On the Riemann Curvature Tensor in General Relativity

Zafar Ahsan
Department of Mathematics Aligarh Muslim University Aligarh-202 002(India) e-mail:[email protected]

Dedicated to Prof. Hideki Yukawa (First Noble Laureate in Physics from Asia)

LECTURE PLAN

1. Introduction 2. The Riemann curvature tensor 3. Electric and magnetic space-times 4. Lanczos spin tensor 5. Space-matter tensor

1. Introduction
Through his general theory of relativity, Einstein redened gravity. From the classical point of view, gravity is the attractive force between massive objects in three dimensional space. In general relativity, gravity manifests as curvature of four dimensional space-time. Conversely curved space and time generates eects that are equivalent to gravitational eects. J.A. Wheelar has described the results by saying Matter tells space-time how to bend and space-time returns the complement by telling matter how to move. The general theory of relativity is thus a theory of gravitation in which gravitation emerges as the property of the space-time structure through the metric tensor gij . The metric tensor determines another object (of tensorial nature) known as Riemann curvature tensor. At any given event this tensorial object provides all information about the gravitational eld in the neighbourhood of the event. It may, in real sense, be interpreted as describing the curvature of the space-time. The Riemann curvature tensor is the simplest non-trivial object one can build at a point; its vanishing is the criterion for the absence of genuine gravitational elds and its structure determines the relative motion of the neighbouring test particles via the equation of geodesic deviation.
3

The above discussion clearly illustrates the importance of the Riemann curvature tensor in general relativity and it is for these reasons, a study of this curvature tensor has been made here.

2. The Riemann curvature tensor

k The Riemann curvature tensor Rijl is dened, for a covariant vector eld Ak , through the Ricci identity

where
k Rijl =

k Ai;jl Ai;lj = Rijl Ak

k il l k + m k m k il mj ij ml xj x ij

(1)

The covariant form of the curvature tensor is Rhijl

1 2 ghl 2 gij 2 gil 2 ghj = ( j i + l h j h l i )+gkm (k m k m ) ij hl il hj 2 x x x x x x x x

Properties: (i) Rhijl = Rhilj ,

h h Rijl = Rilj

(ii) Rhijl = Rihjl , Rhijl = Rjlhi

k k k (iii) Rijl + Rjli + Rlij = 0, Rhijl + Rhjli + Rhlij = 0

(iv) Bianchi identities:

h h h Rhijk;l + Rhikl;j + Rhilj;k = 0, Rijk;l + Rikl;j + Rilj;k = 0

A tensor of rank four has n4 components in an ndimensional space and for n = 4, it is 256. For Rhijl , due to symmetry properties, the number of algebraically independent components, in an n-dimensional manifold, is 1 2 2 n (n 1) 12 so that in 4-dimension, this no. is 20. Geheniau and Debever, have shown that the Riemann curvature tensor can be decomposed as Rijkl = Cijkl + Eijkl + Gijkl (2)

where Cijkl is the Weyl tensor, Eijkl is the Einstein curvature tensor, dened by

1 Eijkl = (gik Sjl + gjl Sik gil Sjk gjk Sil ), 2 with Sij being the traceless tensor 1 Sij Rij gij R, 4 is dened by Gijkl R (gik gjl gil gjk ) 12

(3)

(4)

and where Gijkl

(5)

The Ricci tensor Rij is dened by

k Rij Rijk and R g ij Rij

(6)

is the Ricci scalar. Expression (2) may be regarded as the denition of the Weyl tensor, all of the other quantities may be computed directly if the metric tensor is given. From eqns. (3)-(5), eqn (2) can be written as 1 R Rijkl = Cijkl + (gil Rjk +gjk Ril gik Rjl gjl Rik ) (gil gjk gik gjl ) 2 6 (7)

 Geometric properties:

Ruse, Geheniau and Debever

Algebraic classication of the vacuum Riemann tensor: Petrov and Pirani. Witten and Penrose have redeveloped both algebraic and geometric properties (spinor calculus). The relevance of Petrov classication to the gravitational theory was suggested by Pirani. Petrov classes were further developed: Lichnerowicz, Bel, Ehlers, Sachs and Sharma and Husain In empty space-time, Riemann tensor reduces to Weyl tensor. Thus, in order to have a classication of vacuum Riemann tensor, it is sucient to classify the Weyl tensor. Three main approaches for the classication of Weyl tensor-namely: the matrix method (Synge, Petrov), the spinor method (Penrose, Pirani) and the tensor method (Sachs). The connection between these approaches has been found out by Ludwig. The tensor method is equivalent to the other two.

 Type I (Algebraically general) Type II, D, III, N (Algebraically special) Types III, N : Gravitational Radiation Remark: We have seen that the number of independent compo1 nents of Riemann tensor in n-dimension is 12 n2 (n2 1). The symmetries of the Riemann tensor thus leads to (i) if n=1, Rhijk = 0; (ii) if n=2, Rhijk has only one independent component, namely R1212 = 1 gR; 2 (iii) if n=3, Rhijk has six independent components. The Ricci tensor has also six independent components and thus Rhijk can be expressed in terms of Rij as 1 Rhijk = ghj Rik + gik Rhj ghk Rij gij Rhk (ghj gik ghk gij )R 2 (iv) if n=4, Rhijk has twenty independent componentsten of which are given by Ricci tensor and the Remaining ten by the Weyl tensor Chijk . It may be noted that Weyl tensor Chijk makes its appearance only in a four dimensional space-time through equation
8

1 R Rijkl = Cijkl + (gil Rjk +gjk Ril gik Rjl gjl Rik ) (gil gjk gik gjl ) 2 6 and in empty space-time (i.e., Rij = 0) the gravitational eld is characterized by the Weyl tensor. Thus according to general relativity, if we lived in a three dimensional Universe, gravity could not exist in a vacuum region. Maxwell eqns (vector): E = 1 H , c t H = E = 0, 1 E c t H =0

Invariants:

E H = 0, E 2 = H 2

ij Tensor: Fij;k + Fjk;i + Fki;j = 0, F;j = 0

Invariants:

Fij F ij = 0, Fij F ij = 0 - (em radiation)

The Riemann curvature tensor has fourteen invariants. There is the Ricci scalar R. There are 04 invariants of the Weyl tensor Cijkl . There are 03 invariants of the Einstein curvature tensor Eijkl and 06 invariants of the combined Weyl and Einstein curvature tensors. The component form of these invariants are

Weyl tensor Cijkl :

A1 = Cijkl C ijkl , A2 = Cijkl C ijkl

4 4 ij ij B1 = Cijmn C mnrs Crs , B2 = Cijmn C mnrs Crs 3 3 Einstein curvature tensor Eijkl : E = 2Eijkl E ijkl 1 rs F = 4Eijrs Emn E mnpq E ij E 2 pq 2 rs 3G = 32Eijrs Emn E mnpq Euvpq E uvxy E ij 3EF E 3 xy 1 1 K1 = Cijrs E rs E ijmn , K2 = Cijrs E rs E ijmn mn mn 4 4
2 2 2L1 = 32Cijrs E rs E pqmn Cpqxy E xyuv E ij K1 + K2 mn uv 2L2 = 32Cijrs E rs E pqmn Cpqxy E xyuv E ij 2K1K2 mn uv

Combined Weyl and Einstein curvature tensors:

3M1 = 256Cijrs E rsmn Eijmn C ijkl Eklpq E uvpq Cuvwx E wxyz

3 2 E ij K1 + 3K1 K2 18K1L1 + 18K2 L2 yz 3M2 = 256Cijrs E rsmn Eijmn C ijkl Eklpq E uvpq Cuvwx 2 3 E wxyz E ij 18K1 L2 18K2 L1 3K1 K2 + K2 yz

10

 Importance : For example, the behaviour of the scalar Rijkl Rijkl is studied in connection with the existence of any geometrical singularity. If Rij = 0 (empty space-time) then Riemann tensor reduces to Weyl tensor; and in this case there are four invariants of Riemann tensor which are given by
A1 = Rijkl Rijkl , A2 = Rijkl Rijkl

4 4 ij ij B1 = Rijmn Rmnrs Rrs , B2 = Rijmn Rmnrs Rrs 3 3 Calculating these invariants for the classication of Riemann tensor due to Sharma and Husain and Petrov: If Rabcd = 0 and A1 = A2 = B1 = B2 = 0, then the gravitational radiation is present; otherwise there is no gravitational radiation. Check for validity: (i) Takenos plane wave solution ds2 = Adx2 2Ddxdy Bdy 2 dz 2 + dt2 (ii) Einstein-Rosen metric ds2 = e22 (dt2 dr2) r2 e2 d2 e2 dz 2 where and are functions of r and t only, = 0 and = (r t).
11

(iii) The Peres metric ds2 = dx2 dx2 dx2 2f (dx4 + dx3 )2 + dx2 1 2 3 4 (iv) The Schwarzchild exterior solution 2m 1 2 2m 2 ) dr r2 d2 r2 sin2d2 + (1 )dt r r For the metrics (i)-(iii), all the four invariants of the Riemann tensor vanish and thus correspond to the state of gravitational radiation. For the Schwarzchild exterior solution A1 = 0, B1 = 0, A2 = 0, B2 = 0; and Schwarzchild solution, being a Petrov type D solution, is known to be non-radiative. ds2 = (1

3. Electric and magnetic space-times

The correspondence between electromagnetism and gravitation are very rich and detailed. Some of these correspondence are still uncovered while some of them are further developed. A physical eld is always produced by a source charge. Manifestation of elds when charges are at rest is called electric and magnetic when they are in motion. This general feature is exemplied by the Maxwells theory of electromagnetism from which the terms of electric and magnetic are derived. This decomposition can be adapted in general relativity and the Weyl tensor can be decomposed into electric and magnetic parts.
12

 An observer moving with time-like 4-velocity ua in a relativistic electromagnetic eld measures the electric and magnetic components of the eld, respectively, by
Ea = Fab ub , Ha = Fab ub

Here Fab = Fba is the Maxwell stress tensor with dual Fab = 1 abcd F cd , and the 4-velocity of the observer is 2 normalized so that uc uc = 1 An observer with time like 4-velocity vector u is said to measure the electric and magnetic components, Eac and Hac respectively, of the Weyl tensor Cabcd by Eac = Cabcd ub ud , Hac = where the dual is dened to be Also Eab + iHab = Qab = Cabcd ub ud where Qab = Qba , Qa = Qab ub = 0, a Cabcd = Cabcd + i Cabcd and Eab = Eba , Eab ub = 0, Eab g ab = 0 Hab = Hba , Hab ub = 0, Hab g ab = 0 The Weyl tensor is said to be purely electric if Hac = 0 and purely magnetic if Eac = 0. The Weyl tensor in terms of E and H can be decomposed as
13

Cabcd ub ud
1 2 abef

Cabcd =

C ef . cd

cd Cab = 2u[a Eb] [c ud] + [a Eb] abef ue H f [c ud] cdef ue Hf [a ub]

[c

d]

which equivalently can be written as C abcd = ( acef bdpq g acef g bdpq ) ue up Ef q Now assume Weyl tensor to be of Petrov type I. A coordinate frame can be chosen in which the components of u are (1,0,0,0). For this observer, a frame rotation can be made such that components of Q are observed by that observer to be

+( acef g bdpq g acef bdpq ) ue up Hf q

Qa = b where

0 0 0 0

0 1 0 0

0 0 2 0

0 0 0 3

3 = (1 + 2 ) The Weyl tensor is purely electric if and only if Q is real (i.e., 1 , 2 , 3 are real) and the Weyl tensor is purely magnetic if and only if Q is imaginary (i.e., 1 , 2 , 3 are imaginary). A null tetrad can be chosen such that the NewmanPenrose components ABCD of the Weyl tensor in that tetrad are
14

1 1 0 = 4 = (2 1), 1 = 3 = 0, 2 = 3 2 2 Clearly, if in this frame the Newman-Penrose components ABCD are real (imaginary) the Weyl tensor is purely electric (magnetic). The two independent invariants I and J of Weyl tensor in terms of Q are I = Qa Qb , J = Qa Qb Qc , b a b c a
CD AB CD EF AB (or, I = AB CD , J = AB CD EF )

which can be written as 1 a b a b a b I = (Eb Ea Hb Ha )+2i Eb Ha = (Cabcd C abcd +iCabcd C abcd ) 8 and
a b c a b c a b c a b c J = (Eb Ec Ea 3Eb Hc Ha ) i(Hb Hc Ha 3Eb Ec Ha )

1 ab ab (Cabcd C cdef Cef + i Cabcd C cdef Cef ) 16

From above eqns. 1 1 I = (2 + 2 + 2), J = (3 + 3 + 3 ) 2 3 2 3 2 1 6 1

15

 In terms of the components of the Weyl tensor

2 I = 0 4 41 3 + 32 J =

4 3 2 3 2 1 2 1 0

Types I and D electric and magnetic Weyl tensors have been considered by Mc Intosh et al. For the remaining Petrov types we proceed as follows: For Petrov type II, the Newman-Penrose components of Weyl tensors are 0 = 1 = 3 = 0, 2 = , 4 = 2 2 where 1 = 2 = , 3 = , 2

which, from above eqns. take the form 3 2 3 I= , J= 4 8 and also

2 3 I = 32 , J = 2

I 3 = 27J 2

16

 We thus have Theorem 1. For type II, the Weyl tensor is purely electric (magnetic) if and only if 2 (or ) is real (imaginary). Further, for type N , we have 1 = 2 = 3 = 0 = and 0 = 1 = 2 = 3 = 0, 4 = 2 which gives I = J = 0. While, for type III, we have 1 = 2 = 3 = 0 and 0 = 1 = 2 = 4 = 0, 3 = i which gives I = J = 0. We thus have Theorem 2. Types III and N Weyl tensors are neither purely electric nor purely magnetic. Examples: (a) Plane Fronted Gravitational Waves. The metric is ds2 = 2H(u, x, y) du2 + 2 du dr dx2 dy 2 . This is a space of Petrov type N . (b) The Schwarzchild solution. The Schwarzchild solution in null coordinates is 2m ds2 = (1 ) du2 + 2 du dr r2 (d2 + sin2 d2 ) r
17

(c) The Reisner-Nordstrom Solution. This is just the Schwarzchild space-time with a static non-singular electromagnetic eld and is given by m 2e2 r2 2 ds = 2(1 + 2 ) du + 2 du dr 2 d d. r r 2p
2

(d) The Robinson-Trautman space-time. This is a family of metrics which generalize the Schwarzchild solution to Petrov types II, III and N and is given by ds2 = (2U 0 + 4 0 r + 2
0 2 r2 ) du2 + 2 du dr 2 d d. r 2p

Plane fronted gravitational waves and Robinson-Trautman types III and N metrics: neither purely electric nor purely magnetic. Radiative space-times : neither purely electric nor purely magnetic. The Robinson-Trautman type II metric, the Schwarzchild and the Reissner-Nordstrom solutions are purely electric.

18

4. Lanczos spin tensor

ij Maxwell eqns (Tensor): Fij;k + Fjk;i + Fki;j = 0, F;j = 0

Generated through a potential:

Fij = Ai;j Aj;i

Is it possible to generate the gravitatinal eld through a potental? YES. Through the covariant dierentiation of a tensor eld Lijk . This tensor eld is now known as Lanczos potential or Lanczos spin tensor and satises the following symmetries: (40 conditions ) Lijk = Ljik (4 conditions ) Litt = 0 (or, g kl Lkil = 0) (4 conditions ) Lijk + Ljki + Lkij = 0 (or, Litt = 0)
k Lij;k = 0 or (Lk + Lijjk ); k = 0 ij

The Weyl tensor Chijk is generated by Lijk through the equation (Weyl-Lanczos equation) Chijk = Lhij;k Lhik;j + Ljkh;i Ljki;h + L(hk) gij + L(ij) ghk L(hj) gik L(ik) ghj + where
19

2 pq L (ghj gik ghk gij ) 3 p;q

Lij = Li kj;k Li kk;j From above symmetric relations, the Weyl-Lanczos relations can also be expressed as Chijk = Lhij;k Lhik;j + Ljkh;i Ljki;h 1 1 + (Li pj;p + Lj pi;p ) ghk + (Lhp k;p + Lkp h;p ) gij 2 2 1 1 (Lhp j;p + Lj ph;p) gik (Li pk;p + Lkp i;p ) ghj 2 2 Although the existence of a tensor Lijk as a potential to the Weyl tensor Cabcd was established by C. Lanczos in 1962, there was a little development in the subject for quite some time. Zund (1975): spinor calculus Bampi and Caviglia (1983): proved the existence of Lanczos potential to a larger class of 4-tensors and to a larger class of 3-tensors Illge (1988): spinor formalism, proved the existence of Lanczos potential in four dimension and obtained the wave equation for the Lanczos potential both in spinor and tensor forms Dolan and Kim (1994): wave equations for the Lanczos potential and gave a correct tensor version of the
20

wave equation obtained by Illge Edgar and Hoglund (1997): identities Novello and Velloso (1988): An algorithm for calculating the Lanczos potential for perfect uid space times, under certain conditions Lopez-Bonilla, Ares deParge and coworkers (1989, 93, 97,98,2000): using the Newmann-Penrose formalism, obtained the Lanczos potential for various algebraically special space times Bergqvist (1997): Lanczos potential in Kerr geometry Dolan and Muratori (1998): relationship between the Lanczos potential for vacuum space times and the Ernst potential Edgar and Hoglund (2002): Lanczos potential for Weyl tensor does not exist for all spaces of dim n 7 Cartin (2003): Lpt act as a tensor potential for the linear spin theory Ahsan, Barrera-Figueroa and Lopez-Bonilla (2006): Using the methods of local and isometric embedding, Lovelocks theorem and method of wave equation, obtained a potential for Gdel cosmological model o
21

 The physical meaning of Lanczos tensor is not yet very clear, but the quest for studying this tensor is ON with results of elegance-and the list of workers in this particular eld of interest is very long, we have mentioned here only a few of them. For a given geometry, the construction of Lijk is equivalent to solving Wely-Lanczos equation along with the symmetry eqns of Lanczos tensor; and as seen from the above discussion that there are several ways of solving this equation although none of them are as straightforward as one would like them to be. Using the method of general observers, we have given yet another method for nding the Lanczos potential and found the Lanczos potential for the perfect uid space times in terms of the spin coecients. For a gravitational eld with perfect uid source, the basic covariant variables are: the uid scalars (expan sion), (energy density), p (pressure); the uid spatial vectors ui (4-acceleration), wi (vorticity); the spatial trace-free symmetric tensors ij (uid shear), the electric (Eij ) and the magnetic (Hij ) parts of the Weyl tensor; and the projection tensor hij which projects orthogonal to the uid 4-velocity vector ui . These quantities, for a unit time like vector eld ui such that ui ui = 0 (physically, the time like vector eld ui is often taken to be the 4-velocity of the uid), are dened
22

as follows: (i) The projection tensor hij : hij = gij ui uj , hij = hji , hi k hkj = hij , hi = 3, hij uj = 0 i (ii) The expansion scalar : = ui ;i (iii) The acceleration vector ai : ai = ui = ui;j uj , ai ui = 0 (iv) The symmetric shear tensor ij : ij = hi k hj l u(k;l) 1 hij , ij uj = 0, i i = 0 3
.

(v) The anti-symmetric vorticity or rotation tensor ij : ij = hi k hj l u[k;l] , ij ui = ij uj = 0 and is equivalent to a vorticity vector 1 i = ijkl jk ul so that ij = ijkl k ul 2 where ijkl is completely anti-symmetric Levi-Civita tensor. It may be noted that ai , ij and ij are space like. (vi) The electric and magnetic parts of the Weyl tensor as measured by an observer with a time like 4-velocity vector ui are: 1 Eik = Cijkl uj ul , Hik = Cijkl uj ul = ij mn Cmnkl uj ul 2
23

Also Eik = Eki , Eik uk = 0, Eik g ik = E t t = 0 Hik = Hki , Hik uk = 0, Hik g ik = H t t = 0 The Weyl tensor is said to be purely electric if Hik = 0 and purely magnetic if Eik = 0 and in terms of Eik and Hik , the Weyl tensor can be decomposed as Chijk = 2 uh uj Eik + 2 ui uk Ehj 2 uh uk Eij 2 ui uj Ehk +hi pq up uk Hqj hi pq up uj Hqk +jk +ghk Eij + gij Ehk ghj Eik gik Ehj
pq

ui up Hhq jk

pq

uh up Hiq

(vii) The covariant derivative of ui may be decomposed into its irreducible parts ui;j = ij + 1 hij + ij + ai uj 3

where hij , , ai , ij and ij are, respectively, dene through (i)-(v). (viii) The energy density and the pressure p are given by the energy momentum tensor Tij of the perfect uid Tij = ui uj p hij The relativistic equations of the conservation of energy and momentum are T ij = 0 ;j
24

 Translated the above kinematical quantities and the equations satised by them into the language of spincoecient formalism due to Newman and Penrose and in the process have obtained the Lanczos potential for perfect uid space-times. In fact we have proved the followings Theorem 3. If in a given space-time there is a eld of observers ui that is shear-free, irrotational and expanson-free, then the Lanczos potential is given by Lijk = {m[i where
uj ]

uk

1 m[i 3

gj ]k }

{m[i

uj ]

uk

1 m[i 3

gj ]k }

1 ui = (li + nj ) 2

The Lanczos scalars Li (i = 0, 1, ....., 7) in this case are found to be as follows: 1 1 1 L = , L2 = L , L5 = L , L7 = L 2 3 3 L1 = L3 = L4 = L6 = 0

25

Theorem 4. If in a given space-time there is a eld of observers ui which is geodetic, shear-free, expansion-free and the vorticity vector is covariantly constant (i.e., ai = = ij = 0, i;j = 0) then the Lanczos potential is given by 2 {2(mi mj mimj )uk Lijk = 9 +(mi mk mi mk )uj (mj mk mj mk )ui } where 1 ui = (li + ni ) 2

The Lanczos scalars Li (i = 0, 1, ....., 7) in this case are found to be as follows: 1 L1 = L6 = 9 L = L2 = L3 = L4 = L5 = L7 = 0 Remarks: 1. There is some structural link between the spin coefcients and the Lanczos scalars 2. The Gdel solution is characterized by o ai = = ij = 0, i;j = 0 1 1 ij ij = = constant 2 a 2 The Gdel solution is not a realistic model of the Unio verse but it does possess a number of interesting properties. The matter in this universe does not expand but =
26

rotate. The solution also contains time-like lines, i.e., an observer can inuence his past. It may be noted here that the hypothesis of Theorem 4 are infact the conditions of the Gdel solution and thus obtained a o potential for the Gdel solution. Also it is shown that o Gdel solution is of Petrov type D. o The two parameter family of solutions which describe the space-time around black holes is the Kerr family discovered by Roy Patrick Kerr in July 1963. The two parameters are the mass and angular momentum of the black hole. Kerr solution is just the Schwarzchild exterior solution with angular momentum. Using GHP formalism (a tetrad-formalism), we have obtained the Lanczos potential for Kerr space-time as 2 1 2 1 ) 3 , L5 = A( ) 3 M M which shows that Lanczos potential of Kerr space-time is related to the mass parameter of the Kerr black hole and the Coulomb component of the gravitational eld. L1 = ( A comparison between electromagnetism and gravitation is given in the Table on the next page:

27

Fields Potential Field relations Gauge invariance

Fij Ai Fij = Ai;j Aj;i Ai = Ai + ,i Wij (A ) = Wij (A) Ai = 0 ;i F ij = J i ;j Ai + Ri k Ak = J i

Chijk Lijk Chijk = W (L)hijk Lijk = Lijk + ijk Whijk (L ) = Whijk (L)
t Lij;t = 0 t Cijk;t = Jijk

Gauge conditions Field equations Potential wave equation in matter

Lijk + 2Rkt Lijt Ri t Ljkt Rj t Lkit gik Rpt Lpjt + gjk Rpt Lpit 1 R Lijk = Jijk 2 Lijk = 0

Potential wave equation in vacuo Field wave equation in matter

Ai + Ri k Ak = 0 Fij + Rt i Ftj Rt j Fti 2Rrisj F rs Fi t;t;j +Fjt ;t;i = 0

Rhijk + 4Rhpq[j Rk] q ip Rhipq Rpq jk q +2R [k Rj]qhi +2Rj[i;h];k + 2Rk[h;i];j = 0 Chijk + 4Chpq[j Ck] q ip Chipq Cpqjk = 0

Field wave equation in vacuo

Rt j

Fij + Rt i Ftj Fti 2Rrisj F rs = 0

28

5. Space-matter tensor
Through Sections 1-4, the importance of Riemann tensor is seen. Some other fourth rank tensors which involve Riemann tensor. One such tensor, known as space-matter tensor, is studied. A decomposition of this tensor is given in terms of Riemann tensor and an attempt has been made to express the space-matter in terms of electromagnetic eld tensor. A symmetry of the space-time is dened in terms of the space-matter tensor and studied. Petrov (1969) introduced a fourth rank tensor which satises all the algebraic properties of the Riemann curvature tensor and is more general than the Weyl conformal curvature tensor. This tensor is introduced as follows: Let the Einsteins eld equations be 1 Rab R gab = Tab 2 where is a constant and Tab is the energy-momentum tensor. Introduce a fourth order tensor Aabcd = (gac Tbd + gbd Tac gad Tbc gbc Tad ) 2
29

From the denition this tensor has properties: Aabcd = Abacd = Aabdc = Acdab , Aabcd + Aacbd + Aadbc = 0 which leads to Aac = Tac + R T gac = Tac gac 2 2

Dene a new fourth order tensor Pabcd = Rabcd Aabcd + (gac gbd gad gbc ) This tensor is known as space-matter tensor. The rst part of this tensor represents the curvature of the space and the second part represents the distribution and motion of the matter. This tensor has the following properties: (i) Pabcd = Pbacd = Pabdc = Pcdab , Pabcd + Pacdb + Padbc = 0 (ii) Pac = Rac Tac +
R 2

gac + 3gac = (R + 3)gac

(iii) If the distribution and the motion of the matter, i.e., Tab and the space-matter tensor, Pabcd are given, then Rabcd , the curvature of the space is determined to within the scalar . (iv) If Tab = 0 and = 0, then Pabcd is the curvature of the empty space-time.
30

(v) If gab, the metric tensor, , the scalar and Pabcd are known, then Tab can be determined uniquely. In section 2, we have seen that the Riemann curvature tensor can be decomposed as Rabcd = Cabcd + Eabcd + Gabcd where Cabcd is the Weyl tensor, Eabcd is the Einstein curvature tensor, dened by 1 Eabcd = (gac Sbd + gbd Sac gad Sbc gbc Sad ) 2 1 Sab = Rab gab R 4 being the trace-less Ricci tensor, and Gabcd is dened by Gabcd = These eqns lead to 1 Rabcd = Cabcd + (gad Rbc + gbc Rad gac Rbd gbd Rac ) 2 R (gad gbc gac gbd ) 6 Above eqns. also lead to 1 Aabcd = (gac Rbd + gbd Rac gad Rbc gbc Rad ) 2 R (gac gbd gad gbc ) 2
31

where

R (gac gbd gad gbc ) 12

 Thus space-matter tensor may be decomposed as Pabcd = Cabcd + (gad Rbc + gbc Rad gac Rbd gbd Rac ) 2 +( R + )(gac gbd gad gbc ) 3 which can also be expressed as 2 h h h h h h h h Pbcd = Cbcd +(d Rbc c Rbd +gbc Rd +gbd Rc )+( R+)(c gbd d gbc ) 3 A classication of space-matter tensor, using matrix method, has been given by Ahsan and the dierent cases that have been arrived at are compared with the Petrov classication. It is found that case III(a) correspond to Petrov type III gravitational eld. The algebraic properties and the spinor equivalent of the space-matter tensor have been obtained by Ahsan. Moreover, they have found the covariant form of the invariants of the spacematter tensor and presented a criterion for the existence of gravitational radiation, in terms of the invariants of the space-matter tensor. In general theory of relativity the curvature tensor describing the gravitational eld consists of two parts viz, the matter part and the free gravitational part. The interaction between these two parts is described through Bianchi identities.

32

 For a given distribution of matter, the construction of gravitational potentials satisfying Einsteins eld equations is the principal aim of all investigations in gravitation physics, and this has been often achieved by imposing symmetries on the geometry compatible with the dynamics of the chosen distribution of matter. The geometrical symmetries of the space-time are expressible through the vanishing of the Lie derivative of certain tensors with respect to a vector. This vector may be time-like, space-like or null. Motivated by the role of symmetries in general relativity, we have dened a symmetry in terms of the vanishing of the Lie derivative of the space-matter tensor and termed it as Matter Collineation Denition 1. A matter collineation is dened to be a point transformation xi xi + i dt leaving the form of h the space-matter tensor Pbcd invariant, that is
h L Pbcd = 0

where L denotes the Lie derivatives along the vector . Since every motion in a Vn is a Weyl conformal collineation, we thus have Theorem 5. A Vn admits matter collineation if it admits motion, Ricci collineation and = 0.

33

 Now consider a Vn for which Rij = 0 = and denote this space as Vn0 , then Theorem 6. In a Vn0 every motion is a matter collineation. The energy momentum tensor Tab of an electromagnetic eld is dened by 1 Tab = Fak Fbk gab Fij F ij 4 Representation of space-matter tensor in case of a non-null electromagnetic eld
k x Pabcd = Cabcd gac Fbk Fd gbd Fap Fcp + gad Fbt Fct + gbc Fax Fd

1 +( + Fij F ij )(gac gbd gad gbc ) 2 from which we have

h h h k h h h x Pbcd = Cbcd c Fbk Fd gbd Fp Fcp + d Fbt Fct + gbc Fx Fd

1 h h +( + Fij F ij )(c gbd d gbc ) 2 Since for non-null electromagnetic elds, that L gij = 0 L Fij = 0. Thus taking the Lie derivative of the above eqn. Theorem 7. A non-null electromagnetic eld admits matter collineation if and only if it admits motion and = 0.
34

 The energy-momentum tensor for a null electromagnetic eld is given by Tab = Fac Fbc where Fac = sa tc ta sc and sa sa = sa ta = 0, ta ta = 1, vectors s and t are the propagation and polarization vectors, respectively. The representation of space-matter tensor for a null electromagnetic eld
h h h h t h p h Pbcd = Cbcd + 2(d Fbk Fck c Fbt Fd + gbc Fp Fd gbd Ff Fcf )

Theorem 8. A null electromagnetic eld admits matter collineation along the vector (propagation/polarization) if is Killing and expansion-free. As we are working with the null electromagnetic eld, it is therefore natural to expect that the Lichnerowicz condition for total radiation are satised and we have Tab = 2ka kb where ka is the tangent vector. We now have Denition 2. A null electromagnetic eld admits a total radiation collineation if L Tab = 0, where Tab is dened as above.

35

 Since a null electromagnetic eld with Tab = 2 ka kb is known to be a pure radiation eld, taking the Lie derivative of this Tab , we have Theorem 9. Pure radiation elds admit total radiation collineation if and only if the tangent vector ka denes a null geodesic.

36