# Lectures on Classical Mechanics

by John C. Baez notes by Derek K. Wise Department of Mathematics University of California, Riverside LaTeXed by Blair Smith Department of Physics and Astronomy Louisiana State University 2005

i

c 2005 John C. Baez & Derek K. Wise

ii

Week 4 (Apr. the weekly lectures are outlined here. and derived the Hamiltonian approach from that. While not yet the book I’d eventually like to write. The prehistory of the Lagrangian approach: D’Alembert’s “principle of least energy” in statics. I’ve taught this course three times recently. 30. Week 3 (Apr. Generalized momentum and force. From Lagrangians to Hamiltonians. momentum and angular momentum.ucr. (3) A special-relativistic free particle: two Lagrangians. Examples of conserved quantities: energy. Blair Smith from Louisiana State University miraculously appeared and volunA teered to turn the notes into L TEX . 28. 4. 11. This approach seems more coherent. 22)—More example problems: (4) A special-relativistic charged particle in an electromagnetic ﬁeld in special relativity. but I’ve merged weeks together. (4) A special-relativistic charged particle in an electromagnetic ﬁeld. and sometimes split them over chapter. continued. Week 2: (Apr.iii Preface These are notes for a mathematics graduate course on classical mechanics at U. with a mass hanging on the string. which can be found on my website: http://math. Week 5 (Apr. 8)—Deriving the Euler-Lagrange equations for a particle on an arbitrary manifold. 27. (5) A general-relativistic free particle. with a heavy emphasis on action principles. In 2005 I started with the Lagrangian approach. 6. and how D’Alembert generalized his principle from statics to dynamics using the concept of “inertia force”. 25. Kaluza-Klein theory. Apr. The ubiquity of geodesic motion.edu/home/baez/classical/ Later. Week 1: (Mar. Riverside. to obtain a more textbook feel to these notes. 15)—Example problems: (1) The Atwood machine. Noether’s theorem on conserved quantities coming from symmetries.C. 20. the result may already be helpful for people interested in the mathematics of classical mechanics. . one with reparametrization invariance as a gauge symmetry. Twice I focused on the Hamiltonian approach. Fermat’s “principle of least time” in optics. For reference. A The chapters in this L TEX version are in the same order as the weekly lectures. 1)—The Lagrangian approach to classical mechanics: deriving F = ma from the requirement that the particle’s path be a critical point of the action. Derek Wise took beautiful handwritten notes on the 2005 course. 18. (2) A frictionless mass on a table attached to a string threaded through a hole in the table. 13. 29)—How Jacobi uniﬁed Fermat’s principle of least time and Lagrange’s principle of least action by seeing the classical mechanics of a particle in a potential as a special case of optics with a position-dependent index of refraction.

Weil’s formula. The Hamiltonian version of Noether’s theorem. 6)—From Lagrangians to Hamiltonians. Hamilton’s equations. The cotangent bundle as phase space. Week 9 (May 23. Hamilton’s principal function and extended phase space. 5)—A taste of geometric quantization. Darboux’s theorem. 18. 13)—Waves versus particles: the Hamilton-Jacobi equation. How the Hamilton-Jacobi equation foreshadows quantum mechanics. 4. . The Schr¨dinger picture versus the o Heisenberg picture in classical mechanics. 25. please email me! I thank Sheeyun Park and Curtis Vinson for catching lots of errors. 11. K¨hler manifolds.iv Week 6 (May 2. Getting Hamilton’s equations directly from a least action principle. continued. Poisson algebras and Poisson manifolds. Week 10 (June 1. 27)—Poisson brackets. a If you ﬁnd errors in these notes. Week 8 (May 16. Hamilton’s equations on a symplectic manifold. Liouville’s theorem. 3. 20)—Towards symplectic geometry. Week 7 (May 9. The canonical 1-form and the symplectic 2-form on the cotangent bundle. A Poisson manifold that is not symplectic. Regular and strongly regular Lagrangians.

. . . . . . . . . . . . .1. .1.2 D’Alembert’s Principle and Lagrange’s Equations 1. . .2 Disk Pulled by Falling Mass . . . . .2. . . . . .1 The Principle of Minimum Energy . . . . . . . . . . . . . . . . . . . . . . .1 Gauge Symmetry and Relativistic Hamiltonian .2 Space Translation Symmetry . . 3. . . .2. . . . . . . . .3 Conserved Quantities from Symmetries . . . . 3. . . . 3. . . . 2 Equations of Motion 2. . . . . . . . . . . . . . . . .2 Interpretation of Terms . . . . . .5. . 3. . . .1 Lagrangian versus Hamiltonian Approaches . .1 Time Translation . . . . 2. . 2.1 Comments . .4. . . . .2. . . .4 Example Problems . . . . . . . . . . 1. . . . . . . . . . . . . . . . . . . .3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. . . . . . . . . . . . .3 Rotational Symmetry . . . . . . . . . . . . . . . . . . . . .2 Prehistory of the Lagrangian Approach . . . . . . . . . 1.2 Lagrangian Dynamics . . . . . . . . . . . . . . . . 3 Lagrangians and Noether’s Theorem 3. . .1. . . .4. . . . . . . . . 3. . .2. . 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 Relativistic Hamiltonian . . . . . . .1 The Atwood Machine . . . . . . . . . . . . . . . . . 3. . . . . 1. . . . . . . . . . .3 The Principle of Least Time . . .1 The Euler-Lagrange Equations 2. . 2 2 6 6 7 8 10 12 14 14 16 16 18 20 20 21 22 23 24 25 25 26 28 28 29 31 35 35 36 . . . . . . . . . . . v . .5 Electrodynamics and Relativistic Lagrangians . . . . . . . . . . .3. 3. . . . 3. . .2. .1 Noether’s Theorem . . 3.3 Free Particle in Special Relativity . . . . . . . . . . . . . . . . . .1 Lagrangian and Newtonian Approaches . . . . . . .4 How D’Alembert and Others Got to the Truth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. . . . . . . .3. . 3. .2 Symmetry and Noether’s Theorem . . . .1 Time Translation Symmetry . . . . . . . . . . . 3. . . . . . . . . . . . . . . . . .5. . . . .Contents 1 From Newton’s Laws to Langrange’s Equations 1. . . . . . .4. . . . . . . . . . . . . . . . . . . . . . .1. . . . . . . 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Canonical and Generalized Coordinates . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . 3. . . . .1 Lagrangian for a String . . . . . . . . . . . . . . . . . . . . 3.2 Example: General Relativistic Particle in an E-M Potential . .1 Wave Equations . 4. . . . . . 3. .9. . . . . . . . . . . . . . . 4. . . . . . .2. . .CONTENTS 1 3. .2 Hamilton’s Equations from the Principle of Least Action .9. . . . . . . . . . . . . 4. . . .3 Hamilton’s Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. .2 Regular and Strongly Regular Lagrangians . . . . . . . . . . . . . . .1 The Hamiltonian Approach . . . 4. . . . . . . . . 3. . 4. . .8. .1 Example: A Particle in a Riemannian Manifold with Potential V (q) 4. . . . . . . . . . . . . . . . . 3. . . .2. . . . . . . .3. . . . . .2. . . . .7. . . . . . .4. . . . . .2 Alternate Lagrangian for Relativistic Electrodynamics 3. . . . . . .2 Charged particle in EM Field in GR . . . . . . . . . . . . . . . . . . 37 39 39 41 43 44 45 46 46 48 51 51 54 54 55 55 55 56 57 59 61 61 63 4 From Lagrangians to Hamiltonians 4. . . . . . . . . . . . . .1 Hamilton and Euler-Lagrange . . . . . 3. .4. . . . . . .1 Free Particle Lagrangian in GR . . . .4 Waves versus Particles—The Hamilton-Jacobi Equations . . . . . . . . 3.9 The Principle of Least Action and Geodesics . .3 Example: Free General Relativistic Particle with Reparameterization Invariance . . . . 4. . .8. . . . 3. . . . . . . . . . . . . . . . . . . . . . . . .7 Alternative Lagrangians . 4. . . . . . . . . . . . . . . . 4. . . . . .8 The General Relativistic Particle . . . . . . . . . . . . . . . . . . . . . . . . . .2 The Ubiquity of Geodesic Motion .2. .1 Jacobi and Least Time vs Least Action . . . . . . . . . . . . . . . . .7. . . . . . 4. . . . .3. . .4 Example: A Regular but not Strongly Regular Lagrangian . . . . . . . . . .6 Relativistic Particle in an Electromagnetic Field . . . . . .2 The Hamilton-Jacobi Equations . . .

) Classical mechanics is a very peculiar branch of physics. depends on time t ∈ .Chapter 1 From Newton’s Laws to Langrange’s Equations (Week 1. the Newtonian clockwork). and so forth). q : −→ n . say q. so it deﬁnes a function. So. v=q: ˙ −→ 2 n             . Given that GR and QM are much harder theories to use and apply it is no wonder that scientists will revert to classical mechanics whenever possible. So it is still an indispensable part of any physicist’s or engineer’s education. what is classical mechanics? 1. It used to be considered the sum total of our theoretical knowledge of the physical universe (Laplace’s daemon. The astounding thing is that probably all professional applied physicists still use classical mechanics. March 28. a toy model if you will.1 Lagrangian and Newtonian Approaches We begin by comparing the Newtonian approach to mechanics to the subtler approach of Lagrangian mechanics. atomic structure. 30. April 1. Its position. but now it is known as an idealization.1) wherein we consider a particle moving in n . From this function we can deﬁne velocity. Recall Newton’s law: F = ma (1. It is so useful because the more accurate theories that we know of (general relativity and quantum mechanics) make corrections to classical mechanics generally only in extreme situations (black holes. neutron stars. superconductivity.

a=q: ¨ −→ Now let m > 0 be the mass of the particle. |F (x)| < B for some B > 0. provided the vector ﬁeld F is ‘nice’ — by which we technically ˙ mean smooth and bounded (i. n called the (1. We can then deﬁne a quantity called kinetic energy: K(t) := This quantity is interesting because d K(t) = m v(t) · a(t) dt = F (q(t)) · v(t) So. Why? Because in this case we can deﬁne the total energy of the particle by E(t) := K(t) + V (q(t)) (1.5) F =− V (1. This function is then unique up to an additive constant.. for all x ∈ n ). that is. This is a 2nd-order diﬀerential equation for q : → n which will have a unique solution given some q(t0 ) and q(t0 ). This implies (by Stokes’s theorem relating line integrals to surface integrals of the curl) that the change in kinetic energy K(t1 ) − K(t0 ) is independent of the curve going from q(t0 ) = a to q(t1 ) = b iﬀ ×F = 0. dt and also acceleration. the change of kinetic energy is equal to the work done by the force.1. and goes down when you push it in the opposite direction. That is: m a(t) = F (q(t)) . we call it the potential. t1 ] → n . the integral of F along the curve q : [t0 .e. Newton claimed that the particle satisﬁes F = ma. Moreover. This in turn is true iﬀ for some function V : n → .3)       .1 Lagrangian and Newtonian Approaches 3 where q = ˙ dq . t1 1 m v(t) · v(t) 2 K(t1 ) − K(t0 ) = = t0 t1 t0 F (q(t)) · v(t) dt F (q(t)) · q(t) dt ˙ So.2) (1. A force with this property is called conservative. and let F be a vector ﬁeld on force.4)             n . kinetic energy goes up when you push an object in the direction of its velocity.

bypass the philosophy. if any. The other is to start with Newton’s principles and ﬁnd out what conditions. and derive Newton’s equations from that principle. on S(q) follow from this. (because F = − V ). note that F = ma implies d [K(t) + V (q(t))] = F (q(t)) · v(t) + V (q(t)) · v(t) dt = 0. q(t1 ) = b.8)     n with q(t0 ) = a.4 From Newton’s Laws to Langrange’s Equations where V (t) := V (q(t)) is called the potential energy of the particle. Conservative forces let us apply a whole bunch of cool techniques. and simply use the mathematics of variational calculus to show that particles follow paths that are ‘critical points’ of the action S(q) if and only if Newton’s law F = ma holds. namely: d S(qs )|s=0 = 0 ds (1. and for any curve q : [t0 . and then we can show that E is conserved: that is.6) S(q) := t0 L(t) dt From here one can go in two directions.1) that are critical points of S. We will use the shortcut of hindsight. 1.1: A particle can sniﬀ out the path of least action. we (1. To do this. t1 ] → deﬁne the action to be t 1 (1. constant as a function of time. One is to claim that nature causes particles to follow paths of least action. To see this.7)   . let us look for curves (like the solid line in Fig. In the Lagrangian approach we deﬁne a quantity L := K(t) − V (q(t)) called the Lagrangian. n qs + sδq PSfrag replacements q t0 t1 Figure 1.

forces that can be written as the minus the gradient of some potential. However. and ﬁrst noting that dqs /ds = δq(t) is the variation in the path: F = ma ⇔ d S(qs ) ds = s=0 d ds t1 t0 t1 t1 t0 1 mqs (t) · qs (t) − V (qs (t)) dt ˙ ˙ 2 = = t0 d 1 mqs (t) · qs (t) − V (qs (t)) dt ˙ ˙ ds 2 d mqs · qs (t) − ˙ ˙ ds d V (qs (t)) · qs (t) dt ds d d d d d qs (t) = ˙ qs (t) = qs (t) ds ds dt dt ds and when we set s = 0 this quantity becomes just: d δq(t) dt So. t1 ] → To show that d S(qs )|t=0 = 0 for all δq : [t0 . −m¨(t) − V (q(t)) = 0 q So.9) ds we start by using integration by parts on the deﬁnition of the action. t1 ] → n with δq(t0 ) = δq(t1 ) = 0 (1.. It is a simplifying assumption that has withstood the test of time and experiment. d S(qs ) ds t1 Note that = s=0 t0 mq · ˙ d δq(t) − dt V (q(t)) · δq(t) dt Then we can integrate by parts. δq(t0 ) = δq(t1 ) = 0. noting the boundary terms vanish because δq = 0 at t1 and t0 : d S(qs )|s=0 = ds t1 t0 [−m¨(t) − q V (q(t))] · δq(t)dt It follows that variation in the action is zero for all variations δq iﬀ the term in brackets is identically zero.1.1 Lagrangian and Newtonian Approaches 5 where qs = q + sδq for all δq : [t0 .     n with. s=0 s=0 s=0 .e. that is. i. The above result applies only for conservative forces. this seems to be true of the most fundamental forces that we know of in our universe. the path q is a critical point of the action S iﬀ F = ma.

S. Suppose we did not have the hindsight aﬀorded by the Newtonian picture. Lagrangian and Hamiltonian? They both have their own advantages. some people say that scientists should never ask ‘why’. The key is to understand the integral of the Lagrangian over time – the ‘action’. while the Lagrangian approach takes a while to get used to.1. And.6 From Newton’s Laws to Langrange’s Equations 1. The Hamiltonian approach focuses on energy. We shall see this in more detail soon. the action is minimized. will describe the ‘change in phase’ of a quantum system as it traces out this path. “Why does nature like to minimize the action? And why this action K−V dt? Why not some other action?” ‘Why’ questions are always tough. Our ﬁrst task in understanding Lagrangian mechanics is to get a gut feeling for what the Lagrangian means. t1 S= t0 K − V dt so that slight changes in its path do not change the action (to ﬁrst order). peeking ahead to quantum mechanics. . Often. though not always. and small when most of the energy is in potential form. This seems too extreme: a more reasonable attitude is that we should only ask a ‘why’ question if we expect to learn something scientiﬁcally interesting in our attempt to answer it. The Lagrangian L = K − V is big when most of the energy is in kinetic form. The Lagrangian approach focuses on the Lagrangian. the quantity exp(iS/ ). Why do we need two approaches. In short. while the Lagrangian approach focuses on position and velocity. which is a function of position and momentum — indeed. which is a function of position and velocity. so energy can slosh back and forth between kinetic and potential forms. 1. Then we might ask. Indeed.2 Prehistory of the Lagrangian Approach We’ve seen that a particle going from point a at time t0 to a point b at time t1 follows a path that is a critical point of the action. note that total energy is conserved. so this is called the Principle of Least Action. the Hamiltonian approach focuses on position and momentum. where is Planck’s constant. In the simplest terms. There are certainly some interseting things to learn from the question “why is action minimized?” First. it provides invaluable insights into classical mechanics and its relation to quantum mechanics. ‘Hamiltonian’ is just a fancy word for energy. but before launching into the history of the Lagrangian approach may be as good a time as any. We shall see that this describes the ‘total amount that happened’ from one moment to another as a particle traces out a path. In later chapters we will describe another approach to classical mechanics: the Hamiltonian approach.1 Lagrangian versus Hamiltonian Approaches I am not sure where to mention this.

as in Fig. Historically there were two principles that were fairly easy to deduce from observations of nature: (i) the principle of minimum energy used in statics.1 The Principle of Minimum Energy Before physicists really got going in studying dynamical systems they used to study statics. or in equilibrium. . .2: A particle’s “lazy” motion. but isn’t yet — that’s what the word ‘potential’ means. and we can think of it as doing this in order K − V small here. The rock traces out a parabola.2 Prehistory of the Lagrangian Approach 7 Kinetic energy measures how much is ‘happening’ — how much our system is moving around. to minimize its action. 1.2. So. . This is okay if we don’t take it too seriously. Statics is the study of objects at rest. Potential energy measures how much could happen. the higher the potential energy the less lively. let us recall these earlier minimum principles. we’re being told that nature likes to minimize the total of ‘liveliness’ over time: that is. On the other hand. By putting these together. The perfect compromise is a parabolic path! Here we are anthropomorphizing the rock by saying that it ‘wants’ to minimize its action. consider the path of a thrown rock in the Earth’s gravitational ﬁeld. There is another way to make progress on understanding ‘why’ action is minimized: history. On the one hand.) So. since this is where the kinetic energy is least and the potential energy is greatest. it will need to really rush to get up there and get back down. used in optics. . nature is as lazy as possible! For example. it wants to spend a lot much time near the top of its trajectory. if it spends too much time near the top of its trajectory. good! Spend as much time as possible here PSfrag replacements K − V big. and (ii) the principle of least time. In other words. (Imagine a big rock sitting on top of a cliﬀ. we can guess the principle of least action. So. one of the virtues of the Principle of Least Action is that it lets us put ourselves in the position of some physical system and imagine what we would do to minimize the action.2. and this will take a lot of action. with the potential to fall down. the Lagrangian measures something we could vaguely refer to as the ‘activity’ or ‘liveliness’ of a system: the higher the kinetic energy the more lively the system. Archimedes studied the laws of . bad! Get this over with quick! Figure 1. 1. Indeed.1. minimizes the action. the total action.

0.. imal work done by this motion is zero! He also claimed that the work done on the ith body is.e.3: A principle of energy minimization determines a lever’s balance. . He considered moving the lever slightly.4: A principle of energy minimization determines a lever’s balance.2. −mg) · (0.2 D’Alembert’s Principle and Lagrange’s Equations Let’s go over the above analysis in more detail. 1. −L1 dθ) = m1 gL1 dθ and similarly. He claimed that in equilibrium the inﬁnitesPSfrag replacements dq1 dθ dq2 Figure 1. Later D’Alembert understood this using his “principle of virtual work”. Now D’Alembert’s principle says that equilibrium occurs when the “virtual work” dW = dW1 +dW2 vanishes for all dθ (that is.3). This happens when m1 L 1 − m 2 L 2 = 0 dW2 = −m2 gL2 dθ which is just as Archimedes wrote. inﬁnitesimally. 1. 0. i. dWi = (0. for all possible inﬁnitesimal motions). dWi = Fi dqi and gravity pulls down with a force mi g so.8 From Newton’s Laws to Langrange’s Equations PSfrag replacements m1 L1 L2 m2 Figure 1. and he found that this would be in equilibrium if m1 L 1 = m 2 L 2 . I’ll try to make it clear what we mean by virtual work. a see-saw or lever (Fig.

It’s called ‘virtual’ because it cannot be realized: any actual displacement would occur over a ﬁnite time interval and possibly during which the forces and constraints might change. To displace a system by δr i for each position coordinate. we must deduce that the virtual work vanishes for a system in equilibrium. (virtual work is zero for δq → 0) (no force on it!) If the force is conservative (F = − V ) then this is also equivalent to. the virtual work. However. when Fi = 0. D’Alembert’s principle simply says. Fi · δri = 0. n . a number n of particles in 3 can be treated as a single quasi-particle in 3n . q(t) = q0 satisﬁes F = ma. We also have an analogy between statics and dynamics. The equilibrium will be stable if q0 is a local minimum of the potential V .10) i Note that in the above example we have two particles in 3 subject to a constraint (they are pinned to the lever arm).. Statics equilibrium. without any time interval passing is called a virtual displacement. and if there are constraints it can move in some submanifold of 3n . So ultimately we need to study a particle on an arbitrary manifold. We can summarize all the above by proclaiming that we have a “principle of least energy” governing stable equilibria.e. S = t0 K − V dt critical points of S             . vanishes for all dq ∈ (it’s in equilibrium) dW = F · dq F = 0. For a particle in n . So equal small inﬁnitesimal displacements of the system might result in the forces Fi acting on the system doing diﬀerent amounts of work at diﬀerent times.2 Prehistory of the Lagrangian Approach 9 The forces and constraints on a system may be time dependent. Consider a system in the special state of being in equilibrium. Now call the work done by this hypothetical virtual displacement.1. But. a = 0 potential. (when in equilibrium) (1. and yet remain consistent with all the constraints and forces at a given instant of time t. V (q0 ) = 0 that is. F i · δri . we have equilibrium at a critical point of the potential.. we’ll postpone such sophistication for a while. V critical points of V Dynamics F = ma t1 action. Then because by deﬁnition the virtual displacements do not change the forces. i.

∂qi (1. But this is precisely the path that minimizes the distance of the trajectory subject to the condition that it must hit the mirror (at least at one point).2. there were also hints from the behavior of light. it chooses B such that θ1 = θ2 . What path does the light take? The empirical answer was known since antiquity. ∂qi ∂L = 0. d dt d dt ∂T ∂ qi ˙ ∂L ∂ qi ˙ − D’Alembert’s principle is an expression for Newton’s second law under conditions where the virtual work done by the forces of constraint is zero. which in Euclidean space is the minimum distance. 1.12) .10 From Newton’s Laws to Langrange’s Equations unstable equilibrium unstable equilibrium V PSfrag replacements stable equilibrium Figure 1. As well as hints from statics.5: A principle of energy minimization determines a lever’s balance. hints again that nature likes to minimize eﬀort. Why is ABC the shortest path hitting   n ∂T = Qi .11) − (1. so the angle of incidence equals the angle of reﬂection. In fact light traveling from A to B takes both the straight paths ABC and AC.3 The Principle of Least Time Time now to look at the second piece of history surrounding the principles of Lagrangian mechanics. But more interesting than straight lines are piecewise straight paths and curves. Consider reﬂection of light from a mirror. In a vacuum light moves in straight lines.

2 Prehistory of the Lagrangian Approach 11 A C PSfrag replacements θ1 B θ2 θ2 C the mirror? This follows from some basic Euclidean geometry: B minimizes AB + BC ⇔ B minimizes AB + BC ⇔ A. such that. θ1 medium 1 PSfrag replacements medium 2 θ2 called the index of refraction. The big clue leading to D’Alembert’s principle however came from refraction of light. this is a trick often used in solving electrostatic problems (a conducting surface can be replaced by ﬁctitious mirror image charges to satisfy the boundary conditions). Snell (and predecessors) noted that each medium has some number n associated with it. and in hydrodynamics when there is a boundary between two media (using mirror image sources and sinks). C lie on a line ⇔ θ1 = θ2 Note the introduction of the ﬁctitious image C “behind” the mirror. B. it is also used in geophysics when one has a geological fault.1. n1 sin θ1 = n2 sin θ2 .

Someone guessed the explanation. and he postulated that in dynamics equilibrium occurs when the total force = F +inertia force.4 How D’Alembert and Others Got to the Truth Sometimes laws of physics are just guessed using a bit of intuition and a gut feeling that nature must be beautiful or elegantly simple (though occasionally awesomely complex in beauty). not only is light the fastest thing around. Or symbolically when. One way to make good guesses is to generalize. vanishes. So. ∀δq ∈ F (q(t)) − ma(t) · δq(t) = 0 qs (t) = q(t) + s δq(t) F (q) − m¨ · δq dt = 0 q t1 for all δq. We then take a variational path parameterized by s. for any function f on the space of paths we can deﬁne the variational derivative. D’Alembert’s principle of virtual work for statics says that equilibrium occurs when D’Alembert generalized this to dynamics by inventing what he called the “inertia force”=−m a. realizing that if the speed of light in a medium is proportional to 1/n. But the same is true for the law of reﬂection.2.13) . since in that case the path of minimum length gives the same results as the path of minimum time. not just the path distance. we get 0= t0 t1 − V (q) − m¨ · δq dt q − V (q) · δq + mqδ q dt ˙ ˙ = t0   n (1. d δf := f (qs ) (1.14) ds s=0 Then D’Alembert’s principle of virtual work implies t1 t0 F (q0 ) · δq = 0. In the case of refraction it is the time that is important. then light will satisfy Snell’s law if the light minimizes the time it takes to get from A to C. it’s also always taking the quickest path from here to there! 1. so if F = − V .12 From Newton’s Laws to Langrange’s Equations (having normalized n so that for a vacuum n = 1). where δq(t0 ) = δq(t1 ) = 0 and with these paths.

If you are inside the rotating system and you throw a ball straight ahead it will appear to curve away from your target. We do not really understand why one must introduce the ‘inertia force’. Still.13)).2 Prehistory of the Lagrangian Approach 13 using d V (qs (t)) ds and δ(q 2 ) = ˙ then we have 0= dV dqs m dqs (t) ˙ dt + dq dt 2 ds t0 m d t1 d ˙ dt = − V qs (t) + qs (t) ds t0 dt 2 s=0 t1 d m =δ − V qs (t) + qs (t)2 dt ˙ dt 2 t0 − t1 t1 = s=0 dV dqs (t) dq ds s=0 dqs (t) ˙2 ds therefore δ t0 −V (q) + K dt = 0 so the path taken by the particle is a critical point of the action. there’s something unsatisfying about the treatment so far. Recall from undergraduate physics that in an accelerating coordinate system there is a ﬁctional force = ma. and if you did not know that you were rotating relative to the rest of the universe then you’d think there was a force on the ball equal to the centrifugal force. We conclude with a few more words about this mystery. We use it. which is called the centrifugal force. In general relativity.15) We’ve described how D’Alembert might have arrived at the principle of least action by generalizing previously known energy minimization and least time principles. S(q) = (K − V ) dt (1.(1. for example.1. If you are inside a big rapidly rotating drum then you’ll also feel pinned to the walls. We only see that it’s necessary to obtain agreement with Newtonian mechanics (which is manifest in Eq. This is an example of an inertia force which comes from using a funny coordinate system. one sees that — in a certain sense — gravity is an inertia force! . to analyze simple physics in a rotating reference frame.

So consider an arbitrary classical system as living in a space of points in some manifold Q. and normally we call this the phase space. We’ll look at some speciﬁc examples of problem solving using the Euler-Lagrange equations. April 4.Chapter 2 Equations of Motion (Week 2.1.1 The Euler-Lagrange Equations We are going to start thinking of a general classical system as a set of points in an abstract conﬁguration space or phase space1 . 6. which means you have to disabuse yourself of the notion that we’re dealing with real The tangent bundle T Q will be referred to as conﬁguration space. For example. 1 14 . First we’ll show how the equations are derived. Q”. 2. where Q = S 2 × S 2 .) In this chapter we’ll start to look at the Lagrangian equations of motion in more depth. 2. 8. later on when we get to the chapter on Hamiltonian mechanics we’ll ﬁnd a use for the cotangent bundle T ∗ Q. the space for a spherical double pendulum would look like Fig. So our system is “a particle in Q = S2 × S2 PSfrag replacements Figure 2.1: Double pendulum conﬁguration space.

we have d S(qs ) ds We write.2) . in accord with D’Alembert’s principle of least action. Sometimes to make this clear we’ll talk about “the system taking a path”.1. but we won’t go into that for now. we are not. It is then clear that when we say.) Let the Lagrangian=L for the system be any smooth function of position and velocity (not explicitly of time. Γ = {q : [t0 .1 The Euler-Lagrange Equations 15 particles. q) dt ˙ The path that the quasi particle will actually take is a critical point of S. t1 ] → Q|q(t0 ) = a. We’ll just write is as q(t). q(t1 ) = b} (Γ is an inﬁnite dimensional manifold. for simplicity). instead of “the particle taking a path”. we are dealing with a single quasi-particle in an abstract higher dimensional space. q(t) ∈ Tq(t) Q ˙ to be the tangent vector at q(t) given by the equivalence class [σ] of curves through q(t) with derivatives σ(t) = dq(s)/ds|s=t.2.(2. “the system follows a path q(t)” that we’re referring to the point q in conﬁguration space Q that represents all of the particles in the real system. t1 ] −→ Q and we deﬁne it’s velocity. The single quasi-particle represents two real particles if we are talking about the classical system in Fig. S: S : Γ −→ by t1 S(q) := t0 L(q.1) (2.3) s=0 s=0 =0 as “δ     (2. 2. a path q ∈ Γ such that for any smooth 1-parameter family of paths qs ∈ Γ with q0 = q1 . d ds so Eq. L : T Q −→ and deﬁne the action. So as time passes “the system” traces out a path q : [t0 . In other words.2) can be rewritten δS = 0 (2. ˙ ˙ Let Γ be the space of smooth paths from a ∈ Q to b ∈ Q.

2: Schematic of a 1-parameter family of curves. so it admits a covering of coordinate charts. .2 Lagrangian Dynamics We were given that Q is a manifold. So the 1-parameter family is nothing more nor less than a set of well-deﬁned paths {qs }. or a 1D manifold. each one labeled by a parameter s.16 Equations of Motion 2. 2. So a smooth 1-parameter family of paths will have q(s) everywhere inﬁnitesimally close to q(s + ) for an inﬁnitesimal hyperreal . 2. we can go from q0 to qs by smoothly varying s from s = 0 to s = s What does the condition δS = 0 imply? Patience.1. Next.3: Local path variation.1 Comments What is a “1-parameter family of paths”? Well. A cartoon of this looks like Fig. For now. a path is a curve.3 Then Q a PSfrag replacements U b Figure 2.1. let’s pick coordinates in a neighborhood U of some point q(t) ∈ Q. we are just getting to that. So in Fig. consider only variations qs such that qs = q outside U . We will now start to explore what δS = 0 means for our Lagrangian. 2.2 qs PSfrag replacements q0 Figure 2.

We can use the coordinate charts on U . y n ) where y ∈ Tx Q. Note that we can write δL as above using a local coordinate patch because the path variations δq are entirely trivial outside the patch for U . . (Velocity and position are in the abstract conﬁguration space Q). . q(t)) dt ˙ δL(q. . y) = (x1 . xn ) and we also have coordinates for the 1-forms.2. . . using the coordinates xi . . t1 δS = t0 d ∂L ∂L − δq i (t) dt i i ∂x dt ∂y = 0. t1 ] such that qs (t) ∈ U for t0 ≤ t ≤ t1 . . further. the boundary terms vanish because we deliberately chose δq that           . Let’s just go ahead and rename t0 and t1 as “ t0 and t1 ” to drop the primes. We restrict L : T M → to T U ⊆ T M. in our case the manifold is M = Q. the y i are the associated generalized velocity coordinates. .1 The Euler-Lagrange Equations 17 we restrict attention to a subinterval [t0 . and then we can describe it. The i are generalized position coordinates. ϕ : U −→ n x −→ ϕ(x) = (x1 . y 1 . xn . q) dt ˙ t0 t1 = = t0 ∂L i ∂L i δq + i δ q dt ˙ ∂xi ∂y where we’ve used the given smoothness of L and the Einstein summation convention for repeated indices i. Using these coordinates we get t1 δS = δ t0 t1 L(q(t). If this integral is to vanish as demanded by δS = 0. then it must vanish for all path variations δq. L. . y i on T U . dϕ : T U −→ T n ∼ n × n = (x. Continuing. x2 . t1 ] ⊆ [t0 . . using the Leibniz rule d ∂L d ∂L ∂L δq = δq + δq ˙ dt ∂y dt ∂y ∂y we have. . . y) −→ dϕ(x.

d ∂L ∂L = i dt ∂ q i ˙ ∂q and they call these the Euler-Lagrange equations. That means the term in brackets must be identically zero. physicists write. q) = mq · q − V (q) ˙ 2 1 = mq i qi − V (q) ˙ ˙ 2 TERMS ∂L ∂qi ˙ ∂L ∂q i MEANING (in this example) mq ˙ −( V )i MEANING (in general) the momentum pi the force Fi V we’re assuming the q i are Cartesian coordinates on Q. q(t). indeed the only reason we used those terms was for their analogical appeal. for all δq. Q= 1 ˙ ˙ L(q. So consider. Physicists always give the coordinates xi . or ∂L d ∂L − i =0 (2. When we write ∂V /∂q i =   . despite the fact that these ˙ i i also have another meaning. but in fact it’s also suﬃcient. ˙ So in any case.2 Interpretation of Terms The derivation of the Euler-Lagrange equations above was fairly abstract. 2. namely the x and y coordinates of the quantity. the terms “position” and “velocity” were used but were not assumed to be the usual kinematic notions that we are used to in physics. Now we’ll try to illuminate the E-L equations a bit by casting them into the usual position and velocity terms.4) i dt ∂y ∂x This is necessary to get δS = 0. y i on T U the symbols “q i ” and “q i ”. q(t) ∈ T U.18 Equations of Motion vanish at the endpoints t0 and t1 inside U .

and they say that p=F ˙ (2. and then the Euler-Lagrange equations can be interpreted as equations relating generalized concepts of momentum and force. Another good reason to learn Lagrangian (or Hamiltonian) mechanics is that it translates better into quantum mechanics.5) So there’s no surprise that in the mundane case of a single particle moving in 3 under time t this just recovers Newton II. shorter solutions and more transparent analysis or at least more chance at insight into the characteristics of the system.2. it’s just a pain in the neck to deal with the redundancies in F = ma when we could use symmetry principles to vastly simplify many examples. One can of course introduce simpliﬁcations when solving Newton’s equations.2 Interpretation of Terms 19 So translating our example into general terms. if we conjure up some abstract Lagrangian then we can think of the independent variables as generalized positions and velocities. Simpliﬁcations generally mean quicker. It turns out that the Euler-Lagrange equations are one of the reformulations of Newtonian physics that make it highly convenient for introducing symmetries and consequent simpliﬁcations. it’s just that it’s easier to do this when working with the Euler-Lagrange equations. so it gets closer to the fundamental degrees of freedom of the system and so we cut out a lot of the wheat and chaﬀ (so to speak) with the full redundant Newton equations. The main thing is that when we use symmetry to simplify the equations we are reducing the number of independent variables.   . Of course we can do all of our classical mechanics with Newton’s laws.

q(t) dt ˙ typically will not converge. The rigorous statement of all this is the content of Noether’s theorem. if δq = 0 outside of some ﬁnite interval. When the form of the equations is similarly invariant under time translations then the total energy is a conserved quantity (a constant of the equations of motion).Chapter 3 Lagrangians and Noether’s Theorem If the form of a system of dynamical equations does not change under spatial translations then the momentum is a conserved quantity. then the functional variation. and then deﬁne a new space of paths. since the integral is smooth and vanishes outside this interval. 3. Time and space translations are examples of 1-parameter groups of transformations. So symmetries of a dynamical system give conserved quantities or conservation laws. δS := ∞ −∞ d L qs (t). demanding that this δS vanishes for all such variations δq is enough to imply the Euler20   → Q}. Invariance under a group of transformations is precisely what we mean by a symmetry in group theory. so S is then no longer a function of the space of paths.1   Time Translation To handle time translations we need to replace our paths q : [t0 . Nevertheless. t1 ] → Q by paths q : → Q. dt s=0 . qs (t) ˙ ds will converge. Γ = {q : The bad news is that the action S(q) = ∞ −∞ L q(t). Moreover.

and change variables from q to pi and eliminate pi . The qi are generalized positions. ∂L = pi . dt ∂ qi ˙ ∂qi Recall that. ˙ ˙ 3.1 Time Translation 21 Lagrange equations: δS = = = ∞ −∞ ∞ −∞ ∞ −∞ d L qs (t). or system of particles. So if we have N particles then we’d typically have 6N generalized coordinates (the “6” is for 3 space dimensions. .1. The full ˙ ˙ set of independent generalized coordinates represent the degrees of freedom of a particle.3. qi }. so for example. but sometimes it’s unavoidable. ∂ qi ˙ ∂L = pi . Then the whole thing vanishes for all compactly supported smooth δq iﬀ ∂L d ∂L = . and sometimes it can be eﬃcient—provided everyone is clued in). let’s spend a few moments clearing up some terminology (I hate using jargon. and the qi are generalized velocities.1 Canonical and Generalized Coordinates In light of this noted similarity with the Hamilton equations of motion. is the generalized momentum. by the E-L eqns. is the force. these are the {qi . and at each point a position and a momentum). ˙ ∂qi Note the similarity to Hamilton’s equations—if you change L to H you need to stick in a minus sign. the ﬁrst term in ∂L i d ∂L d ∂L i δq = ˙ δq − δq i i ∂q ˙ dt ∂ q ˙ dt ∂ q i ˙ vanishes when we integrate. qs (t) ˙ dt ds s=0 ∂L ∂L δqi + δ qi dt ˙ ∂qi ∂ qi ˙ ∂L d ∂L − δqi dt ∂qi dt ∂ qi ˙ where again the boundary terms have vanished since δq = 0 near t = ±∞. by defn. These can be in any reference frame or system of axes. Generalized Coordinates For Lagrangian mechanics we have been using generalized coordinates. To be explicit.

and 2 × 3 = 6 velocities. z2 }.     with q0 = q .f. u2 . The canonical coordinates are a special set of coordinates on the cotangent bundle of the conﬁguration space manifold Q. which are 1-forms in the cotangent bundle at the point q in the manifold. By judicious choice of coordinate frame we can eliminate one velocity component and one position component for each particle. later on we’ll see that the relation between the generalized coordinates and the canonical coordinates is given by the Hamilton-Jacobi equations for a system. They are usually written as a set of (q i . w2 } where say u = vx . It turns out that the q i together with the pj . v2 . If we constrain the particles to move in a plane (say place them on a table in a gravitational ﬁeld) then we get 2N fewer degrees of freedom. but if you care to know. y2 . y1 . We want to refer to symmetry transformations (of the Lagrangian) governed by a single parameter. {u1 . Deﬁnition 3. form a coordinate system on the cotangent bundle T ∗ Q of the conﬁguration space Q. overall. v1 . A 1-parameter family of symmetries of a Lagrangian system L : T Q → is a smooth map. pj ) with the x’s or q’s denoting the coordinates on the underlying manifold and the p’s denoting the conjugate momentum. {x1 . w1 . pj ) or (xi .o. 3. q) −→ qs . This makes 12 = 6 × 2 = 6N coordinates. v = vy . matching the total degrees of freedom as claimed. It is also handy to respect other symmetries of a system. z1 . maybe the particles move on a sphere for example. let’s give a useful deﬁnition that will make it easy to refer to a type of dynamical system symmetry. Canonical Coordinates In Hamiltonian mechanics (which we have not yet fully introduced) we will ﬁnd it more useful to transform from generalized coordinates to canonical coordinates.1 (one-parameter family of symmetries).2 Symmetry and Noether’s Theorem First. w = vz are the Cartesian velocity components. x2 . one can then deﬁne new positions and momenta with a consequent reduction in the number of these generalized coordinates needed to describe the system. We will not discuss this here.22 Lagrangians and Noether’s Theorem in a Cartesian frame. and so 4N d. with two particles. in 3D space we’d have the 2 × 3 = 6 position coordinates. hence these coordinates are called the canonical coordinates. F : × Γ −→ Γ (s.

1.2. But δL = dt is a sneaky generalization whose usefulness will become clear. that is.1 Noether’s Theorem in this theorem is the function associated Here’s a statement of the theorem.     qs (t). . Before this can be of any use we’d need to ﬁnd a symmetry F . Suppose F is a one-parameter family of symmetries of the Lagrangian system. d pi δq i − dt = pi δq i + pi δ q i − ˙ ˙ = d dt − q(t). big deal” you might say. in boring detail. qs (t) ˙ ds for all paths q. Then we’d need to ﬁnd out what this pi δqi − business is that is conserved. is conserved.1 (Noether’s Theorem). L : T Q → .3. a symmetry of L in the d most obvious way.e. that is. Remark: The simplest case is δL = 0. in which case we really have a way of moving paths around (q → qs ) that doesn’t change the Lagrangian—i. d ∂L d i q(s)q(s) ˙ q (t) dt ∂y i ds s Proof.2 Symmetry and Noether’s Theorem 23 such that there exists a function (q. qs (t) ˙ pi δqi − s=0 ∂L i ∂L i δq + i δ q − δL ˙ ∂q i ∂q ˙ = δL − δL = 0. Then. = s=0 d dt d dt 3. it’s time derivative is zero for any path q ∈ Γ satisfying the EulerLagrange equations.. Note that with F in deﬁnition 3. d L qs (t). Theorem 3. q) for which ˙ δL = for some : T Q → . So let’s look at some examples. q(t) ˙ =0 “Okay. In other words.

at least not to any extent that we can tell). this time with a speciﬁc Lagrangian. It doesn’t matter what the Lagrangian is.) 3. if Q = 1 ˙ L = mq 2 − V (q) 2 then this quantity is mq · q − ˙ ˙ 1 mq · q − V ˙ ˙ 2 1 = mq 2 + V (q) ˙ 2 The term in parentheses is K − V .3 Conserved Quantities from Symmetries Fs : Γ −→ Γ q −→ qs We’ve seen that any 1-parameter family   n . and the left-hand side is K + V . guaranteed. (The most important example!) All of our Lagrangian systems will have time translation invariance (because the laws of physics do not change with time. and if . ˙ δL = L for d L(qs ) ds so here we take = s=0 d ˙ L=L dt = L simply! We then get the conserved quantity pi δq i − = pi q i − L ˙ which we normally call the energy. For example. if it has 1-parameter families of symmetries then it’ll have conserved quantities. So we have a one-parameter family of symmetries qs (t) = q(t + s) This indeed gives. Let’s repeat this example. Conservation of Energy. The trick in physics is to write down a correct Lagrangian in the ﬁrst place! (Something that will accurately describe the system of interest.24 Lagrangians and Noether’s Theorem Example 1.

3. q) = 2 mq 2 − V (q). L(q. ˙ ˙ This system has ∂L pi q i = i q i = mq 2 = 2K ˙ ˙ ˙ ∂q ˙ so H = pi q i − L = 2K − (K − V ) = K + V ˙ as you’d have hoped. if you are doing Newtonian mechanics then “total energy” is appropriate. we have a 1-parameter family of symmetries qs (t) = q(t + s) because ˙ δL = L so we get a conserved quantity called the total energy or Hamiltonian.2 Space Translation Symmetry       For a free particle in n . L : T Q → . 3. we have Q = n and L = K = 1 mq 2 . This has spatial translation ˙ 2 n we have the symmetry symmetries. H = pi q i − L ˙ (3. 3.1 Time Translation Symmetry   For any Lagrangian system.3.3.3 Conserved Quantities from Symmetries 25 which satisﬁes δL = ˙ for some function = (q.1) (You might prefer “Hamiltonian” to “total energy” because in general we are not in the same conﬁguration space as Newtonian mechanics. so that for any v ∈ qs (t) = q(t) + s v     . qs (t) ˙ ds s=0 Let’s see how we arrive at a conserved quantity from a symmetry.) 1 For example: a particle on n in a potential V has Q = n . q) gives a conserved quantity ˙ pi δq i − As usual we’ve deﬁned δL := d L qs (t).

This gives a 1-parameter family of symmetries qs (t) = esX q(t) which has δL = ∂L i ∂L i δq + i δ q = mqi δ q i ˙ ˙ ˙ ∂q i ∂q ˙ now qi is ignorable and so ∂L/∂q i = 0. and ˙ δqi = ˙ d i q ˙ ds s s=0 d d sX = e q ds dt d = Xq dt =Xq ˙ s=0       .3 Rotational Symmetry     The free particle in n also has rotation symmetry. hence conserved quantities. it describes a rotation.3. (In fact that ˙ n whole Lie group G = is acting as a translation symmetry group. and we’re getting a q(= n )-valued conserved quantity!) 3. here it is a conserved quantity derived from space translation invariance. but earlier it was a diﬀerent thing. and ∂L/∂ q i = pi . that is. to change coordinates so as to make some of them ignorable if possible!) In this example we get a conserved quantity called momentum in the v direction: pi δq i = mqi v i = mq · v ˙ ˙ Aside: Note the subtle diﬀerence between two uses of the term “momentum”. Consider any X ∈ so(n) (that is a skew-symmetric n × n matrix). these ignorables always give symmetries.26 Lagrangians and Noether’s Theorem with δL = 0 because δ q = 0 and L depends only on q not on q in this particular case. namely the momentum ∂L/∂ q i = pi conjugate to q i . It is often useful therefore. (Since L does not ˙ ˙ i i depend upon q we’ll call q an ignorable coordinate. as above. then for all s ∈ the matrix esX is in SO(n). These two diﬀerent ˙ “momentum’s” happen to be the same in this example! Since this is conserved for all v we say that mq ∈ n is conserved.

since there is no potential energy variation for a free particle. The above examples were for a particularly simple system.e. skew symmetric with ±1 in the respectively ij and ji elements. we can then compare our answers with experiments. then we get m(qi q j − qj q i ) ˙ ˙ the “ij component of angular momentum”. for which the Lagrangian is just the kinetic energy.3. I mentioned earlier that we can do mechanics with any Lagrangian. We’d like to know how to solve more complicated dynamics.. the angular momentum in the X direction. L=K−V So we are not using Lagrangians directly to tell us what the fundamental physical laws should be. So we get a conserved quantity. which indirectly tells us something about the physical laws—but only provided the Lagrangian formulation of mechanics is itself a valid procedure in the ﬁrst place. a free particle. i. If n = 3 we write these as. (Note: this whole bunch of math above for δL just says that the kinetic energy doesn’t change when the velocity is rotated. where it’s ±1. But how do we do that? All this theory is ﬁne but is useless unless you know how to apply it. ˙ mq×q Note that above we have assumed one can construct a basis for so(n) using matrices of the form assumed for X. The general idea is to guess the kinetic energy and potential energy of the particle (as functions of your generalized positions and velocities) and then let. δL = mqi Xj q j ˙ i˙ = mq · (X q) ˙ ˙ =0 since X is skew symmetric as stated previously (X ∈ so(n)). pi δq i = mqi · (X q)i ˙ (δq i = Xq just as δ q i = X q in our previous calculation). instead we plug in some assumed physics and use the Lagrangian approch to solve the system of equations. without changing its magnitude. . If we like. or if X has zero entries except ˙ ˙ in ij and ji positions.3 Conserved Quantities from Symmetries 27 So.) We write. otherwise zero. but if we want to be useful we’d better pick a Lagrangian that actually describes a real system.

) × Note that solutions of the Euler-Lagrange equations will only be deﬁned for some time t ∈ .4 Example Problems (Week 3. We have PSfrag replacements x m1 m2 −x 1 d 1 K = (m1 + m2 )( ( − x))2 = (m1 + m2 )x2 ˙ 2 dt 2 V = −m1 gx − m2 g( − x) so 1 L = K − V = (m1 + m2 )x2 + m1 gx + m2 g( − x) ˙ 2 The conﬁguration space is Q = (0. ). 3. as eventually the solutions reaches the “edge” of Q. It’s vastly superior to the simplistic F = ma formulation of mechanics. ˙ here and write Q = (0. The momentum is: ∂L p= = (m1 + m2 )x ˙ ∂x ˙ and the force is: ∂L = (m1 − m2 )g F = ∂x       .1 The Atwood Machine A frictionless pulley with two masses. 13. Apr. let’s do some problems. hanging from it. 15.28 Lagrangians and Noether’s Theorem 3. and x ∈ (0. As usual L : T Q → .4. 11.) To see how the formalisms in this chapter function in practise. and allows us to easily use any coordinates we wish. x). Moreover T Q = (0. ) (we could use the “owns” symbol (x. m1 and m2 . The Lagrangian formulation allows the conﬁguration space to be any manifold. ) x ).

m1 +m2 It is trivial to integrate the expression for x twice (feeding in some appropriate initial ¨ conditions) to obtain the complete solution to the motion x(t) and x(t). The disk is free to move on a frictionless surface. ) × S 1 (r.4 Example Problems 29 The Euler-Lagrange equations say p=F ˙ (m1 + m2 )¨ = (m1 − m2 )g x m1 − m 2 x= ¨ g m1 + m 2 So this is like a falling object in a downwards gravitational acceleration a = m1 −m2 g.2 Disk Pulled by Falling Mass Consider next a disk pulled across a table by a falling mass. ˙ ∂L = (m1 + m2 )r ˙ ∂r ˙ ∂L ˙ pθ = = m1 r 2 θ. minus it’s center = (0. θ) T Q = (0. r. ) × S 1 × × 1 1 d ˙ K = m1 (r 2 + r 2 θ 2 ) + m2 ( ( − r))2 ˙ 2 2 dt V = gm2 (r − ) 1 1 ˙ ˙ ˙ L = m1 (r 2 + r 2 θ 2 ) + m2 r 2 + gm2 ( − r) 2 2 is constant so d/dt( − r) = −r. Note that x = 0 ˙ ¨ when m1 = m2 . and it can thus whirl around the hole to which it is tethered to the mass below. and x = g if m2 = 0. ¨ 3.4.3. ˙ ∂θ pr = having noted that     (r. θ. r m1 PSfrag replacements −r m2 no swinging allowed! Here Q = open disk of radius . For the momenta we get. θ) ˙ ˙ .

2. the minimum of V (r). so then we’ve hit the boundary of Q and our solution is broken. ˙ ˙ (m1 + m2 )¨ = m1 r θ 2 − m2 g r ˙ pθ = m1 r 2 θ = J = a constant. pr = F r . At that time the disk reaches the hole. The forces are. pushing m1 radially out.1 ˙ If θ(t = 0) = 0 then there is no centrifugal force and the disk will be pulled into the hole until it gets stuck. So. So integrate −Fr to ﬁnd V (r): J2 + m2 gr V (r) = 2m1 r 2 this is a sum of two terms that look like Fig. our disc mass m1 will be in a stable circular orbit of radius r0 (which depends upon J). At r = r0 . . and pθ . ∂L ˙ = m1 r θ 2 − gm2 ∂r ∂L Fθ = = 0. 3. ˙ pθ = 0. rotational symmetry.30 Lagrangians and Noether’s Theorem Note that θ is an “ignorable coordinate”—it doesn’t appear in L—so there’s a symmetry. 3. ) of mass m = m1 + m2 feeling a force Fr = J2 − m2 g m1 r 3 which could come from an “eﬀective potential” V (r) such that dV /dr = −Fr . the conjugate momentum. is conserved. which is topologically the center of the disk that has been removed from Q. while the term −gm2 is gravity pulling m2 down and thus pulling m1 radially in. ˙ Let’s use our conservation law here to eliminate θ from the ﬁrst equation: ˙ θ= so (m1 + m2 )¨ = r J2 − m2 g m1 r 3 J m1 r 2 Thus eﬀectively we have a particle on (0. (θ is ignorable) ∂θ Fr = ˙ Note: in Fr the term m1 r θ 2 is recognizable as a centrifugal force. Otherwise we get orbits like Fig. the Euler-Lagrange equations give.

3.4 Example Problems

31

PSfrag replacements

V (r)
attractive gravitational potential repulsive centrifugal potential

r0 no swinging allowed!

r

Figure 3.1: Potential function for disk pulled by gravitating mass.

Figure 3.2: Orbits for the disc and gravitating mass system.

3.4.3

Free Particle in Special Relativity

In relativistic dynamics the parameter coordinate that parametrizes the particle’s path in Minkowski spacetime need not be the “time coordinate”, indeed in special relativity there are many allowed time coordinates. Minkowski spacetime is, n+1 (x0 , x1 , . . . , xn ) if space is n-dimensional. We normally take x0 as “time”, and (x1 , . . . , xn ) as “space”, but of course this is all relative to one’s reference frame. Someone else travelling at some high velocity relative to us will have to make a Lorentz transformation to translate from our coordinates to theirs. This has a Lorentzian metric g(v, w) = v 0 w 0 − v 1 w 1 − . . . − v n w n = ηµν v µ w ν

32

Lagrangians and Noether’s Theorem

where

ηµν

In special relativity we take spacetime to be the conﬁguration space of a single point particle, so we let Q be Minkowski spacetime, i.e., n+1 (x0 , . . . , xn ) with the metric ηµν deﬁned above. Then the path of the particle is, q : ( t) −→ Q where t is a completely arbitrary parameter for the path, not necessarily x0 , and not necessarily proper time either. We want some Lagrangian L : T Q → , i.e., L(q i , q i ) such ˙ that the Euler-Lagrange equations will dictate how our free particle moves at a constant velocity. Many Lagrangians do this, but the “best” ones(s) give an action that is independent of the parameterization of the path—since the parameterization is “unphysical” (it can’t be measured). So the action
t1

   =  

1 0 0 ... 0 0 −1 0 . . . 0 0 0 −1 0 . . . .. . . . . . . . 0 0 0 . . . −1

      

S(q) =
t0

˙ L q i (t), q i (t) dt

for q : [t0 , t1 ] → Q, should be independent of t. The obvious candidate for S is mass times arclength,
t1

S=m
t0

ηij q i (t)q j (t) dt ˙ ˙

or rather the Minkowski analogue of arclength, called proper time, at least when q is a ˙ timelike vector, i.e., ηij q i q j > 0, which says q points into the future (or past) lightcone ˙ ˙ ˙ and makes S real, in fact it’s then the time ticked oﬀ by a clock moving along the path q : [t0 , t1 ] → Q. By “obvious candidate” we are appealing somewhat to physical intuition and
Timelike Lightlike Spacelike

generalization. In Euclidean space, free particles follow straight paths, so the arclength or pathlength variation is an extremum, and we expect the same behavior in Minkowski

3.4 Example Problems

33

spacetime. Also, the arclength does not depend upon the parameterization, and lastly, the mass m merely provides the correct units for ‘action’. So let’s take ˙ ˙ (3.2) L = m ηij q i q j and work out the Euler-Lagrange equations. We have pi = ∂ ∂L = m i ηij q i q j ˙ ˙ i ∂q ˙ ∂q ˙ 2ηij q j ˙ =m 2 ηij q i q j ˙ ˙ j ηij q ˙ mqi ˙ =m = iqj q ˙ ηij q ˙ ˙

(Note the numerator is “mass times 4-velocity”, at least when n = 3 for a real single particle system, but we’re actually in a more general n + 1-dim spacetime, so it’s more like the “mass times n + 1-velocity”). Now note that this pi doesn’t change when we change the parameter to accomplish q → αq. The Euler-Lagrange equations say, ˙ ˙ pi = F i = ˙ ∂L =0 ∂q i

The meaning of this becomes clearer if we use “proper time” as our parameter (like parameterizing a curve by its arclength) so that
t1 t0

q dt = t1 − t0 , ˙

∀ t0 , t1

which ﬁxes the parametrization up to an additive constant. This implies q = 1, so that ˙ pi = m and the Euler-Lagrange equations say pi = 0 ⇒ m¨i = 0 ˙ q so our (free) particle moves unaccelerated along a straight line, which is as we desired (expected). Comments This Lagrangian from Eq.(3.2) has lots of symmetries coming from reparameterizing the path, so Noether’s theorem yields lots of conserved quantities for the relativistic free qi ˙ = mqi ˙ q ˙

34 Lagrangians and Noether’s Theorem particle. hold) so that Noether’s theorem gives a conserved quantity pi δq i − Here we go then: δL = ∂L i ∂L i δq + i δ q ˙ ∂q i ∂q ˙ = pi δ q i ˙ mqi d i ˙ = q fs (t) ˙ q ds ˙ s=0 mqi d d i ˙ = q fs (t) q dt ds ˙ s=0 fs (t) mqi d i ˙ q fs (t) ˙ = q dt ˙ ds s=0 mqi d i ˙ = q δfs ˙ q dt ˙ d = pi q i δf ˙ dt = pi q i δf . Let’s show that δL = ˙. (when E-L eqns. ˙ d where in the last step we used the E-L eqns..e. i.. Here we see it starting to show up in special relativity. diﬀeomorphisms fs : −→ with f0 = .e. Consider any (smooth) 1parameter family of reparameterizations. i. so δL = ˙ with So to recap a little: we saw the free relativistic particle has L=m q =m ˙ ηij q i q j ˙ ˙ and we’ve considered reparameterization symmetries qs (t) = q fs (t) . fs : →         → Q} as follows: given any   ¢ £¡ . These act on the space of paths Γ = {q : q ∈ Γ we get qs (t) = q fs (t) where we should note that qs is physically indistinguishable from q. dt pi = 0. This is in fact called “the problem of time” in general relativity. These reparameterization symmetries work as follows.

5. (since ∂L/∂q i = 0.5 Electrodynamics and Relativistic Lagrangians We will continue the story of symmetry and Noether’s theorem from the last section with a few more examples. Symmetries of L guide us in ﬁnding conserved quantities. we’re seeing an example of what physicists call gauge symmetries. and ∂L/∂ q i = p) ˙ ˙ = pi δ q i ˙ d = pi δq i dt d = pi q i δf ˙ dt d ˙ = pi q i δf. 3. dt and set pi q i δf = ˙ so Noether’s theorem gives a conserved quantity pi δq i − = pi q i δf − pi q i δf ˙ ˙ =0 So these conserved quantities vanish! In short.3. and this is where we begin. via Noether’s theorem. . realizing that a given system may not have a unique Lagrangian but will often have an obvious natural Lagrangian. These are symmetries that permute diﬀerent mathematical descriptions of the same physical situation—in this case reparameterizations of a path. We use principles of least action to conjure up Lagrangians for our systems. This section also introduces gauge symmetry. Given a Lagrangian we derive equations of motion from the Euler-Lagrange equations.5 Electrodynamics and Relativistic Lagrangians 35 we’ve used the fact that δq i := so (repeating a bit of the above) δL = d i q fs (t) ds = q i δf ˙ s=0 ∂L i ∂L i δq + i δ q ˙ ∂q i ∂q ˙ = pi δ q i . 3. in particular Hamiltonians from time translation invariance. This is a good topic for starting a new section.1 Gauge Symmetry and Relativistic Hamiltonian What are gauge symmetries? 1.

and it comes from the symmetry. These symmetries give conserved quantities that work out to equal zero! Note that (1) is a subjective criterion. δL = ∂L i ∂L i δq + i δ q ˙ ∂q i ∂q ˙ = pi δ q i .4. so we see from the previous results that the Hamiltonian is zero! H = 0. Explicitly. and so ˙ ˙ ˙ i δL = d /dt. and (3) is easy to test. (since ∂L/∂q i = 0 and ∂L/∂ q i = pi ) ˙ ˙ = pi 0 = 0       . qs (t) = q(t) + s w where w ∈ n+1 and w points in some timelike direction. it’s glimmerings are seen in the ﬂat Minkowski spacetime of special relativity. (2) and (3) are objective. Well. which implies = pi δq . We have a high degree of non-uniqueness of solutions to the Euler-Lagrange equations.36 Lagrangians and Noether’s Theorem 2. and that’s impossible!) But there is another conserved quantity deserving the title of “energy” which is not zero. These symmetries make it impossible to compute q(t) given q(0) and q(0): since if ˙ q(t) is a solution so is q(f (t)) for any reparameterization f : → .2 Relativistic Hamiltonian What then is the Hamiltonian for special relativity theory? We’re continuing here with the example problem of §3. but in fact it means that there is no temporal evolution possible! So we can’t establish a dynamical theory on this footing! That’s bad news. H = pi δ q i − where under q(t) → q(t + s) we have δ q i = q i δf . (Because it means you might have to solve the static equations for the 4D universe as a whole.5. 3. The result H = 0 follows.3. so we often use (3) to distinguish gauge symmetries from physical symmetries. Now you know why people talk about “the problem of time” in general relativity theory. the Hamiltonian comes from Noether’s theorem from time translation symmetry. In fact any vector w gives a conserved quantity. qs (t) = q(t + s) and this is an example of a reparametrization (with δf = 1). You may think it’s nice and simple to have H = 0. 3.

An dxn   ). We know p = 0 from the Euler-Lagrange ˙ equations. . .6 Relativistic Particle in an Electromagnetic Field The electromagnetic ﬁeld is described by a 1-form A on spacetime. p1 . . but here we see it coming from spacetime translation symmetry. . We’ve just about exhausted all the basic stuﬀ that we can learn from the free particle. So next we’ll add some external force via an electromagnetic ﬁeld. .6 Relativistic Particle in an Electromagnetic Field 37 qs q PSfrag replacements w since δq i = w i . . pn ) p0 is energy. .3) is a 2-form containing the electric and magnetic ﬁelds.4) A = A0 dx0 + A1 dx1 + . δ q i = w i = 0. such that dA = F (3. .3. . A is the vector potential. p =(p0 . . so namely. (p1 . This is our ˙ from Noether’s theorem with ˙ ˙ Noether’s theorem says that we get a conserved quantity pi δq i − = pi w = 0. . the momentum in the w direction. for our free particle. F µν = We’d write (for Q having local charts to ∂Ai ∂Aj − ∂xj ∂xi n+1 (3. pn ) is spatial momentum. 3.

22. . with Ai q i dt the diﬀerential. 20.38 Lagrangians and Noether’s Theorem and then because d2 = 0 dA = dA0 dx0 + dA1 dx1 + .) Note that since A is a 1-form we can integrate it over an oriented manifold. dAi = ∂µ Ai dxµ using the summation convention and ∂µ := ∂/∂xµ . ˙ The Lagrangian in the above action. for a charge e with mass m in an electromagnetic potential A is L(q. The student can easily check that the components for F = F01 dx0 ∧ dx1 + F02 dx0 ∧ dx2 + . q) = m q + eAi q i ˙ ˙ ˙ (3. . April 18. q Note that since A is a 1-form it can be integrated (it is a linear combination of some basis 1-forms like the {dxi }). dAn dxn and since the “Ai ” are just functions. in 4D spacetime   0 E1 E2 E3  −E1 0 B3 −B2   F =  −E2 −B3 0 B1  −E3 B2 −B1 0 where E is the electric ﬁeld and B is the magnetic ﬁeld. agrees with the matrix expression below (at least for 4D). ˙ t0 A = integral of A along the path q. ˙ i i after dq = q dt. The action for a particle of charge e is t1 S=m t0 q dt + e ˙ q A here t1 q dt = proper time.. .5) so we can work out the Euler-Lagrange equations: pi = qi ˙ ∂L + eAi =m i ∂q ˙ q ˙ = mvi + e Ai . So. for example. . but one can also write the path integral using time t as a parameter. (Week 4.

7 Alternative Lagrangians 39 where v ∈ n+1 is the velocity. we ﬁnd the force Fi = ∂L ∂ = i e Aj q j ˙ i ∂q ∂q ∂Aj = e i qj ˙ ∂q So the Euler-Lagrange equations say (noting that Ai = Aj q(t) : p=F ˙ ∂Aj d ˙ mvi + eAi = e i q j dt ∂q ∂Aj dAi dvi = e i qj − e ˙ m dt ∂q dt ∂Aj dvi ∂Ai = e i qj − e j qj m ˙ ˙ dt ∂q ∂q ∂Ai ∂Aj =e − j qj ˙ i ∂q ∂q the term in parentheses is F ij = the electromagnetic ﬁeld. normalized so that v = 1.3. the momentum is an n + 1-vector.7 Alternative Lagrangians We’ll soon discuss a charged particle Lagrangian that is free of the reparameterization symmetry.7. Continuing the analysis. F = dA. Note that now momentum is no longer mass times velocity! That’s because we’re in n + 1-d spacetime. ˙ dt (Lorentz force law) (3.) 3.1 Lagrangian for a String So we’ve looked at a point particle and tried S = m · (arclength) + A   . First a paragraph on objects other than point particles! 3.6) (Usually called the “Lorenz” force law. So we get the following equations of motion m dvi = eF ij q j .

Given that we presumably cannot tamper with the length of the closed string. If the string is also assumed to be a source of electromagnetic ﬁeld then we need a 2-form to integrate over the 2D worldtube analogous to the 1-form integrated over the pathline of the point particle. In string theory we boost the dimension by +1 and consider a string tracing out a 2D surface as time passes (Fig.3). So provided we supply reasonabe physically realistic heuristics then whatever Lagrangian or action that we come up with will stand a good chance of describing some system with a healthy measure of physical verisimilitude.3: Worldtube of a closed string. the physical or physico-philosophical principle we’ve been using is that the path followed by physical objects minimizes the “activity” or “aliveness” of the system. The string action can then be written S = α · (area) + e B (3. But we haven’t checked that this action yields sensible dynamics yet! But supposing it does. where the 1-from A can be integrated over a 1-dimensional path.(3. In string theory this is usually the “Kalb-Ramond ﬁeld”.7) We’ve also replaced the point particle mass by the string tension α [mass·length−1 ] to obtain the correct units for the action (since replacing arclength by area meant we had to compensate for the extra length dimension in the ﬁrst term of the above string action). This may still seem like we’ve pulled a rabbit out of a hat.40 Lagrangians and Noether’s Theorem or with ‘proper time’ instead of ‘arclength’. 3. then the worldtube quantity analogous to arclength or proper time would be the area of the worldtube (or worldsheet for an open string). but it’s tensor components will have two indices since it’s a 2-form. To recover electrodynamic interactions it should be antisymmetric like A. . A generalization (or specialization. depending on how you look at it) would be to consider a Lagrangian for an extended object. Can you infer an appropriate action for this system? Remember. becomes Figure 3. then would it justify our guesswork and intuition in arriving at Eq.7)? Well by now you’ve probably realized that one can have more than one form of action or Lagrangian that yields the same dynamics. call it B.

8) • There’s no ugly square root. What Euler-Lagrange equations does this Lagrangian yield? ∂L = mqi + eAi ˙ ∂qi ˙ ∂Aj ∂L ˙ Fi = i = e i q j ∂q ∂q pi = Very similar to before! The E-L eqns. 1 • It looks formally like “ 2 mv 2 ”.(3. they are handled the same. (3.2 Alternate Lagrangian for Relativistic Electrodynamics In § 3. It’s particularly useful when Newtonian theory cannot give us a head start. meaning that H = 0 and there were non-unique solutions to the Euler-Lagrange equations arising from applying gauge transformations.5) we saw an example of a Lagrangian for relativistic electrodynamics that had awkward reparametrization symmetries.7. in relativistic dynamics and in the physics of extended particles.) The only diﬀerence is that we have m¨i instead q of mvi where vi = qi / q . then say d ∂Aj mqi + eAi = e i q j ˙ ˙ dt ∂q m¨i = eF ij q j q ˙ almost as before. But there’s another Lagrangian we can use that doesn’t have this gauge symmetry: 1 L = mq · q + eAi q i ˙ ˙ ˙ 2 This one even has some nice features. and there’s no trouble with paths being timelike or spacelike in direction. So the old Euler-Lagrange equations of motion reduce to the ˙ ˙ ˙ .3.e. 3. so it’s everywhere diﬀerentiable.(3.7 Alternative Lagrangians 41 That’s enough about string for now. which was a Lagrangian for a charged particle with reparametrization symmetry L = m q + eAi q i ˙ ˙ just as for an uncharged relativistic particle. The point was to illustrate the type of reasoning that one can use in conjuring up a Lagrangian. Recall Eq. i. Eq.6. This freedom to change the gauge can be avoided. familiar from nonrelativistic mechanics.5).. (I’ve taken to using F here for the electromagnetic ﬁeld tensor to avoid clashing with F for the generalized force.

n+1 . With the new Lagrangian it’s not! Indeed H = pi q i − L ˙ and now pi = so ˙ ˙ ˙ H = (mqi + eAi )q i − ( 1 mqi q i + eAi q i ) ˙ ˙ 2 ˙ ˙ = 1 mqi q i 2 Comments. we’re getting “conservation of speed” rather than some more familiar “conservation of energy”. Recall that for our reparametrization-invariant Lagrangian L=m qi q i + eAi q i ˙ ˙ we got H = 0. time translation was a gauge symmetry. That is. so now ˙ H = pi q i − L = 2K − (K − V ) = K + V.4. 3. The reason is that this Hamiltonian comes from the symmetry qs (t) = q(t + s) instead of spacetime translation symmetry qs (t) = q(t) + s w.42 Lagrangians and Noether’s Theorem new ones if we pick a parametrization with q = 1. Let’s work out the Hamiltonian for this 1 L = mq · q + eAi q i ˙ ˙ ˙ 2 for the relativistic charged particle in an electromagnetic ﬁeld. In general then there’s no conserved “energy” for our particle corresponding to translations in time. This is vaguely like how a nonrelativistic particle in a potential V has but now the “potential’ V = eAi q i is linear in velocity. ˙ ∂L = mq i + eAi ˙ ∂qi ˙ H = pi q i − L = (2K − V ) − (K − V ) = K. ˙ As claimed H is not zero. which would be a parametrization ˙ by proper time for example. q) = 2 m q 2 + Ai (q)q i ˙ ˙ ˙ has time translation symmetry iﬀ A is translation invariant (but it’s highly unlikely a given system of interest will have A(q) = A(q + sw)). w∈ the diﬀerence is illustrated schematically in Fig. and the fact that it’s conserved says q(t) is constant as ˙ a function of t. so the particle’s path is parameterized by proper time up to rescaling of t. Our Lagrangian 1 L(q.

. w) = gij v i w j in any basis. 3. 2.8 The General Relativistic Particle 43 PSfrag replacements 4 5 3 4 2 3 1 2 0 1 w qs (t) = q(t + s) qs (t) = q(t) + sw Figure 3. w) −→ g(x)(v. . 0  0 −1 0  gij =  . .  . Q.   1.. but for diﬀerent bases gij will have a diﬀerent form. w) or we could write g(v. 0 0 . is an (n + 1)-dimensional Lorentzian manifold. we have a bilinear map . We deﬁne the metric as follows.8 The General Relativistic Particle In GR spacetime. 3. g(x) : Tx Q × Tx Q −→ (v.. For each x ∈ Q. w) = gij v i w j  1 0 .3. With respect to some basis of Tx Q we have g(v.. g(x) varies smoothly with x. . namely a smooth (n + 1)-dimensional manifold with a Lorentzian metric g. .4: Proper time rescaling vs spacetime translation. w) for short. . . . −1      Of course we can write g(v.

q) = 2 mgij q i q j ˙ ˙ ˙ .44 Lagrangians and Noether’s Theorem 3. q) ˙ 2 = 1 mgij q i q j ˙ ˙ 2 The big diﬀerence between these two Lagrangians is that now spacetime translation symmetry (and rotation. Alternatively we could just as well use ˙ ˙ L(q. and we ˙ need the metric to relate them.8. which in this case start from ∂L pi = i = mgij q j ˙ ∂q ˙ The velocity q here is a tangent vector. In coordinates this would say that the tangent vector v i gets mapped to the cotangent vector gij v j . 2 (where ∂i = ∂ ). w) −→ g(v. q) = 1 mg(q)(q. ∂L = mgij q j ˙ i ∂q ˙ ∂L ∂ Fi = i = i 1 mgjk (q)q j q k ˙ ˙ ∂q ∂q 2 pi = ˙ ˙ = 1 m∂i gjk q j q k . q) ˙ ˙ gij q i q j ˙ ˙ The Lagrangian for a free point particle in the spacetime Q is just like in special relativity but with ηij replaced by gij . −). Suppose then Q is a Lorentzian manifold with metric g and L : T Q → is the Lagrangian of a free particle. We ﬁnd equations of motion from the Euler-Lagrange equations. ˙ Getting back to the E-L equations. the momentum p is a cotangent vector. nor velocity of center of energy) anymore! Let’s ﬁnd the equations of motion. and boost symmetry) is gone! So there is no conserved energymomentum (nor angular momentum. w) which gives Tq M −→ Tq∗ M v −→ g(v. This is lurking behind the passage from q i to the momentum mgij v j . q) = m ˙ =m g(q)(q. ∂q i     1 L(q. via g : Tq M × Tq M −→ (v.1 Free Particle Lagrangian in GR L(q.

among other things.8 The General Relativistic Particle 45 So the Euler-Lagrange equations say d 1 ˙ ˙ mgij q j = 2 m∂i gjk q j q k . Parallel transport is just the simplest way to compare vectors at diﬀerent points in the manifold. a connection is just the rule for parallel transporting tangent vectors around the manifold. which ¨ ˙ jk is really a particular type of connection that a Lorentzian manifold has (the Levi-Civita connection). Then jk qi = gij q j = −Γijk q j q k ¨ ¨ ˙ ˙ ∴ q i = −Γi q j q k . ˙ dt The mass factors away.2 Charged particle in EM Field in GR We can now apply what we’ve learned in consideration of a charged particle. so the motion is independent of the mass! Essentially we have a geodesic equation. in our Lorentzian manifold. Not surprisingly. q ˙ ˙ ˙ If you want to know more about Lagrangians for general relativity we recommend the paper by Peldan [Pel94]. 3. m¨i = −mΓijk q j q k + eF ij q j . a covariant derivative. We can rewrite this geodesic equation as follows d ˙ ˙ gij q j = 1 ∂i gjk q j q k ˙ 2 dt ˙ ˙ ∂k gij q k q j + gij q j = 1 ∂i gjk q j q k ˙ ˙ ¨ 2 ∴ gij q j = ¨ = 1 2 1 ∂g 2 i jk ∴ ∂i gjk − ∂k gij − ∂j gki q j q k ˙ ˙ − ∂k gij q j q k ˙ ˙ where the last line follows by symmetry of the metric. in an electromagnetic ﬁeld with potential A. The Lagrangian would be 1 L = 2 mgjk q j q k + eAi q i ˙ ˙ ˙ which again was conjured up be replacing the ﬂat space metric ηij by the metric for GR gij .8. gik = gki . ¨ jk ˙ ˙ So we see that q can be computed in terms of q and the Christoﬀel symbols Γi . This allows us to deﬁne. This deﬁnes what we call the Christoﬀel symbols Γi . Now let. the Euler-Lagrange equations then yield the following equations of motion. Thorne & Wheeler [WTM71]. of charge e.3. Γijk = − ∂i gjk − ∂k gij − ∂j gki the minus sign being just a convention (so that we agree with everyone else). and also the “black book” of Misner. .

First. If the light s trying to minimize the time. 3. if one imagines a force on a particle resulting from a potential gradient at an interface as analogous to light refraction then you also get a screw-up in the analogy (Fig. Nevertheless. April 18. Suppose it’s in n with its usual Euclidean metric. This analogy is strange.9 The Principle of Least Action and Geodesics (Week 4. ∞). Also.) 3. 0  1   .46 Lagrangians and Noether’s Theorem 3. 1] → Q. the index of refraction n :  δij =       gij = n2 δij n → (0. 0 .1 Jacobi and Least Time vs Least Action We’ve mentioned that Fermat’s principle of least time in optics is analogous to the principle of least action in particle mechanics. 20. times the usual Euclidean metric  1 . PSfrag replacements n high light faster light slower particle slower particle faster V high n low V low Figure 3. 22.5: Least time versus least action. Jacobi was able to reinterpret the mechanics of a particle as an optics problem and hence “unify” the two minimization principles. let’s consider light in a medium with a varying index of refraction n (recall 1/n ∝ speed of light).9. since in the principle of least action we ﬁx the time interval q : [0. its trying to minimize the arclength of its path in the metric that is..5).

Consider a particle of mass m in Euclidean n with potential V : n → . i.10) . Now. where n(q) is proportional to the reciprocal of the speed of light!! But let’s see that it works. ˙ Note: this is precisely backwards compared to optics.e.. m d2 q i = −∂i V dt2 How did Jacobi see (3. v) ≥ 0 So we’ll use the same Lagrangian: L(q.3. L= = = gij (q)q i q j ˙ ˙ n2 (q)q i q j ˙ ˙ 2/m(E − V (q))q 2 ˙       (3.9) gij (q)q i q j ˙ ˙ As before the Christoﬀel symbols Γ are built from the derivatives of the metric g. ˙ ˙ (3. what Jacobi did is show how the motion of a particle in a potential could be viewed as a special case of this.9 The Principle of Least Action and Geodesics 47 This is just like the free particle in general relativity (minimizing it’s proper time) except that now gij is a Riemannian metric g(v. It satisﬁes F = ma.10) as a special case of (3. since 2 (E − V ) = m 21 m q ˙ m2 2 = q . w) = gij v i w j where g(v.9)? He considered a particle of energy E and he chose the index of refraction to be n(q) = 2 E − V (q) m which is just the speed of a particle of energy E when the potential energy is V (q). q) = ˙ and get the same Euler-Lagrange equations: d2 q i + Γi q j q k = 0 jk ˙ ˙ dt2 if q is parameterized by arclength or more generally q = ˙ gij (q)q i q j = constant.

w. ˙ Recall that our Lagrangian gives reparameterization invariant Euler-Lagrange equations! This is the uniﬁcation between least time (from optics) and least action (from mechanics) that we sought. We get the ˙ ˙ ˙ Euler-Lagrange equations. 2. 3.. v · w = δij v i w j . We have 1.r. or m¨i = −∂i V . ˙ d dt 2/m(E − V (q)) · Jacobi noticed that this is just F = ma. The metric at q ∈ Q is.2 The Ubiquity of Geodesic Motion We’ve seen that many classical systems trace out paths that are geodesics.t a basis of Tq Q g(v. i. w) → g(v. t1 ] → Q that are critical points of t1 S(q) = t0 gij q i q j dt ˙ ˙ which is proper time when (Q. w) = δij v i w j   . provided we reparameq terize q so that.e. paths q : [t0 . g(q) : Tq Q × Tq Q → (v. or arclength when (Q.48 Lagrangians and Noether’s Theorem where q 2 = q · q is just the usual Euclidean dot product. g) is a Lorentzian manifold. g) is a Riemannian manifold. that is. q = 2/m(E − V (q)). w) and it is bilinear. pi = Fi = ∂L = ∂qi ˙ ∂L = ∂i ∂q i 1 = 2 2 q ˙ (E − V ) · m q ˙ 2 (E − V (q)) · q ˙ m −2/m∂i V · q ˙ 2/m(E − V q) 1 qi ˙ = − ∂i V q ˙ m q ˙ 2/m(E − V ) Then p = F says.9.

q) with index of refraction n : Q → (0. In fact all three of these results can be generalized to cover every problem that we’ve discussed! (1 ) Light on any Riemannian manifold (Q. e. All of these systems have a Lagrangian which is a quadratic function of position. so they all ﬁt into this framework.. q) with potential V : Q → traces out geodesics w. whereas Riemannian manifolds represent that we’d normally consider as just space. q).3.     where n : Q → (0. (1) In the geometric optics approximation. A particle with charge e on a Lorentzian manifold (Q. light in Q = out geodesics in the metric gij = n(q)2 δij (2) Jacobi saw that a particle in Q = geodesics in the metric n in some potential V : Q → 2 (E − V )δij m if the particle has energy E (where1 V < E). 1 The case V > E. Lots of physical systems can be described this way. We’ve seen at least three important things.r. a rigid rotating body (Q = SO(3)). q) in an electromagnetic vector potential follows a path with e ˙ qi = −Γijk q j q k + F ij q j ¨ ˙ ˙ m where F ij = ∂i Aj − ∂j Ai but this is actually geodesic motion on the manifold Q × U (1) where U (1) = {eiθ : θ ∈ } is a circle. An important distinction to keep in mind is that Lorentzian manifolds represent spacetimes. (3 ) Kaluza-Klein Theory. the Atwood machine. g(q) varies smoothly with q ∈ Q.g.t the metric 2 h = (E − V )g m if it has energy E.   n acts like particles tracing   traces out   .9 The Principle of Least Action and Geodesics 49 3. spinning tops. ∞) is the index of refraction function. would be classically forbidden regions. (2 ) A particle on a Riemannian manifold (Q. and others. ∞) traces out geodesics in the metric h = n2 g. gij = (3) A free particle in general relativity traces out a geodesic on a Lorentzian manifold (Q. if they exist.

Let’s pick coordinates xi on Q where i ∈ {0. . ¨ if qθ = e/m ˙ since F ij is part of the Christoﬀel symbols for h. The metric h on Q × U (1) PSfrag replacements a geodesic Q the apparent path Q × U (1) is built from g and A in a very simple way. To get the desired equations for motion on Q × U (1) we need to given Q × U (1) a cleverly designed metric built from g and A where the amount of “spiralling”—the velocity in the U (1) direction is e/m. . To summarize this section on least time versus least action we can say that every problem that we’ve discussed in classical mechanics can be regarded as geodesic motion! e F ij q j ˙ m . . n} since we’re in n + 1-dimensional spacetime. The components of h are hij = gij + Ai Aj hθi = hθi = −Ai hθθ = 1 Working out the equations for a geodesic in this metric we get qi = −Γijk q j q k + ¨ ˙ ˙ qθ = 0. and θ is our local coordinate on S 1 .50 Lagrangians and Noether’s Theorem Let’s examine this last result a bit further. .

q). position and momentum will satisfy Hamilton’s equations: dq i ∂H = . dt ∂q 4.Chapter 4 From Lagrangians to Hamiltonians In the Lagrangian approach we focus on the position and velocity of a particle. dt ∂pi 51 dpi ∂H =− dt ∂qi     .1 The Hamiltonian Approach In the Hamiltonian approach we focus on position and momentum. and compute what the particle does starting from the energy H = pi q i − L(q. called the Hamiltonian H : T ∗ Q −→ where the cotangent bundle is the space of position-momentum pairs. and compute what the particle does starting from the Lagrangian L(q. But we’re led to consider momentum ∂L pi = i ∂q ˙ since the equations of motion tell us how it changes dpi ∂L = i. q) ˙ ˙ reinterpreted as a function of position and momentum. which is a function ˙ L : T Q −→ where the tangent bundle is the space of position-velocity pairs. In this approach.

q) −→ (q.q) Tq Q ˙ Tq Q q Figure 4. p) ˙ where q ∈ Q. ˙ and p is a cotangent vector in Tq∗ Q := (Tq Q)∗ .52 From Lagrangians to Hamiltonians where the latter is the Euler-Lagrange equation ∂L dpi = dt ∂qi in disguise (it has a minus sign since H = pq − L).1). and q is any tangent vector in Tq Q (not the time derivative of something). given by q −→ pi = ˙ λ ∂L ∂qi ˙ So λ is deﬁned using L : T Q → .. it’s the “diﬀerential of L in the vertical direction”—i. the q i directions.1: Q . 4. We want to deﬁne “ ∂ qi ” in a coordinate-free ˙ PSfrag replacements way. λ can be deﬁned in a coordinate∂L free way. q) −→ q ˙ and dπ : T (T Q) −→ T Q   Tq Q TQ T(q. as follows (referring to Fig. Despite appearances.e. ˙ To obtain this Hamiltonian description of mechanics rigorously we need to study this map λ : T Q −→ T ∗ Q (q. We have ˙ π : T Q −→ Q (q.

q) ∈ T(q. So f gives a linear map p : Tq Q −→ that is.q) Tq Q ˙   q Q . we now know a coordinate-free way of describing the map λ : T Q −→ T ∗ Q (q.q) T Q −→ . p) ˙ given in local coordinates by pi = 1 PSfrag replacements Tq∗ Q ∂L . 4.q) T Q ˙ T(q. 2 The ﬁber Tv V at v ∈ V of vector manifold V has the same dimension as V .q) T Q ˙ ˙ that is.q) Tq Q ∼ Tq Q = ˙ in a canonical way2 .q) T Q −→ . 6. q) −→ (q. so we have ˙ ∗ (dL)(q.       Tq Q V(q.q) : T(q. p∈ this is the momentum! (Week 6. May 2. ∂qi ˙ The kernel of a map is the set of all elements in the domain that map to the null element of the range. dL(q.1 The Hamiltonian Approach 53 has kernel1 consisting of vertical vectors: V T Q = ker dπ ⊆ T T Q The diﬀerential of L at some point (q. T(q.4.q) T Q = T (Tq Q) ˙ and since Tq Q is a vector space. so ker dπ = {v ∈ T T Q : dπ(v) = 0 ∈ T Q}. getting f : V(q. ˙ ˙ We can restrict this to V T Q ⊆ T T Q. ˙ But note V(q.) Given L : T Q → T ∗ Q. q) ∈ T Q is a map from T T Q to .

54 From Lagrangians to Hamiltonians We say L is regular if λ is a diﬀeomorphism from T Q to some open subset X ⊆ T ∗ Q.e.1 Example: A Particle in a Riemannian Manifold with Potential V (q)   has Lagrangian For a particle in a Riemannian manifold (Q. not the result of the operation.. mg(q.2. (q. p) = λ(q. q) = mgij q i q j − V (Q) ˙ 2 Here pi = so λ(q. In this case we can describe what our system is doing equally well by specifying position and velocity. not inserted since λ itself is an operator on tangent vectors. 4. −) ˙ ˙ so3 L is strongly regular in this case because Tq Q −→ Tq∗ Q ∂L = mgij q j ˙ ∂qi ˙ v −→ g(v. q) in a potential V : Q → 1 ˙ L(q. In practice often X = T ∗ Q. then L is said to be strongly regular. which in this case extends to all of T ∗ Q. 3 . the metric is nondegenerate. q) ∈ X. ˙ We call X the phase space of the system.2 Regular and Strongly Regular Lagrangians This section discusses some examples of the above theory. The missing object there “−” is of course any tangent vector. Thus λ is a diﬀeomorphism. i. −) is 1-1 and onto. q) ∈ T Q ˙ or position and momentum (q. q) = q. 4.

q) = f (q) ˙ ˙ All linear transforms are aﬃne. ∂q ˙ This L is still strongly regular. q) = m ˙ gij q i q j ˙ ˙ This is terrible from the perspective of regularity properties—it’s not diﬀerentiable when gij q i q j vanishes. For the example the translation is the “+eA(q)” part.4.) 4.4 Example: A Regular but not Strongly Regular Lagrangian   and Here’s a Lagrangian that’s regular but not strongly regular.2. L(q. we can ask about regularity.2 Example: General Relativistic Particle in an E-M Potential For a general relativistic particle with charge e in an electromagnetic vector potential A the Lagrangian is 1 ˙ ˙ L(q.2 Regular and Strongly Regular Lagrangians 55 4. 4 . the map λ is not 1-1 where deﬁned ˙ since multiplying q by some number has no eﬀect on p! (This is related to the reparameter˙ ization invariance—this always happens with reparameterization-invariant Lagrangians. Where it is deﬁned ˙ ˙ pi = mgij q j ˙ ∂L = i ∂q ˙ q ˙ (where q is timelike). Let Q = L(q. q −→ m g(q. In aﬃne geometry there is no deﬁned origin. Alas.2. −) + eA(q) ˙ ˙ 4. but aﬃne transformations include translations. q) = mgij q i q j − eAi q i ˙ 2 and thus ∂L ˙ pi = i = mgij q i q j + eAi .2. but now each map λ |Tq Q : Tq Q −→ Tq∗ Q is aﬃne rather than linear4 .3 Example: Free General Relativistic Particle with Reparameterization Invariance The free general relativistic particle with reparameterization invariant Lagrangian has. which are nonlinear. and undeﬁned when the same is negative.

H. 1 + q2 ˙ 4. etc. So ˙ p= so that         to some   . This lets us treat q i . all as functions on X (or T Q). q) −→ (q.. take f (q) = eq˙ so f : → (0. thus writing qi ˙ for the function q i ◦ λ−1 ˙ In particular pi := ˙ (function on X) (function on T Q) ∂L (Euler-Lagrange eqn. p) ˙ ∼ Now let’s assume L is regular. q) = eq˙ ˙ G or slope between −1 and 1 G L(q. pi .56 From Lagrangians to Hamiltonians positive slope L(q. will be treated as a function on X. ∞) ⊂ . For example.3 Hamilton’s Equations λ : T Q −→ X ⊆ T ∗ Q (q. Now let’s calculate: dL = ∂L i ∂L i dq + i dq ˙ ∂q i ∂q ˙ = pi dq i + pi dq i ˙ ˙   ∂L = f (q) ˙ ∂q ˙ This will be regular but not strongly so if f : → is a diﬀeomorphism from ∼ proper subset U ⊂ . q) = ˙ and so forth. L.) ∂q i which is really a function on T Q. so This lets us have the best of both worlds: we can identify T Q with X using λ.

∂H . when the Euler-Lagrange equations hold for ˙ ˙ some path q : [t0 .4. q ) as local coordinates on X. So when the Euler-Lagrange equations hold. p(t) ∈ X. In fact. ˙ this gives us dL = pi dq i + pi dq i ˙ ˙ dH = q i dpi − pi dq i . so ∼ . This lets us regard both L and the Hamiltonian H = pi q i − L as ˙ i i functions on the phase space X.1 Hamilton and Euler-Lagrange Though q i and pi are just functions of X. q) −→ (q. The equation qi = ˙ ∂H ∂pi   is regular. Hamilton’s equations describe the motion of a point x(t) = q(t). i ∂p ∂q Since dpi . they will be the time derivatives of q i and pi . to get dH = ∂H ∂H dpi + i dq i . t1 ] → Q. dq i form a basis of 1-forms. and use (q . Hamilton’s equations are just the EulerLagrange equations in disguise. ∂pi pi = − ˙ ∂H ∂qi 4.3 Hamilton’s Equations 57 while dH = d(pi q i − L) ˙ = q i dpi + pi dq i − dL ˙ ˙ = q i dpi − pi dq i ˙ ˙ ˙ = q i dpi + pi dq i − (pi dq i + pi dq i ) ˙ ˙ ˙ ˙ so Assume the Lagrangian L : T Q → dH = q i dpi − pi dqi . in this context. p) ˙ is a diﬀeomorphism. pi ). ˙ ˙ But we can also work out dH directly. ˙ ˙ λ : T Q −→ X ⊆ T ∗ Q (q. As we’ve seen. we conclude: qi = ˙ These are Hamilton’s Equations.3. this time using local coordinates (q i .

˙ m and Hamiltonian H(q. ˙ Given this. inverting the formula ˙ pi = ∂L ∂qi ˙ which gave p as a function of q and q.58 From Lagrangians to Hamiltonians really just lets us recover the velocity q as a function of q and p. p) −→ (q. q) = ˙ 2 which gives p = mq ˙ p q= . ˙       . So we get a formula for the map ˙ λ−1 : X −→ T Q (q. the second is F = ma. i dt ∂ q ˙ ∂q These are the same because ∂L ∂ ∂H ˙ = i pi q i − L = − i . p) = pi q i − L = ˙ 1 p 2 p 2− −V m 2m 1 p 2 + V (q). = 2m p m g ij pj ) m (though really that’s q = ˙ So Hamilton’s equations say ∂H ∂pi ∂H pi = − i ˙ ∂q qi = ˙ ⇒ ⇒ q= ˙ p=− V ˙ The ﬁrst just recovers q as a function of p. ∂q i ∂q ∂q Example: Particle in a Potential V (q) For a particle in Q = n ∂H ∂q i ∂L ∂q i or p= ˙ in a potential V : n → the system has Lagrangian m q 2 − V (q) ˙ L(q. q). the other Hamilton equation pi = − ˙ is secretly the Euler-Lagrange equation d ∂L ∂L = i.

That’s the phase space X. You might have some particles moving on a manifold like S 3 . p(t) dt d we write dt q i instead of q i to emphasize that we mean the time derivative rather than a ˙ coordinate in phase space. we obtained the Euler-Lagrange equations by associating an “action” S with any q : [t0 . More precisely.4. The coordinate-free description of this structure is the major 20th century contribution to mechanics: a symplectic structure. we’ll drop the requirement that X ⊆ T ∗ Q where Q is a conﬁguration space. which is not symplectic. Now let’s get Hamilton’s equations directly by assigning an action S to any path x : [t0 . 4. t1 ] → X and setting δS = 0. X will just be a manifold equipped with enough structure to write down Hamilton’s equations starting from any H : X → . Soon. Let’s show δS = 0 ⇔Hamilton’s equations. p(t) and let ˙ t1 S(x) = t0 pi (t) d i q (t) − H q(t). Note: we don’t impose any relation between p and q. q! The relation will follow from δS = 0. This is important. ˙ Let P be the space of paths in the phase space X and deﬁne the action S : P −→ by t1 S(x) = t0 (pi q i − H)dt ˙ where pi q i − H = L. t1 ] → Q and setting δS = 0. write our path x as x(t) = q(t).2 Hamilton’s Equations from the Principle of Least Action Before. So the Hamiltonian mechanics point of view says that the abstract manifold that you are really interested in is something diﬀerent: it must be a symplectic manifold. δS = δ = (pi q i − H)dt ˙ δpi q i + pi δ q i − δH dt ˙ ˙     dt .3 Hamilton’s Equations 59 Note on Symplectic Structure Hamilton’s equations push us toward the viewpoint where p and q have equal status as coordinates on the phase space X.3. We’ll introduce symplectic geometry more completely in later chapters.

∂pi ∂H ∂q i This vanishes ∀δx = (δq. For paths in conﬁguration space Q. δS = 0 ⇔Hamilton’s equations.60 From Lagrangians to Hamiltonians then integrating by parts. t)       . since X ⊆ T ∗ Q. we might consider a third version based on paths in positionvelocity space T Q. But when our Lagrangian is regular we have a diﬀeomorphism λ : ∼ T Q → X. = = = δpi q i − pi δq i − δH dt ˙ ˙ δpi q i − pi δq i − ˙ ˙ ∂H i ∂H δq − δpi dt ∂q i ∂pi ∂H ∂H dt δpi q i − ˙ + δq i −pi − i ˙ ∂pi ∂q ∂H . δp) if and only if Hamilton’s equations qi = ˙ pi = − hold. Additionally. the really interesting principle of least action involves paths in the extended phase space where we have an additional coordinate for time: X × . So any path x : [t0 . δS = 0 ⇔Euler-Lagrange equations. so this third principle of least action is just a reformulation of principle 2. Recall the action S(x) = = = We can interpet the integrand as a 1-form β = pi dq i − H dt (pi q i − H) dt ˙ pi dq i dt − H dt dt 2. For paths in phase space X. pi dq i − H dt on X × . Just as we hoped. However. t1 ] −→ X × t −→ (x(t). t1 ] −→ X gives a path σ : [t0 . which has coordinates {pi . q i . t}. We’ve seen two principles of “least action”: 1.

13.4 Waves versus Particles—The Hamilton-Jacobi Equations 61 and the action becomes the integral of a 1-form over a curve: S(x) = pi dq i − H dt = β σ 4. ∞) is the index of refraction throughout space (generally not a constant). We’ve talked about geometrized optics. etc.) In quantum mechanics we discover that every particle—electrons. and vice versa. 11. Interestingly Newton already had a particle theory of light (his “corpuscules”) and various physicists argued against it by pointing out that diﬀraction is best explained by a wave theory. 4.4 Waves versus Particles—The Hamilton-Jacobi Equations (Week 7. but we use the new metric hij = n2 gij where n : Q → (0. neutrinos.4. Here we start with a Riemannian manifold (Q. g) as space.4.1 Wave Equations Huygens considered this same setup (in simpler language) and considered the motion of a wavefront: G .—is a wave. an approximation in which light consists of particles moving along geodesics. photons. May 9.

4.62 From Lagrangians to Hamiltonians and saw that the wavefront is the envelope of a bunch of little wavelets centered at points along the big wavefront: 1\$0%×ÐÖÑ)&ÕÒÔÓ1\$0%('×ÐÖÑ)&ÕÒÔÓ (' 1\$0%×Ð1\$ÖÑ)&ÕÒ0%ÔÓ ×ÐÖÑ(')&ÕÒÔÓ ('Ô 1\$1\$0%0%×ÐÖÑ×Ð)&ÕÒÖÑÔÓ)&ÕÒÔÓ ('(' 11\$\$0%1\$0%0%×ÐÖÑ×Ð)&ÕÒÖÑÔÓ×Ð)&ÕÒÖÑÔÓ)&ÕÒÔÓ('('(' 1×1\$0%ÐÖÑ)&ÕÒ\$ÔÓ0%('×Ð1\$ÖÑ)&ÕÒ0%ÔÓ ×ÐÖÑ(')&ÕÒÔÓ (' 1\$0%×ÐÖÑ)&ÕÒÔÓ (' balls of radius centered at points of the old wavefront In short. We can think about the distance function d : Q × Q −→ [0. h). as depicted in Fig.2.) Using this we get the wavefronts centered at q0 ∈ Q as the level sets {q : d(q0 . For larger c the level sets can cease to PSfrag replacements q0 Figure 4. q) = c} or at least for small c > 0. ∞) on the Riemannian manifold (Q. the wavefront moves at unit speed in the normal direction with respect to the “optical metric” h. where d(q0 . (Secretly this d(q0 .2: be smooth—we say a catastrophe occurs—and then the wavefronts are no longer the level sets. q1 ) is the least action—the inﬁmum of action over all paths from q0 to q1 . This sort of situation can happen for topological reasons (as when the waves smash into each other in the back of Fig.2) and it can also happen for geometrical reasons . q1 ) = inf (arclength) Υ where Υ = {paths from q0 to q1 }. 4.

Hamilton and Jacobi focused on distance d : Q × Q → [0. we can approximate the waves of light by a wavefunction: ψ(q) = A(q)eik d(q. They noticed that: ∂ W (q0 .2 The Hamilton-Jacobi Equations k th a uo Ñz rkevhX UPE × Ô{ h Pqqq A7 C !  k P 7 qVq 7 % %1 { qqqq 1 3 )3 7 % 7 I % % %) 7A C QX hk{ Ô ! EP Ua hvekrz Ñ× y ht xgkot We’ve seen that in optics.3: (Fig. This becomes the eikonal approximation in optics5 once we ﬁgure out what A should be. q1 ) = (p1 )i i ∂q1 q1 q0 • • p1 G where p1 is a cotangent vector pointing normal to the wavefronts. Assuming no such catastrophes occur.4.3).q0 ) where k is the wavenumber of the light (i. 4. 4. its color) and A : Q → describes the amplitude of the wave. which drops oﬀ far from q0 . 5   .e. ∞) as a function of two variables and called it W =Hamilton’s principal function. particles of light move along geodesics.4 Waves versus Particles—The Hamilton-Jacobi Equations 63 Figure 4. in optics the eikonal approximation is the basis for ray tracing methods. but wavefronts are level sets of the distance functions: gy Eikonal comes form the Greek word for ‘image’ or ‘likeness’..4.

e. h) is a Riemannian manifold. and we now can go further still. Hamilton went further. in quantum mechanics. its arclength. Suppose Q is any manifold and L : T Q → is any function (Lagrangian). This foreshadows quantum mechanics! After all. h is the optical metric. This is begging to be generalized to other Lagrangian systems! (At least it is retrospectively. the momentum is an operator that acts to diﬀerentiating the wavefunction.) We also saw that ∂ W (q0 . m We’ve seen this reduces point particle mechanics to optics—but only for particles of ﬁxed energy E. k is the frequency and W : Q × Q −→ [0. or Hamilton’s principal function: W (q0 .64 From Lagrangians to Hamiltonians at least while the level sets remain smooth.. In the eikonal approximation. Jacobi generalized this to the motion of point particles in a potential V : Q → . using the fact that a particle of energy E traces out geodesics in the metric hij = 2(E − V ) gij . light is described by waves ψ(q1 ) = A(q1 )eik W (q0 . Deﬁne Hamilton’s principal function W :Q× ×Q× −→         ¤ ψ : Q −→ G   . with the advantage of our historical perspective. ∞) is the distance function on Q. i ∂q1 q1 p 1 q0 • • “points normal to the wavefront”—really the tangent vector pi = hij (p1 )j 1 points in this direction. q1 ) = inf S(q) q∈Υ where Υ is the space of paths from q0 to q and S(q) is the action of the path q.q1 ) where (Q. q0 ∈ Q is the light source. i. In fact kp1i is the momentum of the light passing through q1 . q1 ) = (p1 )i .

q1 ) = (p1 )i . t1 ) 1 (q0 . t1 ) = inf S(q) q∈Υ where Υ = q : [t0 . t1 ] −→ Q be the action-minimizing path from q0 to q1 . let q : [t0 .4 Waves versus Particles—The Hamilton-Jacobi Equations 65 by W (q0 . t0 . q1 . t1 ) are close enough—so let’s assume that is true. (-momentum at time t0 ) i ∂q0 ∂W = −H1 . Then W (q0 . Maybe the best way to get the Hamilton-Jacobi equations is from this extended phase space formulation. t1 ). t1 ). t0 ) to (q1 . q1 . q(t0 ) = q0 . t1 ] → Q. q(t) dt ˙ Now W is just the least action for a path from (q0 . & q(t1 ) = q1 and S(q) = t0 t1 L q(t). The mysterious minus sign in front of energy was seen before in the 1-form. t0 ) where p1 is the momentum of the particle going from q0 to q1 . and ∂W = −(p0 )i . i ∂q1 p (q1 . t1 ) = S(q) • • G   .4. t0 ). In fact. β = pi dq i − H dt on the extended phase space X × . t0 ) and (q1 . Given (q0 . it’ll be smooth if (q0 . (+energy at time t0 ) ∂t0 (H1 = H0 as energy is conserved). we have ∂ W (q0 . at time t1 . (-energy at time t1 ) ∂t1 ∂W = H0 . But for now let’s see how Hamilton’s principal function W and variational principles involving least action also yield the Hamilton-Jacobi equations. These last four equations are the Hamilton-Jacobi equations. (q1 . t0 .

q) dt ˙ ∂L i ∂L i δq + i δ q dt ˙ ∂q i ∂q ˙ ∂L i δq − pi δq i dt + pi δq i ˙ ∂q i ∂L − pi δq i dt ˙ ∂q i t1 t0 = t0 t1 = t0 t1 = t0 the term in parentheses is zero because q minimizes the action and the Euler-Lagrange equations hold. So we δq i have i i δW = p1i δq1 − p0i δq0 and so ∂W = p1i . getting a variation δq which does not vanish at t0 and t1 . . we need to vary t0 and t1 : Now change in W will involve ∆t0 and ∆t1 • • • • t0 t0 + ∆t0 t1 t1 + ∆t1 (you can imagine ∆t0 < 0 in this ﬁgure if you like). i ∂q1 and ∂W = −p0i i ∂q0 These are two of the four Hamilton-Jacobi equations! To get the other two. We get δW = δS t1 =δ t0 t1 L(q.66 From Lagrangians to Hamiltonians Now consider varying q0 and q1 a bit q t0 t1 and thus vary the action-minimizing path.

and C is a curve in the extended phase space: C(t) = q(t). t1 ) = S(q) t1 = t0 t1 L(q. q1 . t ∈ X × . t1 ) = inf S(q) q∈Υ where Υ is the space of paths q : [t0 .4 Waves versus Particles—The Hamilton-Jacobi Equations 67 We want to derive the Hamilton-Jacobi equations describing the derivatives of Hamilton’s principal function W (q0 .   ¥     will now be assumed regular. t1 ) that there is a unique q ∈ Υ that minimizes the action S. q1 ) Then Hamilton’s principal function is W (u) := W (q0 . p) ˙   (Q × )2 → Υ u→q deﬁned only when (q0 . so that λT Q −→ X ⊆ T ∗ Q (q. q) dt ˙ where the Lagrangian L : T Q → is a diﬀeomorphism. q) dt ˙ pq − H(q. . q1 . t0 . and assume that this q depends smoothly on U = (q0 . We need to ensure that (q0 . p(t). t0 . t0 . p)dt is a 1-form on the extended phase space X × . t0 ) and (q1 . We’ll think of q as a function of U : q (t0 . q1 . q0 ) • • (t1 . t0 ) is close enough to (qi . q(t1 ) = q1 and t1 S(q) = t0 L(q. t1 ) ∈ (Q × )2 . p) dt ˙ p dq − H dt β C = t0 t1 = t0 = where β = pdq − H(q. t1 ) are suﬃciently close. q) −→ (q. t1 ] → Q with q(t0 ) = q.4.

. So d d d β= β− β at s = 0. we can approximate it by a path that is smooth)... and the curve As + Cs + Bs has the same end-points.. . . . t1 ) ∈ (Q × )2 .. . . . . .. . ds Cs ds Bs ds As Note d d β(Ar ) dr β= ds As ds = β(A0 ) where A0 = v is the tangent vector of As at s = 0.. .. .. .. ... .. ..... .... . .. .. . ... ... .... . • CB s • . d ds β = β(w) Bs       C d d W (us ) = ds ds β Cs . • .. ...... . . .. q1 .. . .. Bs . . • ... . . Let us be a 1-parameter family of points in (Q × )2 and work out where Cs depends on us as above X× C0 D Let’s compare β Cs and As +Cs +Bs = As β+ Cs β+ Bs β Since C0 minimizes the action among paths with the given end-points. . which in turn depends upon u = (q0 . so let’s diﬀerentiate W (u) = β with respect to u and get the Hamilton-Jacobi equations from β.. . .. . . As ... . .68 From Lagrangians to Hamiltonians Note that C depends on the curve q ∈ Υ. . t0 . . ... we get d ds β=0 As +Cs +Bs (although As + Cs + Bs is not smooth. . Similarly.. . We are after the derivatives of W that appear in the Hamilton-Jacobi relations. ..

Now since β = pi dqi − Hdt.t0 .4.t1 )/ then we get ∂ψ i = − H1 ψ ∂t1 i ∂ψ = p1 ψ i ∂q1 At the time of Hamilton and Jacobi’s research this would have been new. t0 ). but nowadays it is thoroughly familiar from quantum mechanics! . t1 ) = eiW (q0 ..4 Waves versus Particles—The Hamilton-Jacobi Equations 69 where w = B0 .q1 . q1 . we get ∂W = pi 1 i ∂q1 ∂W = −H ∂t1 and similarly ∂W = −pi 0 i ∂q0 ∂W =H ∂t0 So. t0 . t1 ) as we move Cs and v keeps track of (q0 .. p1 . So. d W (us ) = β(w) − β(v) ds where w keeps track of the change of (q1 . if we deﬁne a wavefunction: ψ(q0 . p0 .

A. with generalizations: A review. Actions for gravity.Bibliography [Pel94] Peter Peldan. Thorne. 1994. W. Gravitation. New York. Wheeler. and C. Classical and Quantum Gravity. Freeman. S. [WTM71] J. 70 . Misner. 1971. K. H. W. 11:1087.