continuous index such as x.y. Now if sz is fixed at fi/2 and we vary x. 2) _ _ (14.iSz basis fi "l 0" . (14. ») we denote it as a column vector. i. y.z. In terms of these functions.
z.. rather. Likewise.1 —fi/2) . 5Z> (14.. y.«/2) . it is a 2oodimensional matrix: " 1 l 0 V(oo. let us assume the orbital degree of freedom exists but evolves independently. sz)> of V e is just a direct product x.y basis fi 2 1 (14. which describes a particle with just orbital degrees of freedom.y= 1/2.SZ basis 0 y/+(x. Observe that the Hilbert space V e of the electron may be viewed as a direct product of an infinitedimensional space V 0 .3. to be more precise. which describes a particle with just spin degrees of freedom: (14.3.15b) . Since we already k n o w how to handle the orbital degrees of freedom. P)? Equation (14. z) 0 iftd/d<j> _y/(x.3. 2.17) The basis vector y.1 0 . Formally this means that the Hamiltonian is separable: H—Hq+ Hs (14. W h a t a b o u t the familiar operators Q(R.3. and a twodimensional space Vs. y.19) . let us pretend f r o m now on that only the spin degree of freedom exists. do not describe two particles which are amalgamated into a single system.1 . but. Or. z) (14. y. j r > = xy2>(g>.18) Of course V 0 and V.9) gives P in the compact notation..3.3. two independent degrees of freedom of the electron.16) The forms of these operators are consistent with the requirement that operators built out of R and P commute with the spin operators.378 CHAPTER 10 whereas in its full glory. Lz becomes ifi d/d(f> R.
They are • S.S with eigenvalues ±/z/2.23) If one calculates <S> in the eigenstates of Sz. respectively. y. there are many interesting cases in which H is not separable.\/2y a SZ basis (14. mft) = Is. z. while the evolution of \xA0) is dictated by Hs.3. = \a\2+\Pi (14. basis P. N o w  Vo(0> evolves in response to i / 0 .3. and z. ±> and the verification o f E q .25).. m ) . Sz basis. With this assumption. t In the R. The product form of j y/} ensures that the spin and orbital degrees of freedom are statistically independent.3. We will tackle them in a later chapter.24) One refers to these as states with spin pointing u p / d o w n the z axis. one finds <1/2. Of course.21b) \X> = a\l/2. this means yt(x. szy = js.3. respectively.3. y. Let us address the determination (in the Sz basts) of the components of ]n. A complete basis is provided by the vectors s. We will follow just the evolution of \% s ). in which < « . ± l / 2  S j 1/2. sz.1/2) Any ket \x) in V.20) where  Vo) and \x s } are elements of V 0 and V s . ± > = ±(/i/2)« (14. ±> of n.l)= independent of x. the eigenstates \n. z.basis m) = l/2. y.\/2> + f3\\/2.3. ± l / 2 ) = ± ( # / 2 ) k (14. we have just a (complex) twodimensional Hilbert space Vs to work with. Consequently the state vector factorizes i n t o j (V(0> = iVo(0>®i^(r)> (14.3. More generally. ±  S  « . may be expanded as SZ basis (14.21a) K m ) = l/2. \fto(x.where H0 and Hs depend on just the orbital and spin operators.25) are said to be states with spin u p / d o w n the direction of the unit vector h. t)x(t) where % is a twocomponent spinor . and the orbital and spin degrees are coupled in their evolution.3. (14.22) IP The normalization condition is a i = <*!*> S. 1/2) 1 0 (14.
Another way to state this result is as follows. that nz ~ cos 9 nx = sin 9 cos (j> hy = sin 9 sin (j> The kets \n.3. i. Next pull out a common phase factor so that the spinor takes the form in Eq.3. This verifies the corollary and also fixes n.. (14. Take some spinor with components a = pi e'^ and P = p2e"l>2.1. + ) or if you want  —«.3.380 CHAPTER 10 Let us say n points in the direction (6.2) and to find cos((9/2) _ sin(0/2) e i m sin(0/2) e ^ ' 2 ' h up> = «+> = (14.3.e.} = L cos(0/2)e' w 2 (14. we can give the operator n • S of which it is an eigenvector with eigenvalue fi/2.27) (14. ± ) are eigenvectors of h • S = nx Sx + ny Sy + nz Sz nz nx + iny cos 9 sin 9 J* nx— iriy —n~ sin 9e~£. This has to do with the fact that any element of Vs has only two (complex) components a and j3. .3.28a) h d o w n ) = \ n .3. An interesting corollary is that every spinor in Vs is an eigenket of some spin operator n • S with eigenvalue ft /2.e.3.26) It is a simple matter to solve the eigenvalue problem (Exercise 14.28a).3.) . If we write <S> as (ii/2)n. . constrained by the normalization requirement \a\2 + \fi\ 2 = 1. and <S> contains exactly three pieces of information. Instead of specifying a state by a and j3. i. <j)).29) An interesting feature of Vs is that not only can we calculate <S> given a state. three real degrees of freedom. then the corresponding ket is  n . deduce the state vector given <S>. i.. but we can also go the other way. Exercise 14.28b) Y o u may verify that as claimed < « ±  S  « ± > = ± ( # / 2 ) ( i sin 9 cos $ + ] sin 9 sin 0 + k c o s 9) = ±{fi/2)n (14.p cos 9 (14.e. Let us verify the above corollary explicitly. From = h deduce that we can write pi =cos(0/2) and p2 — sin(0/2) for some 9.
34) and cyclic permutations (14.3. defined by s=\ 2 (14.3.32) (2) From the commutation rules for the spin operators S.) (4) The square of any Pauli matrix equals / : (14.ioz (3) They are traceless Tr cr..36) . <Tj]+ = 0 or ataj=aJal (i^j) (14. Sy.3.35) or more generally. i=xiyiz (14. oxGy .30) 0 1 1 0 0 i i 0 <7 1 0 0 1 (14.31) It is worth memorizing these matrices.3. (1) They anticommute with each other: * [<r. Here are some of their important properties. = 0. It is convenient to introduce the Pauli matrices o. and Sz.3. we get.So much for the state vectors in V s . H o w about the operators on this space? Let us commence with Sx.3. (nff)2= / Proof.3. Since Sz has eigenvalues ±/z/2.3. upon using the anticommutativity of the Pauli matrices. it follows that (14.33) (See Exercise 14.3 for the proof.
t See Exercise 12.J But since what we call the z axis is arbitrary. p = x.2By.40b) By this I mean This equation implies that the er„ matrices are linearly independent.382 CHAPTER 14 in this Hilbert space.3. z (14.3.3. it must beirut that or 4 or ( n • cr) 2 = I (5) W e can combine Eqs.] + = 2 V (6) Combining this relation with the commutation rules [ c f x .3.D ^ (14.3. A . y.3. as usual that £cQcra = 0^cG = 0 a for all a (14.4.3.y. (14. 0.35) into [cr i .37) we may establish a very useful identity (Exercise 14.cr. y.38) Q. (14.33). while i.3. and z.z.3.32) and (14. If we call it <r0.z. then Tr(<ja<jp) = 25ap (a.5.j will run over just x.34). § From now on a. (7) Combining Eqs.4&) Let us view the identity. Gy] = 2/<rand cyclic permutations (14. y. multiply both sides by <rp and take the trace. and (14.3.35) we find that Tr(<7/<7y) . (14.4): (A • <r) (B • <r) = A • B7 + / ( A x B) • <r where A and B are vectors or vector operators that commute with o. U j=x. as the fourth Pauli matrix. I.41) T o prove this for say cp.3.E. 0)§ (14. p will run over four values x.
3.43) (The coefficients ma will be complex in general.It is an interesting exercise to check that this inner product obeys the three axioms. any operator in V s may be expressed in terms of the cr a .42) To find mp.i 8 .45) I The inner product between two matrices M and M' acting on V5 is actually Tr(MMA).) Thus. i. .44) Q = 90 = 0 (14. f/[i?(6)] = txp{~B'S/fi). it may be written as ^ Af=£>„<?„ (14. + ) .3.t Explicit Forms of Rotation Operators The fact that («• <y)2 = / greatly simplifies many computations and allows us to compute in closed form several operators such as U(t) = exp(iHt/fi). However. to find mp=iTr(M(Tp) (14. we get U[R(Q)] = c o s ( ( 9 / 2 ) / .i $m(0/2)0 A Let us put this operator to a test.Since any 2 x 2 matrix M has only four independent (complex) degrees of freedom. and in the next.3. which are intractable in infinitedimensional spaces./ 6 . the propagator. Suppose we have a particle with spin u p along the z direction. Consider C/[/?(0)] = e x p ( .e.S / « ) = exp ( . in the state [i]. which form a basis that is orthonormal with respect to the inner product \ Tr(<7 a a" p ).< r / 2 ) = exp 00 / i0\ 1  W \ 2 2\\ 2) 3! {9iT) + Grouping together the coefficients of / and 0 • <T. If we want to get from this a particle in the state n.. and real if M is Hermitian. the dagger is irrelevant for the HermitianCT'S.3. it is clear that we must rotate [Q] by an angle 9 about an axis perpendicular to the 2 axis and the h axis. we multiply by dp and take the trace. Thus the rotation angle is kx« ik x n\ •a (14. In this section we consider the rotation operators.
(14. Exercise traceless. (1) Write c/cr.28) up to an overall phase. (2) U s e Eqs. (14.3. the first column gives the rotated version of [i]. (14.3.3. 2 [ct.3. < 7 a n d [cr„ cr. z.33) show that the Pauli matrices are Exercise 14. (3 = x. cos <j>.32) and (14.42) and (14.43). Since w = (sin 9 cos <j). ..384 CHAPTER 10 where k is the unit vector along the z axis.3.].47) According to our mnemonic. sin 9 cos <p.3. .3. 0) e x p l 9 .3 * Using Eqs.y.29).i sin I — I 9 a (A •ff)(B• a) = (A • B ) / + i(A x B) • a Exercise 14. 14. 0) = ( . 0) (14. cos (9/2) 1_sin((9/2) e'* sin(0/2)e~''*' cos(0/2) (14.f f j = cosl — I / . <T/] = 2i'£ EijkCTk k (n <r)2 = I Tr e r ^ O Tr(<ja<j0) = 28a0 (a. + ) given in Eq.46) The rotation matrix is.28).3. (7j] + = 218ij [<T.n are given by Eq. it follows that 0 sin 9 (—sin 9 sin <p.2.. Here is a summary of useful formulas that were derived or simply stated. We see that it agrees with n. cos 0).3. (2) Verify Eq.s i n <p.44).3. * (1) Show that the eigenvectors of a .3. from Eq.4. (14.3. (14.39) in two different ways.3. in terms of [ a .3. (14. (14. sin 9 sin <f>. * Derive Eq.
. In the figure. (Relate it to half a certain rotation.1) p=— e c x (14.8* (1) Show that any matrix that commutes with < is a multiple of the r unit matrix. it is preferable to summarize the interaction between the loop and the magnetic field in terms of £ The sense of e± is related to the current flow by the righthand rule. (2) Show that we cannot find a matrix that anticommutes with all three Pauli matrices. is given by (14. Exercise 14. The direction of the arrows in the loop is that of the current. Consider a square loop (Fig. let us recall some basic ideas from classical magnetostatics. Exercise 14.7.4. 14.2) where A is the area of the loop. (1) Argue that \n.1.) (2) (2I+axy\ (3) a . Spin Dynamics Since the quest for the spin Hamiltonian is based on classical analogy. Express the following as linear combinations of the Pauli matrices and /: (1) ( I + i a x y / 2 . Since we finally wish to address a q u a n t u m mechanical problem.6. in a magnetic field B.3.1) carrying a current I. (2) Verify by Exercise 14. Express the following matrix M in terms of the Pauli matrices: a fi M= _7 8 = n/2). Exercise 14. the magnetic moment. B is the magnetic field and ft is the magnetic moment of the loop.5.4. (If such a matrix exists. + > = U[R(<pk)]U[R{0})]\sz explicit calculation. it must equal zero. F r o m standard magnetostatics (force per unit length on a currentcarrying conductor etc.3. and e_L is a unit vector perpendicular to the plane of the loop.) 14.J The effect of T will be to rotate the loop until p and B are parallel.3. ' .Figure 14.4.) one can show that the torque on the loop is T=pxB where p. c is the velocity of light.3.
the interaction energy ist ^int= T(0)d0= nBsin0 dO = pB cos 6 = .j i . Since ji and 1 are parallel. For the particle considered 7 = ~ ~ (14. a stable configuration obtains.e. In this case. when ji and B are parallel. Although we derived the above equations for a square loop.7) 2 mc In the case of the current loop. as is the case for the particle in question. So we may apply it to the following problem.? i As we would expect. T causes a t This is not the full Hamiltonian (for it does not include the kinetic energy of the loop) but just the potential energy of interaction with the magnetic field.6) The ratio of fi to 1 is called the gyromagnetic above.^ l (H4. they are true for any tiny planar loop. charge q..4. K . it was stated that the effect of the torque T is to cause to align with B.4.B (14. over whose extent B is constant.386 CHAPTER 10 the potential energy associated with the torque: If 0 is the angle between (i and B. this energy is minimized. moving in a circular orbit of radius r. This picture changes when ji has its origin in angular momentum. ratio y. The current associated with this charge is 1 = charge flow past any point in the circle per second _ qv 2nr and the magnetic moment has a magnitude fi = qv 2 nr nr qvr ( q \ q — =— = mvr = I c 2c \2 mc 2 mc where / is the magnitude of the angular momentum. Imagine a particle of mass m. . i.
4.2.11) .10) Orbital Magnetic Moment in Quantum Theory These ideas reemerge in the q u a n t u m theory. the tip of the 1 vector moves by an angle (14. We may see this as follows (see Fig.. B = / ( l x B) (14. The Hamiltonian for a particle of mass m and charge q in a magnetic field is (PgA/c)2_Pl2 2m 2m q 2 mc .4.e.P ) 4 (14.4.9) V/sm 6 i.2).21 A I 2 <TIAI 2m c2 H (PA + A . precesses at a frequency to0=rB (14. The equation of motion is T= — x dt So in a small time At. 14. < j precession of p around B. In a small time At.8) Al= y(l x B)At or A / = ylB sin 9 At Since A1 is perpendicular to 1.4. the tip of the 1 vector precesses by an angle A > around the magneticfieldvector.Figure 14.
quadratic in B.) t It is shown in Section 18.12) so that V x A = B = 5k (14.4..388 CHAPTER 10 Let (14.4 that A corresponding to a given B can always be chosen divergenceless. When the middle term acts on any 11//>.±2. We will assume B is small and drop the last term in H. to denote the classical variable and the quantum operator.4.) If we project this relation along the z axis.4.. we get 4 r 2 mc 2 mc (0. so as to follow other widely used conventions. We will occasionally violate our convention this manner.15) exactly as in the classical case. (PA)j !//>•—i/iV • (Ai//) = —i^[(V'A)i// +A'Vi//] = (i/JAV)I//^(AP)I//> since V*A = 0 here.±1.14) 2 mc (14.1 .4..t Thus the interaction Hamiltonian is 2 mc mc 2 =—?LB=fi'B 2 mc so that (14.13) is constant and along the z axis. (We use the same symbol p.
Our present ^formalism does not tell us what g is.) It may be verified.18a) where 7 is a constant. This happens to be the case. We also assume that #int = — H *B = 2mc S • B geh_ 4mc <rB (14.4. by the use of Ehrenfest's theorem. we conclude that ^ \i= / S for the orbital case. We assume tRecall that these two are nearly equal: Mpc2 = 938. Since any operator on V s is a linear combination of the identity and the spin operators. (The nucleon Bohr magneton is also called the nuclear Bohr magneton.4. Since 7 = ~e/2mc  H=g(e/2mc)S where g is a constant. 3 x 10~ u e V / G (14. Spin Magnetic Moment Armed with all this knowledge.4. The electron Bohr magneton.19) The intrinsic magnetic moment due to spin is g/2 magnetons.17) where M is the nucleon (proton or neutron)! mass. and since ^ is a vector operator. We assume once again that there is a magnetic moment operator p. that <L> precesses around the constant field B just as 1 would (Exercise 14.4. to find it we must confront the above H with [experiment and hope that for some value of g it gives the right physics. 6 x 10~8 e V / G 389 SPIN (14. has a magnitude pfj 2 mc ^ 0 .1). simply called the Bohr magneton. we now address the problem of how the electron interacts with an external magnetic field. associated with the spin angular momentum. and the experimental value for g is very close to 2. . while Mnc =939.18b) (14.The quantity qtijlmc is called the Bohr magneton of the particle.4.57 MeV. let us write (14. The nucleon magneton is about 2000 times smaller: eh Bohr I 2 Mc ~ 0 .28 MeV.16) where m is the mass of the electron and G stands for gauss.4.
 R. Van Dyck. Lindquist.0011596521884(±43)] t The timeenergy uncertainty relation allows the production of these particles for short times. predicts that the Dirac result will receive corrections that can be calculated in a power series in a. Dehmelt. 59. a Dirac electron. The physics behind the corrections is the following. an electron will recapture the photon it emitted. The corrections may be calculated in a power series in a : g = 2 1 + — a + o(a2) + • • • which has been evaluated to order a 3 . and several electron positron pairs. which we will not discuss in this book. Q u a n t u m electrodynamics.e. Phys. Lett. P. Schwinberg. since the electron in the twoparticle system has both spin and orbital angular momentum. 1987. we can get a result that corresponds to 2. Occasionally. q u a n t u m electrodynamics predicts that what we call the electron is really a superposition of states that contain one Dirac electron. B. Phys. Between the emission and reabsorption. G. and H. If the magnetic moment of the system is probed at this time. Rev. B.20) Thus the gyromagnetic ratio for spin is twice as big as for orbital angular momentum. will dominate the picture and the complicated states will provide smaller and smaller corrections to the result g = 2. i. But it is irresistible to digress and mention that the Dirac equation.390 CHAPTER 14 hereafter that (14. D42. due to its ability to exchange other quanta such as the graviton. S. this result also receives corrections due to other interactions of the electron. Recall that the interaction between the electron and other charged particles is mediated by the exchange of photons. Kinoshita and W. predicts that g=2 exactly. 1990. Rev. In fact. which we will discuss in Chapter 20. § T. The result is§ gtheory = 2[ 1. Why is g ^ 2 1 A n d why isn't it exactly equal to 2. with just the Dirac electron. the system that originally contained just the electron will contain an electron and the photon.001159652140(±28)] where the error ± 2 8 in the last two digits is mostly due to uncertainties in the value of a itself and in the numerical evaluation of some of the integrals in the calculation. several photons. The experimental value of g is g e x p = 2[ 1. In addition to higherorder corrections..J The reason the observed value of g is so close to the Dirac value of 2 is that configurations of increasing complexity are suppressed by increasing powers of the finestructure constant in the superposition. But these effects are negligible to the accuracy considered above. 636. T h u s the simplest configuration. .4. which would be much prettier? Our formalism doesn't tell us. a Dirac electron and a photon. 26. etc. the finestructure constant.
The state at a later time is M0>= where U(t)=eiH'/n = e+ir'{S. Feynman has pointed out that this is equivalent to predicting and measuring the distance between New York and Los Angeles to within the width of a h u m a n hair! The theoretical situation is bad for the nucleons.ji*B= —/S'B (14..8 (e/2Mc) (ejlMc) Dirac theory predicts y = e/Mc or g=2 for the proton and 7 = 0 for the neutral neutron. i.! Let us now return to the dynamics of spin in a magnetic field B.21) where y= . All we need from now on is the Hamiltonian  "s a H .4.4.B)/n v«»> (14. The reason is that these participate in strong interactions as well.e 2 mc mc Let y(0)> be the initial state of the electron. and the result is (to two places) 7 proton = 5 . the effect of U(t) is clearly to rotate the state by an angle 6(r) = .. and the corrections are n o longer tiny. the magnetic moments of the nucleons are negligible compared to that of the electron.4. In other words.e. We can of course measure g experimentally.23) Since exp(z'8S/#) is the operator that rotates by 0.3 . can emit and absorb pions etc./ B r J End of digression. 6 /neutron = .22) (14.' 2 = .in splendid agreement with theory. The nonzero 7 of the neutron reflects the fact that the neutron can be in a state that has particles with compensating electrical charges but not necessarily compensating magnetic moments.4. and the counterpart of a is large ( ^ 1 5 ) . (14. the state with just the Dirac particle no longer dominates. Because of their large masses.24) .
26) . Let B be along the z axis: B=i?k... + ) • cos ( 0 / 2 ) .e. In this rotating frame.25) Br = B0 + to/r (14. In this case U(t)=exp(iytSzB/fi) = exp(io)0t(Tz/2) Since <rz is diagonal.392 CHAPTER 10 It follows that <S> will precess around B at a frequency (0 0 = .4. . It will precess around B 0 at a frequency to0= . let us consider a concrete example.y ( B 0 + to/y) Thus the effective field in this rotating frame will be (14. the precession frequency will be ©. ()> decreases at a rate &>0 • Paramagnetic Resonance Consider a classical magnetic moment n in a field B 0 = BQk.y B 0 Suppose we view this process in a frame that is rotating at a frequency to parallel to B 0 . = ©o© = .t o = .y B 0 .y B .4.<®0'/2 U(ty 0 (o)0 = yB) 0 e~ °ot/2 i( Consider an electron that starts out in the state + ): v/(0)> = ! « . sin(0/2) e'*/2 in which case V(t)>=U(t)\¥( 0)> cos(0/2) L sin(0/2) e ' ^ ~<B°')/2 i. If this seems too abstract.
3 that fi z oscillates as follows: pz(t) = p cos 2 a sin 2 a cos cort (COoO))2 (mo (o)2+y2Bi2 Y 2 B 2 COS = 0) cott (o)0.5). The z component of the moment oscillates with an amplitude p sin2 a..393 SPIN fi tIn Figure 14.31) . where. In this frame.co)2 + y2I? _ (14. the rotating component of B gets frozen (say along the x axis) and the constant component i?0k gets reduced as per Eq. The amplitude of the fiz oscillation is then at its maximum value of n . The situation in the rotating frame.4.28) We would like to find out the evolution of n(0..29) ootiBsin ^ f j + . timeindependent field is Br = Bir+(B0(o/y)k (14.) C In this frame.3.30b) I It is a simple matter to deduce from Fig. At resonance.Since B depends on time.27) where ir is the unit vector in the J direction in the rotating frame.30a) (14.Bok (B<^:B0) (14.1/2 (BoCQ/y)2] (14. This result is valid even if G and Bo are not parallel (Exercise 14. Consider now > the problem of interest.26) so that the effective.4. The effective magnetic field is B. it proves advantageous to first consider the problem in a frame that rotates at the same frequency co = <uk as the tiny clockwise rotating field B. where a is the opening angle of the cone.4. Br lies along the x axis and JI precesses in the plane normal to it. p will precess around B r at a frequency cor=/Br where ]<ar\=a)r=y[B2 + 2.4.4.4. (k = k r of course. (14. The magnetic moment starts out along the z axis (but is M 60S slightly displaced in the figure for clarity) and precesses around B.4. in a nonrotating (lab) frame B^cos and at ?=0. 14. (14.4.
S=k In Cl = k In 1 = 0 . where it will stay (in either frame). co — co0. As we pump in energy. k is Boltzmann's constant. however. assumed to be fixed. If we pump in energy 2 p B 0 . As a increases from 0. as you may verify by doing Exercise 14. one of the moments can move to the upper energy state. where it is proved in general.4. (Q depends on other variables. These results for the classical moment p apply to the expectation value (fi) in the quantum problem. grows. called a 90° pulse. 0 K. i. all are in the ground state (n parallel to B). At paramagnetic resonance. For instance.. a = n/2. ( ~ . So.) In most systems. Each magnetic moment (or spin) has two states only. where the explicit verification in this case is discussed. a.e. and the amplitude of oscillation. with energies E=±pB0.2 7 3 ° C ) is defined so that nothing can be colder.3.394 CHAPTER 14 This formula for fiz{t) applies in the lab frame as well.1. Sk In Q is the entropy and Q(E) is the number of states available to the system as a function of its energy. if we have a box of gas molecules. What if we apply the rotating field at the resonance frequency. the z component of B r steadily decreases. The system has a magnetic moment M = npk.4. let us quickly sort this thing out. the opening angle of the cone. they can occupy higher states. Consider now a collection of N spinhalf particles sitting on some crystal lattice which is immersed in a field B = B0k. the cone becomes a circle in the yz plane. and pz oscillates with the largest amplitude p at a frequency yB. there are N ways to pick the one that moves . and S= 0. yet here we speak of negative absolute temperatures! There is no conflict. but for a time r such that yBz = 7t/21 Such a pulse. /? is positive because adding energy only opens up more states and increases 0. If we apply a 180° pulse. Negative Absolute Temperature (Optional Digression) 4 The absolute zero of temperature. and S and Q can increase without limit. p sin2 a. choose r such that yBz = n the pulse will reverse the sign of p and leave it pointing down the z axis. will swing the magnetic moment into the xyplane (in either frame). The absolute temperature T is defined as follows: ± j _ s s j m n k dE { E ) v w kT 8E where /? is the thermodynamic temperature. At T= 0 K. increases. and Exercise 14. Thereafter p will precess around B0k at the frequency CD0 in the lab frame. Q = 1. since pz is invariant under: rotations. they all stay in the ground state at T= 0. The behavior for a > Q}0 is obvious. since we will see that negative temperatures are hotter than positive temperatures! Before you give up all faith. Br = Bir. where p is the magnitude of the magnetic moment.
New York (1970). We can also see this by imagining a system at T= . Thus the sequence of temperatures is T= + 0 . there will be population inversion (parallel<> antiparallel). (We should have chosen . (14. Thus M . See R. for it rises monotonically from 00 to +00 as we heat the system. At this point. 0 + . . called the transverse relaxation time. which will eventually cool it down to room temperature. How does one prepare negative temperatures in the lab? One takes a sample at room temperature. In terms of /?. The system has no mean magnetic moment along the z axis. . . . Clearly /? and T are positive.. the transverse components of M . The spin system cannot stay in this hot state (T= .) It should be clear that negative temperatures are hotter than positive temperatures since we go from the latter to the former by pumping in energy. which amounts to a change in the sign of /? and T [see Eq. The return to thermal equilibrium is easier to observe if one applies a 90° pulse which swings M into the xy plane. The temperature now is T— oo K. 3 0 0 . Magnetic Resonance. except for the change. Schumacher. . Benjamin. the former is hotter. S keeps growing until half are up and half are down. when E=NpB0. — 00. 0 = 1 and 5 ' = 0 . . breaches a maximum. Pumping in more energy only reduces S. . 0~. which simulate a bar magnet rotating in the xy plane. Finally. and T= +00. . Since the populations of the two systems are identical. Also note that the final equilibrium temperature is not 0 K but 00 K. . all moments are in the upper energy state (antiparallel to B). . T=0~. . So ft and T become negative.B . . with more and more particles piling up in the upper state. will eventually realign itself with B. .* T = oo. which will initially begin to precess around B = 5 0 k. parallel <»• antiparallel.e.. In this process energy clearly flows f r o m the negative temperature system to the positive temperature system. W. This corresponds to /? = —oo. so that Q = N and S=k In N. there is only one such state. . As we p u m p in more and more energy./ ? as the temperature. The frequency of the (damped) oscillation will be co0 and the halflife will be a time r. fi = dS/dE= 0. besides restoration of thermal equilibrium. A. ./ L . 0 ~ .up. If one winds a coil around it. I at J The transverse components of M decay for other reasons. M = Afyik. . . 00.33)]. . there is more continuity: /?= 0 0 . If one applies a 180° pulse. It will have more moments parallel than antiparallel: ^(parallel) _ 1 ( 1 4 4 3 3 ) t ^(antiparallel) e 13 B " ° and a net magnetic moment M along the z axis. .3 0 0 K brought in contact with identical system at T= +300 K. say at T= 300 K. . since Mz — 0»^(parallel) = ^ ( a n t i p a r a l l e l ) . i. they can increase their entropies by moving toward the state with equal numbers up and down.3 0 0 K) forever.X Exercise 14. .. will induce an oscillating voltage in the coil. The decay of its rotating components in the xy plane may be observed as follows. and B is position independent. —300.4. because it is in contact with the lattice.4. . Suppose the specimen is a long cylinder whose axis lies in the xy plane. —00.1 * Show that if / / = . .
4.4.4. 14.396 CHAPTER 10 C o m p a r i n g this t o Eq.5.4. « Exercise 14.4.4.31).4. W e would like to establish the validity of Eq. H o w m a n y seconds will it take for the spin to flip? Exercise 14.34) and (14. + ) and see w h a t is happening t o the spin for the case o)0 = (o. S l / V ( 0 > (14. Rotate back to the lab and show that cos — +1 V 2 } Q)rt\ 0)o — 0) . (14.4.36) ical/2 \ 2J (Or Compare this t o the state n.2. (14.4.31) by studying Fig. and B is time dependent.(0> = * " t o . that the change as seen in a frame rotating at an angular velocity to.8)]. Calculate </U z (0> and verify that it agrees with Eq.4. (14.4.35) to derive Schrodinger's e q u a t i o n for  ^ r ( f ) > in the S. Notice that this conclusion is valid even if B depends o n time and also if we are talking about spin instead of orbital angular m o m e n t u m .(?)> = Ur(t)\ y/. related to \ f(t)) by a rotation angle cot: ijr. Exercise 14. the propagator in the rotating frame. This state obeys m^\¥(t)> dt = H\y. (14.3* W e w o u l d like to study here the evolution of a state that starts out ( i ) and is subject to the B field given in Eq.y S . is turned on. consider the ket in the rotating frame.29)].27).8).(0)) by c o m p u t i n g Ur(t). / (0rt sin (Or — 2. + i(Ot/2 lv(0> Sz basis (14. Argue.4. Obtain a relation between the time derivatives o f V in the t w o frames.4. Since classical reasoning suggests that in a frame rotating at frequency (—6)k) the Hamiltonian should be time independent and governed by B r [Eq. basis and verify that the classical expectation is borne out. A more explicit verification f o l l o w s in Exercise 14.B . A t f = 0.4. using the results from Exercise 12.4. Derive (14.4. we see that <n> evolves exactly like n. (14.3.4. (1) Consider a vector V in the inertial (nonrotating) frame which changes by AVina time At. A steady field B=Bi. Exercise 14..35) C o m b i n e Eqs.3. (14. is AV — to x \At.3.4. and deduce the formula for the effective field in the rotating frame. 100 G .y (14. \y/r(t)).34) where H= . an electron is in the state with s7 = fi/2. Solve for . ( 2 ) Apply this result to the case of 1 [Eq. .26) when w and Be are not parallel.4. (14.
(j>)x V (14.5. All we have is a doubling of states. If we choose % to be an eigenstate of Sz. . Calculate < g ) . where the Coulomb interaction is independent of spin: H=H. and all that happens is that we attach a constant spinor x to the wave functions we found in Chapter 13. 14. Although both the proton and the electron couple to B. is in thermal equilibrium at temperature T.5. the smallness of the ratio m/M allows us to ignore. It will be dropped whenever it is obvious that we are dealing with spin.1) Here the spin is a constant in time.5.6 (A Density Matrix Problem). Return of Orbital Degrees of Freedom Let us now put back the orbital degrees of freedom.= _1_ \nlmms= \/2}^y/n/m(r. <<r>.5. The simplest case is when H is separable: H~~ Hq + Hs so that the energy eigenstates factorize \w} = \Wo}®\Xs> An example is provided by the hydrogen atom. we have! \nlmms= l/2}^\f/nlm(r. (2) Show that a is the m e a n polarization.2) (14. o (14. <j>)x s The energy levels are of course unaffected. with the electron spin being up or down (the z axis) in each of the orbital states (nlm). Consider next the problem of the hydrogen atom in a weak magnetic field B = Bk. the coupling of the proton's intrinsic and orbital magnetic moments [these are of order m/M and (m/M)2 relative J We use the subscript s on ms to remind us that it measures the spin projection: sz — mfi.4.3) o" x.Exercise 14. Construct the density matrix for this ensemble. 9. (1) S h o w that the density matrix for an ensemble o f spin1/2 particles may be written as p= 2(/+a<r) where a is a cnumber vector. 6. (3) An ensemble o f electrons in a magnetic field B = i?k. in the first approximation.
1* (1) W h y is the c o u p l i n g o f the p r o t o n ' s intrinsic m o m e n t t o B an order m/M correction to Eq. there will now be several. one simply adds the contributions from all the electrons.398 CHAPTER 14 to that of the electron. The details will be explained in the next chapter. different: .4.5. In a multielectron atom. m s = 1 / 2 ) Ry 4 eBft 2 l(m = 0.) (m/M)2 correction? (You may .14) and (14. The splitting of levels leads to an increase in the number of spectral lines.4)? ( 2 ) W h y is the c o u p l i n g of its orbital m o t i o n an order reason classically in b o t h parts.5. namely.6) The second. The eigenstates of H will not be simply products of orbital and spin parts. n/mm 4 ). and consequently the spin and orbital degrees of freedom are coupled in their time evolution. or 1) (14.19). = .1]. we note that it is not separable. This phenomenon is called the Zeeman effect. which was twofold degenerate. Exercise 14. ms=—\/2. and the spacing between them may be varied by varying B.5.l ( m = 0 .5) The degeneracy is greatly reduced by the B field.8) whose origin will be explained in a later chapter. Thus we have. ms= 1/2) . (14. L z .R y ± (14. (14. ms = — 1 / 2 ) and so on.5. ms~ 1 /2)(l~0 0 (m= 1.5.7) mc or m = ~\. m . however. = . Consider lastly the Hamiltonian H=HCo^omb + aL'S (14. H— ^Coulomb .5.5.4. f r o m Eqs.l / 2 ) (/ = 0 or 1) _ — 2(m = — 1. splits into two levels: efiB 2 mc £„. _ —r— H (m + 2 ms) \nlmms} n 2 mc H\nlmms> = (14. \LZ— ( mc IiSj (14. see Exercise 14. this H is diagonalized by the same states as before.5. The eigenvalues are. and S2.I \ 2mc.4) Since the additional terms in H commute with i/c0ui0mb5 L2. but instead superpositions of such states that diagonalize LS. The ground state. which was eightfold degenerate. F o r the present. splits into five levels: 2(m — 1.R y eBft . where there was one.
A beam of particles endowed with magnetic moments enters the inhomogeneous field. traveling along the y axis.4) consists of north and south pole pieces.9) [We have used the identity V(i*B) = ( p .9) is discrete and therefore so is the angular m o m e n t u m along the z axis. (2) Recall that w e have been neglecting the o r d e r B2 term in H.Figure 14. In the present case. w h e n a field B = lOOOkG is applied. A beam of (particles with) magnetic moments. the beam will split into two parts. V x B = 0.5. What one observes is a set of discrete dots (B). What do we expect will happen classically? If we pretend that the magnetic moment is due to a pair of equal and opposite (fictitious) magnetic charges. Estimate its c o n t r i b u t i o n in the n = 1 state relative t o the linear ( — j i ' B ) term we h a v e kept. and by Maxwell's equations. The SternGerlach experiment. the beam is expected to fan out and produce a continuous trace (A in figure) on a screen placed behind the magnet. Exercise 14. which clearly displays the quantization of angular m o m e n t u m (along any direction). This implies the quantization of magnetic moment and angular momentum. between which is an inhomogeneous magnetic field.V)n + i x (V x B) + B x (V x p). This experiment can also be used to reveal the existence of electron spin. if we send in a beam of hydrogen atoms in their ground state.5.V ^ r = V(nB) = ( n .4.V ) B =  / z — k dz (14. Gassically the beam is expected to fan out and produce a continuous trace (A) on the screen.5. by a s s u m i n g the electron moves o n a classical orbit o f radius do.V ) B + (B. This is confirmed if we calculate the force associated with the gradient of the interaction energy • • F = . The actual experiment performed with atoms reveals a series of discrete dots (B in figure). F o r example.] Classically. n is not a function of r. it is clear that any inhomogeneity in B can lead to a net force on the dipole. (14. since fi z is continuous. . The apparatus (Fig. Both Fx and Fy vanish on average due to the precession of spin in the xy plane.2* ( 1 ) Estimate the relative size o f the level splitting in the « = 1 state t o the unperturbed energy o f the n= 1 state. We understand this in semiclassical terms. A b o v e w h a t B d o e s it begin t o be a p o o r approximation? The SternGerlach (SG) Experiment We now consider (in simplified f o r m ) the SG experiment. by saying that ji z in Eq. 14. enters the apparatus in a region where B is predominantly along the z axis and dBz/dz< 0.
V/V100 + WyWlOQ 0 J . Since dBz/dz< 0. in region where y y . Lectures on Quantum Mechani Benjamin. T1 can be undone and the particle restored its original momentum (but filtered wi respect to spin) if we place some more magnets (with B along the z axis) behind tt apparatus. So by Ehrenfest's theorem! we expect the atom to emergei the apparatus in a state (up to a phase factor) Vout= </>V. If we send in just the electron. See. for the associated magnetic moment is too small to affect the dyna ics.+ z is peaked) it will have spin up. quantum fluctuatio would wipe out the effect. its is down. then.+z describes a wave packet that is displaced (relative to the incoming oi along the positive z axis and has also a small velocity in the same direction. Baym.Wy¥ 100 Vinitial . pages 324330 of G. In the present case.e.) Since the electron spin is up. while if we catch it below.+z(rcM)yioo(r) where \fry. the system follows the classical trajectc (approximately) thanks to the massive proton. More generally. we see that the SG apparat has introduced a correlation between the spin and orbital coordinates: if we cat (by placing a screen) the outgoing atom above the original line of flight (i. Suppose the initial state of a hydrogen atom is ^initial = V^(rCM)VlO0(r) al where y/^ is a wave packet drifting along the y axis that describes the CM mot yioo is the ground state wave function. New York (1969). the classical fo on the atom is up. (The prol spin is ignored. for example. and [oj is the electron spinor. ^ .. J . But nc that the filtering process changes the average z component of linear momentum.±z are narrow packets with no overlap. if "o" a a . Likewi if the electron spinor had initially been [ ?]. the C M would have emerged in the stj y y .z (in the same notation). With this modification (which is assumed in the following exercises) tl t Recall the warning at the end of Chapter 6.400 CHAPTER 10 Let us describe the abovementioned hydrogen atom experiment in quant mechanical terms. t spin is down. by the linearity of Schrodinger's equation a Vout= 100 0 100 0 IP Assuming if/y. The SG apparatus can be used to prepare a state of definite spin orientatio to get a pure spin u p / d o w n beam we simply block the lower/upper beam.
e. which lies in the xz plane at an angle 6 relative t o the z axis. w h a t fraction of particles leaving the first will leave the last? Exercise 14. moving along the y axis. w h a t fraction of particles leaving the first will exit the third? If the middle filter transmits both spins u p a n d down (no blocking) the x axis. the first with B along the z axis a n d the second with B along the z' axis.5. Exercise 14. A b e a m of spin1 particles.3. W h a t fraction leaving the first will pass the second? 401 SPIN . * A b e a m of spin1 f l particles moving along the y axis goes t h r o u g h two collinear SG apparatuses. is obtained by r o t a t i n g the first by an angle nj2 a b o u t the y axis). b u t the last one transmits only spin d o w n the z axis. W h a t fraction of particles leaving the first will exit the second? If a third filter that transmits only spin u p along the z axis is introduced. is incident on two collinear SG apparatuses.4.only effect of the SG apparatus with one or the other beams blocked is to filter the spin without affecting the orbital motion. T h e first has its B field along the i axis and the second has its B field along the x axis (i. Both a p p a r a t u s e s transmit only the u p p e r m o s t beams. b o t h with lower beams blocked.5..
What are the possible values for the magnitude and z component of the system spin. 2) (15. the twoparticle Hilbert space V\®2 is spanned by the four vectors \s2m2y = \sim{.15 Addition of Angular Momenta 15. and m. s2m2) Since s( = 1 /2. =S' 1 .1.1.= ± 1 / 2 has freedom only in sign. —+). 1 +> =  s1 = 2ml=ls2 = ^m2 = L2) (15.3) and so on.1. s2m2> which obey S^i/Wi.. and what are the states that go with these values? This is a problem in addition of angular momenta.2b) (15.1. let us use the compact notation j++>.1.  H—). S. For instance.  — ) to denote the states. which is the topic of this chapter.2a) (15. A Simple Example Consider a system of two spin1/2 particles (whose orbital degrees of freedom we ignore). If Si and S 2 J are their spin operators. {In terms of the operators Sf 1 ' and which act on the oneparticle spaces. Suppose now that we choose not to look at the individual spins but the system as a whole. They represent states that have welldefined values for the magnitude and z component of the individual spins. s2m2y ( r = 1.1) ^ l ^ m i . s2m2) = fi2si(si+ l)j]»ii. ®/ < 2 ) and S 2 = 403 . s2m2y = fimi\s\m\. These four vectors form the product basis.
404 CHAPTER 10 Consider the operator 1 S = Si + s 2 (15. i. This is readily verified: sz\++y = (siz+s2z)\++y = a + 2 l + + > 5r+> = 0+> S„+> = 0  . By the method of images (or any other method) (15. rotations of the whole system. • S 2 (15. S 1 z .e.* ! — > Thus the allowed values for the total z component are ti.1. and the eigenspace is spanned by the vectors H—) and ] . aH—> + /?—f>.S22 (15.4} which we call the total angular momentum operator. That S is indeed the total angular momentum operator is supported by (1) our intuition.61 i which commutes with S2u Si. We expect it to be diagonal in the product basis. and S2z. we still get an eigenstate with = but this state will not have definite values for Siz and S2z (unless a or /? = 0). If we form some linear combination.. (2) the fact that it is the generator of rotation for the product kets.+ > . namely.+ > s.1.8) 0 0 Note that the eigenvalue s z = 0 is twofold degenerate.1.1. Consider next the operator S 2 ^= (S.5) as may be readily verified. Consider first ^ SZ = S 1Z + . I — > = . and — fi.9) . iheykSt (15. 0. + S 2 ) • (Si + Sz) = Sf + S j + 2S. Our problem is to find the eigenvalues and eigenvectors of S2 and Sz.1.7) ++ 1 0 product basis +0 0 0 0 + 0 0 0 0 — 0" 0 0 1 (15. (3) the fact that it obeys the commutation rules expected of a generator of rotations.1.
the following linear combinations are: 2t/2 0=1) (15. / 2 [ + . which has SXx. SI. SXz. = 1/2 5 2 = 1 / 2 ) = 2^ . S\).  s = l m = 0.14) which means that the direct product of two spin1/2 Hiibert spaces is a direct sum of a spin1 space and a spin0 space. = 1 / 2 J 2 = 1 / 2 > = I — > s = 0 m = 0.1.1.1."f" S1S2+) (15.1. 0.S2.i/2 Exercise 15.1.+ 0 1 1 0 0 0 0 2 (15.11) (s = 0) . H—) and  —+). (15. S2z) to one that diagonalizes (S2. The corresponding eigenstates in the product basis are s=lm=l. it does not commute with S]z and S2z because of the S] S 2 term. We can describe our findings symbolically as 1/2® 1/2= l © 0 (15. and . However. 405 ADDITION OF ANGULAR MOMENTA ++ 2 0 product basis +0 1 1 0 .13)  j = l r o = —1. It might help to use I S i ' 8 2 .^ 1 ^ 2 2 + 2 (iSi+iS^. namely. The way the dimensionalities work out in . The three spin1 states are called triplets and the solitary spin0 state is called the singlet.11). in it. are not. f r o m one that diagonalizes ( S i .1.! .1 * Derive Eqs. while the allowed values of sz are fi. The problem of adding angular momenta is essentially a change of basis.10) 0 0 Thus we see that although  + + > and  — ) are eigenstates of S 2 [ X s + 1) = 2].> .1. The allowed values for total spin are s= 1 and 0.10) and (15.Although S2 commutes with S i and S\. the states of zero Sz. etc.fi. S\y. S\ = \/2S2 = 1 / 2 ) =  + + ) = 1/2 s 2 = 1 / 2 ) = 2~]/2[\ +>++>] (15. 5. SZ. By explicit computation. they form the totals basis.+ > ] These vectors represent states with welldefined total angular m o m e n t u m .12) This completes the solution to the problem we undertook.1.5.
406
CHAPTER 10
Eq. (15.1.14) is as follows: lefthand side: righthand side: (2s, + l ) ( 2 s 2 + 1) = (2 x 1/2 + 1)(2 x 1/2 + 1) = 4 £ (2s + 1 ) = 1 + 3 = 4 j=0
«
(15.1.flP
The decomposition of the direct product space into a sum over spaces with welldefined total spin can also be viewed this way. The rotation operators for the entire system will be 4 x 4 matrices in the product basis. These matrices are, however, reducible: by changing to the totals basis, they may be block diagonalized into a 3 x 3 block (spin1 sector) and a 1 x 1 block (spin0 sector). The totaks basis is. however, irreducible; we cannot further subdivide the spin1 space into parts that do not mix under rotations. The totakv states have another property: they have definite symmetry under the exchange of the two particles. The triplets are symmetric and the singlet is antisymmetric. Now, the state vector for two identical spin1/2 particles must be antisymmetric under the exchange of particle labels, i.e., under the exchange of their spin and orbital degrees of freedom. We already know that if Q is some orbital operator (built out of coordinates and momenta), then
©2, £ > = 2
1/2
[©1©2>+  « 2 « 1 > ]
and
(D\G)2, A~) = 2
1/2
[<J)i©2>
—
\03203]}]
are symmetric and antisymmetric, respectively, under the exchange of the orbital variable. To form the complete state vector, we simply multiply orbital and spin states of opposite symmetry:
+>+>
©1©2, S ) ® G>]tn\, o)2m2, A) = < fflifi>2, A}<S> (15.1.
These vectors provide a complete basis for the Hilbert space of two identical spin1 / 2 particles. As an example, consider the ground state of the He atom, which has two electrons. In connection with the periodic table it was said that in this state of lowest energy, both electrons are in the lowest orbital state « = 1, / = 0 , m = 0)J and
t If we neglect interelectron forces, the states allowed to the electrons are hydrogenlike, in that they are labeled n, I, m). But the energies and wave functions are obtained upon making the replacement
have opposite spins. We can sharpen that statement now. The orbital part of the groundstate ket is just the direct product, ^ > = 1100)01100) which is already symmetric. So the spin part must be lz.) = 2  , / 2 ( t +  >  !  + » anaso (15.1.19) (15.1.18) (15.1.17)
407
ADDITION OF ANGULAR MOMENTA
Vground> =  V o ) ® ! * * )
In this state, both the orbital and spin angular momenta are zero. Let us now return to the problem of just the two spins (and no orbital coordinates). Now that we have two bases, which one should we use? The answer depends on the Hamiltonian. F o r instance, if the two spins only interact with an external
field B=5 0 k,
H = ~(y ,S, + y 2 S 2 ) • B =  B ^ y ^ + y2S2z) (15.1.20)
the product basis, which diagonalizes S]z and S2z is the obvious choice. (If, however, f\=y2, then HozSz, and we can use the total.? basis as well.) On the other hand, if the spins are mutually interacting and, say, H= AStS2 the total.v basis diagonalizes H. Exercise 15.1.2. * In addition to the Coulomb interaction, there exists another, called the hyperfine interaction, between the electron and proton in the hydrogen atom. The Hamiltonian describing this interaction, which is due to the magnetic moments of the two particles is, H¥=ASrS2 (^>0) (15.1.22) = iA(S2 ~S2X~ S22) (15.1.21)
(This formula assumes the orbital state of the electron is 11, 0, 0).) The total Hamiltonian is thus the Coulomb Hamiltonian plus Hhf. (1) Show that Hkf splits the ground state into two levels: K2A E+ = —Ry + 4 Ry3 t?A
(15.1.23)
and that corresponding states are triplets and singlet, respectively.
408
CHAPTER 14
(2) Try to estimate the frequency of the emitted radiation as the atom jumps frotfflrj triplet to the singlet. To do so, you may assume that the electron and proton are two fj.e and nP separated by a distance a0, with an interaction energy of the ordert
>P Sj
Show that this implies that the constant in Eq. (15.1.22) is
2e (5.6)e 1 al
2mc 2Mc
(where 5.6 is the g factor for the proton), and that
AE=E+E =Aft
2
is a correction of order ( m / M ) a2 relative to the groundstate energy, Estimate that the frequency of emitted radiation is a few tens of centimeters, using the mnemonics from Chapter 13. The measured value is 21.4 cm. This radiation, called the 21cm line, is a way to detect hydrogen in other parts of the universe. (3) Estimate the probability ratio P(triplet)/P(singlet) of hydrogen atoms in thermal equilibrium at room temperature.
15.2. The General Problem
Consider now the general problem of adding two angular momenta J! and J2, W h a t are the eigenvalues and eigenkets of J 2 a n d / z , where J = J t + J 2 ? One way to find out is to mimic the last section: construct the (2/'t + 1) • (2j2+ l)dimensional matrices J2 and Jz and diagonalize them. Now, J: will be diagonal in the product basis itself, for Jz\j\mxj2m2) = fi{mx + m2)\jxmx J2m2> (15.2.1)
It will be a degenerate operator, for there are many ways to build up a total m= mx+m2, except when m = ±(j\ + j2) when both angular momenta have maximal projections u p / d o w n the z axis. F o r instance, if m=jx +j2 — l, there are three product kets: (mi = j\, m2 = j2 — 2), (m, — j\ — 1, m2 =j2 — 1), and (m, = / t — 2, m2 =j2). In each of the degenerate eigenspaces of Jz, we must choose a basis that diagonalizes J2 (and undiagonalizes J\z and J2z). We can do this by constructing the matrix J2 and then diagonalizing it. But this can be a tedious business. (If you have done Exercise 15.1.1 you will know that the construction of S 2 is quite tedious even in this fourdimensional case.) There is, however, a more efficient alternative to be described now. As a first step, we need to know the allowed values for j. O u r intuition and our experience f r o m the last section suggest that j can take on values y'i+;2)
t The description here is oversimplified; both and Hhf are rather tricky to derive. Our aim is just to estimate  A \ and not to get into its precise origin.
JT'i+721, • • • ,./i j'2 (assuming 7, >72)+ Let us check this. The number of product kets is (2ji + l)*(2y 2 + 1). This must equal the number of totaI7 kets. According to our conjecture, this number is
Jl
j:
{2j+l)=Y! j=o
(2j+l)~l
t j=0
1
(2y+ 1) = (2y'i + l)(2/2+ 1)
(15.2.2)
using the formula
I
J\®j2=
£ n= V ^ «=o
N(N+l) 2
We take this to be proof of our conjecture:
<J\+h)®Ux
+ 7 2  ! ) © • • • ©C/i J2)
(15.2.3)
In other words, the total/ kets are
IMA/2>
with
./', +j2>j>j\
j>m>—j
(15.2.4)
Let us write them in the f o r m of an array:
h +J2
\h +j2>j\ +72>
7i + 7 2  1
• • 7i  7 2
\j\ +J2,j\ +72 — 1 > ly'i +72,7. + 7 2  2 )
17i +72,  ( 7 1 + 7 2  2 ) )
l7'i+721,7I + 7 2  1 ) l./'i +721,71 +72 " 2 >
y, + 7 2  1,  ( 7 1 + 7 2  2 ) ) ly'i + 7 2  1,  ( 7 i + 7 2 ~ 1 ) ) 17"i 72,7'i  7 2 )
l7i  7 2 ,  ( 7 1  7 2 ) )
\jt +72, (7'i + 7 2  1 ) )
L7I +72, ~ ( / I + 7 2 ) )
(15.2.5)
(Note that the labels j \ j 2 are suppressed on the total/ kets. We shall do so frequently to simplify the notation.) Our problem is to express each of these kets as a linear combination of product kets. To get an idea of how one goes about doing this, let us consider the problem
J There is no loss of generality, for we can always call the larger one /',.
410
CHAPTER 10
solved in the last section (Ji =j2 = 1/2). In this case the states are
1 11,1) 11,0) 11,1)
0 10,0)
Consider the top state in the first column, 1, 1), which has the largest possible: component. There is only one product state with the right value of m, namely, both spins up. So by inspection,  l , l > = l++> We can multiply the righthand side by a phase factor, but we follow the convention, called the Condon Short ley convention, in which the coefficient of this top state is chosen to be unity. Consider next the state below this one, namely, 11,0). There are two product states with m = 0, namely,  +—) and   + >; and ( 1 , 0 ) must be a linear combination of these. We find the combination as follows. We know thatj SL1,1> = 2 1 / 2 *1,0> so that 1,0) S\ 1 , 1 )
2 ]/2h
But we do not want 11, 0 ) in terms of j 1, 1), we want it in terms of the product kets. So we rewrite the righthand side as
2 l/2h so that l,0) = 2"1/2(+) +   + » in accordance with our earlier result. The next state 1,  1 ) can be obtained by lowering this one more step in above sense, or more simply by noting that there is only one ket with m maximally negative, namely,  — > . So 1 1 ,  0 = 1—> Our phase convention is such that this is what you would get if you lowered  li J Recall J±\j, my = n[U^m)(j±m + 1)] /2;, m± 1>.
This takes care of the 7 = 1 states. Consider next j=0. The state 10, 0 ) has m = 0 and is also a linear combination of H—) and  —+>. We find the combination using two constraints: (1) The combination must be orthogonal to the one that forms the other state with m = 0, namely, 11, 0 ) and have real coefficients.! (2) The combination is normalized to unity. If we call the combination aH—> + j 3   + >, these constraints tell us that a + f3= 0 a2 + j32=l It follows that 0, 0> = 2  1 / 2 (  +  >    + » Note that we could still have multiplied the state by (  1 ) . Our convention is as follows: in each column in Eq. (15.2.5) the top state is given the overall sign which makes the coefficient of the product ket with m\ — j\ positive. Let us now turn to the general problem, Eq. (15.2.5). Once again the top state in the first column, with m equal to its maximum value of /1 + j2, can be built out of only one product ket, the one in which both angular momenta take on maximum possible projections along the 2 axis: \J\ +jij\ +h> = lA/i > A/2> (15.2.6)
411
ADDITION OF ANGULAR MOMENTA
The other m states at this value of 7 are obtained by lowering. Let us consider going down just one step. Since J\h +72,71 +/ 2 > = «[2(/l +j2)]1/2\ji we have, as in the spin(l /2(g) 1/2) problem + j l j l +721)
1
[2(7, +j2)]U2ft
={ — ) \7i +72/
[H2jl)i/2\j\(jl
 1 )J2j2)
+ H(2j2y/2\j\ji,72(72
1)>]
L/IOW)J272>
+( T ^  ] \7i +72/
\J\J\ , 7 2 ( 7 2  1 ) )
(15.2.7)
Proceeding in this manner we can get to the bottom state in the first column.§ Now for the top state in the second column. Since it has m=jx +j2~l, there are two product kets that are eligible to enter the linear combination; they are
t This is a matter of convention. §In practice one goes only to m = 0. The states of negative m can be found using special properties of the expansion, to be discussed shortly.
412 CHAPTER 15
 A / i J2U2  1)) and (7,  \),j2j2). The combination must be normalized to unity. orthogonal t o the other state formed out of these kets, namely, /i + j2,j\ [see Eq. (15.2.7)], and by convention have real coefficients. The answer is, by inspection,
L/I+721,7.+/21>=
/
—T)
• Y /2
UJujiUil)}

777 Vy 1 +jJ
17.(7.l),A/2>
The overall sign is fixed by requirement that the coefficient of the product ket with m, =j 1 be positive. Given the top state, the rest of the second column may be obtained by lowering. Let us go just one more column. The top state in the third column. /i +j2 2,7'i + 72 — 2 ) , can be a superposition of three product kets. The three (real) coefficients are determined by these three requirements: orthogonality to the two preceding total/ kets of the same m, and unit normalization. It is clear that there are always enough constraints to determine the t o p states of each column, and once the top states are known, the rest follow by lowering. Exercise 15.2.1. (1) Verify that 1/1/1,72/2) is indeed a state of jj\ + j2 by letting (2) (optional) Verify that the righthand side of Eq. (15.2.8) indeed has angular momentum j=ji +j2 — 1.
J2 — J2 + J2 + 2JuJ2z + J\+Ji + J\Ji+ act on it.
ClebschGordan (CG) Coefficients
The completeness of the product kets allows us to write the total/ kets as \jm,jxj2) = £ £ l7'i^i J2W 2 ></iWi J2WI2IM/'.72>
m
The coefficients of the expansion O . w 1 ? 72w21 jm, 71J2 ) = O1W1 ,j2m2\jm)
are called ClebschGordan coefficients or vector addition coefficients. (Since the labels jij2 appear in the bra, we suppress them in the ket.) Here are some properties of these coefficients '. (1) O ' i ^ i j 2 / w 2  y m > / 0 only if jxj2<j<jx+j2 (15.2.9)
(This is called the triangle inequality, for geometrically it means that we must be able to f o r m a triangle with sides jx , / 2 , and j). (2) (7'imi ,72W 2 7>n)#0 only if m\+m2 =m
(14.4.31)
(3) they are real (conventional)
(4) <;', ;'i, j2( j
) I j j > is positive (conventional)
413
ADDITION OF ANGULAR MOMENTA
(This condition fixes the overall sign in the expansion of each t o p state and was invoked in the preceding discussion.) (5) <Jm j2m2\jm> = (\y+hJ(Mm1)J2{m2)\j(m)) (15.2.11)
This relation halves the work we have to d o : we start at the t o p state and work our way down to m = 0 (or 1/2 if j is halfintegral). The coefficients for the negative m states are then determined by this relation.
Exercise 15.2.2 * Find the CG coefficients of
(1) 2(X>1 = 2©3 (2) 1® 1 = 2 © 1 © 0
Exercise 15.2.3. Argue that =
If we assemble the C G coefficients into a matrix, we find it is orthogonal (real and unitary). This follows f r o m the fact that it relates one orthonormal basis to another. If we invert the matrix, we can write the product kets in terms of total/ kets. The coefficients in this expansion are also C G coefficients:
t
(jm\j]m]
J2m2y
= ( j ^ ,j2m2\jm)
= <j,m,
,j2m2\jm)
because the C G coefficients are real. As an example, consider the There we have
\mxm2)
problem.
1 jm) ~ 1,1> ~ i,o> 11, —1> _ 0,0> _ "1 0 0 _0 0 1/2
1/2
0 1 /2 0
1/2
0" " l + + > " 0 1
1/2
l+ > I+)
0 1/2V2
—1/2
0_ _ 1 — > _
(Notice that the columns contain not the components of vectors, but the basis vectors themselves.) We can invert this relation to get
"1 0 0 _0 0 1/21/2 1/2./2 0 0 0 0 1 0 " "ll,l> 1/2>/2 1,0> —1/21/2 0 11,1) _ 10,0)
"l++>" l+> l"+> J — >_
414
CHAPTER 15
Thus we can write  +  > = 21/2(l,0> + 0,0>) 
etc. In practice one uses C G coefficients to go both ways, from the product to d totalj basis and vice versa.
Addition of L and S
Consider an electron bound to a proton in a state of orbital angular moma /. Since the electron has spin 1/2, its total angular momentum J = L + S can ha\ values of j=l± 1/2. We wish to express the total/ states in terms of product state  lm0, sms}.% Since m5 = ± 1 /2, at each m there will be at the most two eligible produi kets.§ Let \j—l+ 1/2, m) = a\l, m — 1/2; 1/2, 1 / 2 ) + P\l, m+ 1/2; 1/2,  1 / 2 ) \j = l— 1/2, rriy = a'\l,m1/2; 1/2, 1 / 2 ) + P'\l, m + 1/2; 1/2,  1 / 2 ) (15.2.12 (15.2.1^
The requirement that these states be orthonormal tells us that
a2 + p2= I
a'2 + P'2= 1 aa' + pp' = 0
(15.2.14
(15.2.15 (15.2.16
So we only need one more constraint, say the ratio a / p . We find it by demandini that J2\j= / + 1 /2, m> = ft2(l + 1 / 2 ) ( / + 3 / 2 )  y ' = / + 1 /2, m> Writing J2 = L2 + S2 + we can deduce that + LS+ + L+S_ (15.2.18; (15.2.17
g =f ' + 1 / 2 " f a V/+1/2 + W
2
(,5.2.19)
J Here, , mJ5 and m stand for orbital, spin, and total projections along the 2 axis. § It might help to construct the table as in Eq. (15.2.5). It will contain just two columns, one for j=l+\j 2 and one for j=l— 1/2.
In the presence of spin one changes the notation as follows: (1) Use capital letters S. Exercise 15.2.d. The total/ basis will come in handy when we study the spinorbit interaction [which involves the operator L .. the product basis is adequate. Show that the states with j=2j\ = 1 are antisymmetric. 1/2. or.Given this. to denote the (orbital) angular momentum of a particle.) This pattern of alternating symmetry continues as / decreases.. only one term survives with unit coefficient.2. an interaction with a weak constant magnetic field. obey V>Pj=5IJPj [use Eq. Thus. (3) Append a sw/jmcript 2S+1 to the left of L to indicate the multiplicity due to spin projections. 1/2. (Argue for the symmetry of the top states and show that lowering does not change symmetry. subspaces in the Exercise 15. F o r a multielectron system. (14. Exercise 15.3. m) = [ ± ( / + 1 / 2 ± m ) 1/2 /.2. . for example I 2S+i L j = 2Py2 denotes a state with / = 1.39)]. s = 1 / 2 . Derive Eqs. / = 3/2. The Modified Spectroscopic Notation In the absence of spin. the states with 7 = 2/1 are symmetric.2. . to indicate the value of the orbital angular momentum. (15. m = ± ( / + 1 / 2 ) . . Exercise 15. in addition.e. i. (2) Append a ^ s c r i p t J to the right of L to indicate the j value.19) and (15.. P.7.S2)/fi2 and P „ = (S t S 2 )/fi 2 are projection operators. and our convention for the overall sign.S = \(J2 L2S2)\ in Chapter 17.p. (let us call a typical letter L).20) [Notice that if/ = / + 1 /2.4.2. F o r a single electron the 2S+ 1 label is redundant and always equals 2. D. S and L stand L . Construct the project operators P± for the j=l±\/2 addition L + S = J.5* (I) Show that Pi = h+(S.2.20).. Show that when we add j\ to j\. 1 / 2 ) + ( / + l / 2 T m ) 1 / 2  / .. 415 ADDITION OF ANGULAR MOMENTA \j=l± 1/2.] If the Hamiltonian contains just the Coulomb interaction.6. but is harder to prove.1 / 2 ) ] I (15. (2) Show that these project into the spin1 and spin0 spaces in =>(g) = 1©0. m — 1 / 2 . m + l / 2 . it is sufficient to use a single letter such as s.2.
Vy. 25+ 1 r _ 1 Lj— So 15. Let 11).3.. and J for their sum.4. (15. The relation of Eq. Thus in the ground state of He.3.3. A scalar operator S transforms like a scalar under rotations. (15.S]=0 [J*.5) Notice that we are summing now over the first index of R.3) in an equivalent form.1) By considering arbitrary infinitesimal rotations we may deduce that [Ji. (15. A vector operator V was defined as a collection of three operators ( V x . (15. This seems peculiar. we may deduce that [Eq. If  V} is some vector .3.2) j where R is the usual 3 x 3 rotation matrix.3) where the sum is over the second index.5) is the following.3.S] = 0 Examples of S are rotationally invariant Hamiltonians such as the Coulomb or isotropic oscillator Hamiltonian. i. 2). Replace R by RTX =RTeverywhere to get U[R\Vtu\R\ =Z ^ J j (15.416 CHAPTER 10 for total spin and total orbital angular momentum.3.3.3. for we are accustomed to the likes of Eq.J!} = ihYj£l!kVk k (15.14)]: [Vi.4) Let us rewrite Eq.3) to Eq. remains invariant: (15.e. (12.S]= or in a form that will be used later W±. Vz) which transform as the components of a vector in V\R): V^V\= RyVj (15. By considering infinitesimal rotations. and 13) be basis kets in V3(R) and R a rotation operator on it.3.3.3) 0 (15. Irreducible Tensor Operators We have already discussed scalar and vector operators.
. y = R\i> = l \ j X j \ R \ i > = l R j i \ j > j j (15.3. . under TiJ~>UTijU'*.8) in terms of its components vt and the basis kets  /). Both definitions are of course equivalent.3) defines a vector operator as one whose components transform under Vj—^U^ViU as do components of a vector \V) under \V)~*R\V}.3. its rotated version  Kr> = i? V} has components 417 ADDITION OF ANGULAR V} = <i* V) = Z <W>01 J 0 =L V i J (15. equivalently. = ( / ? " % . may be written as 1*0=1 m i= 1 (15.with components f / = < /  V}.6) MOMENTA If instead we ask what R does to the basis. respond as do the tensor components t^. spanned by the nine kets m\J> I T ( 2 ) ) = t EMi><8>y> i= i j= i (15.7) Since i?7.> UViU f as do the kets ]/> under \i}^R\i). we see that vector components and the basis vectors transform in "opposite" ways. The first played a prominent role in the past and the second will play a prominent role in what follows.e.3. (k— 1). a tensor operator of rank 2 is a collection of nine operators Ty which. . (15. which. or. Of greater interest to us are objects called spherical tensor operators. Equation (15..) We shall call these tensors Cartesian tensors. respond as do the basis kets j / ) ® ! / ) . —k.9) One refers to ti} as the components of !T(2)> in the basis />(8>l.3. while Eq. Tensor Operators We know that a vector  V) is an element of V 3 (i?).3. A spherical tensor operator of rank k has 2k + 1 components Tqk. under Tij*UtTiJU.transform under V. A secondrank tensor I r ( 2 ) > is an element of the direct product space V 3 (/?)®V 3 (J?). i. (Note that a vector may be viewed as a tensor of rank 1. we find z')>/') = i?l/) where \i .q~ +k.3. Tensors and tensor operators of rank n > 2 are defined in a similar way. .5) defines it as one whose components Vt./>As in the case of vectors. .
remains invariant.13) This is because Tqk\ajm) contains only those angular momenta that can be obtained! by adding (k.10) that (Exercise 15.3.) Let us rotate the resulting state and see what happens: U[R]Tqk\jm)=U[R]TqkU\RmR]\jm) m' = YYD%Dliin\jm') q' m' (15. In other words.3.Tqk} = hqTqk ±ft[(kTq)(k±q+\)]l/zn±l (153.13) is an example of a selection rule. This allows us to say the following about matrix elements of Tqk between angular momentum eigenstates: (a'j'm'\Tqk\ajm>=Q unless k+j>j'>\kj\.. so  a ' j ' m ' y is orthogonal to Tqk\jm) unless ( / . Thus„ when we act on a state with T'j. they are also called irreducible tensor operators. m') m one of the possible results of adding (k. we may deduce from Eq.1): [J±. q) and ( j .e.418 CHAPTER 10 under Tqk+UTqkU1  kg}t: respond like the angular momentum eigenkets  j = k.. Why are irreducible tensor operators interesting? Consider the effect of acting on a state  aim) with Tqk. i. For this reason. m ) .M1 Since the 2 k + 1 kets  kg} transform irreducibly. m ) . q) to the state it acts on.12) We find that Tqk\jm) responds to rotations like the product ket \kq}0\jm). t Recall that \kqy^U[R]\kq> =Z Z\k'q'Xk'q'\U[R]\kq> = K \ kq'} .3.3.3. Equation (15. starting with the tensor operator of rank 0. (15. m' = m + q (15.D("iTi (15.3. q) to the state. (Here a denotes labels besides angular momentum. By considering infinitesimal rotations. an irreducible tensor operator Tqk imparts a definite amount of angular momentum (k.m = q) = U [ R ] T l u \ R ] = Y. which transforms like 00>. Tl] = [Jz.11) I Notice that commuting a J with T\ is like letting J act on the ket \kq). q) and ( j . Let us consider some examples. we add angular momentum (k. It has only one component 7"o. so do the operators Tqk.
The selection rule for. We have t In the special case V = J. i. / f = T(Jx±Jy)/2[/2 = T / ± / 2 1 / 2 and / ? = / .4) and the above definitions. the Cartesian tensors happened to have the same number of components as spherical tensors of rank /c = 0 and 1.. 0. .15)$ Given Eq. transforms like T^.3. A Cartesian tensor of rank /jhas 3" components. discussed earlier.3. say rank 2.3.. Cartesian tensors of higher rank are likewise reducible. it is possible to form combinations of Ty such that some of them never mix with others under rotations.3. In fact 7t' = T >1/2 = vi (15.11) with k= 1. and also transformed irreducibly.16b) unless j+ 1 >j" >\j Once we go beyond rank 1. Consider the concrete example of Rq\. (15. Eq. i. m — m' (15. respectively.e. Let us now return to the selection rule. There is one combination that is invariant. .16a) unless j+l>j'>\j\\.e. 0. m m (15. (15. For « = 0 and n= 1. Here we have three objects that go into each other under rotations. The tensor Ty has nine components which transform reducibly. (15.3.3. (15. say.15) like T\\ and finally there are five that transform like T\.3.3. the position operator in spherical form. Since a vector operator V also has three components that transform irreducibly (why?) into each other. V^ajm) 1. it may be readily verified that V f l and V°i obey Eq.13). it is no longer possible to express Cartesian and spherical tensors of the same rank in terms of each other. q = ±1. whereas a spherical tensor of rank k has (2k + 1) components. But consider higher ranks. — 1). Our selection rule tells us that (a'j'm \To\ajmy f 419 ADDITION OF ANGULAR MOMENTA =0 unless /=/'. The tensor Tf has five components that transform irreducibly.14) Consider next Tqk (q = 1. there are three combinations that transform like a vector or in light of Eq.Thus Tq is just a scalar operator S. we conjecture that some linear combinations of the vector operator components should equal each Tqk. We will see what these combinations are when we study the degeneracy of the isotropic oscillator of a few pages hence. We can go a step further and relate the nonvanishing matrix elements. Vx is Vx~V\ ——— (a'j'rri Vx\ ajm> = (a'j'm'\ 0 and likewise (a'j'm'\ Vz\ajm} = <a'j'm'\ =0  ajm} m' = m± 1 (15.
(12.3. a' and A.1.11) follows from Eq. which is essentially the angular integral (up to a factor independent of mi.10) when one considers infinitesimal rotations.y da (15. 1 0 ) = [ j / ( j + 1)] 1 / 2 show that (2) Using J• A = JZAZ + k(JA+ + J+A) (where A± = AX± iAy) argue that (a'jm'\3'k\ajm') = c(a'j\\A\ aj) 4 where c is a constant independent of a. The former is expressed entirely in terms of the C G coefficients.3.3.3. (15. m 2 . It separates the dependence of the matrix element on spatial orientation (on m 2 . (15.h(r) YT^O.3.420 CHAPTER 15 (assuming no spin. and q). (15. say.3. ^ Exercise 15. (1) Using <jj\jj. We have al» used Eq. This example illustrates a general result (not proven here): » f (a2j2m2\Tl\aljtml) = (ajil \Tk\ «i7i> • O ^ f e y ' i m . the v direction and then generalize the result to the other directions.17)t = (a2h\\Rx\\<*xh> where <02/2! I^il ai/>. [Hint Exercise 15. verify that this is just S*T. which appear only in the CG coefficient.3.42) to obtain R*.11).3. the reduced matrix element. m \ . (r) is the radial part of the wave function. (1) Show that Eq. 1 . (15.3.15) obeys Eq.2.\ )qSqkTi~'.5. T n JU+1) <a'jm\J'A\ajm> ' < M V W (15.3. and q. Show that c = H[j(j+ ] )]l/2Sm^. (j>)r2 dr dCl = ^R*2hrRaxhr2 dr • J Y^Y'lY'll1 ihm2\\q.18) This is called the WignerEckart theorem.: t Note that R?x is the tensor operator and Ra. so J = L) (a2l2m2\K{\ad\M\)' = J l ^ r ) Yt{6.) (2) Verify that the spherical tensor V\ constructed out of V as in Eq. ) (15. It is claimed that ( . (2) Prove it in general by considering its response to * a rotation.  (1) For k= 1. show that < a M A < \ a j m > ^ U J ' 2Z . is independent of mum2. Pick SO along.) is a scalar operator.3. and q) f r o m the rest. Exercise 15. (3) Using the above. <f>)r(~j Y\Ra.hm. (Hint: D<$l = <kq'\I(i5QJ)/ti\kq).
.26 nuclear magnetons. 2 ( r i .3 (15.19). 15. 0y. 1=1. Consider for example the (21+ l)fold degeneracy of the different m states at a given I in both these problems.4.* Show that <ym T^ym) = 0 if k> 2j.* (1) Consider a system whose angular momentum consists of two parts Jt and J2 and whose magnetic moment is H = YlJl + Y2>l2 In a state \jm. They are all conserved. (2) If we write this relation in infinitesimal form.. We explain it in terms of the rotational invariance of the Hamiltonian as follows: (1) For every rotation R(9) on V3(R) which rotates the vector operators there exists a unitary operator £/[/?] U*V.3) where L. Explanation of Some "Accidental" Degeneracies In this section the degeneracy of states of different / at a given value of n in the hydrogen atom and the isotropic oscillator (see Section 12.Vj j (15.4. rotations are symmetries of H. Exercise 15. R.6) in a 2 / > 1/2 state and show that ±0.2.e.6) will be explained. that < / O = </iy> = 0 Yi + 72 .4. This is the case for the two problems in question. L(] . F o r every free parameter that defines a rotation ( 0 X .72) jiiji 2 + O J2U2 + 1) j(j+1) 421 ADDITION OF ANGULAR MOMENTA  (2) Apply this to the problem of a proton (g = 5.3. L etc.3.2) i.4. we find [H.4.1) If the Hamiltonian depends only on the "lengths" of various vector operators like P.0.jj2y show. using Eq. (3) For an electron in a zPt/2 state show that <//> = ±5 Bohr magnetons. (15. and 0Z) there is a corresponding generator. are the generators of rotation.Exercise 15. .U=ZRi. But first let us decide what it means to explain any degeneracy. then it is rotationally invariant: UtHU=H (15.3.5.
the RungeLenz vector. Step (3): Construct an operator f r o m these generators that can change / by one unit in the case of hydrogen and two units in the case of the oscillator.5) Since [L.hl} = c\n.4) (15. / + 1).422 CHAPTER 15 (3) F r o m the three generators we construct the operator L.~ Lx — iLy which lowers the m value: L\l.4. The additional symmetry. Unfortunately the only obvious symmetry of the Coulomb Hamiltonian is rotational invariance. Step (3). JV {] = 0) but with higher angular momentum: TV\\nliy behaves as  l l > ®  / / > =  / + 1 .8) .4.7) N°i = Nz Consider the state nliy of the Hatom. So N\\n.) Let us try to d o the same for the two problems in question. This explains the degeneracy in m. N o w we have seen that the Coulomb problem admits an extra conserved quantity. Since we wish to talk about angular m o m e n t u m let us write N in spherical f o r m : ™ i ~ 2 TiTi— (15. Step (2): Find the generators of the symmetry transformations. the lowering operation does not change the energy. starting with the state of highest m at a given /. Acting on it with A^l.4. We will not discuss it. is very subtle and clearest in m o m e n t u m space.my = c\l.4. H] = 0.6) must be the generators of the additional symmetry transformations (or linear combinations thereof). for. 1+1. (We can equally well work with L+. Hydrogen * Steps (1) and (2). We follow these steps : Step (1): Identify symmetries of H besides rotational invariance.ml) (15. the one we are after. Thus the three components of R N=— (PxLLxP) 2m — (X +Y2 2 + Z2)x/2 (15. we get another state of the same energy or same n (since [H. But how then do we go to step (2)? The answer lies in the fact that the generators of the symmetry are conserved quantities.4. we can go down all the way to the lowest m without changing the energy.1+ 1) (15.
4. For example. o j ] = Sy . whose components are complex conjugates of those of a: a 1 (Ipcoh) 1/2 (fifi>RiP) (15. and z oscillators. 0) 3 1 1 3 L 0 3 2 1 1 3 2 (2) The Oscillator Step (1). Let us now rewrite H in a way that reveals the extra symmetry. and using L we can connect all the m states at a given /. 0.9) We say H is rotationally invariant because it depends only on the lengths squared of the (real) vectors P and R.(It will turn out that c vanishes when l=lm= 1. let us look at it again: p2+ p2 + p2 i Fy tz H= +n(D2{X2+ 2p x Y2 + Z2) (15.11) The components of a and a f are just the lowering and raising operators for the x. To find the extra symmetry of H. They obey [at.10) K I and its adjoint. Define a complex vector (operator) whose real and imaginary parts are proportional to R and P : 1 a (2 pcDft)l/2 (pcaR + iP) (15.) Using N! we can connect all the different / states at a given n. y. at n = 3 the network that connects degenerate states is as follows: 0 423 ADDITION OF ANGULAR MOMENTA 3 2 2 3 'N (3.4.4.
15) where £2 is a 3 x 3 Hermitian matrix. under unitary transformations in V 3 (C)..j = x. How many generators of U[C] are there and what are they? The answer to the first part is the number of parameters that define a rotation in V3(C). Consider first the combination Tr T= Txx + Tyy + Tzz = aUx + + = af • a (15.4.e.424 CHAPTER 10 In terms of a and a f .4.12) . i. Thus we find that H is a function of the length squared of a complex threedimensional vector a. i. So it is invariant under "rotations" in V 3 (C). y.14) Step (2). .. ^ t We should really be calling these U. It is clear that in the oscillator case. Just as we denoted the rotations in V 3 (i?) by R.16) are conserved. Ut[C]HU[C] =H (15.17) This is clearly a scalar.4.13) where Q are matrix elements of the unitary operator C in V 3 (C). the number of independent parameters in a 3 x 3 unitary matrix C. Since # depends only on the norm squared of a.e. y. let us call these C. To see what impact Ty has on I degeneracy. What are they? Rather than deduce them (as we did the Vs by considering the effect of infinitesimal rotations on i//) we write down the nine conserved quantities by inspection. we must decompose Tl: into its irreducible parts. transforms like The fact that it commutes with// does not explain the degeneracy in / because it "carries" no angular momentum. the nine operators Ty = a]dj (i. But that will complicate the notation.4.e. It is easy to see that Q has three real diagonal elements and three independent complex offdiagonal elements. Thus it depends on nine real parameters.. i. or z) (15. there will exist Hilbert space operators U[C] which rotate the complex vector operator a: a^a'i = Uf[C]afU[C] = £ j Q a y (15.X For every "rotation" C (unitary transformation) in V 3 (C). So there are nine conserved generators. In fact a f • a is just H up to a scale factor and an additive constant. and z oscillators have the same co (isotropy). Now any such matrix can be written as C=e*x (15.4. The proof is simple: aj destroys a j quantum and a• creates an i quantum and this leaves the energy invariant since the x. H=ftco( a f a + 3/2) (15.
a and only one vector a* x a. Since Ql behaves like 22> and since ]22) = 111 > ®  11 >. then behaves like the direct product of (linear combinations) of two spin1 objects.4.4.Tyx = a\ay . Now a and a are vector operators f r o m which we can form the tensor operators and (af )q] which behave like 11.21) Other components of 0 may be constructed by similar techniques. (It is just a matter of adding angular momenta 1 eg) 1 to get 2. (15. If you do. that is to say. One usually refers to this object as the quadrupole tensor 0 . So we are driven to the conclusion that the five residual degrees of freedom are linear combinations of some T\.1 . (15. you must work out any one of the components of the operator V = a f x a in terms of R and P.a\ax = ( a x a ) . we deduce that Ql = (J)\(a\) _ (al + iay\Iax + iay\ { 2 l / 2 ^ 2 . All we need here is the component Ql. Why? Suppose the contrary is true. This would mean that states differing by one unit in / are degenerate. We argue that these must transform irreducibly. There seems to be a problem here. The product a]a.Consider next the three antisymmetric combinations Txy .) Let us explicitly construct the operator Ql in terms of a]aj to gain some experience.4. and the paradox will be resolved.l+2) (15. We are now left with 9 . since Ql\nll) = c\n.1+1) (15. q). TyzTzy = tfxa)x (15.18) t 425 ADDITION OF ANGULAR MOMENTA TzxTxz = (^yfn)y These clearly transform as a vector V = a f x a .19) is really zero. But we know that given two vectors a f and a we can form only one scalar. But we know from Section 12. The only possibilities are tensors with 1 or 3 components. / 2 ) = i[alax ~ a]ay + i(alay + (15.8).4.19) as in Eq.3 = 5 degrees of freedom out of the original nine Ty s. Then it must be possible to form irreducible tensors with fewer than five components out of these residual degrees of freedom.20) which explains the degeneracy in / at each n. So how do we get out of the fix? T o find out.l+2.6 that states differing by two units in I are degenerate. you will see that c in Eq.) Starting with the smallest value . a + . c vanishes. Suppose we form the operator V\ = (Vx + iVy)/21/2.4. both of which we have already used up. Then we expect V\\nll} = c\n. (When l=n = lmax. 1+1. scalars or vectors.4.
Here again we have a case where states with different /.24 Uo(j>) = C t ( x + iy) . 0 or 1). we can move up in / using \kW) = c(P+)!\my 0 V 2 ) where c is some normalization constant.t which can raise I and m by one unit.23) * where p = kr. however. up to a scale factor (—21/2) which does not change its rotational properties. not "accidental.t* 426 CHAPTER 15 of / at each n (namely.4. The FreeParticle Solutions We examine the freeparticle solutions from Section 12. are degenerate at each energy E= ti2k2/2p. Recall that in the coordinate basis it was easy to find IfcOOWax^^^yg (15. namely." since the extra symmetry of the freeparticle Ham& tonian. Thus. 0 4 4 4 4 2 2 L / (4 0 0) 4 4 2 1 4 4 ' (4) 2'(2) This completes the explanation of the degeneracy of the oscillator. in fact an infinite number of them.— p dp • P J .4.w • .4. . is obvious. we can move up in steps of 2 until we reach at which point c in Eq. 1 d t P+ = Px + iPy is. (15.6 in the light of the preceding discussion. translational invariance. The network for n = 4 is shown below. just P). given the state with l = m = 0. and U0(p) is sin p or —cos p (regular or irregular solutions). We therefore have a conserved! vector operator P from which we can form P * .20) will vanish. This degenetlj ' acy is. It is eas\ to verify that P+\k00} coordinate basis • — ifi(x +iy) 1 r d dr UQ(P) yO * 0 L P J (15.
¥ where C.6. If we operate once again with P+ and use [ / V .4.29) if we set ft = ( .27) \p dpj p \p dp} This agrees with Eq. has absorbed all the factors that have no p dependence.f ) ^ \p dp} p . .25) ( P j W ^ y t M . we get 421 ADDITION OF ANGULAR MOMENTA (pjvm^cAx+v?(if and so finally \p dpj p ^ (15.G i x + i y A ^ .(\ d\ = Q (sin ey e " V \p dpj = QY'tp1 RtY't where U0(p) vp dp) (15. .! ) ' . /?+] = 0 (where R+ = Rx + iRyccR]).4. . .26) (15. (12.4.
Matrix Inversion This brief section is included only to help you understand Eq. i. (1.e. B.5) in the main text and is by no means comprehensive.. A = flji + a^ + a 3 k and so on Consider next a triplet of vectors A/f = B x C B/f = C x A Cj? = A x B (A.Appendix A.l.2) . 1. Consider the inversion of a 3 x 3 matrix a} M = b\ L^i a2 b2 c2 b^ c3_ (A.8. in the notation of vector analysis (which we will follow in this section). 1. and C.1) The elements of M have been named in this way rather than as Mijt for in the following discussion we will treat the rows as components of the vectors A.
1.4) ccR_ Now all three diagonal elements are equal: A'A/f = A ' ( B x C) = B*(C x A) = B'BJR = C .6) (We shall follow the convention of using two vertical lines to denote a determinant. AA/f^O.3) \ If we construct now a matrix M (called the cofactor transpose of M) whose columns are the reciprocal vectors.) Hence the inverse of the matrix M is given by M det M M (A. for if C = aA + /JB then A . is a linear combination of the other two. B.( B x / 3 B ) = B .5) where the last equality follows from the fact that the cross product may be written as a determinant: i B x C = b.( B x a A ) + A . 1.656 APPENDIX which are said to be reciprocal to A. ( A x B ) = C . C\ j b2 c2 k h C3 (A. 1. and C.( « A x A) = 0 . (a*) 1 (bx) 1 M = (ajt) 2 Jaxh then A* A* M •M = BA* AB* BB* C'B/f AC* B*CR A'AR 0 0 0 BB/f 0 0 0 (bR) 2 (c/f)i (cR)2 (cR)3m = (A. 1. In general. AB / e =A*C J R=0 and cyclic permutations (A. C / f = det M i (A. say C.( B x C ) = A .7) When does det M vanish? If one of the vectors.1.
+ 2 x • • • Am x A! x • • • A y ! (A. .8) ••• (fl«l) n (a. with real or complex elements. show that 2 0 1 and "2 4 1 3" 1 2 4 5 J_ 8 4 12 4 . whose n rows may be identified with n vectors.2 1 8 2 1 3" 1 2 1 1 1 2 2 5 1 3 1 4 2 — 0 . A„. If we have a matrix M.9) One tricky point: the cross product is defined to be orthogonal to the vectors in the product with respect to an inner product and not AB^AtB. . . If the matrix is used to represent three simultaneous equations. A„/f. The method can be generalized for inverting n x n matrices.Thus the determinant vanishes if the rows of the matrix are not linearly independent (LI) and vice versa. Exercise A. Using the method described above. . .. and so on.1 2 . A 2 . for the vectors A i .. . just like its threedimensional counterpart. it means not all three equations are independent. . One defines a cross product of n — 1 vectors as i AI x A 2 X • • • A M _ . .1. .1. .)i (fl»l)l ifln1)2 The resulting vector is orthogonal to the ones in the product. changes sign when we interchange any two of the adjacent ones. = 657 APPENDIX j (01)2 k . where AJR = Aj+l x A. then the cofactor transpose has as its columns the reciprocal vectors Aj/f. 1. A]. .. A„ are fictitious objects that enter a mnemonic and not the elements of the space V"(C) on which the operator acts. even when the components of A are complex. (A. There is no contradiction here.1.
) The action of A is then AK') = K>. for acting on  K'). we can think of every \ V ) in V" as arising from a unique source  K> in V" under the action of Q. Upon pulling out fi. > = 0> i with not all at = 0. Let  F .E. we see that every \V)  F ' ) = fi in V" may be written as vy where  F ) is unique. . . If QK> = 0> implies K> = 0> then fi 1 exists. fi V„) is also a LI basis.1. Define an operator A whose action on any vector  V ) in V" is to take it back to its unique source  K>. ( Vn) be a LI basis in V . because it is linear. In other words. let us assume the contrary. We may identify A as the inverse of fi. it would not know whether to give  V{) or  K 2 ). ..D..... that there exists a relation of the form i with not all a. = 0. Proof.658 APPENDIX Theorem A. (If the source of  Vy were not unique—say. To see this. when combined with the assumed property of fi. which is not true. . So we can conclude that every vector  V) in V" may be written as a unique linear combination in the new basis generated by fl as i In terms of  V) = £ at\ Vt). 1. .e. implies that 2 > i  K . i. ) . A = fi_1 or since for any j F'> in V" Afi = / where  V'y = Q\ Vy A vy=Afi vy = \vy Q. we get which. Then another LI basis is generated by the action of fi. because there are two vectors  V{y and  V2) that are mapped into  K'> by Q—then we could not define A. fi Vt).
1) This integral cannot be evaluated by conventional methods. Consider next ~ax2+ftxdx (A. Equations (A. * 0 In 0 Il(a) = •'O = nja Therefore IQ (a) = {7i/a) 1/2 J earpdpd(f> 0 (A.4) . Gaussian Integrals We discuss here all the Gaussian integrals that we will need. a> 0 (A.A.2) and (A.2.2. The trick is to consider I* oo e~ax2dx\ I oo I oo * * e~ ay2 ti(a) 2/„\ _ dy = o oo o Switching to polar coordinates in the xy plane.2.2. »O O 'j /2(a) = JCO x e~ dx=— ^ 2 ax2 Jx: da 2a W (A.3) are valid even if a is purely imaginary.2.2.2) By differentiating with respect to a we can get all the integrals of the form /* oo hn(a) = JC2" <Ta*2 dx For example.3) The integrals / 2 „+i(a) vanish because these are integrals of odd functions over an even interval — oo to +oo. Consider 659 APPENDIX Io(a) = e~ax dx.2.
2. we get ( I0(a.p) =e  \'/2 (A.1.d / d a ) n .5) These results are valid even if a and fi are complex. x and y. Complex Numbers A complex variable z can be written in terms of two real variables x and y.1 ) ! A.l ) 1 / 2 . Finally.3) . we obtain K' r"e~ardr o = a Consider this integral with a = 1 and n replaced by z . y = (z — z*)/2i (A. by applying to both sides of the equation f00 1 e~ardr = a the operator ( .3.3. as z = x + iy Its complex conjugate z* is defined to be z* = x~iy (A.660 APPENDIX By completing the square on the exponent. positive and integral z. and / = ( .3. as x = i ( z + z*). This defines the gamma function T(z) r {z) For real. r2"1 e~r dr T(z) = ( z . where z is an arbitrary complex number.3. provided Re a > 0.1) One may invert these two equations to express the real and imaginary parts.2) (A.
This plane is also called the complex z plane.The modulus squared of z.7) and 0 = tan ~'(y/x) (A. defined to be zz*.4.6) Each complex number z = x + iy may be visualized as a point (x.3.3. The expression z = x + iy gives z in Cartesian form.z ' ./2 (x +y2)]/2 2 J = p(cos d + i sin 0) = pe« where p= (x2+y2) Clearly \A=P (A.5) 2 2 661 APPENDIX (A. From the powerseries expansions s i n x — x — x 3 /3!+X 5 /5! cos x = 1 .3.4) by considering the modulus z = x + iy = (x2+y2y/2 y L(x2+/).iy) ~ x +y* You may verify that z = z' implies that x = x' and y=y' of z .x 2 /2! + x 4 /4! one can deduce that e'* = cos x+i sin x It is clear that e'x has unit modulus (x is real).3.4. A. equals zz* s z = (x + i y ) ( x . The polar form is (A. The is Prescription We will now derive and interpret the formula ^— = 0>~±iji8{x) x + ie x (A.1) . y) in the xy plane.
4) (x — a) T is (x .4.3) The sum of the two integrals in the limit s'>0 is defined as the principal value integral denoted by the symbol In the last term. the integral equals 2ni times the sum of the residues of the poles of / and the pole at z = is.a the pole moves from the origin to x=a a n d / ( 0 ) gets replaced by f(a) so that we may write ±in8{x — a) (A.662 APPENDIX where e>0 is a positive infinitesimally small quantity. g?(l/x).4. (A.2) Viewing this as the integral on the real axis of the complex z = x+iy plane. The principal part is not sensitive to this change of direction and is unaffected. leads to the principal value integral. which brings the pole to the origin. if / has some poles in the upper halfplane and vanishes fast enough to permit our closing the contour in the upper halfplane. and resumes along the real axis from x = s' to oo. there is no problem with the integral. (s' is another positive infinitesimal) goes around counterclockwise. Likewise. For example.4. So we prepare for this as follows. It is clear that if we reverse the sign of the s term. below the origin in a semicircle of radius s'. if we change the sign of the s term.a) It is clear that the limits on x need not be ± o o for the formula to work. The nice thing is that we can now set s = 0. It is clear that if we replace x by x . (A. . which is now in the lower halfplane. The three parts of the integration contour contribute as follows: (A. What if s—*0? N o w the pole is going to ram (from above) into our contour which runs along the xaxis. we simply drop the contribution from the explicit pole.1) is a compact way to say all this.is) leads to two terms: the first. we see that the integrand has an explicit pole at z = is in addition to any singularities / might have. We assume / h a s no singularities on or infinitesimally close to the real axis. leads to inf(0). we change the sign of the delta function since the semicircle now goes around the pole in the clockwise direction. we have set the argument of the smooth function / to zero and done the integral of dzjz counterclockwise around the semicircle to get in. Since the only singularity near the real axis is the explicit pole as z = is.4.4.1) is to be used inside an integral only and that inside an integral the factor 1 j(x . As long as s is fixed. which is restricted to the infinitesimal neighbourhood of the origin. we make the following deformation of the contour without changing the value of / : the contour runs along the real axis from .o o to —s'. Eq. and the second. Consider an integral of the form f ( x ) dx 1 = lim s0 „ — oo x — is (A. It is understood that Eq. inS(x).
2) since in the difference of the two integrals the contribution along the real axis cancels due to opposite directions of travel except for the part near the pole where the difference of the two semicircles (one going above and going below the pole) is a circle around the pole. This too agrees with the present analysis in terms of the integral I in Eq.4. note that according to Eq.4. (A. .Finally.4) the difference between the integrals with two signs of s is just 2nif(a). (A.
1 <L X ) = 0.8.0. L*=1>J/2 _ 1/2 \Lx=\>1/2 1 / 2 1/2 . (1) No.10. Chapter 4 4.Answers to Selected Exercises Chapter 1 5 1.1.1.3.2 (30) 1/2 © = 4> 1 (10) 1/2 (2) 1. = 1/2.0. Yes 1 0 1 (2) fi>=0> 1/2 1.2. . G)=0.2.1.(1) (2) 1.2. AL X = 1/2 1 / 2 ~l/21/2~ 0 \LX = 0 > l/2 1/2 _ (3) "1/2 1/2l/2 .8.8.A = 2. no. (1) fi> = l>\co = 2}.
This does not mean that when the vector \y/) appears as a part of a linear combination it can be multiplied by an arbitrary phase factor. .2. H]y = (P(t)yjm. (a) (b) R = (maV0)2/(fi4k2 + m2a2Vl)\T=\R T= (cosh2 2ica + a2 sinh2 2Ka) A where itc is the complex wave number for JC<A and a = (V02E)/[AE(V0E)]1/2. (1) (2) (3) (X(t)y = (ift)\[X. then. / 2 )(0> e~ia"/2+ 11) e~3. P(L^= 1) = 3/4.8 1 are physically relevant but the overall phase is not. ( n + \ / 2 ) f i / m c o . 0. />(£Jf=i) = i/4 (5) 1/2 0 = projection of  y/) on the L2Z = 1 eigenLl/2 1 / 2 space. Chapter 5 5. mco>(ma>)~x is at work here. To see this right away note that if 51 = 52 = 53 = 0.l / 4 .. 0. {P(t)y = mco\X(t)y.) \Lz=iy is physically equivalent to Although  y/)' has different coefficients from  y/) in the linear expansion. By eliminating <P> we can get an equation for <X(r)> and vice versa and solve it using the initial values <Z(0)> and <P(0)>. Yes. In our example one can only say. In summary. P(LX = 0) = l/2. <P(t)> = (mcoH/2)1/2 sin cot (n+l/2)fi. e. 1 ( 1 / 4 + 1 / 2 ) 1/2 No. P(LZ=1) = 2/3.] The vectors  y/) and e'°\ y/} are physically equivalent only in the sense that they generate the same probability distribution for any observable. for instance.a"/2) <JT(0> = (H/2ma))1/2 cos cot.2.  ^ ) = 1L X =1) and if 5i = <53 = 0 and 82 = 71. Note that the recipe 7.4. the state changes.666 ANSWERS TO SELECTED EXERCISES (4) P(LX= l ) . <Af(f)> = <X(0)> cos cot+ [(P(Q)y/m(o] sin cot.g. [See answer to part (3).'•(S2S1) J(Si8. it has the same "direction" as 11//).4. \y/y = \Lx= — 1). (n+\/2)mcofi.4. that (6) .5. (l/2 . as you will have seen in the calculation of P(LX = 0). If Lz is measured />(L*= 1) = 1/3. the relative phases <52 — and 8 3 . Chapter 7 7.
4.1. 2i/4(COS 7T/8 + I(sin TI/&)Gx). (1) <1 1.0.4. Chapter 15 15.5. Roughly 200 MeV. Roughly 2 x 10~9 second.3. 10'°G. 13.4.6.Chapter 10 10. <1 1. (1) (2) (3) 14. E=fi2/2/^. 3~ I / 2 [ 334) +1343> +1433)] 667 ANSWERS TO SELECTED EXERCISES Chapter 12 12.2.7. V=ft2/naGr Chapter 13 13.1/4. Chapter 14 14. 1/2.2. l / 2 ( . .1.2.4. <10.2.2.6.( l / 3 ) I / 2 . (eH/2mc) tanh(eftB/2mckT)k 14.3. <1 0.5.3.5.3. 14. 2/3/l/3crx.l / 2 )  3 / 2 1/2) = (l/3)l/2 1/2 1/2(3/2 1 / 2 ) = (2/3)' 1 / 2 l/2(—1/2) 1/2 1 / 2 ) = ( 2 / 3 ) l / 2 1/2 l / 2  l / 2 l / 2 ) = . Roughly one part in a million. Roughly 1 A. (1) (2) 14.3.4.
1 > + (  ) 1 / 2  0 .3~ 1/2 0.2. 15.1. or follow from Eq.1) 10. 16. 0> + 3 _ l / 2  —1. (15.0> = 3~ 1/2  1. = 21+ 1 p 2/+ 1 Chapter 16 16.668 ANSWERS TO SELECTED EXERCISES (2)  jm) = 12.2.1.6.3.1) 1> 1) = 2~ 1/2 1. .1 > .4. obvious. / 2  1.5 x 1017 seconds or 1010 years. E(a0) = IQEo/n2 malvl/nfi2 £(flb) = 3 « a ( { f ) I / 2 Roughly 1. 1) 11. 0> = 2~ . 1> l/:l =0. / 2  —1.11). 0> + ( g ) .m2= 1 > The others are either zero.1. . —1> —2~ 1/2  —1.2. 16.m2 = 0} + 2~ \mx 2. . 1 > = 2 ~ i / 2 K = 1 .4. 16. 0> —2 _ l / 2 0.0> = 6 _ 1 / 2  1.2.
Table of Constants %c= 1973. T.511 A e#/2wc = 0. v/c = p = a T. E„ = Ei/n 2 .3 eV A a = e2/hc = 1/137.58 x 10" 8 e V / G & = 8.04 mc2 = 0.62 x 10~5 e V / K k T ~ 1 /40 eV at T = 300 K 1 e V = 1.28 MeV (m is the electron mass) ( M is the proton mass) a Q = n 2 ! m e 2 = 0. E\ = T=mvaQ = fi In higher states.511 MeV Mc2 = 938. 1 2 itnv 1 2 2 —jifica .6 x 10~12 erg (room temperature) (Bohr magneton) Mnemonics for Hydrogen In the ground state.
80 electromagnetic case. 373 in three dimensions. 364 ClebschGordan coefficients. 352. 25. 368 Accidental degeneracy freeparticle case. 592 Berry potential. 69 Canonical momentum. 280 Adjoint. 603 Bohr magneton. 364 Bohr radius. 263 Bound states. 445 energy quantization in. 497 Angular momentum addition of J + J . 160. 59 Complete set of observables. 92 point transformations. 139 Commutator. 319 eigenfunctions. 11 BreitWigner form. 122. 85 Centrifugal barrier. 261 Anyons. 160 Bra. 29. 13. 244. 371 Active transformations. 340 Characteristic equation. 423 hydrogen atom case. 422 Actinides. 324. 414 S + S. 271 Bosons. 389 Bohr model. 448 Born approximation timedependent. 426 harmonic oscillator case. 543 Canonical commutation rule. 20 Compatible variables. 112. 616 tunneling in. 607 spin. 534 validity of. 321 spin. 656 Coherent states fermionic. 333 eigenvalue problem of. 408 L + S. 403 commutation rules. 318 in two dimensions. 367 Basis. 94 regular. 131 Canonically conjugate operators. 33 Characteristic polynomial. 616 Balmer series. 640 AntiHermitian operators. 412 Cofactor matrix. 366 Double well. 642 oscillator. 529 timeindependent. 129 Completeness relation. 133 . 23. 636 Collapse of state vector. 97 introduction to. 179 Classical radius of electron. 551 de Broglie waves. 6 Berry phase. 27 Antisymmetric states. 26 Aharon ovBohm effect.Index Absorption spectrum. 641 Classical limit. 97 Center of mass (CM). 33 Chemical potential. 357 BohrSommerfeld quantization rule. 359. 308 Anticommutation relations. 607 BoseEinstein statistics. 84 Canonical transformations active.
362 Finestructure correction. 465 Dirac delta function. 367. 493. 54 Gauge Coulomb.Complex numbers. 496 Gaussian integrals. 180 Eigenket. 529 for Gaussian potential. 533 for hard sphere. 246 electronic. 63 eigenvalue problem for. 659 Dalgarno and Lewis method. 499 quantization of. 614 Euler angles. 559 partial. 29 Diagonalization of Hermitian operator. 64 Destruction operator. 526. 11 Ehrenfest's theorem. 634 Correlation length. 127 FermiDirac statistics. 60 derivatives of. 197 Coulomb scattering. 426 Functional. 466 Fourier transform. 62 Freeparticle problem cartesian coordinates. 278 Exclusion principle. 588 Dirac equation electromagnetic. 605 Direct product of operators. 220 Electromagnetic field interactions with matter. 85 Correlation function. 133 Derivative operator. 548 photoelectric. 125 mixed. 566 free particle. 483 Fermions. 605 Dirac notation. 63 threedimensional. 312 for translations and rotations. 363 Condon Shortley convention. 572 Degeneracy. 125 Euclidean Lagrangian. 205 Determinant. 526. 463 Dipole selection rule. 502 Dipole moment. 43 Differential cross section. 263 Field. 123 Compton wavelength. 90. 660 Compton scattering. 76 Consistency test for threedimensional rotations. 506 review of. 306. 61 integral representation of. 318 for translations. 462 Darwin term. 565 Dirac monopole. 506 Rutherford. 531 Coupled mass problem. 628 connected. 529 Dipole approximation. 44. 108 quantum explanation of. 493. 30 Eigenspace. 61 Gaussian approximation for. 81 relative. 557 differential. 549 in lab frame. 642 Fermi's golden rule. 30 Eigenvector. 496 transformations. 492 Ensemble classical. 38. 133 quantum. 494 invariance. 249 Doubleslit experiment. 310 Coordinates canonical. 270 Fermionic oscillator. 120 Density matrix. 205 Cross section in CM frame. 66 matrix elements of. 40 simultaneous. 410 Configuration space. 37 Eigenvalue problem. 333 EulerLagrange equations. 2 Filling factor. 629 Correspondence principle. 531 for Yukawa potential. 85 cyclic. 591 Finestructure constant. 46 Creation operator. 79 Exchange operator. 250 spaces. 342 . 83. 30 Einstein temperature. 151 spherical coordinates. 3 Dirac string. 264 Expectation value. 640 thermodynamics of. 77 Functions of operators. 60 definition of. 531 Cyclotron frequency. 175 Dual spaces. 94 centerofmass.
27 diagonalization of. 263 Identity operator. 554 Infinitedimensional spaces. 385 Magnetic quantum number. 86 Hamilton's equations. 265 fermionic. 67 Incompatible variables. 10 Lagrangian. 20. 369 Ideal measurement. 316 Heisenberg picture. 592 Laughlin quasihole. 359 energy levels of. 490 Hermite polynomials. 122 Identical particles bosons. 588 Laughlin wave function. 619 Hydrogen atom degeneracy of. 35 eigenvectors of. 65 Hilbert space. 3 KleinGordon equation. 40 simultaneous. 57 Inner product. 241 Multielectron atoms. 76 for electromagnetic interactions. 69 derivatives of. 640 isotropic. 19 Impact parameter. 408 wave functions of. 67 bosonic. 55 functions of. 533 Generalized force. 523 Improper vectors. 490 Hermitian operators. 27 conjugate. 189 in the energy basis. 147. 322 for harmonic oscillator. 655 Ionic bond. 18 Lorentz spinor. 553 Minimum uncertainty packet. 574 Landau Level. 587. 370 Irreducible space. 202 energy levels of. 8 Inner product space. 43 eigenbases of. 263 definition of. 4 Linear operators. 196 wave functions of. 18 adjoint of. 655 Mendeleev. 128 Induced emission. 9 Normal modes. 194 propagator for. 361 Operators. 351 quantum. 314 Matrix elements. 367 Magnetic moment. 80 Generalized potential. 642 Green's function. 185 in the coordinate basis. 88 Harmonic oscillator classical. 25 antiHermitian. 265 't Hooft. 566 Lowering operator angular momentum. 87 Linear independence. 607 Least action principle. 356 Lamb shift. 356 21cm line. 54 . 77 Legendre transform. 370 Metastable states. 418 Ising mode!. 83 fermionic. 357 Hyperfine interaction. 593 GramSchmidt theorem. 195. 52 Number operator. 205 see also Destruction operator Lowest Landau Level. 588 Lyman series. 265 normal mode problem in. 84 Geometric phase. 202 thermodynamics of. 83 Laguerre polynomial. 394 Norm. 627 Hamiltonian formulation. 207 Numerical estimates.Gaussian potential. 260. 407 Ket. 260 fermions. 521 Negative absolute temperature. 14 Grassmann numbers. 564 Kronecker's delta. 36 infinitedimensional. 260. 356. 70 for two particles. 36 eigenvalues of. 534 Gyromagnetic ratio. 20 Matrix inversion. 330 Irreducible tensor operator. 351 twodimensional. 219 threedimensional. 386 Inelasticity. 7 Inverse of operator.
193 Reduced mass. 121 Product basis. 614 phase space. 310 Runge Lenz vector. 67 Pictures Heisenberg. 478 periodic. 198 Photoelectric effect. 118 Recursion relation. 166 Probability density. 92 invariance of. 154 for harmonic oscillator. 167 of two particles. 28 Orthonormality. 639 Quadrupole tensor. 425 Quanta. 499 Photons. 18 matrix elements of. 168 Resonances. 403 Projection operator. 339 in two dimensions. 420 Reflection coefficient. 477 timeindependent. 63 inverse of. 229 statistical mechanics. 545 Particle in a box. 223 and Schrodinger's equation. 485 Schrodinger. 23 Paramagnetic resonance. 582 definition. 115. 550 Rotations generators of (classical). 110. 555 Orthogonality. 355 Scattering general theory. 96 Polarizability. 482 sudden. 205 Range of potential. 88 Phonons. 111 Poisson brackets. 147. 147. 531 Rydberg. 610 configuration space. 72 Proper vectors. 20 projection. 9 Outer product. 451 Phase shift. 316 Raising operator for angular momentum. 160 Quantum Hall Effect (QHE). 568 Pauli exclusion principle. 646 free particle. 546 Phase space. 29. 555 Scattering amplitude. 586 recipe. 111. 223 fermionic. 381 Periodic table. 67 Pseudospin. 578 for free particle. 280 Path integral coherent state. 197 Quantization of energy.Hermitian. 516 Physical Hiibert space. 100 invariance under (quantum). 624 Pauli equation. 86 Reduced matrix element. 23 Propagator for coupled masses. 225. 264 Pauli matrices. 259 Paschen series. 100 generators of (quantum). 111. 198 quantum theory of. 392 Parity invariance. 157. 22 infinitedimensional. 297 Partial wave amplitude. 367 Passive transformation. 490 interaction (Dirac). 370 Perturbations adiabatic. 211 Probability amplitude. 27 identity. 308 invariance under (classical). 615 for (classical) string. 484 Planck's constant. 116 Population inversion. 422 Rutherford cross section. 527 . 9 Orthogonal matrix. 523 of identical particles. 371 Ray. 153 for Gaussian packet. 121 Probability current density. 525 Rare earth elements. 20 linear. 22 unitary. 339 in two dimensions. 607. 360. 222 for harmonic oscillator. 560 from step potential. 316 Radial part of wave function in three dimensions. 21 product of. 51 Feynman's. 464 P operator. 28 Optical theorem. 589 Radial equation in three dimensions. 582 imaginary time. 548. 545 expansion. 395 Postulates.
417 irreducible. 39 eigenvectors of. 325. 67 inner product of.412 Triplets. 98. 138 Unitarity bound. 116. 280 Translation finite. 175. 547 Spectroscopic notation. 94 regular. 263 Symmetries classical. 425 second rank. 529 partial wave. 350 modified. 5 field of. 27 Translated state. 348 Spherical Hankel functions. 289 generator of. 279 spontaneous breakdown of. 148 Timeordering symbol. 284 point. 625 Thomas factor. 301 Time translation invariance. 2 of Hilbert. 336 Spherical Neumann functions. 279 Transmission coefficient. 264 Spontaneous decay. 468. 468 Spin statistics theorem. 116. 237 energytime. 9 orthogonality of. 348 Spin. 418 spherical. 269 StemGerlach experiment. 67 infinite dimensional. 375 Spinorbit interaction. 128 Uncertainty principle applications of.Schrodinger equation equivalence to path integral. 17 Superposition principle. 30 Transformations. 283 operator. 140 Uncertainty relation. 29. 9 outer product of. 399 Subspaces. 418 operator. 412 Vectors components of. 264 determination of. 168 Transverse relaxation time. 458 Shell. 410 Total S basis. 97. 294 Top state. 92 generator of. 521 Squarewell potential. 459. 245 physical basis of. 417 quadrupole. 405 Tunneling. 283 identity. 571 ThomasReicheKuhn rule. 57 subspace of. 517 Spontaneous emission. 458. 651 Timereversal symmetry. 548 Unitary operator. 39 Variational method. 67 Vector operator. 319 cartesian. 29 active. 117 Symmetric states. 405 Trace. 164 Stark effect. 280. 143 timeindependent. 29. 229 timedependent. 2 basis for. 98 quantum. 27 eigenvalues of. 415 Spherical Bessel functions. 99. 292 in quantum theory. 6!6 Twoparticle Hilbert space. 395 Triangle inequality. 247 Uncertainty. 198 derivation of. 335. 373 Spinor. 459 dipole. 280 Translational invariance implications of. 370 Singlet. 429 Vector addition coefficients. 25 proper. 620 Tensor antisymmetric (£ijk). 459 general. 147. 405 S matrix definition of. 457 Timeordered integral. 98 passive. 100. 280 canonical. 313 Vector space axioms for. 484 Schwartz inequality. 465 Stationary states. 6 dimensionality of. 145 Schrodinger picture. 97 unitary. 6 improper. 348 Spherical harmonics. 8 norm of. 16 Selection rule angular momentum. 146 Statistics. 633. 17 575 INDEX . 417 Thermal wavelength.
645 WignerEckart theorem.676 INDEX Virial theorem. 212 for hydrogen. 445 introduction to. 359. 108 Wick's theorem. 198 . 398 Zero point energy. 112 plane. 121 Waveparticle duality. 420 WKB approximation and bound states. 531 Zeemann effect. 113 Waves interference of. 68 Yukawa potential. 471 Wave functions. 108 matter. 444 X operator. 435 and path integrals. 449 and tunneling. 68 matrix elements of. 438 threedimensional.