5
PHYSICAL REVIEW D VOLUME 15, NUMBER 8 15 APRIL 1977 Gravitational Lagrangian and internal symmetry Elhanan Leibowitz Department of Mathematics and Department of Physics, Ben Gurion University of the Negev, Beer-Sheva, Israel (Received 9 September 1976) The field equations of general relativity, when interpreted as a set of equations in a Riemannian manifold endowed with internal space at each point, are shown to be derived from a variational principle. The failures of other approaches to recover the full general-relativity theory are traced to the absence in their Lagrangians of direct coupling between space-time and internal structure. Examination of this observation sheds some light on the fundamental implications of the intimate relation between space-time covariance and internal symmetry. I. INTRODUCTION In the classical formulation of in stein's theory of gravitation the dynamical variables are the components of the metric field gp,(x) correspond- ing to the Riemannian manifold of space-time, to- gether with the physical fields ("matter field") present. The field equations are derived from a principle of least action with a Lagrangian L . This Lagrangian decomposes into two parts: the gravitational Lagrangian, Lo, which is equal to the Ricci scalar R, and the mat- ter Lagrangian L,, viz., (K is a coupling constant). Variation with respect to the metric field in this action integral leads to the Einstein field equation It has been found, however, that from a theo- retical as well as a computational point of view, it is desirable to recast the theory as a formalism in space-time with internal structure, posessing SL(2, C) symmetry. This is the conceptual back- ground for the two-component spinor theory in curved space-time.' This approach is also the basis for the tetrad calculus and the Newman- Penrose spin-coefficients formalism,' and is use- ful for the analysis and classification of gravita- tional fields. Furthermore, it has been shown by Carmeli3 that, using such a framework, the theory of gravitation can be cast into a Yang-Mills-type formalism. Yang in his recent integral formalism4 again relates gauge fields and gravitation when the latter is expressed as an SL(2, C)-covariant theory. This is one step in the current tendency to bring general relativity more in line with the prevalent trends in particle physics. It is therefore of importance to find out whether in formulations of that type the Einstein field equa- tions can be derived from an action principle under the variation of the fundamental quantities up- pearing in the partzcular formulation. More spec - ifically, the question arises whether an action principle can yield the field equations without add- ing in the process ad hoc quantities. A partial answer to this question has been given in a few article^,^ where it is demonstrated that an appro- priately chosen Lagrangian leads to field equations which are related to Einstein equations. These equations, however, are weaker than the full Ein- stein equations, and must be augmented by addi- tional conditions in order to recover general rel- ativity. Furthermore, auxiliary quantities must be artificially introduced in order to facilitate the derivation. Carmeli's method6 of first-order form suffers from the same latter deficiency. The merits of the aforementioned schemes lie in their Lagrangians being functions of gauge fields, an approach which is not attempted in the present article. Likewise, the equations obtained by Yang4 from a variational principle constitute a vast generalization of general relativity, and ad- mit solutions which are physically ~nacceptable.~ It is the purpose of this paper to show that the full Einstein field equations are obtained from an SL(2, C) -invariant Lagrangian, which is consid- ered as a functional of the dynamical variables alone, without any necessity to introduce any auxiliary quantities. This Lagrangian, in contrast to the Lagrangian considered by others, mixes up the space-time and internal structures. In Sec. I1 the framework is set and the notation is established. A basic decomposition of the cur- vature is represented in Sec. 111, and in Sec. IV the Lagrangian is introduced and the resulting Euler-Lagrange equations are derived. The equa- tions as applied to vacuum and to space-time ad- mitting a neutrino field are discussed in Sec. V. The last section is devoted to concluding remarks,

Gravitational Lagrangian and internal symmetry

  • Upload
    elhanan

  • View
    213

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Gravitational Lagrangian and internal symmetry

P H Y S I C A L R E V I E W D V O L U M E 1 5 , N U M B E R 8 1 5 A P R I L 1 9 7 7

Gravitational Lagrangian and internal symmetry

Elhanan Leibowitz Department of Mathematics and Department of Physics, Ben Gurion University of the Negev, Beer-Sheva, Israel

(Received 9 September 1976)

The field equations of general relativity, when interpreted as a set of equations in a Riemannian manifold endowed with internal space at each point, are shown to be derived from a variational principle. The failures of other approaches to recover the full general-relativity theory are traced to the absence in their Lagrangians of direct coupling between space-time and internal structure. Examination of this observation sheds some light on the fundamental implications of the intimate relation between space-time covariance and internal symmetry.

I. INTRODUCTION

In the classical formulation of in stein's theory of gravitation the dynamical var iables a r e the components of the met r ic field gp,(x) correspond- ing to the Riemannian manifold of space-time, to- gether with the physical fields ("matter field") present . The field equations a r e derived f rom a principle of l eas t action

with a Lagrangian L . This Lagrangian decomposes into two parts : the gravitational Lagrangian, Lo, which i s equal to the Ricci s c a l a r R, and the mat- t e r Lagrangian L,, viz.,

(K i s a coupling constant). Variation with respec t to the met r ic field in this action integral leads to the Einstein field equation

It has been found, however, that f rom a theo- ret ical a s well a s a computational point of view, it i s desirable to r e c a s t the theory a s a formalism in space-time with internal s t ruc ture , posessing SL(2, C) symmetry. This i s the conceptual back- ground for the two-component spinor theory in curved space-time. ' This approach i s a l so the b a s i s fo r the te trad calculus and the Newman- Penrose spin-coefficients formalism,' and i s use- ful fo r the analysis and classification of gravita- tional fields. Fur thermore , i t has been shown by Carmeli3 that, using such a framework, the theory of gravitation can be c a s t into a Yang-Mills-type formal i sm. Yang in his recent integral formalism4 again re la tes gauge fields and gravitation when the la t t e r i s expressed a s an SL(2, C)-covariant theory. This i s one s tep in the cur ren t tendency to bring general relativity more in line with the prevalent t rends in par t i c le physics.

It i s therefore of importance to find out whether in formulations of that type the Einstein field equa- tions can be derived from an action principle under the variation of the fundamental quantities up- pearing in the partzcular formulation. More spec - ifically, the question a r i s e s whether a n action principle can yield the field equations without add- ing in the p r o c e s s ad hoc quantities. A part ia l answer to this question has been given in a few article^,^ where it i s demonstrated that a n appro- pr iately chosen Lagrangian leads to field equations which a r e related to Einstein equations. These equations, however, a r e weaker than the full Ein- s tein equations, and mus t be augmented by addi- tional conditions in o rder to recover general r e l - ativity. Fur thermore , auxiliary quantities must be artificially introduced in o rder to facilitate the derivation. Carmeli 's method6 of f i r s t -o rder fo rm suffers f rom the s a m e la t t e r deficiency. The m e r i t s of the aforementioned schemes l i e in their Lagrangians being functions of gauge fields, an approach which i s not attempted in the p resen t ar t ic le . Likewise, the equations obtained by Yang4 f rom a variational principle constitute a vast generalization of general relativity, and ad- mit solutions which a r e physically ~ n a c c e p t a b l e . ~ I t i s the purpose of this paper to show that the full Einstein field equations a r e obtained f rom a n SL(2, C) -invariant Lagrangian, which i s consid- e r e d a s a functional of the dynamical var iables alone, without any necessi ty to introduce any auxiliary quantities. This Lagrangian, in con t ras t to the Lagrangian considered by others , mixes up the space-t ime and internal s t ruc tures .

In Sec. I1 the framework i s s e t and the notation i s established. A basic decomposition of the c u r - vature i s represented in Sec. 111, and in Sec. IV the Lagrangian i s introduced and the resulting Euler-Lagrange equations a r e derived. The equa- tions a s applied to vacuum and to space-time ad- mitting a neutrino field a r e discussed in Sec. V. The l a s t section i s devoted to concluding r e m a r k s ,

Page 2: Gravitational Lagrangian and internal symmetry

2140 E L H A N A N L E I B O W I T Z - 15

and in the Appendix a few identities used in the a r t i c le a r e stated.

11. RIEMANNIAN STRUCTURE IN TERMS OF SPINORS

In the s tandard spinor algebra in curved space- time8 a one-to-one correspondence i s se t between vec tors 5, in the Riemannian manifold of space- t ime (endowed with met r ic tensor g,,), and Herm- itian spinors <,A, viz.,

= ~ $ g 5 , , 5" = U " A B . tAB . The connecting quantities U X ~ sat isfy the relat ions . .

+U;~U$;) = gPV€, B J (2.1)

p , v = O , 1 , 2 , 3 , A , B = O , l .

The approach to be followed h e r e i s to take the spinor veco t r s u i i a s the fundamental quantities of the theory, and to construct the Riemannian met r ic out of the spin s t ructure. Consequently, Eq. (2.1) i s the definition of the met r ic g,, .

Covariant der ivat ives of spinors a r e defineds by

where the spin connections rAB, a r e determined by the usual requirements f o r differentiation (Leibnitz rule , etc.) together with

The spin curvature R,,,, i s defined by the com- mutator of the covariant derivative

v A I F u - v A J v P = R A B p u ? 7 B , and i s expressed in t e r m s of the connections

~ ~ ~ ~ v = r ~ B ~ , u - r ~ B u , ~ + r A c u r C B p - r A c p r C B u

The spin curvature i s equivalent to the F,, ma- t r i ces in Carmeli 's gauge f ~ r m u l a t i o n , ~ and i s r e - lated to the Riemann curvature tensor Rx,,, by

where

111. DECOMPOSITION OF THE SPIN CURVATURE

The three bivectors S,,,OO,S,,O'=S,V'O, S,," and their complex conjugates a r e s i x independent com- plex bivectors ( a s can be shown with the aid of the "completeness relations" in the Appendix), and hence s e r v e as a bas i s in the space of complex bi-

vectors . The spin curvature R,,,, , fo r fixed val- ue of (A, B ) , i s a complex bivector and therefore can be uniquely expanded in t e r m s of the basis

where the coefficients a r e given by

Contracting (3.1) with S p V and using the "com- pleteness relations" of the Appendix and (2.4), one finds

Fur thermore , a s imi la r but longer calculation shows that the coefficients RABbb a r e related to the t race- f ree Ricci tensor:

where $pUA,;; i s the totally symmetr ic spinor ten- s o r defined in the Appendix.

Infact , RA+b i sequa l tothe spinor - QAB6b in the Newman-Penrose formalism.*

F o r fur ther reference i t i s mentioned here that the symmetr ies of RABcD (RABcD= RAmc = R,,,) entail

N. THE GRAVITATIONAL LAGRANGIAN AND ITS VARIATIONAL DERIVATIVE

If the theory i s confined to quantities which a r e of, a t most, second differential o rder in the upAb, and l inear in the second derivatives, then the only concomitant of the fundamental quantities u p A i which i s s c a l a r under coordinate t ransformations in space-t ime and invariant under the action of the group SL(2, C) i s (up to a constant, which contri- butes only a cosmological term)

Variational pr inciples of this type have been con- s ideredg f rom other points of view. Our calcula- tion, which a g r e e s with the aforementioned r e - sul ts , will be based on the decomposition scheme of Sec. 111, thus gaining in simplicity and brevity. Thus l e t u s consider the change of

under the variation

F o r this purpose we have to express the variat ions of the quantities appearing under the integral sign

Page 3: Gravitational Lagrangian and internal symmetry

15 - G R A V I T A T I O N A L L A G R A N G I A N A N D I N T E R N A L S Y M M E T R Y 2141

in t e rms of 6 a 5 , g o r 6apAB. Using (2.1) one finds The last relation, which i s coordinate- and SL(2, C)-invariant, can be most easily verified in

A B A B 6 g l l u = a p ~ i 6 a u + a v ~ i b a p , a "geodetic" coordinate system and spin frame, A i 6 J - g = 3 J - g e p U 6 ~ p u = q C T ~ ~ ~ ~ U ~ , adapted to the point under consideration. In such a

. . coordinate system and spin frame the fundamental 6 a p A B = -apCDuVAB6aUcb , quantities apA; a r e constant (at the point) and r e - ~ S P U A B = a [ ~ ~ d d o v ~ ~ ~ + a [ ~ ~ . b a v l ~ d duce to the Pauli matrices, and r = O (at the

C point). s ~ ~ ~ ~ ~ R ~ ~ ~ , = v ~ ~ ~ , with TIP = 2 ~ ~ ~ ~ ~ 6 ~ ~ ~ . Taking these relations into account one finds

Finally, using (Al), (3.4), and substituting (3.2), we conclude

Now a s usual we consider variations 6 0 , ~ ' such that they and their f i r s t derivatives vanish on the boundary of integration. Under these assunlptions the vector Vp too vanishes on the boundary and by Gauss's theorem the divergence c g vP 1, = (Ggvp),, does not contribute to the Euler- Lagrange equations. Thus (4.1) leads to the field equations

2RAc %aPCD + noP,& = - K T ' A ~ , (4.2)

where

That (4.2) i s indeed equivalent to the Einstein equations i s confirmed with the aid of (3.3), where- by we find that the left-hand side of (4.2) i s equal to

V. APPLICATION TO VACUUM AND NEUTRINO

In the case of a vacuum, the field equations (4.2) reduce to

~ R , , ; ~ U C' + 510uA~ = 0 .

Contracting with a," we find th is equation to be equivalent to the pair

W = O ,

RAcib = 0

which in turn a r e equivalent to

R = O ,

and these a r e the Einstein vacuum field equations R,, = 0.

For a space admitting a neutrino field qA, the matter Lagrangian L, i s taken to be

- qAq'lA&) - Variation of qA and 77A in the action integral leads to the Euler-Lagrange equations

q A I A i = O (and c.c.) , i.e., the Weyl equations for the neutrino field. This equation can now be used to find the "energy- momentum spinor vector" TLLAg, to be substituted in the right-hand side of (4.2), namely in order to calcylate the variational derivative 6 (G L,)/ 6oLLAB. The variational variables oLLAB appear in

(the contribution of this t e rm i s known from Sec. IV), and in the spin connections rAB, in L,. Thus

6LM =i(VAI .Vi - 71A7?BILL)60LLAB

+ iqA77B(aLLcir61?cALL - auA;6 rci,). (5.1)

Firs t observe that the indentity

AB A& o =a,, I,,= u ~ ~ ~ , ~ - rALLv~X + rACv~LL CB

+ rBEva, AE

yields the following expressions for the variation of the spin connections:

6 rABW = - $ U " ~ ~ [ ( ~ C T , ~ ~ ) ~ ~ - oXAE6rXWV]

=+.+ovAC[(60VBE)lLL + 0ABE6rYLLAI.

(Again this relation i s most easily verified in a "geodetic" coordinate system and spin frame.) Substituting in (5.1), the t e r m s containing 6rX,, cancel, and we have

Page 4: Gravitational Lagrangian and internal symmetry

2142 E L H A N A N L E I B O W I T Z - 15

After some manipulations the following expression is obtained:

with

= - i(SpYAC77i)77C - quuBE77ATE)6U,,ai.

Again, the divergence t e rm does not contribute to the variational equations.

The contribution of the variation of 6, i.e., L , 6 e , vanishes also due to the neutrino equa- tion. Hence we conclude from (5.2)

+gcl~i77d - q c r l ; l ~ ~ ) u " ~ * . (5.3)

From the field equations thus obtained, viz.,

2 ~ ~ ~ ~ ~ o ~ ~ ~ + f b U A g = - K T ~ ~ ~ , , (5.4)

with TpA;, given by (5.3), the familiar form can be recovered. Contracting (5.4) with a p A B yields

and therefore

VI. CONCLUSION

It has been demonstrated that general relativity can be viewed a s a theory involving the fundamen- tal quatities upA* alone (along with external phys- ical fields). The field equations a r e derived from a variational principle, the gravitational Lagran- gian being

It will be instructive to compare this Lagrangian to other Lagrangians considered r e ~ e n t l y , ~ ' ~ " j which have not led directly to the full general- relativistic field equations, but have nevertheless been adopted by proponents of the gauge-field ap- proach to gravitation. A typical Lagrangian pro- posed by those authors i s (in our notation)

In such a Lagrangian the space-time indices ( p , v) a r e contracted with space-time in- dices, and internal-group indices (A, B) a r e con- tracted with internal-group indices, and no mixing occurs. The only way to couple directly indices of one kind with indices of the other kind i s to recas t

RABbv into a form where all the indices appear on the same footing. This i s accomplished by the usual way of changing tensorial indices into spin- orial indices, viz.,

R A B ~ ~ - R A B P ~ x ; = ~ ~ ~ P ~ ~ ~ x ~ R A B ~ ~ . Now the coupling emerges naturally: . ,

E E B X ~ Q Y ~ A B P 6 X P = S p u A B ~ A B p u , and this i s precisely the Lagrangian of the present work, which indeed leads to the full general-rela- tivity field equations. (We mention in passing that the program outlined here can be car r ied out, mutatis mutandis, within the framework of spinor calculus in the five-dimensional unified theory of relativity," thus tying up the electromagnetic field with internal degrees of freedom.)

The las t observation may provide more insight into the centrality in general relativity of the in- separability of space-time aspects and internal group, and into the fundamental role played by the intimate relation between general covariance and internal symmetry.

APPENDIX: ALGEBRAIC PROPERTIES OF THE S r v A B

Several properties of the basic quantities sPuAB, which a r e used in this work, a r e listed here. All of them can be verified from the definition (2.6) and the relations (2.1).

"Completeness relations" a r e

(e,.,, i s the alternating tensor e p v X 7 = c g t J J U 1234 = '1' Single-index contraction yields

+ S P U a ( c ~ D ) ~

+ E A ( ~ E D ~ B ~ " ' ) , S, XAB sU = -$fi A B ~ ; ,

where +pVAB&b i s the totally symmetric spinor tensor

+ p U A B d b = ~ ( p a ( b ~ ' " ~ b )

( $ p U A B ; ~ = @ v p a e ~ b = $ b V s ~ ~ b = $ " ~ ~ , gpv+'" 'a~tb =O). Another useful identity re la tes products of a 's with S's:

upA;uVc~- o u A i ~ p c i = S p u A C ~ & + S " ' & E ~ ~ . (A l l

Page 5: Gravitational Lagrangian and internal symmetry

15 - G R A V I T A T I O N A L L A G R A N G I A N A N D I N T E R N A L S Y M M E T R Y 2143

' ~ f . P. G. Bergmann, Phys. Rev. 107, 624 (1957). 2 ~ . Kewman and R. Penrose, J. Math. Phys. 3, 566

(1962). %!I. Carmeli , J. Math. Phys. 11, 2728 (1970). k . N. Yang, Phys. Rev. Lett. 33, 445 (1974). %I. Carmeli and S. I. Fickler. Phys. Rev. D 5, 290

(1972). "1. Carlneli, Nucl. Phys. B 3 8 , 6 2 1 11972).

? ~ f . R. Pavelle, Phys. Rev. Lett. 34, 1114 (1975). 'F. A. E. Pirani, i n Lectures on General Relativity,

Brandeis Summer Institute in Theoretical Physics, edited by S. Deser and K. W. Ford (Prentice-Hall, Englewood Cliffs, New Jersey . 1965).

9 ~ . Prese , Nuovo Cimento Suppl. 9, 389 (1962); J. Wainwright. Tensor 1 9 , 2 1 7 11968); 9, 265 (1968)

'OE. Leibowitz, J. Math. R y s . 2, 306 (1974).