EINSTEIN'S UNIFIED FIELD THEORY M. A. Tonnelat Maitre de Conférences & la Sorbonne With a Preface by André Lichnerowicz Translated from the French by Richard Akevib Queens College, N.Y. GORDON AND BREACH Science Publishers New York London ParisFrench edition originally published as La Théorie du champ unifié d'Einstein et quelques -uns de ses développements, Copyright © Gauthier-Villars, 1955. Copyright © 1966 by GORDON AND BREACH Science Publishers, Inc. 150 Fifth Avenue, New York, N.Y. 10011 Library of Congress Catalog Card Number: 65-16831 Distributed in the United Kingdom by: Blackie & Son, Ltd. 5 Fitzhardinge Street London W1, England Distribution en France par: Dunod Editeur 92 Rue Bonaparte Paris 6, France Distributed in Canada by: The Ryerson Press 299 Queen Street West Toronto 2B, Ontario Printed in the United States of America Com osition by A.F, Michael Publications Service, Inc.Preface to the French Edition Some time ago, in a small office at the Henri Poincaré Institute, I met Mrs, Tonnelat. We used to dis~ cuss our views on fields and sources within the frame of classical relativistic theory. At the time, it did not enter my mind that I would some day have the privilege of present- ing the following book. Its author was already exhibiting a highly synthesizing mind and an intellectual honesty which make this book. As a legacy, Einstein has left an enigmatic theory that scientists view with suspicion and hope. A large amount of work is necessary to either prove or disprove the theory. Even if, as is probable, this theory does not yield the final understanding of physical fields, the work will still be fruit- ful in that it will lead to a better understanding of the unsuc- cessful geometrization of the fields and of the ambition itself. This bock does not pretend to be a complete treatise on the theory. We are, by no means, at the stage where such a treatise is necessary. However, the book may prove to be a good research tool te further work on the theory. Cutting across the multitude of original papers, vvi PREFACE Mrs. Tonnelat has succeeded in synthesizing and criticiz- ing the various points of view. She starts with an exposi- tion of the principles of the theory and analyzes the nature of the geometrical synthesis of an affine connection and a fundamental tensor of rank two which satisfies the field equations derived from a variational principle. These equa- tions exhibit the necessary geometrical and physical in- variance. The first sides of these equations satisfy also conservation identities whose importance has been exhibited in general relativity. The difficulties begin when one has to interpret the ge~ ometry to find, within the theory, gravitation and electro- magnetism and to compare these equations with those of general relativity. Two approaches are available: to find rigorous and as simple as possible solutions of the field equations or to construct approximate solutions, With re- spect to the first method, the spherically symmetric solu- tions are the simplest and the most amenable to interpreta- tion and discussion. With respect to the second approach, the manner of approximating is equivocal and implies an a priori interpretation. This labor was started all over the world and it was quite difficult to obtain an idea of the total work carried out with various methods, ideas and implica- tions. Mrs. Tonnelat's book put in evidence the considerable difficulties involved. Appendices detail certain delicate cal- culations and discuss the isothermic coordinates which seem, to Mrs. Tonnelat and myself, most important for the future developments of the theory. It was up to the scientist who, in addition to her other achievements, was the first to explicitly solve the equations tying the affine connection to the fundamental tensor to give us this book. André LichnerowiczAcknowledgments "God is sophisticated but not malicious'' —Albert Einstein I wanted to assemble here the principles and some of the developments of the Einstein-Schrédinger Unified Field Theory. This is not a complete exposition of the theory. Despite their intrinsic interest, I have systematically left out certain original work of which the point of view would have detracted form the homogeneity of this book. Notably among these researches are those referring to Cauchy's problem treated in previous texts. In a large measure, I have unified the notations used by various authors. However, in some particular cases, I have adhered to the original notation so as to facilitate ref- erence to the papers. Most of the chapters arrive at conclusions which raise more or less important difficulties. This is the common lot of all the theories. The unified field theory continues the simplicity of its principles with a profusion of calcula- tlons and a wealth of formalism. It is thus difficult to sur- mount the mathematical complications and to decide between vitviii ACKNOWLEDGMENTS the various physical interpretations to enable one to com- pare the hopes and realizations of the theory, One must not seek here a didactic exposition of the ac- quired results but the more or less happy and complete development of a theory in the process of formation. This book is just a collection of work whose unique goal is to facilitate research on the subject. The fragmentary con- clusions which are reached can be only headings of chapters for further work. M, A, TonnelatContents Preface... ee es v Acknowledgments... 20... 2.2... 2220200. vii INTRODUCTION-- Purpose and Methods of Uni- fied Theories Field and Charges in the General Theory of Relativity. 2... ee ee 1 Role and Possibilities of Unified Theories. . . 5 Unified Theories with more than Four Di- mensions... 6... ee ee ee 7 Four-Dimensional Unified Theories. Spaces with General Affine Connections. ..... 9 Einstein's Theory... ......-.--25- 11 1. MATHEMATICAL INTRODUCTION Relations between the Symmetrical and Anti- Symmetrical Parts of the Tensors g uy and ee, eee 15 Absolute Differential calculus in a Space With a General Affine Connection. ... 2... 0. . a1 The Choice of Ricci's Tensor in a Space With a General Affine Connection, ......... 25 2. FIELD EQUATIONS, VARIATIONAL PRINCI- PLES, CONSERVATION EQUATIONS Method of Application of a Variational Princi- Plena 31CONTENTS Application of the Variational Principle to Ricci's Tensor Ruv(r). .. Conservative Equations. .... THE FIRST GROUP OF EINSTEIN'S EQUA- TIONS--Expression of the Affine Connection as a Function of the Fields... ......... THE SECOND GROUP OF EINSTEIN'S EQUA- TIONS The Rigorous Field Equations .......... Approximate Equations... ........0-. SPHERICALLY SYMMETRIC SOLUTION Differential Equations of the Spherically Sym- metric Case... ee ee The Various Forms of the Spherically Sym- metric Solutions... ..........00- THE FIELD AND THE SOURCES Choice of the Metric and Fields in the Unified Theory... ee ee ee Principles of the Born-Infeld Theory The Non-linear Relations Between the Dis- placement and the Electric Fields in the Unified Theory. 2... ee eee Definition of Conjugate Fields . Current and Charge. . . . dee SOME PROBLEMS RAISED BY THE UNIFIED FIELD THEORY Energy-Momentum Tensor. ..... Geodesics and Equations of Motion Appendix I. RELATIONS BETWEEN DETERMI- NANTS 0.0... ee eee Appendix ff. APPLICATION OF THE VARIATION- AL PRINCIPLE TO A DENSITY CON- STRUCTED WITH RIGCI'S TENSOR av 65 81 95 111 112 116 122 129 135 145 155CONTENTS Appendix HI. PROOF OF THE RELATION Az=-2A,-3, As. Mo *(e- y- 9) Appendix [V. CALCULATION OF THE AFFINE CONNECTION IN THE STATIC SPHERICALLY SYMMETRIC CASE. . Appendix V. ISOTROPIC COORDINATE SYSTEM IN THE UNIFIED FIELD THEORY . . xi 165 167 175 179 185Introduction PURPOSE AND METHODS OF UNIFIED THEORIES i. FIELD AND CHARGES IN THE GENERAL THEORY OF RELATIVITY. In developing the General Theory of Relativity, Einstein succeeded in giving a purely geometrical interpretation of gravitation. Newton's law of gravitation was replaced by six independent conditions imposed on the structure of the universe and the trajectories of material particles became geodesics in a four dimensional Riemann space. This Riemannian space which serves as a vehicle for the general theory is completely determined by the line ele- ment: * ds? = gy, dx! dx”. (1) The affine connection between Riemannian spaces is well de- {ined in terms of the metric tensor gy). It is equal to the Christoffel symbols: * We use here the usual summation rule that any repeated index {ndicates a sum over the index,2 EINSTEIN S UNIFIED FIELD THEORY e a {uv} = 6 fou gay + 8p Bue ~ 80 Burbs Oy ox (2) The structure of a Riemannian space is completely determined. It differs from the euclidean structure by the existence of a curvature which is given by the Riemann- Christoffel tensor. p a Chivom Polfet ~ af fobs {Ab L2F THEE) and the curvature is thus defined in terms of the gy). 1°. In empty space (exterior case), one can obtain the field equations by imposing the following conditions on the space: Supe” (4) where S,,,, is a tensor of rank two subject to being a function of the gu p and its first two derivatives. The ten equations given by (4) are not all independent since theoretically at least, they would then determine completely the metric tensor and would restrict the choice of the reference system. There can therefore be no more than six independent condi- tions between the ten components of §),,, to enable us to choose freely the coordinate system in a four dimensional space. We will therefore assume that the S,,,, must satisfy the four conservation equations V Su Pea, is the covariant gradient. formed with the aid of the Christoffel symbols. We thus have four arbitrary condi- tions in the determination of the metric tensor. E. Cartan (2) has shown that the only tensor Sy, p Satis- the above conditions must have the form Buy = h {Guy $ Buy (G - 2}, (6) h and A being constants and G,, ,, being the contracted Riemann tensor known as the Ricci tensor: Guy = Guop = af} -3 vliph + p iit Spt teh wk Pt (3)INTRODUCTION 3 If, from (4) and (6), we form the invariant S= gly §,,=0 we obtain = ltl = G=¢ Guy = 4, (8) Thus, in vacuum, the condition (4) is equivalent to Guy = Abu (9) 2°, In matter or in the presence of an electromagnetic field (interior case) one must balance the effects of the conser- vative tensor S,,,, by those of another conservative tensor: this tensor will represent the energy-momentum of the field or the matter. We will then have Sy) XT), With Vp T,P = 0, (10) x being a constant. For h-=1, we have 1 Guy x Buv(G - 2) =xTyy (11) The above relationhas aparticular character, for while the interpretation of S,, ,, is a purely geometrical one, that of Ty,» is not. Further T,,, contains: a. The energy momentum contribution of the electro- magnetic field which in Maxwell's theory is represented by 1 p Tuy = Suv Foo FP? - Fup Fy. (12) b. The contribution of a distribution of matter of den- sity u. Moo = # Yp Yor (13) Uy being the four-vector velocity of space: LB ul = Be, wus (14) If we assume the simultaneous presence of charged material particles and an electromagnetic field, the con- dition4 EINSTEIN'S UNIFIED FIELD THEORY Pp. Pl. ep VoTy =O or Vp Mu = Vp Tu (15) becomes Vy(uu, Wy) =-F a (16) Xr o or if Tyy is the Maxwell tensor as expressed in (12) and Dag a Fe, an Equation (16) is identical to the first group of Maxwell's equations if we assume that J* is the four vector current. If this is true and if we accept Lorentz's hypothesis that all currents are convection currents, then we always have: DP =put (18) p being the charge density. If we multiply (16) by u? and sum over o taking into account (14), we have (uur) = 0. et) This is just the continuity equation; (16)can then be written as A p A w Va Ug = ap oa Ue (20) 1°. The trajectory of an uncharged material particle (p= will be wy) u,=0 (21) which can be rewritten . view of (14), in the form: ax?” ri on ds = = 0, and can also be obtained from the variational principle 3 fds = 0. This is the equation for a geodesic in a Riemannian space defined by the line element ds* given in (1): thus the tra- jectories of material particles coincide with the geodesics of a Riemannian space.INTRODUCTION 5 2°, In the case of a charged particle (p #0), Eq. (20) shows that the trajectories differ from the geodesics of a Riemannian space. They could be interpreted as geodesics of a Finsler space in which the metric tensor is determined from the element 1 2 & ds" = (Buy dx" ax”)? im axl, by putting e/m = p/u and assuming that Pu is the four vec- tor electromagnetic potential such that Fy, = 6, Py - ah The ratio e/m varies from one particle to another. We must therefore associate with each particle a particular Finsler space whose geodesics will be the trajectories. 2. ROLE AND POSSIBILITIES OF UNIFIED THEORIES. General Relativity separates quite radically the gravi- tational field from a, the electromagnetic field which has not had any ge- ometrical interpretation and forms, with the gravitational field, a heterogenous ensemble. b, the sources of field which conserve a phenomeno- logical interpretation even when one talks about uncharged particles. The so-called unified theories attempt to remove the heterogenous features of the combined gravitation and elec- tromagnetic fields. Other theories, neglecting this problem, have attacked the duality of fields and particles in the purely electromag- netic domain. Such are, for example, the theories of Mie and Born-Infeld (122). Their theories have not attempted a geometrical interpretation of the fields. Despite this, they have often been labeled unified theories. To remove any confusion in the use of this term, we shall refer to them as non-dual theories. If we were to take into account all the new problems which would arise in the integration of material tensors in a purely geometrical scheme, it would seem that the6 EINSTEIN'S UNIFIED FIELD THEORY difficulties, which would be encountered by a non~dual uni- fied theory, would be insurmountable. It was Einstein's ambition to resolve these difficulties. This, perhaps, is not impossible, since, quite often, a question properly asked would remove in one step all the difficulties which, separately, are irreducible. We shall say that a unified theory is one which will unite the electromagnetic and the gravitational fields into a single hyperfield whose equations are the conditions imposed on the geometrical structure of the universe. 1°. The equations which will describe the behavior of the gravitational and the electromagnetic field in empty space are, by definition, relative to a unified ''exterior'' case and are always written as 8,79 or R pe uy = Rj» being the Ricci tensor formed with some given affine connection. These equations correspond to the 'tinterior™ case of gravitation and electromagnetism proposed by the general theory. 2°. Inthe presence of matter, the equations of a wmified theory will include an additional term since we are not, in general, attempting to integrate the field with its sources, To impose on the structure of the universe the addition- al conditions that will lead to the electromagnetic field, we cannot stay in purely Riemannian space. The description of such a space is complete when the metric tensor is given. The conditions imposed on the curyature tensor in such a space are sufficient to lead to the interpretation of the gra- vitational field. The electromagnetic field cannot be de- scribed in it. Additional conditions on the curvature can be obtained by going to a more complex space than the one used by Einstein for General Relativity. This can be done in one of two ways; a. by increasing the number of dimensions of a Riemannian space.INTRODUCTION 7 b. by going into a more general variety of spaces with any affine connection which will give us more latitude in the definition of the parallel displacement of a vector along an infinitely small closed contour. 3, UNIFIED THEORIES WITH MORE THAN FOUR DIMENSIONS, The theories with {in spaces of) more than four dimen- sions occupy an important place in the development of uni- tary theories. The first of these was advanced by Kaluza in 1921 (1), (10), (12). This was further developed by Einstein and Mayer (1), (12), in 1932, A recent modifica- tion of the basic postulates has led to theories with 15 field variables by Jordan (1947) and Thiry (1948) (10), (12). All these theories are developed in a five-dimensionai space whose geometry leads to a formalism which is convenient for the interpretation of the laws of the generalized fields. For this reason, these theories are all subject to a cylin~ drical condition which displays clearly the peculiarity of the fifth dimension. Only those coordinate transformations, which satisfy the cylindrical condition, will lead to covari- ant field equations. A further development of the five-dimensional formal- ism has been carried out by the ''projective'' theories. The cylindrical condition imposed on these theories can be interpreted in a natural way as a projective condi- tion and shows more clearly the purely auxiliary role of the five-dimensional space. Along with these theories, other developments have been carried out in a different way. To circumvent the in- troduction of a pure formalism, one can assume that the physical space actually is five-dimensional. The difference between this postulate and the above (i.e., between a geo- metry and a geometrical formalism) lies in the interpreta- tion of the cylindrical condition. The condition becomes a structural hypothesis which must be satisfied by the five- dimensional physical space. The Einstein-Bergmann- Bargman theory (1), (12), assumes that the five-dimen- sional space is closed by the coordinate x°®, Thesix-dimen- sional theory of Podolanski (12) gives the space a sheet8 EINSTEIN'S UNIFIED FIELD THEORY structure, These assumptions, although gratuitous, are nearer to the spirit of General Relativity which lead toa geometrical character of gravitation. The great advantage of five-dimensional theories, or at least of some of them, is the following: in a four-dimen- sional Riemann space, the trajectories of charged particles are not geodesics; it is possible to interpret these trajec- tories as geodesics in a Finsler space under the condition that for each type of particle characterized by the ratio e/m, a different Finsler space is used. It is possible to show however that one can obtain a parametric representa- tion in a five-dimensional space corresponding to a family of trajectories relative to a given e/m: these trajectories then coincide with the geodesics of a five-dimensional Riemann space. This remarkable result would push us to develop a uni- fied theory in a five-dimensional formalism. Einstein's and Maxwell's equations can be brought into a satisfactory unified formalism. Aside from criticisms particular to each theory, a general criticism that can be applied to five- dimensional theories is the introduction of an additional con- dition, namely, the necessity of using the cylindrical con- dition. In particular the criticism that has been leveled against them is that they lead to a simple codifying in afive- dimensional formalism of the equations of Einstein and Max- well. No matter how interesting such a synthesis is, it should lead, as General Relativity did, to predictions that will confirm or reject the theory. It is in answer to these objections that the Jordan-Thiry theory with 15 field yariables was developed. Essentially, the consequences of these theories are: i. The gravitational constant x becomes a weakly varying factor, the variations being a function of the varia- tions of e/m. 2. The laws of Einstein and Mawell introduce addition- al terms due to the variation of x. If x is constant, one re- verts to the classical laws.INTRODUCTION 9 3. There is a fifteenth field equation relative to the variations of x which implies that in the absence of any charge (p= 0), the presence of matter (u 4 0) leads to the creation of a magnetic field. In this way we arrive at the prediction of the existence of a field due to matter in motion and particularly for a rotating body (Blackett effect) . 4. FOUR-DIMENSIONAL UNIFIED THEORIES. SPACES WITH GENERAL AFFINE CONNECTIONS, Along with five-dimensional theories, other theories, since 1918, have attempted the synthesis of gravitation and electromagnetism in the context of a four-dimensional space- time. They must therefore start from more general multi- plicities than the Riemannian variety in order to enable one to impose supplementary structural conditions that will coincide with or better modify, the classical electromag- netic equations. The Riemannian space, which is described by a single type of curvature, will be replaced by a more general continuum formedby considering a space witha more gen- eral affine connection, Cartan (3) has shown that the structure of sucha space defined by the coefficients Ty? will, in general, have a. A rotational curvature defined by the tensor af which generalizes the Riemann- Christoffel tensor* tie p_it veo Qf =~ F Rare Lax” 6x] ly, Pp p A pP nr we 3 Poluy - PP ue + Pav Tao “Pho Pavt 4 [dx” x7] . * The [dx" 6x”] denote the differential increments and are: fax" ax”] = dx 6x” 6x4 dx”.10 EINSTEIN'S UNIFIED FIELD THEORY pb. A (homothetic) curvature which suffices to character- ize the invariant 2 = ot and which is zero in a Riemannian space. ¢. Atorsion which is zero in a Riemann space and is given by p_ pe Hp v OP = -T i, [dx” 6x"). Most authors did not immediately perceive the nature of the elements a, b and c at their disposal. However, any affine four-dimensional unified theory assumes the existence of one or several of these elements. All these theories, al- though apparently different, can be classified in three dis- tinct categories depending on whether they assume the ele- ments a-b, c, a~b-c, which are used to characterize a Riemannian space and develop the general theory of relativ- ity. The first theory of this type was developed by Weyl (13). He assumed that a parallel displacement does not only modi- fy the direction of a vector but also its magnitude. This as- sumption is equivalent to postulating the non-vanishing of the homothetic curvature and the conservation of elements a and b. Analyticaily, this assumption introduces gauge trans~ formation in addition to coordinate transformations, The affine connection is then expressed in terms of the metric tensor and a four vector ¢,, tied to the gauge transforma- tions. Thus, to determine the structure of the universe and the field equations, it is necessary to use the ensemble (Su %y) and not simply ds’, Weyl's theory does not modify the symmetrical charac- ter of the space: this is equivalent to assuming a non-tor- sional space. The torsion will lead to a non-zero value for a closed infinitely small contour. The torsional spaces in- troduced by Cartan (3) were used in unitary theories by Eyrand (1926) (12), by Infeld (1928) (12), and finally by Einstein (5), (12), (125). A theory developed by Einstein (12) has even been based on the existence of a strictly tor- sional space which permits the definition of absolute par- allelism. In many of their features, the theories of EyrandINTRODUCTION 1 (a, b, c) and Infeld (a, c) and Einstein's theory (1923) (5), (125) are forerunners of Einstein's unified theory which we shall describe. Unfortunately, they have un- necessarily limited the formalism at their disposal and did not succeed in exploiting the richness of the structure of a general affine connection to derive new laws. 5. EINSTEIN'S THEORY. Einstein's theory starts froma general formalism which, from the beginning, conserves without restricting properties a, bande. If afour-dimensional unitary theory willgive satis- factory results starting from a variational principle, it will be within the context of this theory. As we shall see the price that has been paid in return for the completeness of the theory lies in the ambiguities at its foundations. A serious difficulty arises in attempting to resolve these ambiguities and to choose, in a consistent way, the contracted curvature ten- sor, the metric tensor and the fields. The theory has the advantage of not restricting the possibilities offered by any affine theory. We stated earlier that the often justified criticism leveled at unified theories was that they restricted them- selves to synthesizing existing and well-established laws. However, although the theoretical interest in a geometrical synthesis of gravitation and electromagnetism is undeniable it is nevertheless true that it is rather useless to recover Einstein's and Maxwell's laws in theories that introduce a complicated formalism, One would wish to go further. Einstein's theory opens new perspectives for us. These are: 1. The laws of Einstein and Maxwell are modified. They contain additional terms which represent the interaction of the gravitational and the electromagnetic fields. It seems thus that one can justify the creation of a magnetic field by a purely material distribution. 2. Inthe general case, this type of effect is difficult to bring out due to the complicated nature of the unified12 EINSTEIN'S UNIFIED FIELD THEORY field equations. In the particular case of spherical or cy- lindrical symmetry, one can obtain rigorous laws for this particular case. One can thus determine precisely the laws obeyed by the field of a pulsating or rotating body. This would permit a justification or a modification of Blackett's empirical formula. 3. The laws of electromagnetism which result from Einstein's theory are not linear. They seem to predict, theoretically at least, some new effects, such as the scat- tering of light by light and the possibility of defining a field in the center of a particle. 4. The relations between inductions and fields in Einstein's theory (or at least between certain quantities tied to inductions and fields) have the same form as in Born's theory. In the case of a static spherically symmetric solution, one can show the existence of a field which is finite in the center of the particle. One would think that Einstein's hope to integrate with the field its singularities and the sources of the field ina geometrical synthesis can be realized. It is difficult to be so positive about this hope. Assuming that even this first objective has been attained, it is difficult to see the next step which will lead the theory to a reinterpretation of quantum theories, a step which Born's theory was unable to bridge in a satisfactory way. Perhaps as we stated ear- lier, the problem considered as a whole, namely unifying the fields and unifying fields ahd sources, will be more easily solved if stated more judiciously. . In an immediate way, the other difficulties of the theory result on the one hand from the richness of the formalism and on the other hand from the complexity of the solutions, On the one hand, one encounters ambiguities which must be resolved, interpretations which are not always clear: the choice of the tensor R,, ,, used in the variational principle, the ''true'' metric, the determination of the energy-moment- tum tensor and the derivation of the geodesics. All these points raise problems, a few of which are on the way toINTRODUCTION 13 being solved. On the other hand, one must handle compli- cated solutions which, except in particular cases and some~ times even in these, hide the physical contents that one could find. Despite this, a number of satisfactory results have al- ready been obtained, Others need to be studied further. Perhaps in this fashion, we will be able to proceed toward a clearer conception of the nature of the sources of a field and from this to the scope of the theory. One can detect in both Einstein and Schrédinger a mixture of discouragement and great hopes in reference to this point. Be that as it may, Einstein's theory unites the realiza- tion of a satisfactory synthesis, obtained from a general principle, to the possibility of new predictions in the classi- cal domain, Such are almost always the signs which charac- terize a fruitful and important physical theory,Mathematical Introduction A. RELATIONS BETWEEN THE SYMMETRICAL AND ANTISYMMETRICAL PARTS OF THE TENSORS Buy and g” (54), (57). 1. DEFINITIONS. Consider a tensor of rank two g,,,, whose determinant is g and let gg” be the minor relative to each element gy p. We have by definition: uo OU _ so Eup & Bou = & (LD dg = ge? dg, = - By, ag” (1, 2) whence we have 8B ny = ~ Epy Bug d8°%, agh” = - gPP gHO dgig. (1.3) Let us now write g,, , and g"” in terms of their symmetric parts yy, and h! Vand antisymmetric parts? , and fH, respectively: Sup = Yuv t Pup (1.4) gh a nly hh (1.5) 1516 EINSTEIN'S UNIFIED FIELD THEORY We will now adopt the following convention: the same letter will be used to denote the covariant components of a tensor, its determinant and the minor relative to each of its ele~ ment; thus y and ¢ will denote the determinant formed by the elements yj y and Yj, y and yy” and pgHY the minors re- lative to these elements. In an analogous way h and f will be the determinants formed by the elements buy and fay and hh” and ff/ will be the minors relative to these ele- ments. We always have: og. AO LO _ so Uo so Yupy? = 6 > Pup PO = OF by, We =F, (1,6) Mo go. fup 1° = og | = yy = hy dy = Ay yyy IP = PPT” Pay pple pple (1.7) dh = hh’ dha» gi = ff" df. Let ey ypo = «4 ¥P° be the Levi-Civita tensor (equal to +1 as to whether we have an even or odd permutation with respect to the indices and zero if any two indices are equal). The relations between the covariant and the contravariant components of an antisymmetric tensor are: 1.8) vg p uv. 1 wyps, ( Pure % EuypoMs P= Fe Fog i t aE po YF coupes 9) wy ="Z €uypo* » wr whence the determinafts ¢~ and f are: 1 VG =e POY Gag = (Pr2 Poa t Yor Pea * Pa Pu) (1, 10) 1 VE = cAMP Ey fog = (fue fos + fas fos + fe fs) (111)MATHEMATICAL INTRODUCTION 17 For a symmetric tensor of rank two, we will have:* BAL VT BPAT, YEuypoY Y =© Nptus (1. 12) MAL yt _ PAT yp Te NT hg (1. 13) We wili always assume that g, y and h are different from zero, Thus, we can always compute y#¥ and hU¥ starting from y,,) and h,, [fg and 1/f are zero, the def- initions of g#¥ and fyy as given in (1. 8) and (1.9) have no meaning. In this case, we shall consider the expression V0 gHY and tu ypAT which will always have a meaning since they represent 4 «po Pp and 3 €uvpg 1%. he i 2, FUNDAMENTAL RELATIONS. The determinant g can be expressed in terms of the symmetric and antisymmetric determinants y and y. In the same fashion, one can express the minor gg" in terms of yvE¥ and gp4¥, Thus one can write: = x bP Ww BaytOrtg yy Puy ?po (1,14) by we by up w Lp ow se = yy +t PP + yy YY Pog + OF e Yoo" (1. 15) In a reciprocal way 1/g, 1/n, 1/f are the determinants of g#¥, hUv and f{#¥, We will thus have relations analogous to (1.14) and (1.15) between 1/g, 1/h and 1/f: Liijioa LY ppc, B7 RE top up bot? (1. 16) Lau wy Luv po po ze h tri +h Bu phvsf +f tiptyoh. (117 * We note here a typographical error in the corresponding formu- las (12) of article (57).18 EINSTEIN'S UNIFIED FIELD THEORY Let us separate (1.15) and (1.17) into their symmet- ric and antisymmetric parts. This leads to (1.18) and (1,19). One can also obtain the reciprocal relations for (1.18) and (1.19). (See Appendix 1.) This leads to (1. 20) and (1.21). These points are summarized in the following table (54), (57): Symmetric part (s) Antisymmetric part (a) pl Kye 2 gHP py, teY 8 ob BELG MP Won, g po ete (1. 18) gr po ve uy B Bup By f 8 2 - Yay Bayt F fupte Puy= 1) heyy yt Pup? wv? favs Puy Muplne (1. 20) v. _ply +P Pn ph v a gh, pUP pW 5 o (1. 21) If we multiply (1. 18s) by (1. 20s) and (1. 18a) by (1, 20a), we obtain the following conditions eg =yhate (1, 22) We note here the relation (1.7) (see Appendix I): eget OM y My = OF (B-¥- 9) - vr Ty OL Pap (1. 23) which reduces foro =Tto OP OP ay Xap = ¥YHO YY Puy Po_= WE-7- %. 24) REMARK (case y = 0, 1/f= 0). (1. 20a) and (1. 21a) in the form given above hold only if g 4 0 and 1/f 40. If both g and 1/f are zero, one must rewrite (1.8) and (1.9) in the more meaningful wayMATHEMATICAL INTRODUCTION 19 We ph = eHupa %p¢ £7, (1, 25) 2 ft = JP BY * Euypo From Appendix I, one then obtains: uv, £ yuo ,ve ay Ne Pi Vp o # ne? HPO tng F faye : Yuptg? (1. 26) These are just (1.20) and (1.21) where we have used V9 HY and fyyI/VF instead of g#Y and fy, This is the recipe for obtaining the general forms of Nequations (4. 20) and (1.21). As we shall see later, it is always the products VO oHY ana fy,»/VE that appear in the theory. 3. CONJUGATE VARIABLES, Consider the expressions L=Wog= Woy (1. 27) with = fe. f. @ 1 ue, ve 3 L & : fis2a5y y % nv Pog} (1. 28) 1 -fpahat wy spol? = [L+F +3 Buy Bygt pot From dg = gg de, = -V=eg,,, age” (1. 29) one can verify that er eT 1.30 6 guy Sue” oie ghey “ef Separation of (1.30) into symmetric and antisymmetric parts leads to; raw Hee, 22 | gave _ 82 t 2 Beap eByp 2)0eny 8p) (4, 31a) 1 al aL 1j aL al =e meet agi +) (LS Yuv 2 agus pit Puy Bg sal (1.31b)20 EINSTEIN'S UNIFIED FIELD THEORY On the other hand, £ is a function of gy» and one has by definition: ak tj ac | ee } df = = dey = 3\so— + dy, 3g, AD ise ag ae aad wy Buu J (1, 32a) 1s ol al +5 - di 2e8uy re | Puy aL 1({ aL al | solv dé == ght = Ho + aguy ByeGhY "age | (1. 32b) 1) aL al Ly ~asph - pir} as ; A comparison of (1.31) and (1.32) leads to:* al 1f ae al bY aes ee ta ay, ke og \ , Ly LY bE (1. 33a) al if aL ae Logue Guy 228uy BS al ij af | ae of) aIY * 2)yeh? vu pth logh? “ag (1. 33b) ae af ek el. agh¥ = 2)agRy agua) ~ fue As wehave not taken account of the symmetry property Of yy y= Ypy and ICH = x¢¥" and the antisymmetry property of Puy =-Pypand gv an Gh the relations (1.33) are quite general. In particular the fields UV and Puy are conjugate to each other with respect to the scalar density v=g. One can see from the expressions for h#Y, fH ¥, yyy * We note here that the notation 02/aA,) represents the partial derivative of £ taken as if the ,,,, OF A,,, was an independent variable.MATHEMATICAL INTRODUCTION 21 and @,,, as given in (1. 18) and (1,19) that the relations (1. an”, are correct when £ is given by (1.27). We note also that (1.33) can be written as* uy 2 ak avrg al ar Puy? 3 aa (1. 34) B. ABSOLUTE DIFFERENTIAL CALCULUS IN A SPACE WITH A GENERAL AFFINE CONNECTION. 4. COVARIANT DERIVATIVES: For a symmetrical affine connection Tey , the parallel displacement of a vector can be defined unambiguously by pau. - Ip & ox". (1. 35} On the other hand for a general affine connection re » we can define the parallel displacement by one of the two re- lations: Bw nH Ao ap bA=-TY, AY 6x (1. 36) Bop Ao ae 6A = Too 4 ox (1.37) * By letting L He op HY G ee e x minorG ay mdb ; ae 2 eR = x minor 1, a KH {he determinants @, 3¢ and ¢ of mn and Fay are 1 Y 7 ee Fe y aS She st«ebel Geo e Whence we have: 228 Lan {1s E15 yp ww THY pt22 EINSTEIN'S UNIFIED FIELD THEORY which conserve the tensorial character of 89 AU + TH, A®. We can thus define two types of covariant derivatives: ALD = a,ae +Th, ae (1. 38a) = Hpk ao AL =BpA" +T hg A. (1. 38b) In the same fashion, we have: Ausp = 2pAy > Tilo Ag (1. 39a) + Aus = 8pAy ~ Toy Av: (1, 39) If the connection is symmetrical, then (1. 38a) coincides with (1. 38b) and similarly for (1.39) and we have: a o Ao = OpAl + 1h, A (1. 40a) = -To Asp = 8pA, ~ TE, Ay. (1. 40b) Lett fy and rf. p be respectively the symmetrical — Vv and antisymmetrical parts of Vip : Pp pe p Thy = Toy Puy (1.41) If we contract the indices p and v, we have Ty, =Thy , Ty v being the torsional four vector of the space. The transpose of PP is defined as By Pe cP Tt ip Tou (1. 42) such that re = Civ - ry. (1. 43) VvMATHEMATICAL INTRODUCTION 23 5. TENSOR DENSITIES. ; From (1. 39) the covariant derivative of a tensor Sup is; Co Oo Euvjo = *p8uy ~ Pup Sev ~ Tay Sus ey (1. 44) Multiplying (1.44) by g””, we obtain: », = gt “pe Ht Ben yp = 8° pSuy~(Thp + Top) ayV¥ 8. j aly 20NE pel. (1. 45) ve 7 He y¥~g is a tensor density. Its covariant derivative will then be defined by (¥=8) jp = ~VE Tho - (1. 46) We will then have: BMY giuasp = # = (V8) sp. (1.47) If we multiply (1. 44) by gid gT’, we obtain: Bro = ane + Vp gr? ye gh, (1. 48) If we define the tensor density GhY 2 ime gh, (1.49) its covariant derivative will be Gey = paGhhs rh, Gm +rZ, gue - gh? rg, (1. 80) by taking into account (1. 46).24 EINSTEIN'S UNIFIED FIELD THEORY 6. FUNDAMENTAL IDENTITIES BETWEEN THE ght op and between the gy uv;p]- ge BP pp i a. Let us form oan gh =o: By contracting (1.50), we have the identity: Lippe _ PR VL LP _ sche a1 Sep Ge aspp® Op EO" HOT, (1. 51) by using (1.5) and the corresponding notations for tensor- ial densities, GY = 5ch¥y GHY (yeh) = Vag nb? GEV VTe hy, (1. 62) b. Let us form the difference between the circular permutation of Suvio and Byuse- This leads to: +e ~ 1 gl (Burp * Bpyjy * Srp;u) ~ (Byu;zp * Bpvsp * Sup;y)t +> ++ +. + +- + = 12 rp t% Puy ty Pout *2T uy, 0 Tony (1. 53) + Typ yl Vv where we have used (1. 4) and defined* = o Purp > Yoo Pity - (1. 54) Vv Vv The two identities (1.51) and (1.53) are completely analogous. One refers to the vector f¥ and the torsion Tp: ue 1 f =Fe eh, Tp) = Thy (1. 55) g v * We note here that the comma in Tay, p simply separates the symbols and is not a sign of differentiation.MATHEMATICAL INTRODUCTION 25 the other fo the pseudo-vector I! and the pseudo-torsion * Tp T (Bo Ba}! 1. uypo Mage (8) Png + 8g Pyp +p Yard (1.59 eo ge vy Ayer po Pp = Ty Be] ar) pony yPT Py, 1 —— BUNT oT WS {ypu Yov € y aa! (1.87) i MAT, =- € (r +7 +T ). AT Y x 6i-y ye vr, T a C. THE CHOICE OF RICCI'S TENSOR IN A SPACE WITH A GENERAL AFFINE CONNECTION. 7. POSSIBLE CHOICES FOR THE CONTRACTED CURVA- TURE TENSOR. In general Relativity, the contracted curvature tensor of order two can be defined without ambiguity. This is Ricci's tensor as given in (7). In the case of a general affine connection, we must take into account the two possible forms of the parallel displace- ment of a vector: aap = - rey AM gy” (1. 36) P_ ope ak avi pe oak a oAT = - Th, AM oy? = - PP, AM oy”. (4.37) These two forms permit the definition of two tensors suscep- tible of generalizing fhe Riemann-! Christoffel tensor (one being derived from Thy and the other from Pity: : pe. 58) Rio (P) = agT uy - aT he +r, Ty The re, (1.58)26 EINSTEIN'S UNIFIED FIELD THEORY p = p PPA RP FA FP Raw!) = afy - ale + Py Tele Thy e pe A aP AP 28g 0T lout yn Tor > Toul ya (1. 59) By contracting p and o and p and up, we obtain from (1. 58) two contracted tensors known as tensors of the first and second kind, A similar process applied to (1.59) leads to two other tensors, . a re) aA pe Re Rp) = Ty yl ap * Pay Pap _ph (1. 60) up Av? = R? 3 rP p Puy REAP) = 80 p,- 8nl oy (1.62) pe Mog PP op phar pe Ri y= Ruy OP) = ep ou - Pou * Typ Poy » Toy re (1. 62) P =R° (De Po p Puy = Rouy T= 8 pT ip ~ 2:0 bp - (1. 63) If the connection is symmetrical, it is clear that (1. 60) and (1. 62) coincide and that (1.61) and (1. 63) are zero. For a general connection, this is not the case and, a priori, any combination of the above four tensors and the tensor Tuv=TyuTy (1. 64) can be considered. 8, CHOICE OF Ry) a. Hermitian principle. To limit the arbitrariness in the choice of the tensor generalizing Ricci's tensor, Einstein first proposed theMATHEMATICAL INTRODUCTION 27 adoption of a hermitian principle (14). To this end, he de~- fines the transposed quantities Buy = Bvp (1. 68) P . pP Py = iu (1. 68) According to Einstein, a tensor Aut; g) is said to be hermitian with respect to the indices . and y, if, upon in- terchanging u and y and changing g and T into ¢ and f re- spectively, the new tensor is equal to the old one Au p(T, 8) = Ay (T, 8). (1. 67) The tensor is said to be antihermitian if Ay p(T, g) = - Ay (Bs z). (1. 68) According to this scheme, we can then reduce the basic tensors by considering only those linear combina- tions of (1,60), (1.61), (1.62) and (1.63) which are hermitian. We therefore would consider only the following tensors: 1 oy 1 Un = 5{Ruy + Ruut - GPuy + Pout G. 69) Buy 3 {Puy +P, (1.70) a (1. 64) Explicitly: U.= Pay “75. te Pipt (1.71) + rh ON - ri re, Buy = uly - oy Py (1.72) It seems somewhat advantageous to consider instead of U ep the combination28 EINSTEIN'S UNIFIED FIELD THEORY 1 i Une tg Huy Flay (1.78) and we can see that this combination whigh, a priori, is really artificial, is exactly the tensor Ruy given in Appendix Ii. Ors Rub (LT) @) Ryyp being the Ricci tensor formed with the affine con- nection (),e opp lege p Liv = FR, + 5 {oy T,- of rt (1.75) whose torsion-vector is zero. p= Mis <0 (1.76) v b, Connections with null torsions. ‘These results lead us to consider more or less compli- cated linear combinations of the Ricci tensor formed with the help of special affine connections, namely those with zero torsion vector. In fact, starting with the general con- nection rf py» one can form the two connections 2 (),e pe p Liv =Tiyt x82 Ty (1.09) Qe 2 re, Lege p uPy= Thy + lof Uy - of ri}. (1.78) The anti-symmetrical parts of these connections are the same and in particular the torsion vector is null: _Q),o _ (),0 _ by = Bog = “po =0. (1.79) We can then define, from the affine connection Lf, py » three tensors of the second order (54) OR ay = Rhy CL) (1.80)MATHEMATICAL INTRODUCTION 29 (2) a Ruy = Rey CL) (1.81) R= FRE yp CL) +RP pL} (1.82) of which only the third (1. 82) is hermitian and is the tensor suggested by Einstein in the third edition (1950) of the Meaning of Relativity. In short, if we do not retain the hermitian principle, it is possible to adopt a basic tensor which is some linear combination of eqs. (1.60) through (1,64) and, in particular one of the four tensors R,,,, (1) Ruy» (2) Ry, (3) Ry. It is to the basic tensor that’ we will apply a variational principle In what follows, we shall use the tensor p Roy = Thy OThp + Th, rh, - re Thy (4. 60) which we will call the Ricci tensor and to which we will apply the variational principle. However, we shall examine in Appendix IL the results of a variational principle applied to the tensors L Ruvs R,,, and (3)R,,,, which are analogous to Ricci's tensor but which are formed with an affine connection whose torsion is null.2 Field Equations Variational Principles Conservation Equations A, METHOD OF APPLICATION OF A VARIATIONAL PRINCIPLE. 1. MIXED THEORY (EINSTEIN). VARIATION OF A FUNCTION (5(g#”, Thy ag Ty). The mixed theory assumes that the variations of the affine connection and the metric of the space are independ- ent of each other. We start from a variational principle applied to the scalar density ©: 6 fedr=0 (2.1) © being a function of the affine connection T. P py, and its first derivatives and the Poneralized field density g4¥, We assume that rh and g’” are independent variables and that their variations are, and dg” vanish at the limits of integration. With these assumptions, we obtain an p as sop Tay + see 6 S| Se, MY” a(agr py) oP fv) . (2, 2) as + agEr oct” Jar=o 3132 EINSTEIN'S UNIFIED FIELD THEORY If we integrate the first two terms by parts, we have: as ae p ) or? + 89( ap 8 i; Tey yw (secre) ue as p | -fa or ar. ( aa orh, 7 By The second term of (2.3) will not contribute to the integral since 61, y® is zero at both limits. Thus (2.1) can be written as (2. 3) Sian” orf, + Jay igh} ar = 0 (2.4) where Ev aH 3a ns) G Bee ese loos (2, 5) P ork, axe Goss ) : (2.6) The basic equations of the theory will therefore be: . Gh? =0 { a Sp (2.7) be Jpy = 0 (2. 7a) are Euler's equations relative to the variation of the re vy REMARK IL If we had imposed some conditions on the GHY, for example ap @HP = 0, the variation 5g” will not be independent and this would have to be taken into account in the variational principle. To this end, one can use the method of the Lagrange multipliers. If we impose the above condition to the G4, we would have to add to H the scalar density -20,, &p GHP, o, being an arbitrary four vector. The variational principle leads to Sigg’ ory + (uy +2yo, ~249,)6gh} ar=0 (2, 8) which leads to the field equationsVARIATIONAL PRINCIPLES 33 a ay” =0 ba, THY a0 oo I Tuy ty Fp, “8 y oy =o REMARK I: a. If we vary the scalar density o=g'’R,, (2. 10) Ry p being some tensor which is a function of the rf py and their derivatives, we would obtain, since Ry) = 3 $/8 gh! = for the field equations (2.7) ve a. GHY 20 e (241) b, Ryyp= 0 b, In varying the scalar density H' =H - 2.7, (2, 12) the variation of the @#” gives the additional term 22, 2B 5 guy -rguy gh, — (2,18) aghy since avme 1 aghv “g8ur The field equations will thus be: hy . 20 a G p (2, 14) b. Ryy - A Sup = 0 2. PURE AFFINE THEORY (SCHRODINGER). VARIA- TION OF A FUNCTION 9(Tfi, ,®¢Tf,). a. Instead of assuming that I iy and G4” are inde- pendent variables , we can assume that the only independent34 EINSTEIN'S UNIFIED FIELD THEORY variabies are the rh . Thus © depends only on rhy and their derivatives and for an increment arf, , the action integral leads to faby arf, dr=0. (2. 15) Thus the only field equation derivable from the variational principle is equation (2. 7a) oh =0, (2, Ta) b, We can obtain a second group of equations by de- fining the GH” from the scalar density 9: cHy = (2, 16) Re being a function of re, and their derivatives. It Hows then that 5 must 4 the homogeneous function = R,, (2.17) with the understanding that the variation GUY cannot yield an additional relation since the GH” are not independent variables but uniquely defined by (2. 16). To proceed further, we must use (2. 16) and conse- quently, we must give the explicit form of the action func- tion in terms of Ry. Most of the time, we adopt the func- tion 2 D = det Ry (2, 18) We can see then that (2. 16) is equivalent to: Gl! = - ete 1. a (det Ryy) (2, 19) w-detRyy Ruy which leads to: minor of guy _ 1 minor of Ruy vg dY-det Ry Ruy = uyVARIATIONAL PRINCIPLES 35 wy ja Gp =0 (2, 20) Lb. Ruy =A up B, APPLICATION OF THE VARIATIONAL PRINCIPLE TO RICCI'S TENSOR Ryy{T). 3, FIELD EQUATIONS DERIVED FROM A GENERAL AFFINE CONNECTION, The strong system. Starting with the principles of a mixed theory, we shali determine the variation of an assumed homogeneous scalar density. 8 (gM, Thy, agT fy) = GY Ruy (2.21) We shall not impose a priori conditions to the variables. The field equations will then be given by (2, 11) a) 8 as a G = - ee (2,11) \ p arp, ax? 3 (o.0 ft) | b Ry =o. Let us write explicitly these equations in the case where R, ,, is Ricci's tensor My = Pp. p o Ply p Ruy =2pTav~ Tap + TY, Th - re, rey. (2, tla) then becomes:* by = cv 2Rw 8 fey Ry) _ Gyr eG arg axe aG TH) ® (2,22) * We can also generalize Palatini's method for 2 nonsymmet- rical connection, We then have 8R - (8) 3 (or 2) ~2 fap? we . . or Pepe BV bef "OVE ve fu. Seo Elnstein (14) p. 143,36 EINSTEIN'S UNIFIED FIELD THEORY with oRuy € -B -Y P_ pe ph pY xp acre ° fp 8; 8) 8g - by By Op Of, By (2, 23) BRypy PY .F 0 BY Papo re, = 8, by GT gy +85 55 Of Thy Bey go pp BY gP po - by 85 by To, - 8 8 8 Tip (2, 24) whence = Y 2 gt 9 eh¥_ P _ sv - Gy =0,g°%- oF a gh -ghrrf - a} gure, oP? rZ, + oty rh, 20 (2.25) contracting @ and y and then @ and f leads to: - 394 GP2 - p gee Tp - ag 8, =0 (2,26) 4 92%=0, (GAB = set +74), (2.27) We solve (2, 26) for GHY¥T ww and substitute its value in (2. 25), obtaining: a Y 0g Gy - GRY rh + 5 ay Gri + Girt, + G°rr8, <0 (2, 28) whence Y 2 CM sate oF chr - GaY ry =0. (2.29) The field equations in a space with a general affine connection will then be Be 2 a. @* p=-3 Op gh r+ gt? rs, ay, gtP -0 q b. Ry = 0.VARIATIONAL PRINCIPLES 37 Equations (I) are deduced from a variational principle and are necessarily compatible. Einstein suggested that one should add to I, the condition Tp = 0. by This leads to G@*";p = 0 and as a consequence of this con- dition (1, 51) leads to au guP = 0, Thus we have the fol- lowing system: uy @* sp =0 Tp = 0 (a, 74° = 0) (F) Ry = 9 These are the equations of a strong system and are not deducible from a variational principle. 4, CHANGE OF AFFINE CONNECTION. The weak system. We shall define a new affine connection Lfy such that Lo = Lg = OQ and refer all the field equations to it. We shall adopt the following notations: p p Upp Aly Affine connection {Lv} Tih, (Lp=0) (g yy; 79) Covariantdifferenti- Vp ; Dp 3p ation (2.30) Ricci's tensor Gup Ruy Wyy Einstein's tensor Suv Hpy Ky Let (2.31) such that L,=LP, =0. (2, 32)38 EINSTEIN'S UNIFIED FIELD THEORY Equations (2. 28) can then be written as agG8 - PY LB, + GPP LY, + gre, =0 (2.33) or By D,G* =0 (2.34) Ruy 48 given in (1.60) takes the form: 2 Ruy = Way > zur, -8yTy} = 0 (2.35) where Wyy is the Ricci tensor formed with Liv 3 = p p a ,P rn 7P Way =8plyy ~2vLup + Li Lap ~ Llp Lay (2.36) We can, from (2.34), determine all the Liy. But (2,31) shows that rh can be defined up to some arbitrary Tp; thus the antisymmetrical part of (2.35) must satisfy the relations 8» Wuv + %y Top + 894 Wyp = 0- 2. 37 Wap + 8v Tey * 2 Wop (2,37) Finally, the field equations can be obtained from 1* with the understanding that the affine connection in Lf, (Ly = Lg’ = 0): v a. Dy GH =0 a GHP =0 ° p (u) ». Wyy =9 apWuy +2,Wou + aWyp = 0 wy ana: * The equations equivalent to (I) but referred to Ly’, are (Ia) and (2. 35) Dp eft=0 a, gHP-0 2 a) Wuy=% (uly - pty) ‘They identify the arbitrary vector for which W b is a rotational ten- sor with the torsion (10) page 267, (43). aVARIATIONAL PRINCIPLES 39 There are the equations for a'weak' system, They are completely derivable from a variational principle.* If we add a cosmological term to the action function or if, ina purely affine theory, we choose the density (2. 18), we would obtain equations of type I on the condition of re- placing W),, by the tensor Way Way 7% Buy (2. 38) We would then have by a, DyGt-= 0 ap FHP = 0 a) b. Wyy = x Yu 8 pWuy + eyWoy + en Wyp = APuup This is the system of equations adopted by Schrédinger. 5. REMARKS ON THE SYSTEMS OF EQUATIONS DEDUC- IBLE FROM THE VARIATION OF 6 = GHY Ruy» In Appendix II, we will apply the variational principle to a Ricci tensor (4. Ryp(a= 1, 2, 3) formed with the affine connection L whose torsion vector is zero (54). We obtain then a system of equations which are weaker than (I). Infact, atransformation of the affine connection permits us to express them in terms of a connection A (See (II - 7) App. i) and to give them the form by fe etd ee « Ub. Ru ay +x, (4) = 0 * Mrs. J. Winogradski has shown that all hamiltonians of the type GH Ry , are equivalent for the formulation of the field equations. In fact the densities G4Y Ruy, GY” Ruy, GUY Ry, and G4” Ry, can be deduced from each other by a or X transformation, i.e., in substi- tuling for Ppp the connection FE, +f ay (orTRy + oP Py). The field equations obtained by varying the metric under the condition that ap THO = 0 and the connection are invariant under these A-trans- furmations, The transformation of the affine connection (2.31) which leads to the field equations (II) is a particular case of the A transfor- mations (cf, (43) and Einstein (19)).40 EINSTEIN'S UNIFIED FIELD THEORY We can make two remarks with respect to these equations. I - From the variational principle, it is clear that re can be determined up to an arbitrary vector Ip. Il ~ If we define a connection shy such that the co~ variant derivative of GHY is zero with respect to this connection, we have the following results: a. From theidentities (1.51), we have: Bp TMP ~sc#P an = 0 (2.39) If we multiply by 1//-g hyo and sum, we obtain h, ry 4,2 ap (ng £HP) = by fH. 2. 40 og (-g fh") = hug (2, 40) We can substitute for hy, its value as given in (1, 20s) and we are led to Ag = Orig + Our Gor v7) EF = fy = (2. 41) where 1§= qv", f° b. From the identities (1.53), we have 1 Auvp"-3 Pup (4) with = (2. Au Sie * Sou, v* Sup, a (2, 43) Puyp = 8p Puy tv You +2u Pup: (2, 44) It is clear that the 64 coefficients of the affine connection are determined unequivocally from (Iia). * This is due to the fact that Pf, never satisfies the equations eM so = 0, which would then determine it completely, unless addition- al conditions are imposed on it.VARIATIONAL PRINCIPLES 4l SPECIAL CASE: In the special case where we start from the Ricci tensor formed with a generalized affine connec- tion, the connection A such that Dp Ge = Q is the same as the connection (af, = rp, + 2/3 8 Ty whose torsion is zero. In this case, it turns out that the four vector 7# is zero, (1.51) written in terms of Lhy is automatically satisfied by the simultaneous vanishing of L, and 9?, The connection Lop which satisfies equations (II) is also deter- mined unequivocally. The vanishing of its torsion vector results directly from these equations, In all our further developments, we shall use Ricci's tensor as the basic tensor. The condition 7? = 0 results directly from the variational principle and the connection 4), coincides with the connection Lj, whose torsion vector is zero. The field equations which result from a variational principle and which we shail use are given by (m): a. Dp Git =0 Lp =Lpg =0~9?-0 . Vv (I) b. W, 0 % Wry + Wot 3 Wyp = 0. Vv Vv v It will be in this context that we shall treat most of the ap- plications and that we shall discuss the possible forms of the spherically symmetric solutions.* ¢. Conservation equations (41), (115), (116). In a Riemannian space, the tensor S{/ (which is a function of the Sun and its first and second derivatives) sat- isfies the identity p yo Su = (2. 45) * Nevertheless, we must not forget the possibilities resulting from the use of a different basic tensor (such as ‘+)R) than the Ricci tensor. The field equations will then be given by eq. (Illa) with 2 fy - fj. The disappearance of the conditions fp = 0 or Ap= 0 might entail useful modifications for the determination of solution.42 EINSTEIN'S UNIFIED FIELD THEORY and will necessarily have the form sf =GP -% of (G- 29). (2. 46) If) = 0, (2,45) takes the form p i an Sh +3 Spo % GP = 0 (2. 47) These identities follow uniquely from the definition of Ricci's tensor as a function of the g,,,and their derivatives, i.e. , from the definition of the Christoffel symbols equivalent to ApGhY = 0, In unified theory, the definition of the affine connection in terms of the metric is equivalent to equations ye HP Dp Gt = 0 a FHP =O. (Ta) The question arises as to whether Eq. Wa is sufficient to lead to identities between the components of Ricci's tensor corresponding to the connection rp, . We will give here the proof of Lichnerowiez (41). Con- sider the integral A= J Ruy GY ar. (2, 48) Generalizing Weyl's method, we determine the variation of GHY resulting from an infinitesimal change of coordinates x'P = x°+ ¢P determined by the vector field EP. We will have: ise) Bets {0 98H) EP + BHP an E+ gPMayEH (2. 49) These equations determine the corresponding variations: ~6V-g = ve guy Og'Y = (apV=e )gP+ YE ape? = a,(v=BE%), (2. 50)VARIATIONAL PRINCIPLES 43 (Ry, Ghar) = -{V-g 6R + RO-g}de = {V-g (8g R)EP + Rap(V-g 8") } ar = 8p (RV7BEP) ar. (2.51) The variation 6 Ais expressed in terms of the integral of adivergence. From Stokes' formula, this integral is zero unless the increments EP are zero at the limits of integra- tion, We assume that the latter holds and obtain aA= J { OR) GHY+ Ry, 5GHYbar = 0. (2. 52) However, Eqs, (Ila) which we assume to hold are precisely the necessary and sufficient condition for Jor) GUY dy =0 (2.53) for an increment ore, which vanish at the limits of inte~ gration. This will be the case if we assume that the ePand their first two derivatives also vanish at the limits, We will then have: J Ryd gh’ ar=0 (2.54) + ég4¥ is then determined by the difference between the g¢” computed at the two points which have the coordinates x? in the two reference systems, Og = geY x) = BEY) = fe HM(!) -eP M(x) fe BM) 2 HPO] ax’ axty agh” [25 par Ow) - oY) a ee Hw by (ots 28") ( 14285) e540) 28 ge agho 2k , gov ak aghY p Brand 8 Qe aye A For a scalar R: BR = R(x) - REx) = -[R'Ua!) RGD] = - 2B? (cf, Von Laue (9), p. 208).44 EINSTEIN'S UNIFIED FIELD THEORY But, from (2, 49); HRyy OGHY = Ry (B_GMr) EP - (RyyG’ + Ry.GrP) ope" + Ry, Ghv a p Ep = Oo [Ruy Glre? ~ (RypGP” + RypG)e*] +18 (RuyG?” + Ry G') -6,PG*apRgp ke. The integral of this expression must be zero independently of &4, The first term being a divergence will vanish. We will thus have: ap(RuyGPY + RG) - 6 Ga, Rog = 0. (2. 55) Let a p ai v 1 5p. seh =Vreap, HP = 3 (R,eP"* Re?) -3 OLR, (2. 56) We then have;* p it = apie + HF Ry pu G%F = 0. (2. 67) This is an identity of the type (2.47). It refers to the ten- sor Ry, = BR, aul Tr). To ‘write it in terms of the tensor Wyy = Ryy(L), we make the substitution Ruy = Wy ~Z(8,F,-8Py)- (2.38) We will then have: , sep =«KP - Fe, 7? “5 i gp B vag FI, (2.58) by letting Yury = uly - aly, (2. 89) * cf Schrodinger, Proc. Roy. Ir. Acad. 51A, 6, 1948 Eq. (4.1); Lichnerowicz, C. R., Acad. Sc. 237, 1383, 1953, Eq. (4).VARIATIONAL PRINCIPLES 45 p_ p pi v 1 KP ave KP, BP = 5 fw, "+ Wye’? bs af w. (2. 60) Whence, substituting (2.35) and (2.58) in (2.57) and re-~ calling that ap GHP = 9, p ii ap _ Op KP +5 Wap Py GF = 0. (2.61) We will thus have identities of the same form in the case of Ricci's tensor expressed in terms of the LP3 The First Group of Einstein's Equations EXPRESSION OF THE AFFINE CONNECTION AS A FUNCTION OF THE FIELDS. (53), (54), (55), (57). The group (IIa) or (Ia) of Einstein's equations permit the determination of the affine connection in terms of the fields guy. We shail solve the system Be “30 = 0 (3.1) without making any assumptions about the affine connection Ady in terms of which (3.1) is expressed. We shall deter- mine* the most general affine connection satisfying (3. 1). ‘The solution of the system (Ila) Dy Git =0, a GHP = 0, Ly = 0 (Ta) follows immediately from those of (3.1) by setting the four- vector f p equal to zero, * phis is the solution of G#4p = 0 which we labeled PP, , if no confusion arises when a single affine connection is introduced, (J. Phys, Rad., 13,177, 1952; 16, 21, 1955.) 4748 EINSTEIN'S UNIFIED FIELD THEORY NOTATIONS; Consider a four-vector with components A°, We will use the following notation: Ap = Yoo ae, Ag = Ppg A = P9g VTA (3. 2) AB = pa AY = po PT AZ = Ppa PT Pay Aw = o AB = Poa A” = Pog PT pay Og? TAG, AB = Opa A” = MoT Pay Ou gy Veg ¥TY Ay and AP = po = yh Boy AP = PT ag = Pg AT, = PO p_ PS a AP PO ngs PO gy Tay, AM, ae FAG = Og yy TGA, @. 3) % HH ay 0 " PO BE PT ey em yo yl yy A”, One can establish the following identities between these expressions (see App. I) 2.-2a Lege oye = 7 he 7 (é y> 9) AG. (3. 4) Consider an antisymmetric tensor Ay, p: * Let Vv * x A 4,0 Auy,p =" we 1 ¥ "Ang, p- (3.5) "In the folowing, the comma (appearing in A yy, o) simply sepa- rates the subscripts and has no relation to the derivative symbol,FIRST GROUP OF EINSTEIN'S EQUATIONS 49 Then aK ¥~ Aun, p= FS cum YY TART, p =~ Ayy,p 6.8) A. The Equations Eupip = 0. The density equations wy v v Gr ip ang» ab, gm aat, Gh. ghag. =o are equivalent to either of the following systems: Vy aii =a vs af, ef” + ax, gl = 0 6.7) ss ce oT = Suvip =®p8uv- Ano Boy ~ Any Suc *° (3.8) We will solve the first group of Einstein's equations starting with this last form. 1, CALCULATION OF Afip AND AP, . _ v We multiply (3.8) by g4” and sum obtaining: in BM ao euy-Aup - Apu =o or 1 Apu = gy ud = % log V-E. (3.9) P, 2 We form g* pret p. From (3.7), we have: 1, He au, ° Vie se - pp} =a, MP4 agyth? ~ af, nl? =0 (gta ts ahh) &-19) GH may (agit), tH a ee ap (ye MP), (3. 11)50 EINSTEIN'S UNIFIED FIELD THEORY We multiply (3. 10) by hy, and sum over #. We obtain, using (1.6) and (3.9): p = HP. AP gue Ay = By (jp EMP + aby HP) 1 im 7 (3. 12) uy Fe MoN=EE PY hyp £ Using (1. 20) and (3.2), we have: ‘tp = (uy t Pup Mr?) tH =f) - t= Ay B13) 2. EXPRESSION FOR THE AFFINE CONNECTION As A FUNCTION OF ITS ANTISYMMETRICAL PART.” Consider again Eq.(3,8). Separation into its symme- tric and antisymmetric parts in», v leads to: o o g o BD 2% Hy > (Aue Yop + Sip %p- (App opt Ayp Fp) =O - o ~ Ae _ o ~ Ae (An) 2% Puy - Bue Yay ~ Sup You) (Sup Ye Avp Yop) =9 A permutation of p and 1 and then p and v leads to four ad- ditional equations analogous to the above two. Labeling these equations (Sz), (Az), (Ss) and (As), we form the quantities (S) = (S2)+ (Ss) - (Si) and (A) = (As) + (As) + (Ai). This leads to: oC o o an Yypt By 7p ~ 8p uy *Auy Yop *2(Auo ey +A Pop) =O Vv Vv (8) ara Ae 8 Pov tev Pup tp Puy” 2Oup Sov“ Avp Fou) -2A iv Yop = 9 (A) Vv * gee (53), (54), (55) and (57).FIRST GROUP OF EINSTEIN'S EQUATIONS 51 We introduce the symbols: ( 1 \{oyel = 3 0, Yup * 8v tp ~ % uy) p toe (3. 14) Vib} =F Py Met By Yo ~ 8 Yd and let ‘p represent the covariant derivative with respect to these symbols, In particular, we have: o o Vp Gur = % Puy ~ {Het Pov - {Wt Pg and by definition Vp yy = 0. Tf u, - represents the symmetrical part of the affine connection other than {uv}, then aby = {lv} +uky (3.15) and, if we lower the superscript p of apy? and Ay P by let~ ting = oC = oo Uuy, p * Yoo Yly » Auv,p =Ypo Any» (3. 16) v we note that the relations (S) and (A) can be written as oe Uny, p = Aup Foy + Mop Pour (3.17) 1 o Any, p= 3 Ppt Vp Guy ~ up By Bp You), B. 18) with Fuvp = 80 Fuv * ®v Fou > uy Pyp- (3. 19) The equation Suvi o = 0 equivalent to the systems (S) and(A) can thus be divided into the 40 equations (3. 17) which deter- mine the symmetrical part of the affine connection and the 24 equations (3.18) which upon substitution of (8.17) deter- mine the antisymmetrical part of the affine connection in terms of the field quantities. The problem thus is the deter- mination of the antisymmetric part of the affine connection.52 EINSTEIN'S UNIFIED FIELD THEORY If we were to form (Ag) + (As) ~ (Ax) and (Sz) + (Ss) + (81), we would obtain the system o o Oy Yyp + vy Yup + % Yu ~ 2(Apy Yop * Avp You (s) oO + Auy Yop) = © ~ o o au Pup +8, Pup + a, Puy ary %p + 4p Yeu Vv Vv (A) o = + Aw Yop) =0 which is a consequence of (S) and (A). With the notations (3.16) and using (3. 15), we have: Upyp = Ywv,P *Ypy, y+ Yyp, wp = 9 (3. 20) 1 Sump = Auv,p * Soury * Sup, =~ 3 Suv. (3. 21) 3. THE 24 EQUATIONS IN Aly . Vv To solve (3.17) and (3. 18), it is sufficient, in princi- ple to substitute the uh from (3.17) into (3.18) and solve for AP, . Starting from (3.17), we form the expression ay Wp Yop ~ Wp Fou in (3,18). We have: Ye Sy -%p fou May Lap ov Son Pru fav (3. 22) - Sun Pap Fou ~ oN ry out Thus, using the notations of (3.2), we have:FIRST GROUP OF EINSTEIN'S EQUATIONS 53 = Co Suv, 0 = Ypo Any Vv Vv = o - o (3. 23) 4Suy,p = Yoo 7 Auy, r= Yoo Sty > Vv Vv v =e = oT A Aun, = foo YT yy 2 Fog YF Pra Aity - Vv v v Let Suvp * Auv,p + Apu, + Arp, a Vv v Vv Auvo = Auy, 6 + Apu, Dt Av, & (8. 24) Auyp = Auy,p + Soup + Ayo, i Vv Vv v We then obtain from (3. 22} (see (57), App. 11): ‘ o = yo ~ ~ Wp You" tp Yon =~ 8 Spa, Bev Apa, Fou] Vv Vv ~ YB uae Fav" Bing Foul . (3. 25) ~ Auup * Auy, 5 v Substitution of (3.25) in (3, 18) leads to: OAR A Any, -Y (Burp Fou ~ Ayn Fou) OA, ae 7 (Spa, i You Apa, v%oy)* Aw, yp 1 =~ 5 Fup By Pu (3. 26)54 EINSTELN'S UNIFIED FIELD THEORY It is then sufficient to solve these 24 equations to entirely determine the affine connection. 4, DERIVATION OF THE 24 EQUATIONS IN Suv, p» Any, p D Any, B We shall transform (3. 26) in such a way that the 24 antisymmetric coefficients Aly will be represented by the following terms: = o uy, * Yoo Aty (3. 27) Vv v * = o x Auv.e = Yoo YT Buy, z= fog?” Py Suv + (3. 28) v v * oy aiY ed or Auv,p =F3 Eves YY aro (3. 29) v To this end, we shall introduce the intermediate nota- tions: 1 a= phar Ag eg ve yt Py Mag p (3. 30) v \ 1 oy . =39 Aw, o (3. 31) We must compute the three expressions R or = Aip ¥ (pa, h Foy 7 A pred Fy) and Auyp, which appear in (3, 26) as a function of the quantities given in (3,27) through (3.31) (with the same u,v, p indices in the same equation). After some calculations (see (57), Notes IV, V, VI), one obtains: Binp = ~(Fux Ap * Fon dg + ApAy)* V8 Eu yoo 9B, (3,32)FIRST GROUP OF EINSTEIN'S EQUATIONS 55 ye (Soy 4 Or,” You Son D or =o $Y yo Enver * for Puy Spar we (3. 33) TS Suv, p ~ Puy Ap» 4VGe Biot fl-2+2 up *1 1-5 +7] Aun A GAY Hepe (3.34) VF euypo 19% 70, -By ve + Furry Gye PF apo: Substitution of (3.32), (3.33) and (3.34) into (3. 26) leads to: =. NP, Suv, p * Au, p - Ty * ip (3. 35) = - 4%. Vo Fup 7 YOO fev - Mv Fon ~ Foo Cupar “YB (ay Mp +Oyq Ay + Py Ag) -Y YE eure ~ Yor Pav) Sprx +L 4 +f) Sup PE €prV TY F970 bans VP euag YY 78 gp B, +12 nyo oA, ~ 2P yy Ap- To obtain the desired equations, we must compute Ap and By56 EINSTEIN'S UNIFIED FIELD THEORY a. Calculation of B,;* We multiply (3. 18) by uy and contract obtaining: 1 1 By =~ 79 Pup +3" Vo uy - Up (. 38) But 1 1 g ohh v Suv = op ¥ 5% 7 = By Log | (3. 37) and from (3.9), we had ep if a Aup * {fo} +Wp = J dy Loss (3. 38) or eup od g Wy = Up = 7% boey (3. 39) ‘Thus, upon substitution in (3,36), we have: B 1 2 -igi¥ i g = GT? Fup + 5% Loe E- (3. 40) b. Computation of A,: We multiply (3.26) by 4 gh and contract. This leads to (see (57), Note Vil): i Bo + Ay - 5 2H yBY gig A p+ Ap 5 a3 Suvp 41) 1 1 = -t gu = phy GO yp FEO 0 Pur or using (3. 21) and (3.37): 1 1 g Ay = Zr vPY ae Fup +3 % LOB, - * It might seem preferable to compute By = VF By = 7H Any p since Bp has a meaning when @ = 0 whereas By does not, The results obtained are equivalent since it is By and not By which occurs in (3,32) and (3.34).FIRST GROUP OF EINSTEIN'S EQUATIONS 57 One can now substitute A_ and B, in (3.35). After some computations (see (57), "note VII), one can write We HY, P “Ly Ap? Ay, = Ru, p (3. 43) * where 1 we. ve R, m- > +, uy,p = 3 Fup * Vp Puy + wy Pt iw]o* Wy Sy PO Poap ve a Log ¢ + F eupe or a Log & (3. 44) Oxy - Be es 42 9* a poe ary “Lion” “ary “He bhEo » +H NF € rpg Ax *P ap Aa* Mody Fig yd]: All the terms occurring in R, ,p are determined in terms of y, vy and Puy We recall that Vp» the covariant derivative in terms of y,,, is a function of y. The Ap are known from (3, 13); Ap = fo Further from (3,5): ot ad 3° aap? PF Org * L Pl wip =F VF Eye B yO bo Prop" * 1 uv] oa = 3 YY euvag ¥°7 Y €ropn- “ We must bear in mind that the letter R used here (always desig- nated by three indices) cannot be confused with R,,,, Ricci's tensor (always two indices). We also recall that the comma represents a lowering of an index and has no relation with the ordinary derivative.58 EINSTEIN'S UNIFIED FIELD THEORY * 5. SOLUTION OF THE EQUATION Aw, p- WeN-y Aw, p * Any, p > “uy, p* v Vv Let (6*) andresent the relation S uy, p> 0 correspond- 7 ing to (3,43) and(&*) and ® the equations corresponding to 6fy,p and 6,5. These equations can be deduced from 6 p in the usual way: We * 1 Su, =507 €uvap yr PO 826, 0° (3. 45) Sun, * to YT oy OT Sw, r (3. 46) x Let us now form (6*). Recalling that from (3.6) Auy,p = Any, 0 we have: * we * _p* * Aun, * Ty Any, p + Aue = Ruy, p- (6*) and We A* a = = Sun A Aun + Any, 6 = Ruy, p- (6) Thus the quantity 2/@A/Ty (6%) + (2) yields: He We * 4p = Tay pre “9 Supe * Sen * Sap 3.47 -2e Rivet Buy o Yo v If we apply (3. 4) to Any, p> we have: v 2..¢2 ~842 = Ayo * y Sune * yy) Say P (3. 48) Comparison with (3. 47) leads to:FIRST GROUP OF EINSTEIN'S EQUATIONS 59 Wo * _ Sy g Stal 4 2-BL)q = ay “ep Aue + (2 -oe0) 1, B * (3. 49) = 2 Rly,» + Ruy, p. Yoy v Subtraction of the quantity (2 - g/y + @/y)(6) leads to: £489 avo 2 (2- Fe Sw, p-e (8 = gs 7) Sup (3. 50) =(2- £+£)p _ 24g R* Rw,o Vy we ~®, ay, B* v We can form (3. 50)* and then compute the quantity 6?, Wo g¢ 2 -842%)(3, (3 -245)(3.50)*]. [12 -E eS) (3.50) + 22 (8 - 5 +2) (3, 50)>] a This yields: g, 8? 4/3 _¢,0) [e-£-%) y (:- “yt ) Au (3. 51) This relation determines completely the antisymmetric part of the affine connection aw, e in terms of the field quantities Buy Yu Puy The Rup p are uniquely determined as a func- 7 tion of these quantities from (3. 43). By setting 6y Es a=2-£48 pio @ (g- 3. 52 yy x ( ‘) (3. 52)60 EINSTEIN'S UNIFIED FIELD THEORY and - NP pe wp "Fa Bune (3. 51) can then be written in the form: (Pee yA -R, Lp, p (3. 53) = g Swp = (2-£+2)n, = 9 8iy,6 Psu, Hy, P 0" v In summary, from the fundamental equation ByyiP = 0 1° The 64 coefficients defining the affine connection can be expressed in terms of the antisymmetric part of the con- nection by the relation p at, = {fh} tu, + hy @. 55) p = {tr} * PMaly Poy + AOp Fou* Suv, ale Vv v v 2° The antisymmetric part of the connection can be ex- pressed in terms of the Ruy, p by (3.51) or (8. 54). Ruy, 0 is determined completely wy (3. 44) in terms of the fields Suv Yy and Pyyt =-i we * vo oe oT Ruy, ~~ Suv Xp Cu * ays, “Tuvle ay Sav | Foro © ~ Puy Log’ + Fe pupo PO Log (3. 56) vo g -—— ur) a €{usjpr 9 AgLoeg + Way uv {M8 g + PIF € yup by - £5) + Pro Puy Oo - B) + Ay Ey - £9) + Ppp(by ~ £9) It If one wants to determine the connection Lf up such that Lp = = Lou = 0, all the above relations hold but the last line of Russ in (3.56) must be omitted (Ap = i - ).FIRST GROUP OF EINSTEIN'S EQUATIONS 61 6, CONDITIONS FOR THE EXISTENCE OF SOLUTIONS. These are evident. We always assume y 40. From (3.54) Au, p can be expressed uniquely in terms of R, it My Hy, P a+b £0, On the other hand the Ruy, p are completely determined if g#0 We must thus have g(a” +b") £0 (3.57) e((2 ~B. ee - 2(s -Es 2 £0 (3. 58) 1, THE CASE ¢ = 0 (56). or In the solution of io = 0, we never required the con- Suy3p =" dition g 40. The terms in y#¥ which have no meaning when gy = 0 seem to enter into our answers through Bo and appear in (3, 43) or (3,56). But in fact, we are always dealing with quantities of the form ¥~ oY or vy gH” which represent expressions of the form 3 €#¥P° vpg or V9/2 EH¥PO gg. These expressions are finite (eventually zero) when ¢ = 0, Similarly terms of the form V@ 8 Logy represent ex- pressions in 299/¢@ and vanish. From (3.52) and (3.54), the general solution of Bue 0 in the case ¢ = 0 reduces to 4 4uv0 = Suv,o = 9 Rin ~ Ruy, B (3.59) with62 EINSTEIN'S UNIFIED FIELD THEORY OTAT, 1 Lo Ruy, p = 3 fav * Vp tay * Se Yaw © Yan Soro - g,i g Pup ®p log y +5 (Ow 2% + Pp evt yp %.L08 + YP Mol Guy(fo 18) + Pou (Ey 89) + Yo (Ey -t)] (3,60) the last line of Ry disappears if {,, and hence Ap» is v, 0 zero. v If g # 0, we must assume that a=2-Euo (3.61) or Qo (3. 62) 1 wp yw l-gy y™ Fup Cag * In the case of ¢ = 0, the affine connection can thus be obtained directly from the general solution of the equations. 8. INTEGRATIBILITY CONDITIONS (24), (28). The equations (3.7) eps BY AB OU AV gt ef. ip ap B+ Agp BY + Bog Bi = 0 (3.7) imply certain conditions which can be obtained by forming (2% 8 - agap) BHP = 0 (3.63) Schrédinger (23) and Bose (28) have obtained these condi- tions explicitly by substituting in (3. 63) the expressions for 8g gH” and ag!” obtained from (3.7). One has then be be v LLY Roop oh +R g! 0 (3.64) where Boo BL B T Ab AT Ah Ryo = % Aygo - 8g Np + Ayg Aap - Ap Ato (3.65) is the curvature tensor formed with the A's and R is the same tensor formed with 4.FIRST GROUP OF EINSTEIN'S EQUATIONS 63 The identities (3.64) (when (3.7) are satisfied) can also be written, Riga fpvt Roa Eup 0 (3. 66) Multiplying (3. 64) by Sup and summing, we have Riiop + Riso =0 (3. 67) that is . % Buc ~% Ali = 0. (3. 68) It can be easily verified that this is immediately satis- fied by (3. 9) He = Alp * 9, Logv=8 Contracting o and u in (3. 64) and then o and v, we have: Rhy @? + Rey gs 0 (3. 69) or (Ry - Riu) ev = 0 (3, 70) These conditions can also be written as o = sY{@, A+, 4) - 247, Ags 0 (3. 71) These are evidently satisfied for all connections Lie such that Ly = 0. (i.e., starting from © = guy Ry,)- One can also verify that they are satisfied when A),,, is any con- nection such that @ tS p=0 (i.e, such that x = {fp - fp).4 . The Second Group of Einstein's Equations A, THE RIGOROUS FIELD EQUATIONS, i. THE RIGOROUS FIELD EQUATIONS DEDUCED FROM #. If one starts from a variational principle applied to the density # = GUY Ru» Ryy being Ricci's tensor RAT), one can obtain the second system (II) of the field equations. They are expressed in terms of an affine connection, P J ph 2 ge =Lo. = Lin = Tupt ¥ 5g Ty Up = lpg = 9- (2. 31) Let us consider system II in which we shall include a cos- mological term * a. Dp ct = 0, Lp = 0 equivalent to g? = au gts 0 a, Way +3, Wout ay Wrp = A Puyp- v v v "Ray = Eup ees from (2. $5) Wiy = uy Woy . i (Ty > Ty) = uy 6566 EINSTEIN'S UNIFIED FIELD THEORY InII, we must express Ricci's tensor W y,(L, ) in terms of the fields and their derivatives by substituting the value of Li, obtained from (3.51) and (3.56). Following the notations (3.55) adopted in the preceding chapter, we can write: Lhy = ip} + uf, + Uy = {f} + hy (4. 1) where Of, = ul, + Ly and the {iv} are the Christoffel symbols obtained from Yuy Substitution of (4, 1) in Wyy as given in (2. 36) -3 iP P dah ah ye Way = 8p Lay ~ vlan + iy Lip - Lip yy» lead to: (2. 36) 7 pot p A AP Wy = Guy + Vp Opp - 5 Vu V, Loge +Ouy By - Op Oy: covariant derivative with respect to {/1,} and up = ug, (see 3.39). We can now separate Wup in terms of its symmetric and antisymmetric parts: p (4, 2) where Guy is the Ricci tensor formed ae wf, Vp is the 1 Wy = Guy + Vptay - 3 Vu Woes + = (4. 3) p a ea a B + Uy Up ~ (np wy + Lup ty ). . 1? a oP auf ph Wyy = Vu Lop + Lay Up > (uip In, +B, Lup)- (4. 4) v v v v v Xr In the above relations, uf, is expressed in terms of Lg, From (3. 17), we have v Pp a a Wp ee PU ey Lug + Fur Lye ) (3. 17) - yh - : 7 Bug, Lag,SECOND GROUP OF EINSTEIN'S EQUATIONS 67 The field equations (II) can then be written in the form: -ve (up, 5+ Lap, wt) - $0 Vy log 1 Pik (4.9) “ho * hyp, a) Log y +PO PT [Lup chin, 27 Luo, a+ Ups, 2) Vv ¥ v v (Lye, *Lag, ad] = yw ep p g 0% Ey * 3 2 Uy 8, Log y (4.6) p e + ATLL Lye, 6 + Lia Lor, p - Lia Lr, »| Vv Vv vVoev Vv v + Circular permutation of 0, Uy v= > Pyp- VP =F 7, 3 = PF ag. (4.7) In these expressions, one can substitute the solution of Ly obtained from (3.51) and (3. 56). In this form, the field equations (4,5) and (4. 6) seem quite complicated. A resolution similar tothe one obtained by Foures (118) for Guy =0 does not seem possible, Insofar as the physical interpretation of these equations and their applications, one cannot consider more than the following: with 1° One can keep the rigorous solutions (3.51) -(3. 56) and apply them to special simple cases. An example of this is the search for a spherically symmetric solution (see Chapter 5). 2° One can seek an approximate form of the relations (4. 5) and (4.6) (see § B this chapter).68 EINSTEIN'S UNIFIED FIELD THEORY 2. DEFINITION OF THE CURRENT FOUR VECTOR. The theory permits the a priori definition of two current densitites. Mig gue gr = a er, (4.8) t gt « HMO ge PO ¢y 6V-y 0 Bay vipe (4.9) + Pp + % Pay) But the current density g satisfies the identity 1 pu gate, - G4.) = gM? -achPr,, (4.51) In particular, if this identity is expressed in terms of an affine connection A such that @* oo = 0, we always have: 5 4p = Nipg £° (2. 40) or Ap = ip - 5. (2. 41) In the case where the variational principle is applied to a density formed by the Ricci tensor Rup(T), the relation GH = 0 is valid for the connection Ley = Tey + 2/3 oft whose torsion is zero. Thus Ip = f, - it is satisfied by Lp = 0, fp =0, these relations following from the variational principle. Thus, since #4 is zero, we are led to believe that the existence of an electric current is necessarily tied to the definition of the pseudo vector g#. If we assume for simplicity that the only non-zero components of the antisym- metric field* are ¢,, and ¢,, 9, reduces to in polar coor- dinates * his is the case for the spherically symmetric static field.SECOND GROUP OF EINSTEIN'S EQUATIONS 69 -1 rr v-y We are then led to identify @3 with the electric field and P24 With the magnetic field. This identification is different from the usual one (in Maxwell's theory, Yy, is identified with the electric field and pq (p, q = 1, 2, y with the mag- netic field). We shall see however that this identification is better from the point of view of the unified theory. In fact, Einstein noted (7) that it is possible and prob- ably preferable to identify the electric field with the com- ponents Ppq(Pp, a = 1, 2, 3) of an antisymmetric tensor and the magnetic field with the components Pps. In this man- ner, the sign of the magnetic field and the current auto- matically change sign when one reverses the direction of time. In discussing this question Schrodinger (75, 76) points out that the disymmetry between the electric and magnetic cases which manifests itself in unified theory is one of the most satisfying characters of the unified theory. This disymmetry manifests itself in the particular case of spherically symmetric solutions. Nevertheless, it does not seem evident to us that the only possible definition of the electric current is (4.9). We shall give in Chapter 6 another possible definition from which it will not be evident that the pq represent the elec- tric field. The latter can be identified with Pp* B, APPROXIMATE EQUATIONS, 3. DERIVATION OF THE APPROXIMATE EQUATIONS, If the fields Py are weak and if the y uy are close to the “'m this case, a spherically symmetric solution for such a Papapetrou solution can still be associated with the electric field case when one chooses the equations of the weak system. The correspond- ing equations for the strong system cannot represent the electric field case (cf.Chapter 5, p. 86 and (61), (63)}. But the natural equations here are the equations of the weak system.70 EINSTEIN'S UNIFIED FIELD THEORY flat space values 7,,,(-1, -1, -1, +1), we can expand the Sup in powers of €: = 2 3 Yay = Tuy t Pug *€ Yur *€ Put (4. 10) ° ure ~Porp > ~ [ay ]apyy7 + [,0] yyy, + [09, A] fay, 7] 1: (4. 48) - [ov, ][up, T] - q 8? Poort * PrP OP yg L 1 . ~ZPova Prup t A Pyp%p Prua * Puc Pr) Einstein and Kaufman have shown that to each solution of (4. 43) corresponds a solution of the second order equa- tions (4. 44). b. Equations of the antisymmetrical field. These are equations (4,18), (4.19), (4,24) and (4.25). The first order equations (4. 18) and (4. 19) will be:SECOND GROUP OF EINSTEIN'S EQUATIONS 75 WY ou, = 0, (4. 18) WP p8y Purr = 0. (4, 46) The second order is obtained by substituting (4. 40) into (4.19) and taking (4.18) into account. They can then be written in the form WLP Pav @v-B)a ~ 2° (yor Pua) » = - I Osho Yul =% (4. 47) aca TapL, wy, oom "QorPoPyy, xP, rater - Po, 0] by, 7 tle, TILyy, 6 v (4, 48) +[ep, T]Lya, o) ] any + circular permutation of u, v,€ = 0. The significance of these equations was discussed by Einstein and Kaufman in the case of the strong system, that is, when the circular permutationin (4,19), (4.25) and (4. 48) is replaced by that quantity which is controlled by 3. (or a). The physical consequences of the conclusions reached seem to exclude the strong system. In fact, in this case, the first order equations (4.18) and (4,19) reduce to nfo, Paw = % (4, 18) PrEp%y Quy = % (4. 49) since po, = 1% uve = 0. In order to insure the existence of solutions of the second order approximation, Einstein and Kaufman have shown that the condition76 EINSTEIN'S UNIFIED FIELD THEORY uv? y You + 0gdy Yor =0 (4. 50) must always be satisfied. Tofirst order, Yuy and Puy satis- fy (4.18) and (4, 20) and are independent of each other. We could then arbitrarily superpose the two fields. This con- dition is also applicable to the weak system. But to second order (4.50) leads to an important restriction to the super- position of these fields. One can show that in the case of a Schwarzschild field that (4. 50) will impose restrictions which are physically inacceptable ((18) p. 334). The addi- tive property of the weak fields cannot be restricted by the stringent condition (4,50). For this strictly physical reason, the strong system must be rejected. Schrédinger (20) has shown that the equations of the weak system lead to the following conditions for the anti- symmetrical field in second order: yy = n Pup * Po 0, (4, 51) 1 3 Lup = Vip? (4.52) where Purp =F Pyp + PoP uy + Pv Spur (4. 53) Vuvp =u ¥ vp * % Vu + 8, Vou (4. 54) and Vy yp being quadratic expressions. We then ha ave identically 18 ePp 20° (4, 55) On the other hand, if we choose a coordinate system such that 1 po -~= = (85 Vig Zn Yoo) = 9 (4, 56) we can reduce (4.52) to L r (4,57) q _bupo W T ay a, Puvp = eho a, YuSECOND GROUP OF EINSTEIN'S EQUATIONS 77 or 1 _uvpo Fea) Bg Pruyp = ~e HPO MAT HE OC, P78) Grepys (4. 58) where G is the curvature tensor formed by the y to first order* The right-hand side of (4.58) does not vanish. Thus, even to the second order, the condition Purp = 9 cannot be realized. On the other hand, the divergence of the right hand side of (4. 58) does vanish and leads to the exist- ence of solutions eMUPe a, Pup = 9 (4. 59) that satisfy the continuity equation. The order of magnitude of Yyyp could be determined from that of the Riemannian curvature if one knew the value of ¢g yr Unfortunately, we do not even know the units in which the electromagnetic field is expressed in terms of Pup Schrédinger has applied (4.58) to the static fields. He shows that a purely electric or magnetic field in the first order remains unchanged in the second order. An applica- tion to the terrestrial field leads to the existence of a second order field which is a/R times the first order, being the gravitational radius of the earth. But the first order field D/R® is of the order of a gauss, D being the magnetic dipole moment of the earth. It is thus not probable that one can demonstrate experimentally the existence of this field and the current g vp If the relationship between the current and the two types of field is of theoretical importance, it nevertheless is outside the reaim of experimental measure- ment. * phe difference in sign with Schrédinger's (26) Eq, 27 arises in the-definition of the Riemann-Christoffel tensor.78 EINSTEIN'S UNIFIED FIELD THEORY 5, THE APPROXIMATE EQUATIONS Yur > 0. If we assume that the y,,, differs from the 7), in the second order only (i. e., Yu» the second order equations simplify considerably. Schrédinger (117) has discussed them for the strong system. We shall derive them for the weak system. a, First order equations: L, uy vanishes and the first 1 order equations refer only to the antisymmetric field. From (4. 18) and (4, 19) we have: Pao? uy =0, (4. 18) Papa Pur = 0. (4, 46) Using Schrodinger notation: Muy = Ey Su» fore, = - 4, -1, -1, +1 (4, 60) where we do not sum over », (4.18) and (4. 46) become Epp Yup = % (4. 61) epee Yarn = % (4, 62) b, Second order equations: These reduce simply to (4, 44): Luv * Quy = 0 (4.44) From (4, 45) and yw =0, Quy reduces to 1 Quy = ~ WP OT 17 29, Yoo Par ~ 2r Pav % fue (4. 63) 1 1 + ZPorPrup ~g {wr Prur 1 “2 fue eptral :SECOND GROUP OF EINSTEIN'S EQUATIONS 79 We then have, in Schrédinger's notation: 1 Eu = €p€o it 2u®y foo foo ~ 8s foro P ue ai 1 +5 You Loup ~ 3 we Youp (4. 64) 1 3 Puc’ Pop | - We will discuss this relation in the study of the energy tensor, * “In the strong system, (4. 18) is not changed, (4.19) must be re- placed by Ws 0 or P¥ ay Luv 7 & (4. 19") v v We then have for the antisymmetric field PM Puy = 0, (4, 18) PPA A Yyy = 0. (4. 46") From the definition of Fup? we have , Pay Pup = 9% (4.46) that is, if we use the notation (4, 60) =O. (4, 61) = a 0 %p Yup = 0 “op Sue Lup 38 then the expression (4.23) which Schrodinger denotes by yy. In the strong system, (4. 61") is valid and £',,, reduces to 41 { Pay = po} g2u8y foo * % Soy % Pout F fom fous which is the same as Schrddinger's expression (ci.(117), p. 21 Eq. (2,17)80 EINSTEIN'S UNIFIED FIELD THEORY 6. QUASI-STATIC SOLUTIONS. We can assume that, in certain problems, the time derivatives will be of a lower order of magnitude than the space derivatives. We must then distinguish between g, and gp«(p = 1, 2, 3) in an expansion in terms of ¢. These solutions have been studied by Einstein and Infeld (94). Callaway (87) has applied this study to the weak system. We shall discuss his results in Chapter 7.5 Spherically Symmetric Solution A. DIFFERENTIAL EQUATIONS OF THE SPHERICALLY SYMMETRIC CASE. 1, SPHERICALLY SYMMETRIC CASE IN GENERAL RELIABILITY. We shall seek the solutions Sup of Einstein's equations (Il) for the case where the source possesses spherical symme- try. a. Neutral particle, In general Relativity the prob- lems is stated as follows: the field equations in the neigh- borhood of attractive masses py = 0 (5.2) can be written as a function of the Christoffel symbols which are determined in the terms of the metric au We thus have asystem of ten differential equations which satisfy four identities. When the given masses are spherically symme- tric, it is convenient to work in polar coordinates, Then the line element is ds* = - adr® - 8(de" + sin? ody?) +6 dt? (6. 2) where the independent variables are: 4 ter, x =0, ®=g, xt=t (5.3) 8182 EINSTEIN'S UNIFIED FIELD THEORY a, a and & depend on r and eventually on t. Computation of Ricci's tensor in terms of the 3,,,(-8, -B, -4 sin’ ¢, 8) leads to three differential equations. These can be reduced to two owing to the identity relations (cf.(4) p. 000 and (5) B 85). We can then choose arbitrarily one of the unknowns a, 8,0. It is convenient to choose f= r%. Then the solu- tions in the static case are 2 2: g-r, $2%--y? 22, (5. 4) a r V being the speed of light which we shall set equal to unity. This is the Schwarzschild solution of the field equations, It defines completely the gravitational field in the neighbor- hood of attractive masses and permits the determination of the trajectories of particles moving in it. Although the field is computed for a purely static case, Mineur has shown that it is the samein the case where r and t are aiso time depend- ent. b. Charged particle. If weconsidera charged particle, we must take into account the electromagnetic field produced by the particle. Then instead of (5. 1) we use Guy = XE (5. 5) ‘ey? where E,,,, is the electromagnetic energy tensor endowed with a geometrical significance. If we postulate a maxwellian form of the electromagnetic field, we have ((5) p. 185). E oF F (5. 6) er eid Fug py = Tay = 4 Sup¥p Lp Fup» with Fup = 8,%y - Gy (5.7) In the electrostatic case, the only non-vanishing compo- nent of the field in polar coordinates is Fi. It follows then that & sle-de-dat-S 4,SPHERICALLY SYMMETRIC SOLUTION 83 Substitution into (5.5) leads to the following form instead of (5. 4) eet, (lL 2m, tet (5.8) p-r’, yf sprit Sr where x has been set equal to 87 and V = 1. If we do not limit ourselves to the static case and if we recall that the nonvanishing components of the field are Fis and Fz in polar coordinates, the solutions will not be identi- cal to (5.8). In this case, the metric depends on time. Narlikar and Vaidya (69) have obtained the solution 1 r (5, 9) where m and m' designate 2m/ot and am/or, and m is a func- tion of r and of t. The above solutions assume that the electric current vanishes, that is we can place ourselves outside the charge distribution and conserve the electromagnetic field. 1-2 1 in? 2m =r tn = Pups : ; , -psin® @ QO -using 0 Q 9, -w 0 0 0 (5. 23) where usty. The various determinants areSPHERICALLY SYMMETRIC SOLUTION 89 y = - ang" sin® 6, (5. 24) g=w uv sin® 8, (5. 25) g=- (ao - w’) (6 + u*)sin® 8. (5. 26) Use of Eqs. (1.21) leads, when applied to this case, to ra 1 a w zl —i B usiné yh = -1 » ave x1 , B sin? 6 using 1 at a Ww 0 0 ae 5.27) 0 §=6B¥ sino 0 vy vac NY sl Pe € wpov PTA, se ind 0 feo 0 0 =e 0 0 0 To determine the affine connection, we substitute the above values into Eqs. (3.52) -(3. 56) which express the affine connection in terms of the Sup. The computations are carried out in Appendix IV. The results are summarized in the following table (57) (58).90 EINSTEIN'S UNIFIED FIELD THEORY Static condition Non-static condition Symmetric group A si 2 = Ab BAe UBs sin’ 20 ay 2 = Lactoga (1 -¥) " ix) es & it So oa Q ra ' ae Ts = - sin 6cosé, Aa, = Ade - A Ah = Ad = Al, = Ah, = Ae w= As = 9 Ba = Aga = 9 A3,= -ASgsin’o ZB sing | a2, = - Alzsin?o-%B4 sina = = 2a = = 20 Antisymmetric group A e Ab =~ Ad BBE UAL sing | a8 =- ads = ~BBa + As sing Al =g—aL ( *) At, = 5% agLo: ( ae 1 By OBE oe, At Fwy OeMOB\! og Ai, =A%, sin’o=-Btsine |a3,=- ab, sin®o=-B¢ sino Vv v v Vv w A, 3 Ww Aba = Ay =- oA = aot coy 2a Age A ag As (5. 28)SPHERICALLY SYMMETRIC SOLUTION 91 where . 1 wv uut + Ay = 3 a Log 3 (5) = we? ; (5. 29) B= -Latog (1+ ¥-)- 22 oe ve Bg 8 ( ) wae / t Ww : / (5. 30) w ug - pu, Ba = ~ ft Log (1 x) Ss which is a generalization of Bonnor's notation to the non- static case. Here 3 a Oss be and the primes indicate differentiation with respect to r, and the dot with respect to time. The general condition for the existence of a solution (Chapter 3, $6) g(a’ +b’) «0 (3. 57) or ~£ 488) 49 (3 8, 2) e| (2 Be 2) f(s £.2) #0 (3. 58) becomes in our case (aisles) ele) & ag, gs ace ao awe? (5. 31) #Q, ao p + We have deduced the affine connection and the conditions for existence of the general solution (3. 52) -(3. 56) and the conditions for existence (3, 58). However, if we limit92 EINSTEIN'S UNIFIED FIELD THEORY ourselves to the purely static case, that is if we keep only the left hand side of (5.28), we are led to the expressions computed by Bonner ((63) Eq. 2.1) starting directly from Bu vs sp = 0. These equations simplify and can be resolved easily. Bonner also established directly the condition (5,31) which we deduced from general existence conditions. The agreement with Bonnor's results is thus obtained in a special case and serves to show the validity of our solution. 5. THE DIFFERENTIAL EQUATIONS OF THE SPHERI- CALLY SYMMETRIC FIELD. The affine connection A given by (5. 28) is now substi- tuted into the expression er) p {dP r» 1? Wun = Spbpy - ely + Layla - Lipbay (5, 32) o such as Ly = Log = 0, the connection L being the same as A with its torsion vector set equal to zero. Let wi) wie » and wo? be the parts of the components such that the superscripts designate: (1) purely static (4) purely dynamic (14) mixed The only non-vanishing components of Wy are (66):SPHERICALLY SYMMETRIC SOLUTION 93 Static contribution Wy. Non-stalie contribution Wy cabsBas te) 1 gta. ALB dee - a 6G Bs ay -2oo+4o0Ay + {aq Log U) Stee wo | fase (2) av x 2 : 7 (- s) (a Log) +H boge 1) (24 - Loge) _ . + sak Log at? _ Bg? * boea! 148, obyteas +34 BBy Ban) 2 BAe = uBs) pe Bete. Wes * Pe Be “sin 6 ~ ~ + BBLS a Loge -w") + Bg BBs 9, Log (ao -w*) cA BI oo (ee Bectee ay) we 4a’ tA, Abe Bi wage ae ALL BE ag -2atg +4a0 A, Be 2 + ByLog UI (ot = Bate “hay Ay) 4a L = We : + PaLog a + 7o-a Loga| a3 (Se-Z joven 1 6.39) Awl 2a + Ooze uy? + gh ALog ot? ‘eB, sh) UB, - BA, . (aBajtae ~ g, BAe eB Be We 0 (C224) -n 2B a RBa eet) - 3, Bas ain 9 ABsceay a; Log (ae -w*) Betts Loplao -w) i : w ALL BE ay MBS ta” 4 Bo ALM AHO OY Log u Aeaw ewe 2H Loe y awi w 49% a Log U ifAsy , Aaa! . (Ay Ac+ Bi Be) + Hue, Beet. A > Ay Wae* 1 + Feteloay -4(3- a Log Uae Log U ada + gh) as toet a 7 Bi ao / *94 EINSTEIN'S UNIFIED FIELD THEORY where w? ua1-¥ (5. 34) We must also impose on these equations the condition or oS Lp = Xoo = 0, equivalent to (5, 35) age #4”) = 0, that is to say a: (V-gf") = a4 (V=ef*) = 0. From (1.18), we have -w ps to Fy G0 *gorw? (5. 36) Thus (5. 35) becomes: (235) * -w sin @) = a) * -w sin@)=0. ag - Ww (5.37) The solution of this equation is: 2 2 ae w =k, (5. 38) k being an arbitrary constant. Eq. (5,38) can also be written eat awe Bee or ue—2 tt (5,39 (B+ wy ye ree f ) The condition 3,, ¢'? = 0 is thus satisfied if wee kK? ao Powe? (5, 40) which is always satisfied byk=0 if w=0. If w #0, (5, 40) can be written as ao 1 1-Sh= ei (f? + u?) (e.= = - 2):SPHERICALLY SYMMETRIC SOLUTION 95 In the case of the spherically symmetric solution where Wuv has the form (5,33), the field equations (II) reduce to Wu = 0, Wa = 0, a We = of or We; = const. Was = 0, O41 Wa = 0 v (5.41) Wu = 0, with either w = 0, w Ke ss . 5. 40) or ao Ky +u ( ) In addition the two equations 82 Wu = 0 and & Wig =0 which occur in (I) are automatically satisfied since Wy is only a function of r and t. B. THE VARIOUS FORMS OF THE SPHERICALLY SYMMETRIC SOLUTIONS. 6, THE VARIOUS CASES OF RESOLUTION OF THE DIF- FERENTIAL EQUATIONS LEADING TO A SPHERICAL- LY SYMMETRIC SOLUTION. The equation ang? = 0 has given the following conditions: either w=0; (5. 40a) w k? or a POP De’ (5. 40b) Three cases are distinguishable:96 EINSTEIN UNIFIED FIELD THEORY Case i: (w #0, u= 0) this leads to w kK? ad ae Case 2: (u #0, w= 0) (5, 40) is automatically satisfied. Case 3; (w ~0, u +0) the condition is: 2 kK? o Baw +k We shall treat these three cases separately. alg Case 1: The particular case u = 0, w #0. 7. PAPAPETROU'S SOLUTION (71). In the particular case where u = 0, we have = 0 . Wa (5, 42) The field equations (5. 41) are: Wu =0 i a (5.43) Was = 0 a KP ag . Wu = The equation Wi, = 0 enters only in the non-static case. On the other hand, from the identity relations, one of the func- tions a, 6,¢ or wis arbitrary. Let Bar, We then have 2 A =4 Bi = Ai = By =0, Vege 44)SPHERICALLY SYMMETRIC SOLUTION 97 and Wig = 0 is always satisfied (cf. (5.53) if a=0 of if AL- 3 a, Log U = 0. From the value of U in (5. 44), it follows that @ = 0 and the field equations must be identical in the static and the non- static cases (cf. Mavrides (66)). From (5, 33), we have Wu = - za Log o- 5%: Logo a Logs sha, Log @ 1 1 - Foi Log U +74 Log Ua, Logs o U= 0, 1+2(tapog2-4 ~~ a: Log U = 0, ( a\2 2a Wea or 1 1 ail 2 Wis = x01 Log a+ qe Log oa: Log at por Log o { 1 o*x® +a; Log U + qe: Log Ua: Log aria 2w? + gor = % (5.45) with = 5. 46) U=2y (5. 46) We now compute Wi, + 2 Was. This yields: 2 2 ao=t+h 14k, (5. 47) B rx We can then deduce from (5. 44) i 2 w 3 \ Usher BE? | kK kK. (5. 48) w=a0 ee Be98 EINSTEIN'S UNIFIED FIELD THEORY On the other hand, the equations Wi = 0 and Wa = 0 reduce to a single equation since 8; Waz = Bera . We thus have a single equation to determine a ora. If we use Wez = 0, we have 1+ pha Logag ~ Za, Logs - 4 - ya, Log U = 0. (5. 49) ‘Taking into account the values of go and U as given by (5.47), (5.34) and (5. 48)and letting en zy k b+ e (5. 49) can be written in the form 2nt oe? r +e =e (2rn'+ 1) = V? (5. 50) whose solution is: an _ 2 | 2m et ay - =, (5. 51) In the case of u = 0, the spherically symmetric solution is wer 1 am is eo, ashe oe ( - ( +E) 32) Wr This was the solution determined by Papapetrou (cf.(71) pp. 74-75, Eq. (15) and (16)). If r—o, we have a-l, por, or-Vv, woo (5, 53) which shows that (5. 52) satisfies the condition that, as ro, the affine connection is euclidean,SPHERICALLY SYMMETRIC SOLUTION 99 8. CRITIQUE OF THE SOLUTION. The first criticism of this solution lies in the fact that, as r becomes large, (5,52) does not agree with (5.8) which results from the introduction of Maxwell's tensor 7, _,, into the equations of General Relativity. This may be itgonven- ient in that one would expect that, at large distances from the sources, the results of the unified theory should reduce to the results of the general theory of relativity combined with Maxwell's theory, even though the results are different near the sources. We think that this conclusion is wrong since we believe, as suggests Papapetrou, that the separa- tion from Maxwell's theory must take place earlier. Following Born let us imagine for a moment that, ina purely euclidean electromagnetic theory, we can substitute for the description of field due to point charges, the defini- tion of a ''free current density'' with the aid of an electric field which remains finite as r->0. As in Born's theory, this density p will not be a function of the coordinates but will be expressed in terms of the fields. In this case, the as- ymptotic condition p= 0 can be realized only if the fields van- ish, In a geometrical interpretation, this is equivalent to Saying that the space is euclidean. It follows then that to a weak field would correspond a weak but non-vanishing value of pand could not possibly go over into the maxwellian de- scription which gave (5.8), since this description is equivalent to placing ourselves outside the charges to cancel the current. In other words, the non-euclidean character of a point in space will be tied, not to the presence of the charges in the neighbor- hood of the point, but to the existence, at this point, of a density p which is a function of the field. The results ob- tained from such a theory could not go over into other theories which, in the limit, are valid only outside the sources, This limit corresponds here to a euclidean space, We shall try to crystallize these considerations by examining the correspondence between Einstein's and Born's theories. ‘The second objection to the solution (5. 52) is its de- pendence on two arbitrary constants mand k. This leads one to think that the identification of the metric with (5, 52) is arbitrary. In fact, Wyman (85) has shown that it is100 EINSTEIN'S UNIFIED FIELD THEORY possible to choose a metric which with the help of (5, 52), —which depends on two constants—goes over into Schwarzs- child's solution—which depends on one constant. Adopting Wyman's definition of the metric, Buy = Yup * Wd (5. 54) where o. g nl Ug q, =< (5, 55) Le A 1 po +3 pt with hy Ry Apt u, = eee = oe, hh =, ATT, (5.58) # Ver hah V pon oh, a AT v we obtain, with the help of the values (5, 52)for Su» the following results: The only nonvanishing components of the affine con- nection corresponding to (5. 52) are: 2 ag, Ww 1 _ -2w Ay = Ass = ren? Aus ren (5.57) This leads to the following values of bu uy and dy as de~ fined in (5. 55) and (5. 56) -k? h, =| 0 0, 0}, (5. 58) , por’, o>], veo w-?0, when r 7, An examination of these limits shows that L r will satisfy them. It follows then that the sets a and b re- duce to (loc. cit., pp. 431, 432, Eqs. (3. 2), (8.3) and (3.4)) x=108 EINSTEIN'S UNIFIED FIELD THEORY a(2) and a(i) (4) (2) ¢- Bho (l= cosh sx cos sx) pth 0 ~ “Teosh sx = Cos x)" “"E GOsh SR” cos” SK |“ * __ 8? sinhsx sinsx ai o1 peat 8 * Yosh sx= COs 8x)? BSD Coen eR GOs TR | ORE ay -2 a =8"x! (cosh sx- cos sx) at 1 (6.78) 2 Wk nn oer fast o = 142 "(cosh ex - cos sx)? cosh SixCOs' Sx weet o= t+ sink? s'xsin?s'x| o = 142x* ve week? weal? where s=V2h) ifu is real and h imaginary st=v{h| if u is imaginary andh real. An examination of the solutions .a (1) and b (1) of (5. 78) shows that if u does not vanish, then u will approach a finite limit as r-> «. The strong condition given by Wyman cannot be realized except in Case 2. It does not lead to a valid construction of a field u «0. It is improbable but not certain that equations (5.74) for the weak system will not lead to a more encouraging result in Case 3. Further- more, (5.78) cannot represent the properties of a change and mass distribution. In particular, no arbitrary constant in (5. 78) can correspond to the mass of a particle. It would therefore seem that the case u ~ 0, w = 0 does not lead toa satisfactory solution for the strong system. It may be that more satisfactory conclusions would result if a similar res- olution is applied to the weak system as expressed by (5. 74)SPHERICALLY SYMMETRIC SOLUTION 109 14. CHOICE OF A SPHERICALLY SYMMETRIC SOLUTION. The possibilities due to the choice of a spherically sym- metric solution are the following: I. One could, in the particular case u = 0, w #0, adopt Papapetrou's solution (5.52). In this case one should note that the solution as given by (5.52) assumes that u = 0 and conserves only the field component 4. This, coupled, with the definition of the current as given in Chapter 4 (Section 2) would tie ¢., to a*magnetic field and (5. 52) would then correspond to the existence of isolated magnetic poles which is unacceptable. We shall discuss this objec- tion in Chapter 6 (Section 1). We think that it can be re~- moved if we define the current in such a way as to the P14 to the electric field and to represent the charges by the solution u = 0, w #0. In addition, a certain arbitrariness exists in the choice of the metric and the electromagnetic field. Here, it seems again possible to remove some of this arbitrariness by con- siderations which will be discussed in Chapter 6 (Section }). The removal of these two objections would then render the solution (5, 52) a perfectly good one for the static case. II. One could try to solve the equations for the weak system in the general case u «0, w #0. One obtains then a spherically symmetric solution tied to both types of fields. TI. All solutions discussed in this chapter assume: a. the choice of a purely static solution vb. F* or L po = 0 which results automatically from a v variational principle applied to the density 6 = @HY Rup where Ry, is the Ricci tensor. c. The form (5. 14) for the Sup110 EINSTEIN'S UNIFIED FIELD THEORY These assumptions could be enlarged by a, Consideration of the non-static case (this will go into the static case for u= 0). This could eventually be tied to the assumption of magnetic poles due to the creation of a magnetic field by rotating matter. b. Deduction of the equations of the theory from a density 5@ formed with a tensor R, other than the Ricci tensor. In particular if we use RR, R® , or R® | the condition gH =0, Ly =0 does not follow from the variational principle and can be ignored. The spherically symmetric solutions would then involve less stringent assumptions, c. Assuming a different form for the jy for example to the conditions (5. 19) suggested by Vaidya (83) (it seems that this should be associated with ¢4 = 0 in view of Vaidyats recent conclusions) .6 The Field and the Sources A, CHOICE OF THE METRIC AND FIELDS IN THE UNIFIED THEORY. The separation of the fundamental tensor into symme- tric and antisymmetric components Buy = Yuv * Puve gh? = phy 4 gy leads to an indeterminacy in the choice of the metric and the fields. We could—at least in particular cases—compare the results of the solution of y,,, and h#” in unified theory to those of general Relativity. However, to do this, we must know which fields or combination of fields are susceptible of representing the 'ttrue't metrics. We could postulate, a priori, that the "true"! metric a,,,, must be a function of Spy T f that is of gy, and its derivatives. Keeping in mind the pertinent remarks of Wyman (85), we could assume that the metric has the form Auy = luvlBpor Tho) = Iyy(Bpq> 280) (8 1) provided that | y 18 such that ay), still represents a symme- tric tensor. In particular, Wyman has shown that a certain choice of the metric, although quite artificial, will permit 11k112 EINSTEIN'S UNIFIED FIELD THEORY us to write Papapetrou's solution (static spherically sym- metric solution with u = 0, w =0) in the form of a Schwarz- schild solution with one arbitrary constant. This example illustrates the fact that the choice of a metric in Einstein's unified theory is an important but unresolved question. A desire for simplicity has led most authors to yy, (Sy, in Einstein's notation) to the metric. However as proposed by Lichnerowicz (cf. (10), p. 1288), it is as simple and more justifiable to associate hH” with the metric (hHY = gl) In this fashion the waves ¢ satisfy the charac- teristic equation: Arg = ¥ awa, p= bs, v ay9=0, (6.2) In any case the metric can be defined by the y,,, or hn?” up to an invariant. To guide ourselves in the choice of the metric and the fields, we can try to adopt certain definitions which will permit us to find in the unified theory certain results of the Born~Infeld nonlinear electromagnetic theory. We shall give a short summary of the theory before discussing the above question. B, PRINCIPLES OF THE BORN-INFELD THEORY (122), We define in a euclidean space, a coordinate system by introducing the metric a,,. We shall adopt the convention that a bar under the letters as to whether the index has been raised or lowered by use of a,,,. If s,,, and s~=ala’7g, represent the covariant and contravariant components of the electromagnetic field, then the two invariants of Maxwell's theory are F-4s,, ott. 5 ae a” Suy Spgs (6. 3) i wav. 1 pups 8 4 Bag Sues = OS Sy Spo = yg (8-4)THE FIELD AND THE SOURCES 113 where s and a are the determinants of s ey and auy and the asterisk defines the conjugate fields: uy vo +1 yvpo s*HY 2 aup a” shy € s pos ma po» x w-a Sty = 3 €uvpo she (6. 5) Here <#¥PO = + 1, 0 is the Levi-Civita tensor. We consider now the scalar density * Lp = 2v-a (6. 6) where 7 is the determinant of the tensor Tyy = Ayy + Sy, (6.7) Calculation of the determinants similar to those of Chapter T leads to qearsrgall aw Spy Spo > (6. 8} that is Lp = ra L, (6,9) where Le (1+S +3 ath as Spo) 2 4R-G%)? ~ Lv “po, ~ . aa m (6. 10) We can now fens the conjugate field from £;* . . ab wu , aL Fa pie = SEB Vea (22% SAY 42% gal) (6.11) or uv —s v puv = $-7.2.G sty (6. 12) L *Born and Infeld do not start from the density yy = V7 but from Lavon -V7a, * The derivation of 2 and L are carried out by assuming that Si and 5, ,, are independent variables. (See footnote p. 20, )114 EINSTEIN'S UNIFIED FIELD THEORY A variational principle applied to Lp leads uniquely to Euler's equations TH 2 a (la pl) = 0, (6. 13) A second relation is cbtained by assuming arbitrarily that Sup = uSup * Sup t Soy =O (6,14) or ap (v=a s*#P) = 0, Instead of considering 2p as a function of Quy and s i) we can define a Hamiltonian Hy = WG (6, 15) x being the determinant of the tensor Xuy = yp +Phy (6. 16) We have then the following results: Sey = 2V-a H, (6. 17) where He Be (eB ogame og Phe) a? (6, 18) with 1 wt prllpx .-tglP gw # Pag prltpa= ga” a” ps Py (6. 19) QqzelpHps = 1 .uupoyx p* =F 4 Hi wy a (6, 20) for phy = — HYPO yxTHE FIELD AND THE SOURCES 115 We then define the conjugate fields s/f, from p*H¥ Yaa st = = Va (28 pity + 22 puy) (6, 21) uy or s¥ly = (6, 22) prike - Qplv i : In the particular case of a static spherically symmetric solution, the non-linear relations between the displacement and the electric fields lead to the conclusion that the dis- placement D(p™) tends to infinity while the electric field E(su,) stays finite. We thus have e Dr=-3, E = r (6. 23) Born's theory proposes two equivalent interpretations of charged particles: one interpretation considers them as point charges producing the displacement D which is dis- continuous at the origin and the other assumes that the particle can be replaced by a continuous charge distribution characterized by the field E remaining finite at the origin. The latter conception leads to the definition of a free charge density p and a free current which are functions not of the coordinates, but of the field quantities. The current density is then defined by Hol, cg git g “Tt a6 as (6. 24) while the four-vectors 3p(v-a pHP) and ap(v=a s*H?) noth vanish. In the spherically symmetric static case, integration of (6.24) leads to the charge of the particle.116 EINSTEIN'S UNIFIED FIELD THEORY C. THE NON-LINEAR RELATIONS BETWEEN THE DISPLACEMENT AND THE ELECTRIC FIELDS IN THE UNIFIED THEORY. The interpretation of the notion of a charged particle suggested by Born's theory seems particularly adapted to the principles of a field theory such as the unified field theory, The hamiltonian in Einstein's theory is expressed in terms of the Ricci tensor R v and introduces two types of tensors: on the one hand the g#¥(hH¥, 14?) and the Suv (Yup Py) and on the other hand R,,, Thus there are four tensors (£4, Pum Ruy nr) that can be associated with the electromagnetic field and as we have seen (cf. § 2) two tensors Or hl) that can be associated with the metric in a manner which is still to be specified. In unified theory, the most common form for the hamiltonian is the homogenous density. 5 =v oY” Ruy (2.19) If we want to define conjugate fields with respect to 9 and associate them by nonlinear expressions as given in Born's theory, it is necessary to adopt a purely affine theory by specifying conveniently §. In this case, the variations arp, of the only independent variables lead to corresponding variations 9R,,, such that 8 foar= fet” oR, ar=0, where . gh =e (2.16) yy Without writing explicitly the form of © , we obtain only the group of equations (cf. Chapter 2, § 2) which, in terms of the connection P _pP 2 - Lay = Tuy +5 8 Ty (Ly = 9)THE FIELD AND THE SOURCES 117 can be written Me we up. D, Gt =0, 8, 8h =0 The other field equations are obtained from (2.16) and assume consequently a convenient form of the hamiltonian In 1943, Schrddinger had tried to obtain nonlinear re- lations between the conductions and the fields analogous to those of Born's theory (22). To this end, he used instead of the density $ (Ry,) the expression One”, Ry) = V¥=g heY Ruy ~ $ (Ry) - (6. 25) He defined the conjugate quantities in terms of 8: - 28. wy 2 8 Ruy" ogg? PR (6-28) wy To derive the second group of the equations of the theory, Schrodinger specifies 6 (hH”, Rw) in the following manner (cf. (22), p. 51) B= 2 ef V~det(hyy, + Ruy ~ ¥~det yh (6.27) This definition is very close to the one Born proposes. Sub~ stitution of (6.27) into (6. 26) leads to:” gv _ 2h (ue nYR,, - RTH), (6. 28) v Ruy = ETP? Rup Ryg ~ buf A > 1], (6. 29) voy¥ where Ey _i pee RW = L= qk Bw a = Ale pv ar A=(1+4 he Ruy Rag” BR),118 EINSTEIN'S UNIFIED FIELD THEORY Equations (6.28) form asystem of non-linear relations between inductions and electromagnetic § fields and permit us to express the tensor R,,,, ~ 3 buy nee Roo as a function of a maxwellian energy tensor. Schrédinger (75) has rewritten the definitions (6, 28) and (6, 29) in the particular case of a static spherically symmetric solution, Using polar coordinates, he seeks a solution of the type hn=-e, ha =~ 22, hye = 3? sin?0, hay = e% Ra = - Ru = ¥(x), v Vv A, v, and wheing functions of r. The metric and the elec- tromagnetic field have a form which is identical to the forms of Papapetrou. However, here the various quantities are associated with hy, and Rup (and not yyy and Y,,) which do not form the symmetric and antisymmetric parts of g,,,, (or g#¥), Integration of . gH = 0 and (6, 29) leads to the solutions ener = a Ser rt - r*)dr, ve a oe™ + rt where f:, b, « and c are constants. This is identical with the electric field proposed by Born and Infeld and conse- quently finite at the origin. Schrédinger also shows that for r >>ro, we have instead of which result in general Relativity from Schwarzschild's solution with the electrostatic case taken into account.THE FIELD AND THE SOURCES 119 This theory, while quite attractive in view of the phy- sical interpretations, has the great inconvenience of de- parting considerably from the more general and more sym~ metrical form of Einstein's initial theory. This is essential- ly due to the hybrid character of the field variables, the ex- pression of § associating Ruy to Huy it follows therefore that the true hamiltonian 6 ( R yp) is a complicated expres ~ sion of the functions R,,, and Ry In addition, (6.26) applied to 6 will not lead to the second group of the field equations. Wuy = Yup 8 Ww +8, “ou +8 “ip = Pup The relations (6, 28) and (6. 29) which replace them will evidently lead to different solutions. This will be clearer in the determination of the static spherically symmetric so- lution which will not be the generalized Schwarzschild solu- tion we obtained in Chapter 5, Nevertheless, we have seen that it is possible to main- tain in a purely affine theory the general relations of Einstein's theory. It is sufficient to specify § in general- izing the expression 2 Oe Se 6 x det Guy proposed by Weyl and Eddington where G,, ({ })is Riccits tensor relative to a Riemannian variety. We can define as in (2.18) (Chapter 2) the scalar density. 2 S=> v-det Ry, (2. 18) as done by Schrédinger. Substitution of (2. 18) in (2. 16) leads to (cf. Chapter 2 § 2) ngeye - minor Ruy (RTE oreppw, V7aetR,, v-TR] (6.31)120 EINSTEIN'S UNIFIED FIELD THEORY with [R] = det R,, (6. 32) Ruy => Buy (6. 33) The resultant field equations are those occurring in the mixed theory (cf. Chapter 2) by adoption of a non- homogenous hamiltonian = GM R,, - Ie, (6. 34) With (6.31), Eqs, (6.33) are the non-linear relations be- tween the conjugate variables (with respect to 9) GUY and Rup. In fact, splitting into symmetric and antisymmetric parts Ruy = Put Puy (6. 35) HY Bl’ y pey (6. 36) and letting [R], I, @ be the determinants of Ruy, Ty), and Pup we have the standard relations of determinant theory: [R] =P + bes Pe TS, 2ogs (8, 37) HUY Bi re 4. SHOSYO TL (6. 38) ply ree og (6. 39) TR “Tey Thus (6,31) becomes ave hl’ es - ee [ey sEplParey ohh ve a ‘6. 40) rvag 1 - ART jae Maer rhe r¥o oy, [R] (6, 41)THE FIELD AND THE SOURCES 121 It is then preferable, as Einstein believed, to eliminate all heterogeniety from the theory and consequently to exclude with the A terms, Eqs. (6. 40) and (6.41) which result from the preceding theory. If we were then to start from a mixed theory without a cosmological term, the tensors R y and GH” will remain conjugate with respect tothedensity ,, On the other hand, the 2 My PR +5 GT, - a0) and the G bw will satisfy the equations (see Chapter 4 § 4) BP _ = o,f =0, +e Wout ° yp = 0 (6. 42) analogous to those of Born's theory. But there are then no non-linear relations between them. These non-linear relations between GHV and g),,, exist by definition and it is possible to assume that the fields and the inductions are associated, in a manner to be specified, to Gy,(Byp) and to the tensor density vue fMY( HY). How- ever, these quantities (although conjugate with respect to £ = 2V-g and associated by non-linear expressions) are not conjugate with respect to the homogenious hamiltonian of the theory. In addition, they will not satisfy the fundamental relations (6.13) and (6.14) of Born's theory since one of them only, /-g f#”, has a vanishing divergence. Thus, the interpretation of the antisymmetric fields introduced by the theory is in part uncertain. Kursunoglu (107) has re-examined this question by asso- ciating the electromagnetic field and inductions with the tensor gy, and its dual p+” =(/ W-7) HPS @,. He intro- duces a nonhomogeneous lagrangian which isin our notation £1= GY Ry, - ap*(v=y - v2) =GMPRyy- By ( -5) The quantity v B/y is the scalar 1+ F - G’ defined in Chapter 1 aw ew 2 * Kursunoglu's b is e in our notation ana & = y (1,22)122 EINSTEIN'S UNIFIED FIELD THEORY and p is a fundamental constant which has the dimensions of a reciprocal length. This theory which starts from a new version of Einstein's theory (39) uses approximate solutions for the static spherically symmetric solution and attempts to find the following results; finite self-energy for the election, equations of motion, bending of light, etc. Recently, Mavridés (65) has re-examined this question in trying to remove the indeterminacy of the metric and the fields. His results are discussed in the following section, D. DEFINITION OF CONJUGATE FIELDS. Consider a tensor which is split into symmetric and antisymmetric parts: Ty = Ay + Sup (6. 43) Let 7, a, s denote the determinates of 7, a, and Sy, and let a4”, aal” and ss#¥ be the minors relative to the corre- sponding elements, yy can be written in terms of its symmetric and antisymmetric parts as THY op HY 4 gi? (6. 44) The relations between ay), Sy), b#” and q#” are those de- fined for yj, Yu», he” and fv in Chapter 1. Let = any = Ty, Typ = F8uy ad (6. 45) Sup = TPu where g is an invariant. We then have m=94 8, (6. 46) and i gf. 1 qh = minor ayy = 77g minor By, =F gl”. (6, 47) The relations (6. 47) are equivalent toTHE FIELD AND THE SOURCES 123 jones 3 a, 1 ay 2 guy « nevoe gi (6, 48) g / ww 2) guy as PHY ery a, Let us assume that the true metric is aj, and one of the conjugate fields is s,,. Then 1 1 = gi BY. = mi =s a= @Z'y, al = Z minor ayy ag ve (6, 49) and we can define the quantities 1 sll = ghP gw Spg =F Py bog, (6. 50) skUY = eHypt g eHUpe op (6. 51) av=a Oo 2 Ivy ald and the two invariants 1 Syy SA% = 5 HP y Guy Ypos (6. 52) 1 1 G=ZS,, sth “G Pl 1 tage [4 . (6.53) If we consider the invariant density L(y, ¢) defined in (1.27): 4 La®Vg=2 fore +E yey uv oe] 2). (6. 54) that is £2 V-yL, (6. 55) where ./8. eye L - [£-(+ +b yny pe Pup S90 * oye (eF-@y? (6. 56) The relations (cf. footnote p. 17) py 1 BL _? ov, Yup we gy at i. ve Ay 8 ee gy YX Pogs (6. 57)124 EINSTEIN'S UNIFIED FIELD THEORY can then be written ob sth - Gs 1 x 9g 4 Y/Y pp. £ obi Foe hg TY git HPv? ong + 9%) 98uy (1 +E - G2)? -4 feu (6. 58) or Eu iy canes" pins (6. 59) where v 1 fe uy == /25¢”, 6. 60) 3 |S (6. 60) b. Let us now assume that the true metric b,,, and the fields are associated with et Let 1 bay = G a 7H a gh (6.61) 1 by 2) guy. nrg The determinant b of by, will then be = 4 = beth by = TR, (6. 62) We then have Quays Bup Pye IP7 = F hyp hy g f°, (6, 63) ~~b po -J V~h P97 (6. 64) Vy = FT Epps WU = EE wpe and the following expressions of the two invariants 1 =+ up Liv spo Food, yA = 38 up hyo i (6, 65) (= dyue 1S Puy Pog) >THE FIELD AND THE SOURCES 125 G =): (8. 66) Let us consider the function £ (Buys Fay yp) Gefined in (1. 28) e=avg=20h (+Bs th, hyo tt » PY? (6.67) that is 222th (8. 68) with ‘ $ 4 L - f- fie +b hyp hyo fEMEPF a ere +F-G)? (6. 69) The relations 1 aL 2 g * Ey ator * 5 patty = pia ty up tig 17 = 9, Hy can then be written (see footnote p, 21) aL _ ql - Gg t¥* L *duy (4-6)? i h w = Beat + Bat? 2 Soe 1 up yw (6. 70) -4 Jin bh” ‘bo or (6. 71) where (6. 72)126 EINSTEIN'S UNIFIED FIELD THEORY Choice of the invariant. It is natural to assume that the densities formed from the field p#” and the metric a—or from the field q#4¥ and the metric b—will satisfy the usual divergence relations. However the relation GP . 9 (fap tP) = 0 25 ay(v=e £4?) can be written in the form with the choice ayyi I9(IV=¥ pHP) = 9 ove) = 0; with the choice b!”: op(V=e 9 aHP\(-0, = qu?) = 0. (6.73) We will then have 2% pHP £0 if J =1inthe casea (PHP = V-a pHP), (6,74) ay QHP =O -[E = fein the case b(QUP =V=bqHP) . (6.75) The metric and the fields will then be defined in the following way per fz ne a aoo=y , Be fy b. or V-b bHY = v=g be, Sup = Pu Tuy = Pus (6.76) guy = /E eH, 4 Y Thus, in both cases, the fields and inductions can be expressed in the same way as a function of y,,,, and £4¥, On the other hand, the metric a,,,, and by, are such that a=b-y. Application to the case of spherical symmetry. Let us as- sume that a static spherically symmetric solution has onlyTHE FIELD AND THE SOURCES 127 a single antisymmetric field vis so that we can use Papapetrou's solution (5. 52) since this is the only usable solution. We then have (by putting V = 1 in (5. 52); 1 “@ w an om? 1-2 g,,=( 8 with ame, Ke Le + ? mm -p sin’ 9 o= e ) (Ql +a), -w o g-r’, w=4,, (6.77) that is g=- (ao - w’)p” sin’@ = ~- r* sin’a, (6. 78) . Ke y=-aop’ sin’@ = - (a +5) r* sin’6, (6.79) [PE F Using (1.18), we can deduce the components g#” niiy = Ey + E gta grb Yap (6. 80) 1H 2 Fol 4 E yay gay that is “oO “Ww wi = B gv = 1 f sin? 6 w a 14 Yt ae _k. (6. 81) i =a” yo Pia128 EINSTEIN'S UNIFIED FIELD THEORY Depending on whether we choose a or b, the metric tensor will have the components: a, Choice ayy = Yup b. Choice bHY = ghey = 2 (orb, Puy): o1 1 2m Ke au =p Das = ar - > 1+) - =m 3 2 Dag 2 2 aes Sy ba=sea 7? Livy 2 a au = (1-2) 0) r (6. 82) and the components of the electromagnetic inductions and the field will be the same in (a) as in (b). k Su = Tu =u =F, (6. 83) k 1 pi =q™ = . ms POTEET: (6. 84) 1+ These expressions can be rewritten in the form bsu=-=D, (6, 85) r -pp = ——=7- = £ (6, 86) 4 r where kb=e, k= re. (6.87) While the induction D becomes infinite as r tends to zero, the field E remains finite at the origin. These conclu- sions are those predicted by Born's theory. The value of E—THE FIELD AND THE SOURCES 129 defined here in terms of its contravariant components pHY— due to a generalization (w ~0) of Schwarzschild's solution, is the one predicted in Born's theory. E, CURRENT AND CHARGE. a, Let us assume that we have chosen the metric ayy = Ypy and reconsider the equation gH 2 a,(v=g HP) = 0, that is a (=aplP) = 0. (6, 88) From (6. 58), we have ap (=a pH?) = dy [ea a2 22k she 8 +k © sxup)] = or fe. 89) dima jH = a (v=a sty, (6. 90) where . 1 Lp aL * “An = [2 st? ap (22) + sto, (22) (6.91) + z ap ORa sep] The term ap(v-a s#¥P) = % 1 MWAT(a, 8,7 +8) Sry + 278yy) brings in the pseudo-vector Puyp = a ou and does not vanish when we maintain the usual assumptions of the unified theory which do not assume that jj) = 0. Inthe particular case of a static spherically symmetric solution with only x4 (Guvp = 0), this term does not enter. In this case j! as defined by (6, 90) reduces to j* such that130 EINSTEIN'S UNIFIED FIELD THEORY fay =A anGrastt), (6, 92) But from (6, 82) and (6, 83) stt =a al sa 5 Le > (6. 93) 1+ 7 % V-azV¥- =r sing liz (6, 94) Substituting in (6.92), we have 4njt =«—bee 2 — = 5 (==) - r? 1+5 or +k 7 If we let as in (6. 87) kb =e, k=x, (6. 96) then (6.95) can be written a A\3 3 (x r Lt 2are @ ( + 5) r Integrating (6.97) over all space we have co 7 ,at fp av= J J, J, bj*V-adv, with dv=dy dé ar (6, 98) orTHE FIELD AND THE SOURCES 131 oT 28 2 os Joav= ff’ f° —2E 88 ay ae ar 0 70 ; (6. 99) aad yr = 28 f =ef cosydp=e, rT r r\3 0 me \ ry where r. tey = = % Thus, the charge distribution, whose expression as a function of the fields is characterized by the free density j*, jeads to a finite expression when integrated over all space. In this case with ayy as the metric, we can associate p' and Si, or Vg/y £ and 4 with the electric field. b. The expression (2, 27) Mag GHP. g 3 5 g! 0 cannot lead to a valid definition of the current in case b. In fact, (2.27) can be written as [2a] = 8% = at e.9)) =0 Pup Fup (6. 100) 5 and leads us to define a current 0) (V=y yt *)P%q 4) = an(\-a st) deduced from L (v,y). This current is the diver- gence of a field syg = yg associated with the metric yyy = ayy which permits the definition of the contravariant compo- nents pl! = yA yO org, This definition is not valid in the case b of the metric bHY = hy, To obtain a satisfactory definition of the current, we start from the pseudo-vector ? uv = 29? uv +8y Pout 82 Pup132 EINSTEIN'S UNIFIED FIELD THEORY which, in general, doesnot vanish in unified theory. We then have using (6. 76b) and (6. 70) Pup * 8 Tuy t 2 voy + 8y Typ» (6. 101) with aL aL kg Pw Bagg = 2aR4 Ey Vy (8. 102) Tuy = 2bup L (f, h) being expressed by (6,69). Substitution into (6.101) leads to aL aL aL Yup ~ Grp % Vay *2u rose * ye 5G (6. 103) + circular permutation of u, v, p. But from (6. 73) 1 3 EMT 9, Sq = ap(V=b q#P) = 0 (6. 104) Hence (6. 103) becomes - 4 V7 iH = 8p (¥=b gtt®) 1 =§ HPS (a, dng +8 Apr +2, Ggp) 5 where (6. 105) 4a oh Pearl aE +atMy BE + 9,(V=b rey] ae , (6. 106) We can immediately see the similarity between (6. 106) and (6. 91) but, in case b, the current i+ is the divergence of the field q*#, In the particular case of spherical symme- try, the definition of the charge density V-b i* is associated with 9, v-b q**! that is the component = baz bys = [2 hashes. (6, 107)THE FIELD AND THE SOURCES 133 This assumes a satisfactory and possible definition of the induction {* and the field @ 3. But, if it seems prefer- able to associate the electric field to the components Ypq * (p. a= 1, 2, 3) and the metric to huy**, it neverthiess remains true that Wyman's solution in ¢: does not lend to physically acceptable results in the limit of the so-called ' strong conditions, '’ It is possible that this problem can- not be solved without associating toy 14 the field 23 in the spher- ically symmetric case,! In any case, if we adopt the defini- tion (6. 101) for the current, we must, in the spherically symmetric case, examine all the possibilities discussed in Chapter 5, Wr If we limit ourselves to a spherically symmetric solu- tion associated with the antisymetric field ~, and to the choice ayy = yyy for the metric, it is possible at the present stage of the theory, to define a field which remains finite at the origin and to interpret the charges as a spatial con- tinuous distribution whose free density is only a function of a field everywhere finite. Thus, we come close to a con- ception of charges an example of which is given by Born's theory in a limited domain. This conception seems neces~ sary to a pure field theory as Einstein's unified theory proposes to be. ‘ cf. § 2, Chapter 4. * ef. $1, Chapter 6 and Lichnerowicz ((10), p. 288). t As carried out by Bonnor for the strong system (§ 12, Chapter 5). it cf. § 14, Chapter 5.7 Some Problems Raised by the Unified Field Theory A. ENERGY-MOMENTUM TENSOR. 1, ENERGY MOMENTUM TENSOR IN GENERAL RELA- TIVITY AND THE CONSERVATION EQUATIONS. In General Relativity, the effects of the gravitational field described by Einstein's tensor Tuy are balanced by the energy-momentum tensor Sup While Sup is expressed uniquely as a function of the geometrical properties of the universe, Tuy has purely phenomenological origin. The fundamental equation relative to the interior case 1 Suv = Guy - 5 uy G@= xX Ty, leads to the identity oL Vp Ty =0 or Plto P 169, 2G, = {it =5G Pou B eps p p (LY where ©, eV -g Ty. In any coordinate system, the energy density 6)? does not satisfy a conservation equation. But, we can define a pseudedensity te which, added toG?, satisfies the diver- gence equation (ef. (5), p. 135). 135136 EINSTEIN'S UNIFIED FIELD THEORY apy + ty) = 0 (7.2) From (1.2), we have 1 aGHY = V=g(- GY” dg. +5 G BM dg...) (7.3) =~ (GH ~ 5 5 etG)VEdg.,, =~ x6” dey, But, we always have the relation oA aA Guy" °0 5G ~ 5G (7. 4) where =o ELEY tA} - GRP GRH (78) Substitution of (7. 4) into (7.3) leads to aa p “XGEY 8g Buy = % Og OM emmy = BF A o Puy { a a (a, ¢ o tia. or, if we use (7. 1) p e %Go =~ ep tes (7.7) where = 0,9) 24. 5P a (7.8) p 2x ts 2 (a GHEY o We will then always have (7. 2) by the introduction of iff which is not a tensor density** and can be made tocorre- spond to apotential energy by analogy with classical mechan- ics, * ‘This is the reason that the basic equation G,, , = 0 which is equivalent to the 6 { 4dr = 0 (7. 5) can be deduced from a variational principle, ** | P can always be made to vanish at a point by a covariant choice of the coordinates (cf. e.g., Eddington (5), p. 136).UNIFIED FIELD THEORY PROBLEMS 137 2. THE ENERGY TENSOR IN THE UNIFIED THEORY. CONSERVATION RELATIONS. We have established the conservation relations (2. 61) (ef. Chapter 2, § C) (115), (116), (41) in the unified field theory: ai AT By Ke + 5 Wye where (2, 56) Ba SOW, PY + Wiel?) -5 OP w CW gw.) y being Riccits tensor (2. 36) formed with the connection hy? whose torsion vanishes. Schrdédinger has shown that these identities can be written in a form which generalizes (7.2) (115). He makes use of the development given in Chapter 2 without making the assumption that the ¢? which define the transformation vanish at the integration limits. We shall follow Schrédinger's method as developed in (115), (116), (117). We consider the expression A= GH Ay, (1.9) with LP LP Lees Lg = 0. (7. 10) Au = Lip uy Meo» po It can be shown that Ap, 6(39G4%) = = 2A yy OMY - GHP 6 Ay, (7.11) where a) Ags Ly Pe Ly. . (7.12) We then have from 1 9) and (7.11) 5A =-A,,, 5H" - Au 8 GHP) (7, 18)138 EINSTEIN'S UNIFIED FIELD THEORY and, for the variations 6G'” which do not necessarily van- ish at the limits of integrations 6 fAdr= Jp Aji, ~ Ayp) 64" ar - Sap ayy 6G#4 ar = fw, OGY ar - fep( ah 0G!) ar (7.14) =X [Gy, deh ax - foot Ayh G4) ar. In fact, we have as in (7. 3) 1 W,,, 8GHY = Way vag( 6glY ~ xe” Epo 68°) (7. 15) 1 = V8 (Wuy - 3 Sup W) de” = xG,,,, det”, where 1 Way Fou = XT uw Spy Ve Thy and from (2. 60) aX =K” 2 Cup Bi? +S pue) = Ky We identify now 6g4” with the Lie derivative obtained for the infinitesimal charge of coordinates x10 = xP 4 ¢P, and obtain (cf. footnote p. 43) v Lv Ly. gp 2b” v eet ag?» 6 gl = ghP Sie + @hY Seg - EOP, (716) ae a 8 alt ygp gh? - ghee + GP? (ee) . The corresponding variation of [ A dr is then 8 SAdr= fon drt fap(AkP)ar (7.17)UNIFIED FIELD THEORY PROBLEMS 139 where the last integral f acs) art - f ats ar pt D arises from the variation of the limits of integration. If we assume that the variations 6G” which occur in (7.13) can be expressed in terms of (7,16) we can replace 6 A in (7.17) by its value (7,13) (See Schrédinger (115)) and obtain (7. 18) BSA dr=x Joy, oe ar (7,18) A AP + fa (AE - Ay 8G HY) ar. Using (7, 16) and integrating by parts, we have 3 fAdr=—2f [Kp + hE, ee Je? 1) + falas? - af, agh’ + 20% £0 Jar. If we take into account (7,15) and the identity (2. 61) et a8 ay Ky! + 5 Wag 2, 928 = 0. derived in Chapter 2, then the first integral on the right- hand side of (7.19) will vanish. Substitution of 6GH” in the second integral by its value (7.16) leads to: BS A dr= f a {(GFA + Ag, HY + 2xGB)EP (7. 20) =(Apvo™” +A$,G" - ah An, GhY a Ear, However, in a linear transformation, the components of the affine connection transform as a tensor and consequently A behaves asa scalardensity. (7, 20) must be invariant inde- pendently of the £?. In particular if the £? are constant, (7. 20) reduces to140 EINSTEIN'S UNIFIED FIELD THEORY dglGp + ty) = 0, (7. 21) with o 1 o x= 3 (HA + AY, MY. og. 99) This is exactly eq. (7.2). On the other hand, substitution of (7. 21) into (7. 20) leads to Totty Ab, GY + At, Gh - ag, MY (7, 23) where we have taken (7, 22) into account and assumed that So £P are arbitrary constants. Papapetrou has established relations, analogous to (7, 21), for the tensors 8) and sf (112) [sp being express- ed as a function of Ruy just as co is a function of Wyy and ooo 8 Od . 8, = tp 3x! & E°")T,]. He then shows that for sufficient 1 = 52, ly weak static and spherical fields to admit the approxima- tion 1 Six = CK +6 as against &' y, ~P: (7, 24) we can write m= JG" -Gt -G3 -3) av, (7.25) m being the mass which occurs in Schwarzschild solution. For a static field such that T,,,,= 0, we must have m = 0. This generalizes for the unified theory the theorem proven by Einstein and Pauli for general relativity (9). AU static non-singular fields (T,,,, = 0) correspond to a vanishing total mass (cf, (112) p, 1108). 3. THE TENSOR ve IN UNIFIED FIELD THEORY. According to Einstein, an essential principle of the unified field theory is to develop a complete geometrical description of the unified field. Thus, there is only one case that is unified and the tensor T,,,, must be extracted from theUNIFIED FIELD THEORY PROBLEMS 141 first set of the field equations, Referring to the field equa- tions (4,3), we have for the symmetrical part Guy = Pps (7. 26) with poi p ay = Yo Uy 7g Vy Vy, Loge + af, Up ~ uh Uy 127 Na) ~ Lup Liv: v v These equations are rigorous. Gyy is Ricci's tensor formed with the Christoffel symbols {/f,} relative toy,,,,and satisfies the relation p 1 Pp . Vp (Gr -% 8, G) =, where Vp is the covariant derivative in terms of the Christoifel symbols and the indices of G,,, are raised with the help of y. We must therefore show that $1, satisfies the identity 1 Vol @P - 550%) = 0. (7. 28) This equation simplifies considerably if we use the approxi- mate values obtained by assuming that y,,,, is smalland of the order (~ € ). There the y y differ from the galilean values to order ¢?. We thus obtain (4. 44). = 0. Suv * Say Using the notation of (4.60), we can write 1 Quy = Puy =~ €p&o (7 ur Pookpa - 6 ¥pie Yuo 1 1 (4.64) ‘7 PovpPoup “3 Pe Poup 1 “2 Puc'aPowp)142 EINSTEIN'S UNIFIED FIELD THEORY and from (4. 22), we have L Luv =a € (8p? Yup ~ 20? oY uy - ®Puyop + 2,297 uo) (7. 29) The above is valid in the weak system. In the strong system, the first order equation €p pp Pa = 0o (4.62) 1 can be replaced by stronger equations (cf. footnote p.79); p08 p Pav = 0, €p®@ oP up =0, (4. 61) and thus the last two terms of &,,, vanish. Schrddinger shows that, in the strong system, the divergence of (af = yPob no? = Pa) (7. 30) vanishes identically ({117) p. 13 and 22). One can show that this holds also for the weak system (by writing @y» in the form (4.64) and using (4.18) and (4. 46)). Finally, Schrédinger seeks to simplify (7. 26) Guy = - Sup (7.32) by choosing a convenient coordinate transformation Xy = Ry + Ay (x) (7. 32) whose coefficients a, are of the second order ((117) p. 14). With this assumption, we have +€,,9,a, (7, 33) os - Yu ~ Yu * €v uhy Eye y ay: Choosing ay such that . 1 en @uYun - By Yun) + Eye 2nyay = 0, (7. 34)UNIFIED FIELD THEORY PROBLEMS 143 we have eu Outhy- 4% %,) 20 (7. 35) Baty Fv Ya) FY : Substitution of (7. 33) into (7. 29) leads to the usual expression = =-2 oy = Guy 2-9 2% Yip (7. 36) Ou and consequently 6,,,, remain unchanged. We then have 1 1 1 Guv ~ 3 Bur S = 5 €pAp% (thy - 7 Pu wea) =~ Bax (7.37) The expression Yuv ~ ; Sup {uoYoo corresponds to the re- tarded potentials of ¥,,,, whose divergence vanishes. The divergence of this expression therefore vanishes. This con- dition is precisely (7.35). Thus the retarded potentials corresponding to (7. 37) constitute an approximate solution of the rigorous equations (7.26). This solution is not unique 4. THE PSEUDODENSITY tup IN UNIFIED THEORY. The symmetric tensor ¥,,, which occurs on the right- hand side of (7.3% could be identified with the energy ten- sor whose properties it has. However, according to Schré- dinger ((117), p. 23), there are two other tensors that fulfill these conditons. The first is the pseudodensity, up to a function i ty = 510, CAL, + OF Al, (7, 38) where A, Aty are defined by (7.9), (7.10) and (7, 12). The other is a second pseudodensity L : tf = - 5 GM ay Af. (* 39) A