Vol.37 (1996) REPORTS ON ~ATH~~ATICAI, PHYsIfS No.

WSTORS AND NEARLY AUTOP~LEL MAPS

SERGIU I. VACARU
Institute of Applied Physics, Moldovan Academy of Sciences, 5 Academiei str., Chiginau, 2028
Republic of Moldova (e-mail: lises@cc.acad.md)

and
SERGIU V. OSTAF

Department of Mathematics and Physics, Tiraspol State University, 5 Iablochkin str., Chi$&u, 2062
Republic of Moldova

(Received December 6, 1994)

The purpose of the present paper is to investigate the problem of definition of twist-
ors on generic curved spaces. Firstly, we consider nearly geodesic (autoparallel) maps of
(pseudo)-Rieman~an spaces. Secondly, we shall define nearly autoparallel hvistor equations
which are compatible on nearly ~nfo~aily Aat spaces. Finally, we shall study nearly auto-
parailel twistor structures generating curved spaces and vacuum Einstein spaces.

1. Introduction: Spinors and maps of curved spaces with deformation of connections

Our geometrical constructions will be realized on pairs of 4-dimensional (pseudo)-
Riemannian spaces (V, IL) with the signature (- + + +) and i-l local maps of
spaces f: V + g given by functions f’(x) of class CT(U) (r > 2, T = w for analytic
functions) and the inverse functions f?“) with co~esponding non-zero Jacobians at
every point x = {x”} c U c V and : = {@} c U c IL (U and g are open regions).
We shall attribute the regions U and u to a common, for a given f-map, coordinate
system, when every point q E U with coordinates d‘(q) is mapped onto a point q c U
with the same coordinates xp = c&‘(q) = x”(q), i.e. f: d‘(q) -+ x”(q). We note that
all calculations in this work will be ha1 and will refer to common coordinates, for
given f-maps, on open regions of spaces into considerations.
The metric tensor, the connection, differential operator and the tetrads (frames)
are denoted on U as g~~(x),~~~(x), I), and h;(x), respectively, where

s/N(x) = qxY$M?as, v,b = const, (1)

and on u as gPy(x)=gPy(x),
-- &~(x)+~(x),QP = D, and &t(x) = h;(x), where

(2)
(the Greek indices ~1,v, -p,y, . . . range from 0 to 3). Throughout the present paper
310 S. I. VACARU and S. V. OSTAF

we shall use the terminology and definitions of the abstract and coordinate tensor
and spinor index formalisms described in the monographs [l, 21. For simplicity, we
shall write the Latin symbols a, b,a,b, . . . for both the abstract and the tetrad indices
implying that in the first case Latin indices are abstract labels and in the second
case (for decompositions on tetrads) we shall specify their explicit values (a, b, . . .
= 0, 1,2,3). We consider spinor decompositions of metrics (1) and (2):

= cpAA’(Z)~,BB’(Z)E~~t~,~,,
S&AU (3)
where O,““‘(X) = h;(~)~,A~‘,cr~~’ = cons& are the Infeld-van der Waerden coeffi-
cients and EAB = -CBA, EA~B/= -EB/A~(A, B, A’, B’ = 0,l) are spinor metrics

9 =fl_;A’(x)G?fB’(x)~AB~AtB5 (4)
-P
where azA’(z) = O,““‘(X) = $(z)a,AA’; if necessary, we shall write, for example,
A,A’ A-
EABYEA~B/~& ,W or 5:: in order to point out that these spin-tensor values are
associated to the spinor decomposition (4) on the space E.
For mutual transformations of the tensor and spinor indices one introduces the
inverse Infeld-van der Waerden coefficients $J,,(z) and azA,(z), for example, B”
= GA, BAA’ and &B, = AE~EB,.
Covariant derivation of spinors on V is defined in terms of the spin coefficients
Y&B and r~$?~, :

DAA’~ B - azA,(~)D,eB = ~AAJ<~ + y2zCEB,

and DAA~<B~= u~~,(x)D~<B~ = ~AA~[BJ - ~;4.,$i,<p.

In a similar manner the spin coefficients r,,“,,, and r,,“:,, are considered in
order to define the covariant derivation of spinors on v.
Below we present formulae interrelating spin coefficients r;,” and y;$ with Chri-

stoffel symbols in the case of orthonormalized bases when CA/B’ = ~,JB =

and, inversely, for (T:,’ = const,

p.AA’ (&&&, = _BB,Cgg:
yA
BB’CC, = ~i;.&q$’ + ~&&, E;CA + r,&c;CA),

where ~2 = 6:, 6;: = 6: and T&$~, denotes the complex conjugation of Y;~,~,.
Our first objective in this paper is to study deformations of spinor objects as
a result of the superposition of local l-l maps f: V+y with the deformation of
connection
r,;(x) = T$(Z) + P$(x), (6)
 TWISTORS AND NEARLY AUTOPARALLEL MAPS 311

and, as a consequence of formulae (5), deformations of the spin coefficients, for

example,
r,“,(x) = Y,S) + *Y;&)J (7)

where P$(s) and *r,“, = P..?’

PBC - ~(&B,~pc$‘“’ - nk,,~?,oz~‘) are called the
deformation tensor and the deformation spin tensor, respectively. Deformations of
the covariant derivation operator, caused by splittings of type (6) or (7) will be
denoted as
& = D, + *Dll, DA.& = DAA’ + *DAA’. (8)
In a particular case of conformal maps c: U + U, when

--ab = Q29atl,
9 CAB = acAB, tA’B’ -- f%A’B’,

L?(cc) is a nonzero real function on U,

*D, = fT’D,fi’ = D&R. (9)
Conformal transforms are largely used, for example, in the twistor [2] and conformal
field theories.
We note that there are classes of l-l local maps with the deformation of con-
nection ((6) or (7)) more general than that for the conformal maps (9) (see, for
example, (r~- 2)-projective spaces [3], nearly geodesic maps (ng-maps [4]) and nearly
autoparallel maps (na-maps) of spaces with torsion and nonmetricity [5, 61, of fibre
bundles [7] and of Finsler and Lagrange spaces [S]). In our papers [5, 9-111 we have
proposed to apply the ng- and na-maps for the definition of conservation laws on the
curved spaces. Na-maps were used for definition of the nearly autoparallel twistors
in connection to a possible twistor-gauge interpretation of vacuum gravitational fields
[7, 1l-141.
The second objective is the investigation of na-map deformations of the twistor
equations [2] (for our purposes written on space v)

DcAuB) = f(&,gB
--A’-- + &g”) = 0, (10)
where ( ) denotes symmetrization.
Because for the uncharged twistors we have

D A’(C&Jr) = _,+r,C$Q~,
(11)
where
C =@
-abcd--ABCD~AJBfkDt + EA~Lv~~LI~&~BSCLI (12)

is the conformal Weyl tensor on space v, there is a hard compatibility condition

for twistor equations (IO), namely, g_ABcDg D = 0, which characterizes, for example,
conformally flat spaces. That is why a mathematically rigorous, and generally accepted
definition of twistors was possible only for conformally flat spaces and this fact is
the main impediment to the twistor interpretation of general gravitational fields (for
details see [2]).
312 S. I. VACARUand S.V. OSTAF

Our main idea [ll, 7, 13, 141 was to define twistors not on the generic curved
spaces V, where twistor equations are incompatible, but to remove the problem on
auxiliary conformally flat (or more simply, flat M) background spaces V, interrelated
with the fundamental space-time V by means of chains of na-maps (nearly confor-
ma1 maps, nc-maps). On the space M, twistor equations (10) become compatible; we
can define twisters in a standard manner as pairs of spinors, z* = (gA,7_rA). Then,
transferring z* on V, by using nc-maps, we can define nearly autoparallel twistors,
na-twistors, as satisfying the na-twistor equations, being the na-images of equations
(10). For simplicity, in this paper we shall restrict ourselves only to the nearly geo-
desically flat (ng-flat) spaces V which admit ng-maps to the Minkowski space M.
We shall analyse conditions when na-twistor equations contain information on the
vacuum Einstein fields.

2. Nearly geodesic maps and spinors

The aim of this section is to present a brief introduction to the geometry of
ng-flat spaces. We shall specify basic ng-map equations and invariant conditions [4]
to the case of the vacuum gravitational fields on V. Proofs are mechanical, but, in
most cases, calculations are rather tedious, and similar to those presented in [4, 131.
They are omitted.

2.1. Definition of ng-maps

Let us parametrize curves on U c V by functions ZY = z?(n), nr < 71< 112,with
the corresponding tangent vector field defined as up = F.
DEFINITION 1. A curve 1 is called a geodesic on V if its tangent vector field
satisfies the autoparallel (a-parallel) equations
UD@ = u%pu” = p(q)@, (13)
where ~(77) is a scalar function.
We note that for (pseudo)-Riemannian spaces the extremal curves, the geodesics,
coincide with the straightest curves, a-parallels, and that is why we shall use the term
geodesics for both classes of curves (for spaces with locally isotropic or anisotropic
torsion and nonmetricity we have started with a-parallel equations [5-81).
On the space v we consider a new class of curves: Let a curve 1 c v be given
parametrically as zQ = zY(rj),nr < 9 < QZ,U* = s#O. We say that a 2-dimensional
distribution E$) is coplanar along 4 if in every point 1c E 1 there is defined a
%-dimensional vector space I&(z) c T,y (T,V is the tangent space at x E V) and
every vector p%$$ c E2(I), ~(“0, E I, is contained in the same distribution after
parallel transports along I, i.e. p*(zp(v)) c: IQ(l).
DEFINITION 2. A curve [ is called a nearly geodesic on space E if along i there is
defined a coplanar distribution I??&) containing the vector field ~~(7) tangent to 1.
Ng-maps are introduced [4] according to the
 TWISTORS AND NEARLY AUTOPARALLEL MAPS 313

DEFINITION 3. Nearly geodesic maps (ng-maps) are focal l-l mappings of (pseudo)-
Riemannian spaces, ng: V-v, changing every geodesic 1 on V into a nearly geodesic
1 on y.
Let a geodesic 1 c U be given by functions xa = F(n), ua = T, ~1 < 7 < 772,
satisfying equations (12). We suppose that to the geodesic 1 there corresponds a
nearly geodesic i c u given for a chosen ng-map by the same parametrization in a
common local coordinate system on U and u. This requirement is satisfied if and
only if vectors zF, g;, = uQF and U+ = U&J;) are linearly depended in every
point 2 E U, i.e.,
UC, = a(77F + b(n)u&
for some scalar functions ~~(17)and b(n). Putting splitting (6) into expressions for 2;)
and I& on U, and from the just presented linear dependence we obtain

u~uyu6(DpP~~ + P&?P$) = bUyU6P$ + aua > (14)

where b(q, U) = 6 - 3p and
a(% u> = C?+ bp - ubabp - p2 (15)

are called the deformation parameters of the ng-maps.

2.2. Classification of ng-maps

Ng-maps were classified [4] by considering the possible polynomial dependencies
on 2~~ of deformation parameters (15). We shall consider the maps ng : V --f r satis-
fying the reciprocity conditions (ng-‘: V --f V is also an “g-map). This requirement
is fulfilled if
P&P;; = dCqP+; + c&$),

for a vector de, and a tensor cap on V.

THEOREM 1. Four classes of the ng-maps are characterized by the corresponding
paramettiations of the deformation tensors and basic equations:
-for the trivial ng-maps, the geodesic maps (or 7r(o)-maps)

P$W = ~(P~,*,~ (16)

where 6p” is Kronecker symbol and +p = $0(x) is a covariant vector field;
-for the 7r(l)-maps, Pi;(x) is the solution of the equations

3D,P$ = 2Riz,r,a - 214i,ra, + 6bcc,P&j + 6a(,&); (17)

-for the n(2)-maps,
Pi; = 24(&, + 2cr(,F& (18)

where F; = F;(x) is such that Ft = F; = e@ (e = fl):

8,pF,iFx’ - d,xF,jF; = 0 (19)

314 S. I. VACARU and S. V. OSlXF

([ ] denotes antisymmetrization) and solves the equations

D,jF,’ + pLyF$Y; - pQF; = 0 (20)

for a covariant vector field pLr = &x);
-for the n(3)-maps,

P$- = 2&S,; + Oy6(PT, (21)

where the contravariant vector field yT = q’(x) satisjies the equations

D,cpp = uS,p+ pL,cpB, (22)

for some scalar field u = u(x), covariant vector fields pLr = p?(x) and (p,, = (pv(x),
and a symmetric tensor field gap = a,p(x).

We emphasize that for (pti = 2 = gpvpy and a,o(x) = gap(x) we obtain a

particular case of conformal maps, TV: gpLv = &'gpv (the so-called concircular
maps [15]).

2.3. Invariant criterions for ng-flat spaces

DEFINITION 4. A (pseudo)-Riemannian space V is ng-flat if it admits a map
ng: V-a.
We shall consider four classes of na-flat spaces denoted respectively as nCij-flat
spaces, where (i) = ((0), (l), (2), (3)).
It is significant that the na-maps are characterized by the corresponding invariant
conditions for values similar to the Thomas parameters and the Weyl tensor (the
invariants of conformal maps [16]). Below we present the criterions for a space V
to be ng-flat.

PROPOSITION 1. For the ng-flat spaces there are satisfied the following conditions:
-for the 7r(())-spaces,

W$, = R.”
u.y6 - ;R ,,&, = 0; (23)

-for the 7rCl)-spaces,

3D-,Pi,$ = 2RiLp), + 6b(aP,2y, + ~u(,,S,‘,; (24)
-for the T(~)-spaces,

where

(26)
R,, = - R,, + eF,“(F,pR&, + F,pR&, - 2D,D16F,q -

- 2eFiDc,F:D,FF + eFiDc,FgDrF,“); (27)

 TWISTORS AND NEARLY AUTOPARALLEL MAPS 315

-for the r(+puces,

Rprvys= Atgu,m - gamy) + 4L7ptgarm - gawr) - vGhw45- mw~)l~ (28)

where

A=-1[~--2(~~~‘)1; A+B+($+ru2) =O, e=&l, (29)

for scme gradient vector p, = -$f$and scalar v(p) welds.

We note that for the Ricci tensor of the ;?s-flat spaces from (28) and (29) it
follows the expression

2.4. The integrability conditions for q-maps equations

Ail presented in this paper basic equations for the ng-maps (equations (17) (18)
and (ZO)-(22)) are systems of the first-order partial differential equations with alge-
braic constraints of type (19). The integrability conditions for the ng-map equations
have been studied in [4] and, in the language of the Pffaf systems [17-201, in [5,
6, 131. The most important conclusion made in the mentioned works is that we can
always verify, by using algebraic methods, whether a given system of the ng-map
equations on V is, or is not, integrable for maps to the Minkowski space. Let us
illustrate this for maps nl: V -+ &J specified by the equations
3(&P;-; f P;;P<;) = 2Ri;‘a,p)r + 6a(,&) (31)

(these equations can be obtained from (14) by using auxiliary rro-maps, v ZV --i hf,
see [4]). The integrability conditions of (32) can be written as

where
K&E = - ~~~~~.~~~ + LV&), f 6P~~~R~.~~ + 4”&&‘& +
-f- fiP;lS&ap,) - 6~~~~~~a~~~. (33)
From (32) we obtain
3&a& = $f-,(,,), - s.‘,o,,. (34)
Equations (31) and (34) form a Cauchy system of the first-order partial differential
equations (with the coefficients given by such geometrical objects as connection and
curvature on V) for the unknown variables P&T and a,B.
The first set of integrability conditions for the system of equations (31) and (34)
can be found from (32) by excluding the covariant derivations of a,@ according to
(34). As a result we obtain the integrability conditions in the form of linear equations
for the deformation tensor Pii. Introducing the second, third and higher sets of the
316 S. I. VACARU and S. V. OSTM

integrability conditions, we are able, in general, to clarify the question of existence

of solutions of (31) and (34).
We note that in a similar manner we can construct sets of the integrability con-
ditions for the nc2)-, and r(j)-maps, cf. (20) and (22).

2.5. Spinor formalism and the ng-map theory

This question was studied in detail in [13, 141 by means of deformations (by
ng-maps) of spin coefficients (7) in spinor covariant derivation operator (8). Using
the g-coefficients we can transform basic ng-map equations (18), (20) and (22) and
flat projectivity conditions (23), (24) and (28) into a spinor form. We omit these
considerations here. For our purposes it is important that for every deformation of the
spin coefficients *r,:(z) (see splitting (7)) we can define a corresponding deformation
tensor (see expressions (5)-(7))

(35)
Putting (35), for example, into (33) we obtain a system of algebraic equations, if
necessary in spinor variables with a spinor representation of the curvature and de-
formation parameters, which permits us to answer the question whether the given
deformation of the spin coefficients generates, or not, a map rC1): V + M.
Finally, in this section, we note that every curved space V, if corresponding con-
ditions on differentiability of the components of metric, connection and curvature on
V are satisfied, admits a finite chain of ng-maps, i.e. a nc-transform, to the Min-
kowski space M [5-9, 131. So, it is possible a new classification of the curved spaces
in terms of minimal chains of ng-maps characterized by the corresponding sets of
invariant conditions of type (23)-(25) and (28). This ng-map classification of curved
spaces differs from the well-known Petrov algebraic classification [21].

3. Nearly conformal twistors

The purpose of this section is to define twistors on ng-flat spaces.

3.1. Spinor equations for massless fields with spin irn (m = 0, 1,2,. . .) and twistor
equations
Let a spinor $AB...L have m indices and be symmetric:
+A&L = #(A&-L). (36)
The dynamic equations for a massless spin $rn field are written as

DAA’~~~...~ = 0. (37)
The compatibility conditions [22, 23, l] of (37) for uncharged spinor field (36) can
be written as
(m - ~)~AJ~M(cT.,.K*~~~~ = 0, (38)
 TWISTORS AND NEARLY AUTOPARALLEL MAPS 317

where @LABM is the Weyl spinor on space V.

Because conditions (38) are not satisfied on the generic curved spaces, there is an
essential difficulty in definition of physical fields (36) as solutions of equations (37).
The same difficulty appears for twistor equations (lo), rewritten on the space V:

DjqqwB) = 0 > (39)

with the compatibility conditions
DA’(CDA wB) = _@A.BJ’,
A’ ..D

3.2. Systems of first-order partial differential equations

The above-mentioned field and twistor equations ((37) and (10)) are systems of
the first-order partial differential equations. We shall study the general properties of
such systems of equations using methods of the geometrical theory of differential
equations [17-201.
Let us consider, in general form, a system of the first-order partial differential
equations on a space V”, dim V/” = n:

(
fs x1 ).“) xn,yl,..., y$$ )...) g) =o, s = 1,2, . . .) q,

where x1,. . .,x” are independent variables and y’, . . . , yT are unknown functions. In-
traducing new unknown variables (functions)
dy”
P%=@ 6 = 1)“‘) r; & = l)... ;n,

we reduce equations (40) to a Pffaf system

8” = dy” _ #dxb = 0, (41)
where the variables p$ satisfy a set of finite relations:

f&a, y”,&) = 0. (42)

Solving (42) for q independent values p” = {pi} and putting them into (41) we ob-
tain a system of T Pffaf equations on i = T + nr -q unknown functions of independent
variables z? (differentials dz& play the role of distinguished variables).
Let ti be an open region locally isomorphic to Ri+n. We write the new Pffaf
system as
0” = @i(z’, p”)dp” + bf (z”, p’)dz” = 0, (43)
where A = 1,2,. . ., r; iL = 1,2,. . ., T + TV_- q. Equations (42) are linearly independent
if rank I]c$]~ 5 r at every point x& of an open region U c V”. We mention that
the integral varieties IO of the system (43) should be defined by the equations (the
closure of (43)):
318 S. I. VACARU and S. V. OSTAF

0” = 0, DOA = 0, (44)
where the quadratic exterior forms are written as

Dt+ = a;&dp” A dp’ + C&&dp” A dx” + b/i,bdx:a A dx!

If a solution
ya = yQ@“) (45)
of equations (40) (or equivalently (41) and (42)) is found, it must satisfy the inte-
grability conditions
dy” dy”
-=-
dy”ldxP dxfldya ’
or, equivalently,
a,p; = a&p;, (46)
i.e., if equations (40) are compatible, the Pffaf system (41) can be reduced to total
differential relations
dX” = dy” -p%dz* = 0.

In this case solution (45) should be obtained from the relations

P(yb,xb) = c”, C” = const,

where rank I/$$ #O. If conditions (46) are not satisfied, one tries to solve equation
(40) by introducing new unknown functions
X”
jyg = pi - e (47)
Xa

and considering a new Pffaf system

-e” = dy” - pzdz” = 0, (48)
where yS(xir, y&,5:) = 0. To obtain a total differential relation, we multiply (48) by
a nondegenerate matrix function 1_1j(x”, y”):
dXb = +jdyb - Cl&&&+ = 0. (49)
Integrating system (49) we obtain relations
-.
Xb(x”, y”) = C”, Cb = const,
from which the solution y” = y&(x”) of equations (40) can be found in an explicit
form. We note that if the deformation functions <‘” from (46) and the integrating
matrix p$ from (48) exist, the Pffaf system (41) can be expressed as
0” = dy” _ && = d<” + (p-1);&“,
 TWISTORS AND NEARLY AUTOPARALLEL MAPS 319

where (p-‘)b” is inverse to phi,.

b As particular cases we can consider the trivial inte-
grating matrix, ,L$ = Sk, and (or) vanishing deformation when <” = const.
We also emphasize that introducing new deformed variables (47) and a matrix
bf into relations (42) we obtain new finite relations fs(za, @,pE) = 0 which, as a
matter of principle, reflect deformation of the first-order partial dffferential equations
(40) into another ones (actually, for every given system (40), there exists an infinite
number of deformations (46) and integrating matrices pf). Considering equations (40)
as fundamental physical equations of type (37) or (39) one arises the problem of
the definition of a unique deformation, motivated from the physical and geometrical
point of view, transforming mentioned, in general incompatible on the curved spaces,
equations into the compatible ones, on some auxiliary spaces,

3.3. Ng-deformations of twistor equations

As was shown, the twistor equations happen to be incompatible on a given curved
space V. Our aim is to formulate an algorithm of transporting the above-mentioned
differential equations, by using ng-maps, from a space V to another one, y, on which
compatibility conditions (11) will be satisfied. In this case we extend our system of the
twistor equations by introducing into consideration additional variables (deformation
parameters (15) a,p and b,, and the deformation of connection, Pi;), and we
supplement the initial twistor system on V by a system of basic ng-map equations
on E. Let us illustrate this construction for the twistor equations (39) rewritten as a
Pffaf system (41):
dwA - pC’CAdzcc, = 0, (50)
where the unknown functions
A
i3W
C’CA _
P
dXCC~
satisfy finite relations of type (42):
A’(AWu~ = o.
P A’(AB) + y,,D 7 (51)

&f” are spin-coefficients on space V. Twistor equations (39) and their associated
Pffaf system (50) and (51) are incompatible on the generic curved space-time V. We
suggest to extend this system of equations by considering a new system of differential
equations:
DA’(A;B) = n~$A.Wzn
> (52)
or equivalently, a new Pffaf system associated to (52):
-A
dw -p -C’CA&CC, = 0,
(53)

where the unknown functions jjCfCA = * must satisfy the relations

azc,,
A’(A.QD + I1~;A-B)ij~ = o.
ijA’(AB) + y,,n (54)
320 S. I.VACARU and S.V. OSTAF

Spinors II,.,
A’A B from (52) can be considered as those resulting from a deformation
of type (47) and a multiplication by an integrating matrix as in (49). Introducing the
ng-maps we identify A-spinors with the deformation of spin-coefficients *$ (see
relation (7)):
A...B
‘J'AD = dye&). (55)

PROPOSITION 2. Deformed twistor equations (52) (and associated twistor PfJkf sys-
tern (53) and (54)) are compatible if spinors (55) solve one of the ng-map equations
((16), (17), (18)-(20) and (21)-(22)) and satis& one of the corresponding ng-put cri-
terions ((23), (24), (25) and (28)).

Proof: Let us define new spin-coefficients

which according to our proposition- become the trivial (with the vanishing curvature)
spin-coefficients on a flat space M. In this case equations (52) can be written as
fj~'(~p) -.
_ o
(56)

Equations (56) are compatible because on the flat space % the Weyl tensor vanishes
(see relations (11) (12)). The proposition is proved. n

Instead
of the ng-maps, we can consider chains of ng-maps (nc-transforms)
nc: V -+ M. Nc-twistors are defined as solutions of the deformed twistor equations
with A-spinor, being a superposition of the spin tensors

(57)
associated to a finite chain of ng-maps. In a particular case when (57) reduces to
(55), we obtain ng-twistors.

3.4. Ng-images of twisters

On the flat space M twistors are defined as a pair of spinors, z” = (gA, EA),
where
-WA = &‘),A _ izAB’U’)~B,, $,, = (o)~A, = const,
is a general solution of the twistor equations

D A’(AWB)
- = 0.

Nc-twistors on space V, being an nc-coimage of the space M, for a given map

nc : V t iW, are defined as a pair of spinors 2” = (uA, TA,), where wA is a general
solution of the nc-twistor equations
DA’(A~B) _ AA’tA.B)W~
- ..D > (58)
with A-spinors defined by (57). F or a local common spinor coordinate system on
spaces under consideration, we can write tiA = wA and define a second spinor TAl
 TWISTORS AND NEARLY AUTOPARALLEL MAPS 321

as TAG = ~DAA’w~. Taking into account that ‘A, = iDAA,wA, we have

TA, 3) 1 .n...A C
TA’ - 28 AA’& . (59)

In a similar manner we can define the dual nc-twistors on V as the pairs of

spinors W, = (AA, p”), where
AA = (‘)x/i + $~.;i&,~A’ 7 (‘)xA = const,

and p,A’ = (0)pA’ + ixAA’(‘)XA is a general solution of the dual nc-twistor equation
DA(A’~B’) = npl;j’J’$D’.

The spinors AA$~, are defined as a superposition of the ng-transformation like in

equation (57).
We end this section by considering the question of the geometrical interpretation
of nc-twistors. To an isotropic twistor Z” = (gA, TA’) # 0, Z”Z, = 0 (I, = (?&, aA’)
denotes the complex conjugation of z”) one associates [2] an isotropic line on the
space M:
xa = CO),,+ n<a, 711< rl < r/2, (60)

where cO)za = (i (“)~“.(Ill?rg,)-lO)~A~O,~~~, <” = (‘J),A(a)#‘.

_ The nc-coimages of z”
and z” on the space V are defined as 2” = (wA, ?r,&) and, respectively, as z,
= (?TA,gA’), where
WA = -WA = (QWA
- _ izAA’(0),rA,, =A, = (‘&A, - ;i&‘&#, (61)
and ~~ =(a)~~ _ +;i,,$,,;C’,wA’ = y-4’. U sing spinors (61) we can verify that

Z”Z, # 0, (62)
i.e., an nc-twistor Za defined by an isotropic twistor z* is not isotropic. So the
nc-twistors on space V parametrize a class of curves on this space as theirs nc-images
(of type (60)) on the flat space A4, but unlike to the usual isotropic twistors, to
an nc-twistor one must associate a nearly geodesic on V being a corresponding
nc-deformation of an isotropic line in the flat space. Indeed, the nc-image of iso-
tropic line (60) on M is a curve I on V (because of (62)) with the tangent vector
ECZ= KATA’ and a complementary 2-dimensional distribution defined, for example, by
<;“1,= EbDbE” and t(az, = [bDb[[l,, where Db = & + *Db,*Db is a deformation of
connection associated to the map nc: JJ + V.

4. Deformation of spinor and twistor structures and generation of curved spaces

Let on a flat space M with given primitive spin-coefficients ~~i,o, or a connection
F$ ( with vanishing curvature) be defined a global twistor structure as a solution of
twistor equations on M. Our task is to specify conditions under which a deformed
322 S.I.VACARU and S.V. OSTAF

twistor structure, obtained as a solution of deformed twistor equations (58) with

the deformation A-spinor of type (55) will generate an ng-flat (pseudo)-Riemannian
space-time.
Firstly, we fix a spinor Ai,:o. It is still not clear if the new connection $;,
defined as to satisfy the relations

(63)
where
p(.$q = _A,.$.$ _ X-C’ .c
aBtEB (64)

(for simplicity we consider torsionless connections) will generate both the compati-
ble ng-twistor equations (56) and the basic ng-map equations associated to a mutual
transform ng: V + M. We try to answer this question in the following way. Calcu-
lating auxiliary curvature and the Ricci tensor for connection (63) and putting both
these expressions and deformation tensor (64) into relations (32)-(34), we obtain an
algebraic system of equations. If this system is satisfied for some deformation par-
ameters a,p and b, (see formulae (15)) it is clear that we have obtained a 7rclj-flat
space V.
PROPOSITION 3. Deformation spinor A.;i2,o and its corresponding deformation ten-
sor P&T (see (64)) generate a vacuum Einstein field if and only if it is compatible
with the system of partial differential equations:

8, Pijy = 2(bc,.Ptiyj + awq (65)

(b, and a,0 are some covariant vector and symmetric tensor fields, respectively).

Proof: We sketch the proof by observing that equations (65) can be obtained by
contracting the indices Q and T in equations (18) written for a map ~(1): M, where
R aB = 0 and RorPyb = 0. Of course, to find general solutions of equations (65) in an
explicit form is also a difficult task. But we can verify, by solving algebraic equations
(see Subsection 2.4), if equations (65) are, or are not, integrable. w
In a similar manner we can analyse the problem of generation of the rr(z)-flat and
r(s)-flat spaces. Let us consider, for example, the 7rc2)-transforms. In this case we shall
parametrize the deformation spinor (55) in such a form as to induce a deformation
tensor of type (18). A-spinors should be also chosen in such a way as to induce a
deformation tensor (64) satisfying conditions (25) for the r(2)-maps. Calculating an
auxiliary curvature and the Ricci tensors for connection (63), and putting both tensors
into (26) and (27) and taking into account the basic T(z)-map equations, we obtain
that criterion (25) is an algebraic equation on the tensors Rap+, Rng, F,$ and the
covariant vector field p_,.
It is evident that the foregoing considerations point out at a to mutual interre-
lation between integrable deformations of the twistor equations and the criterions of
invariance and integrability of basic equations for the ng-maps rather than constitute
a method to solve the Einstein equation because no explicit constructions of metric
have been considered in our study. Perhaps the twistor-gauge formulation of gravity
 TWISTORS AND NEARLY AUTOPARALLEL MAPS 323

on flat nearly autoparallel backgrounds [7, 12, 141 is more convenient for the twi-
stor treatment of gravity. The interrelation between nc-twistors and gauge gravity is
a matter of our further investigations.
Finally, we remark that this paper contains a part of results (concerning the defi-
nition of nc-twistors on locally isotropic spaces) communicated by us during the Col-
loquium on Differential Geometry (Debrecen, Hungary, 25-30 July 1994) [24]. There
we have also presented some generalizations on the spinor and twistor calculus for
locally anisotropic spaces (which generalize the Lagrange and Finsler spaces) [25, 261.
The geometric constructions developed in this paper and in [12, 71 (in the frame-
work of the so-called twistor-gauge treatment of gravity) admit a straightforward
extension to locally anisotropic spaces if we apply the formalism of locally aniso-
tropic spinors and twistors [27, 281 and use the gauge-like formulations of locally
anisotropic gravity [26, 11, 10, 13, 141.

REFERENCES

[II R. Penrose and W. Rindler: Spinors and Space-time, V.l., Cambridge University Press, Cambridge
1984.
PI R. Penrose and W. Rindler: Spinors and Space-time, V.2., Cambridge University Press, Cambridge
1984.
[31 G. Vranceanu: Lectii de Geomettie differentiala, V.2., Ed. Didactica si Pedagogica, Bucuresti 1977
(in Romanian).
[41 N. S. Sinyukov: Geodesic Maps of Riernannian Spaces, Nauka, Moscow 1979 (in Russian).
PI S. Vacaru: Romanian J. Phys. 39 (lYY4), 37.
161 S. Vacaru and S. Ostaf: Buletinul Academiei de Stiinte a Repuhlicii Moldova, fzica si tehnica 3
(1993) 4.
[71 S. Vacaru: Buletinul Academiei de Stiinte a Republicii Moldova, fizica si tehnica 3 (1993) 17.
PI S. Vacaru and S. Ostaf: in Lagrange and Finsler Geometry, P. L. Antonelli and R. Miron (eds.).
Kluwer, Dordrecht 1996, p. 241.
PI S. Vacaru and S. Ostaf: Buletinul Academiei de Stiinte a Republicii Moldova, fizica si tehnica 1
(1994) 64.
[lOI S. Vacaru, S. Ostaf and Yu. Goncharenko: Romanian .f. Phys. 39 (1994) 19Y.
1111 S. Vacaru: Contr. Int. Conf “Lobachevski and Modem Geometry”, Part II, V. Bajanov et al. (eds.).
Kazani University Press, 1992, p. 64.
1121 S. Vacaru: tistnik Most. University 28 (1987) 5 (in Russian).
[I31 S. Vacaru: Applications of Nearly Autoparallel Maps and Twistor-Gauge Methods in Gravity and
Condensed States, PhD Thesis. “Al. I. Cuza” University, Ia$, Romania 1993 (in Romanian).
[I41 S. Vacaru: Twistors and Nearly Autoparallel Maps of Curved Spaces (subm. to Rep. Rom. Phys. ),
WI K. Yano: Concircular Geometry, I-IV. Proc. Imp. Acad. Tokyo, 16 (1940), 195; 354; 442; 505.
[I61 S. A. Schouten and D. Struik: Einfihrung in die neucren Medoden den Differentiol Geometrie, Bund
1, 2, 1938.
[17] E. Cartan: Les Systems Differentielles Exterieurs et Leurs Application Geometriques, Herman, Paris
1945.
[1X] M. Haimovichi: Selected Papers, Finsler Geometry, “ALL Cuza” Univ., la$i, Romania lY84.
[lo] S. V. Finnikov: The Method of the Cartan Exterior Differential Forms in Differential Geometry,
Gostekhizdat, Moscow 1948 (in Russian).
[20] P. K. Rashevsky: Geomettic Theory of Differential Equations, Gostekhizdat, Moscow 1947 (in
Russian).
[21] A. Z. Petrov: Einstein Spaces, Pergamon Press, Oxford 196Y.
324 S. I. VACARU and S. V. OSTAF

[22] H. Buchdahl: Nouvo Cim. 10 (1950), 96.

[23] .I. Plebadski: Actu Phys. Polon. 27 (1965) 361.
[24] S. Vacaru and S. Ostaf: in Abstracts of Contributions to the Colloquium on Differential Geometry, 25-30
July 1994, Lajos Kossuth University, Debrecen, Hungary 1994, p. 56.
[25] R. Miron and M. Anastasiei: The Geometry of Lagrange Spaces: Theory and Applications, Kluwer,
Dordrecht, Boston, London 1994.
[26] S. Vacaru and Yu. Goncharenko: Int. J. Theor. Phys. 34 (1995) 1955.
[27] S. Vacaru: Spinor structures and nonlinear connections in vector bundles, generalized Lagrange and
Finsler spaces (to appear in J. Math. Phys. 36 (1995)).
[28] S. Vacaru: Spinors in multidimensional and locally anisotropic spaces (in preparation).