= -A,, (RJ
(5.5a)
= -L,.
The connections are now further simplified
by imposing two final conditions:
G:,(R) = G%(L) = lG:,,
(5.6)
AN
EFFECTIVE
where Gg, is the gravitational
CURVATURE-LAGRANGIAN
285
connection field, and tr P, = tr R, = 0,
(5.7a)
tr (1, = tr L, = 0.
(5.7b)
It is readily shown by utilizing transformation rules (5.3) that Eq. (5.6) is completely invariant. Since the inhomogeneous parts of transformations (5.4~) are linear in the gauge group generators, and therefore traceless, any nontraceless parts of the connection fields would be gauge tensors. Hence conditions (5.7) eliminate the last vestiges of tensor fields from the vector connections, and they represent the last application of the minimality principle. The result of the assumptions made in this subsection is a theory which is more minimal than the previous theories [6] it is modeled upon in the sense that no fields are present that do not perform an essential function. At this point the complement of boson fields consists of gravitation and four vector gauge fields, two in the R connection and two in the L connection. B. The Complete Lagrangian It is finally possible to utilize the various relations developed previously and assemble a coherent theory. Basically, the boson Lagrangian provides the equations of motion for the irreducible parts of the vector connections, while their interactions with the spinors are found through the spinor connection-vector connection relations. When Q,, K, expansions (2.16), spinor-vector connection relations (3.17), local-global connection relations (3.11) and global connection expressions (5.1) are all substituted into the spinor Lagrangian, Eq. (2.5), it becomes
=%= (i/2)(-W” [XV% + To+ (k/2) PJ # - XE + L + t&/2) 4 ~“$1, (5.8) where r, = (rd/4)
hi,a,Pj
(5.9a)
I, = -A”jauXi,(yjyi/4).
(5.9b)
and Note that G”,,(f), d, , and L, do not enter into this expression for -Es . G%(R) and G;,(L) cancel each other as a consequence of condition (5.6), so the spinors interact with the gravitational field only through the factor ( -$j)1/2 and the X terms given by Eq. (5.9). The disappearance of (1, and L, is due to the asymmetry of the connection terms in which they appear (see Eqs. (5.1)). The contracted curvatures associated with the world model of interest are found
286
T.
A. BARNEBEY
by using the assumed forms of the connections, Eqs. (5. I), in curvature definitions (4.8) and (4.9). When the results, together with metric forms (4.6) and (4.7), are substituted into boson Lagrangian (4.12), it becomes
- @/W-kY’*
Wf’,y - WV - W’, - 4. , p, - WI g’W
+ x
WE~Y2
tr{g,,(R)gY’-‘(R) + g”-‘(L)g&l
W>(-W2
Wp, - RuIIPv- &I g”’ + gWL - Ll[~y - Llh (5.10)
where the constants denoted b, c, d are simply related to the numbers a and b occurring in Eq. (4.12), GUYis the usual contracted Riemann tensor and
puv= a,p,- u, + (WW, ,pJ,
(5.11a)
R,, = %& - a,& + WW,
, %I,
(5.11b)
4,
,-A),
(5.1 Ic)
= 44,
- ad6 - WW,
Luy = auk - up
- WW,
, a.
(5.1 Id)
By employing the transformation rules for the P,, , R, , A, , and L, fields, Eqs. (5.4), it is easily shown that the new quantities defined above are tensors under all groups. Explicitly: transformation
effect p,Yw>’ RJx’)
ay
P,(X),
-Iax”’
R,(x),
ax0 aY
A,(x),
ax0
= w
global coordinate fL”W
(5.12a)
Lpy(x’)’ I = -33 -Iax”’ L,(x);
local Lorentz
local gauge
Pw’ = P,Y 3
4’
= 4,
,
R,w’ = R,, ,
L’
= LY ;
(5.12b)
(5.12~)
AN
EFFECTIVE
CURVATURE-LAGRANGIAN
287
The complete Lagrangian JZ stemming from assumptions (5.1) is simply 9 = 3s + =% ,
(5.13)
Z’s being given by Eq. (5.8) and 6pB by (5.10). 2 is simultaneously a global scalar density, a local Lorentz scalar and an invariant under local guage transformations. C. Conjugation and Local Lorentz Reflections In analogy with spinor connection expansions (2.17), P,, and A, may be expanded in terms of y6 projection matrices.
p, = al.P,tr) + alp,(l)
(5.14a)
A, = a,&>
(5.14b)
+ aJ,tO.
Of course, the components P,(I) and A,(i) are not all independent; conjugation relation (55a) together with definition (2.24) of conjugation and the hermiticity of P, and A, implies
pu0; = -A, (,‘).
(5.15)
The behavior of the components introduced above with respect to an improper local Lorentz transformation may be inferred from the vector connection expressions, Eqs. (5.1), the global-local connection relations, Eqs. (3.1 l), the postulated vector-spinor connection relations, Eqs. (3.17) and the Lorentz transformation properties of the spinor connections, Eqs. (2.19). These verious equations combine to yield the following behavior under a local Lorentz reflection:
PLa (3’ = pw(I, and (1, (1)’ = (1, (f)
(det a = -l),
which, with the help of component be rewritten in the form
relations (5.15) and expansions (5.14), may
P,’ = -A,
(5.17a)
and A,’ = -P, 595/82/x-19
(det a = -1)
(5.17b)
288
T. A. BABNEBEY
When the Lore& reflection rules above are compared with the effects of conjugation as given by Eq. (5.5a), it is seen that each of the two operations duplicates the action of the other. In this sense, conjugation and local Lorentz reflections correspond. Similarly, it may be shown that the analogous statement regarding the fields R, and L, is also valid. And in fact, by continuing the development in this direction, the conjugation invariance of the theory is found to be equivalent to local Lorentz reflection invariance. D. Local Vectors and Axial Vectors Consider yet another pair of fields pu and au , which are introduced relations:
by these
P, = pu + y5au
(5.18a)
4
(5.18b)
= -pu + Pa, .
With the help of expansions (5.14) and the definitions is found that pp = U/W,(r)
+
of a, and al , Eq. (2.9), it
P,(Ol
and
a, = U/NP,(r) - p,W and from the appropriate properties of components P,,(a, Eqs. (5.16a), the improper Lorentz transformation rules of p,, and a,, are easily derived. ‘-
Plb - Pu
a,’ = -a,
(det a = -1).
Clearly, pu is a local scalar and a, is a local pseudoscalar. Expressed differently, the components pi and ai, which appear in the local Lorentz frame expansions PP = hu”ipi
and au = Auia i form a local Lorentz vector and axial vector respectively. Of course, a similar vector and axial vector could also be associated with R, and L, , but they shall not be explicitly displayed because R, and L, may be used instead to represent pseudoscalars, as is shown in Section V1.C. This subsection completes the discussion of the world model and its general properties. Its relationship to effective Lagrangian theories is contained in the following section.
AN
EFFECTIVE
289
CURVATURE-LAGRANGIAN
VI. THE PHENOMENOLOGICAL
LAGRANGIAN
A. Introduction It shall be shown below that a successful phenomenological theory is contained within the scalar curvature Lagrangian given by Eq. (5.13). In order to obtain the theory of interest, two steps are taken: (1) Attention is restricted to the flat spacetime limit of the Lagrangian, and (2) the terms in the boson Lagrangian, Eq. (5. IO), which involve the commutators (6.la) and (-44 - L, 9 A” - -L)
(6.lb)
are dropped. Step (1) is taken simply because there are as yet no phenomenological theories known to the author which include the effects of gravitation, and since it is with such theories that the present results are to be compared, the curvature of spacetime shall be ignored. Step (2) also leads to a final Lagrangian which is more directly comparable with phenomenological theories. It cannot, however, be justified as a limiting process. Rather, the attitude adopted in taking this step may be stated as follows. An invariant boson Lagrangian is to be constructed and interaction terms are needed; they are to be selected from among the several separately invariant terms found in the generalized scalar curvature, Eq. (5.10); all terms except those containing the quantities (6.1) are chosen. Of course this is a tentative procedure which is being followed for the sake of obtaining a familiar effective Lagrangian. There is always the possibility that the omitted terms provide a direction in which the usual theories could be usefully expanded, but this direction has yet to be explored. B. The Basic Lagrangian In accordance with (1) and (2) above, the limits Gv + 0,
h,i -+ 8 I4
and g IA”+ %v (T,,” is the Minkowski metric) are taken in Eq. (5.13) and the appropriate commutator terms are dropped. Also, specific values are chosen for the constants b, c, d. The Lagrangian then becomes 3 = -Ep, + % ,
(6.2a)
290
T.
A.
BARNEBEY
with 3s = m)[xr”(a;T and gB = -(l/8)
tr(P,,
+ &/2) PJ * - XK + W2) 4) e/4 - +RJ(P
+ (mz/4) tr(P, - R,)(P
- $W) - (l/8) tr(ll,,
(6.2b)
- &,y)(~~y
- RU) + (m2/4) tr(A, - L&P
- +P’)
- P).
(6.2~)
Note that g,,(R) and g,,(L) have been eliminated via their equations of motion; this simplify&g step is p>rmissible in the sense that it leaves the Lagrange equations of the vector fields unaffected. Obviously, the newly introduced symbol m stands for the bare mass of the vector mesons. In the limit above the x”-dependent structure of spacetime disappears and global and local Lorentz transformations become synonymous. All tensors become Lorentz tensors and the spacetime transformations under which 8 is invariant are the usual constant Lorentz rotations. Also, the fields pU and czrrwhich compose P, and (1, (see Eq. (5.18)) now transform as a vector and axial vector, respectively. The effect of a local gauge change upon the various fields in the Lagrangian is already known. The spinors transform according to Eqs. (2.lc), the vectors according to Eqs. (5.4~) and the tensor transformation laws are given by Eqs. (5.12~). Of course, the theory is still invariant under local gauge transformations that satisfy condition (2.7), and this is true even though the vector fields have finite masses. C. Special Gauges and Pseudoscalar Fields
It is evident upon inspection of Lagrangian (6.2) that R, and L, do not directly interact with the spinors, and it is tempting to try to eliminate them from the theory. One way of accomplishing this would be to make the covariant assumptions P, = R, and A, = L, , but then the vector fields in the theory would be massless. Since vector mesons have measurable nonzero masses, this idea could not lead to a phenomenological Lagrangian. An alternative procedure is to assume that R, and L, vanish. However, because of the inhomogeneity of transformation laws (54c), such an assumption is not gauge invariant; R, and L, can only vanish within one special class of gauges at a time (the members of such a class differing from each other by x”-independent rotations in gauge space). It is hereby supposed that such a class does exist. A member of this class shall be called a vector gauge, and fields in a vector gauge shall be distinguished by the symbol ‘. The assumption being made may then be written & = L, = 0, (6.3a) Is,, = I?+, = 0. (6.3b)
AN
EFFECTIVE
291
CURVATURE-LAGRANGIAN
Together with rules (5.4~) and (5.12c), it implies (6.4a)
R, = (2/ig) R-V,R, L, = -(2/ig)
Lap
(6.4b)
and R,, = L,, = 0,
(6.44
where R and L are now to be interpreted as representing a transformation between a vector gauge and any other gauge. Now consider a gauge which differs from a vector gauge by a purely chiral transformation, that is, a transformation satisfying R = L EE Q-l.
(6.5)
The matrix Q introduced here may of course be written as a function of spacetime dependent group parameters 4”(x), k = 1,2,..., n2 - 1, that is,
and it is shown in Appendix B that as a consequence of the chirality of the transformation represented by Sz, the 4’s are pseudoscalar fields. From this fact and expressions (6.4a, b) it is seen that pseudoscalars appear in Lagrangian (6.2~) whenever gauges obtainable from a vector gauge via chiral transformations are contemplated, and it is natural to designate these special gauges as the physical gauges. Then the pseudoscalar group parameters 4”(x) may be regarded as measuring the “angle” between a vector gauge and a physical gauge. With all of the above assumptions and ideas taken into account, i.e., Eqs. (6.3-6.5), the boson Lagrangian, Eq. (6.2~) is found in a general gauge to be -9” = -(l/8)
tr[P,,P’
+ n,,+4““] + (m2/4) tr[P, - RJ2 + (m2/4) tr[cl, - LJ2,
whereas in a vector gauge it is (6.6)
and in a physical gauge it becomes -% = -(l/8)
tr[Pw,PuY + A,Jlry]
+ (m2/4) tr[-4, + (2/ig)
+ (m2/4) tr[P,, - (2/ig)
sz-la,w.
Qa,Wy (6.7)
Since the fields R, and L, do not enter into spinor Lagrangian (6.2b), it clearly has the same appearance in all gauges.
292
T.
A.
BARNEBEY
D. Gauge Covariance It is interesting to note that the special gauge forms of the boson Lagrangian given by Eqs. (6.6) and (6.7) are actually equivalent to expressions derived in the more usual approaches [3-51 by simply adding a mass term to the Yang-Mills theory [12]. However, the present development does not agree with the earlier theories in so far as gauge transformation properties are concerned. The massive Yang-Mills Lagrangian and equations of motion are not gauge invariant, whereas the generalized scalar curvature is invariant and yields covariant equations of motion. But the curvature contains more fields than the Yang-Mills Lagrangian, and it is by placing constraints upon some of these fields that agreement between the two approaches is reached in any particular gauge. The discovery that certain gauges can have special physical significance even though the Lagrangian is invariant has a precedent in the massless Yang-Mills case. The author has found classical solutions to the massless Yang-Mills equation which exhibit localized energy densities only in one particular class of gauges [13]. E. Conclusion The physical gauge Lagrangian developed in this paper and contained in Eqs. (6.2a), (6.2b), and (6.7) may be easily cast into the following simple form: 2 = i [xWV;#
- xVzy’“*]
+ -$ tr(V;Q)(V*W)+
- t tr[P”,Puv + A,&“] + -$ (sZV;)(sZV-W)+,
(6.8)
where the covariant derivative operators V; and VL are given by
and v,e 3 (a, + + q, respectively. $P represents a theory which describes fermions, vector and axial vector mesons and pseudoscalar mesons, and which possesses vector dominance in the sense that the pseudoscalars do not directly interact with the spinors. Lagrangion (6.8) is essentially identical to the primary Lagrangian from which Finkelstein et al. have derived a well-determined phenomenological theory that reproduces all the usual low energy W(2) and W(3) results [4, 5, 141. The present
293
AN ETFECnVE CURVATURE-LAGRANGIAN
formulation differs only in that the right-left (i.e., P, - ll,) symmetry is more explicit; compare, for example, Eq. (6.8) above with Eq. (1.2) of [4]. To summarize, the local invariance group has been expanded to include chiral SU(n> x W(n) transformations, and it has been shown that the generalized scalar curvature contains a useful effective Lagrangian.
APPENDIX
A: LOCAL LORENTZ FRAMES
1. Local Lorentz Tensors The formalism in the text is required to be independent of the labeling of spacetime points, i.e., it is to possess global or world covariance, but the arbitrary coordinates xU thus allowed are not the directly measured ones. Rather, measurements are made with respect to an arbitrary Lorentz frame constructed in the neighborhood of a particular spacetime point. The theory is also required to be independent of the choice of these local frames, that is, locally covariant. Local Lorentz frames and their associated local tensors and spinors are discussed in this appendix, the general approach being adopted from a paper by Robertson [15]. Choose four contravariant world vector fields XUi(x), i = 0, 1, 2, 3, and define four covariant vector fields X,,(X) by A~&
= qij
(A.la)
and huj&j7p = au,,
(A.Ib)
where 7;1~,and qii are the Minkowski metrics. Let Au and B,, denote a pair of arbitrary world vectors. They may of course be expanded in terms of the x’s; explicitly: A” = )cuJ’, B, = &Bi, 64.2) and with the help of Eq. (A.la) these expressions may be inverted to yield the following formulas for the world scalar components A* and P:
By utilizing expansions (A.2) and orthonormality condition product of global vectors A” and B, is found to have the form AuB 11= A”B$,
(A.la),
the scalar
= A”Bo - A - B ,
which is seen to be the Lorentz scalar product with A* and B’ playing the roles of Lorentz vectors. This identification is now made more complete.
294
T.
A.
BARNEBEY
Note that Eqs. (A.l) do not completely determine Pi and hut ; linear combinations of them can also be orthonormal and form complete sets. But if a new collection of basis vectors h~‘i and Xui’ is defined by (A.4a) and (A.4b) then the coefficients a$ are restricted by the requirement that Eqs. (A.l) must remain valid among the X’s. In particular, this condition implies
Clearly, since qij is the Minkowski metric, a,j is a Lorentz rotation matrix. Furthermore, as a consequence of Eqs. (A.3) and the invariance of world vectors with respect to changes in the local Lorentz frames, sets of components such as Ai and Bi are really Lorentz vectors as hinted above; for instance, Eqs. (A.3) and (A.4b) combine to give the following Lorentz transformation rule for Ai: Ai’ = aijAj ,
(A.3
where dj
f
qakakzq$j.
The basis vectors h and the matrices a( represent local Lorentz frames and Zocal Lore& transformations, respectively, in the sense that, in general, both sets of quantities are functions of the spacetime coordinates xU. Many of the formulas above become neater if the metrics qij and $1 are used to raise and lower local Lorentz indices in the usual way; e.g., Ai
= u--r)
ijh
Ai = qijAj,
Y3 9
f&j
G
fZik7)kj,
etc.
Then, for example, Eqs. (A.l) become
It is now a simple matter to develop a local tensor calculus. Local vectors transform according to Eq. (A.5), higher rank tensors transform as AiAj,
AfAjAk,...,
etc *,
and all such quantities are global scalars if they comprise the local components world tensors.
of
AN
EFFECTIVE
295
CURVATURE-LAGRANGIAN
2. Local Lorentz Spinors
The local spinor calculus may be taken over directly from the usual spinor theory, which already contains the mechanics of Lorentz transformations. The only innovation is that the transformations are now allowed to be spacetime dependent. Reviewing briefly, if 16 is a spinor, then under a local Lorentz transformation represented by the spinor matrix S(X), # changes according to the rule
S is related to the vector transformation plyis
matrix
aij by
= aij(x) yj
with solution qx)
=
e(l14wqiLd,
where the o’s are spacetime dependent Lorentz group parameters satisfying
and the y’s are the usual constant Dirac matrices which obey the anticommutation rules yiyj + yjyi = 2.p. 04.7) The spinor fields are assumed to be global scalars, and of course the constant Dirac matrices are scalars under both global and local transformations. The hermitian spinor matrix y5 is defined by y5 = (i/4!) ~~~~~~~~~~~~~ = iy”y1y2y3,
and its properties include (y512 = 1, y5yi + ygy5 = 0, and S-ly5S = det ay5. In the text, some additional Dirac matrices are employed which transform as world vectors. They are defined by yu(x) = &i(X) yi
and
Yu(X)
z hifi(X)
Yi
*
T. A. BARNEBEY
296
These expressions, along with Eqs. (A.6) and (A.7), imply Y”YY+ y”y’ = 26P”) so the y@‘s appear to be the natural generalizations of the usual Dirac matrices. Note that these objects are neither constants nor local scalars, however.
APPENDIX
B: LORENTZ REFLECTIONS OF THE
CHIRAL TRANSFORMATION PARAMETERS
According to rules (5.4c) a chiral gauge transformation, i.e., one which satisfies condition (6.5), produces the following changes in the fields P, and /1, : P,’ = J2PJ2-l
+ (2/ig) GVJF,
A;
- (2/ig)
= Q-lA,Q
i-2-1au52.
(B.la) (B.lb)
In this appendix the local Lorentz reflection operation shall be denoted by * in order to distinguish it from the gauge transformation. With the new notation, reflection Eqs. (5.17) become Fu = -A,
(B.2a)
and (det a = -1).
cr, = -P,
(B.2b)
By reflecting Eqs. (B.l) and applying rules (B.2) above, A,’ = ztkl,~-l
- (2/ig) s”ia,a-1
P,’ = &lPuG
+ (2/ig) C%,fi
and
are found. But the latter relations must be consistent with Eqs. (B.l); therefore, si = 52-l.
(B-3)
The fields c$“(x) which parameterize the chiral transformation are coordinates in the group space, and it is always possible to choose the origin of coordinates such that a<-$) = In-l(4).
297
AN EFFECTIVE CURVATURE-LAGRANGIAN When this is done, Eq. (B.3) is seen to imply QC&
= Q(-$1
or C$ = -$ so the group parameters stated in the text.
(det a =
of the chiral
-1),
transformation
are pseudoscalar
fields
as
ACKNOWLEDGMENT The author is grateful to Robert J. Finkelstein for suggesting this work and for the abundant, valuable advice and guidance he provided while it was in progress.
REFERENCES 1. 2. 3. 4. 5.
J. SCHWINGER, Phys. Rev. 152 (1966), 1219. S. WEINBERG, Phys. Rev. Left. 18 (1967), 188. J. WESS AND B. ZUMINO, Phys. Rev. 163 (1967), 1727. R. FINKEL~TEIN AND L. STAUNTON, Physica 47 (1970), 182. R. FINKELSTEIN, L. STAUNTON, AND J. HILGEVOORD, Phys. Rev. D 1 (1970), references to other effective Lagrangian
6. R. FINKEL~TEIN
AND W. RAMSAY,
Ann.
2832 (Extensive
theories may be found in this paper.) of Physics
(NY.)
21 (1963),
408.
(Relevant
earlier
works are referenced in this paper.) 7. M. GELL-MANN, Phys. Rev. 125 (1962),
1067. 8. R. FINKELSTEIN, Physica 44 (1969), 262. 9. E. SCHROEDINGER, “Spacetime Structure,” Chap.
XII, Cambridge
Univ. Press, London,
1950. M. BAZIN, AND M. SCHIFFER, “Introduction to General Relativity,” Chap. 5, McGraw-Hill, New York, 1965. 11. A. EINSTEIN AND B. KAUFMAN, Ann. of Math. 62 (1955), 128. 12. C. N. YANG AND R. L. MILLS, Phys. Rev. 96 (1956), 191 . 13. T. A. BARNEBEY, Special Solutions to the Yang-Mills Equation, unpublished, 1968. 14. L. STAUNTON, “Group Geometry and Effective Lagmngians Generated by a Massive Gauge Field,” Ph.D. dissertation, University of California, Los Angeles, 1969. 15. H. P. ROBERTSON, Groups of Motions in Spaces Admitting Absolute Parallelism, Ann. of Math. 33 (1932), 496. 10. R. ADLER,