Documente Academic
Documente Profesional
Documente Cultură
4(2017), 46-59
§1. Introduction
L2 (x, y)
L(x, y) = .
β(x, y)
In [3], Prasad, Shukla and Pandey have proved that the theorem A is also valid for expo-
nential change of Finsler metric given by
L(x, y) = Leβ/L .
Recently Prasad and Kumari Bindu [5] have proved that the theorem A is valid for
β−change [7] given by
L(x, y) = f (L, β),
In all these works it has been assumed that bi (x) in β is a function of positional coordinate
only.
The concept of h−vector has been introduced by H.Izumi. The covariant vector field
∂bi
bi (x, y) is said to be h−vector if ∂y j is proportional to angular metric tensor.
In 1990, Prasad, Shukla and Singh [4] have proved that the theorem A is valid for the
transformation (1.1) in which bi in β is h−vector.
All the above β−changes of Finsler metric encourage the authors to check whether the
theorem A is valid for any change of Finsler metric by h−vector.
In this paper we have proved that the theorem A is valid for the β−change of Finsler metric
given by
L(x, y) = f (L, β), (1.3)
∂bi
= ρhij , (1.5)
∂y j
where ρ is any scalar function of x, y and hij are components of angular metric tensor. The
homogeneity of f gives
Lf1 + βf2 = f, (1.6)
where the subscripts 1 and 2 denote the partial derivatives with respect to L and β respectively.
Hence, we have
f11 f12 f22
2
=− = 2
β βL L
which gives
f11 = β 2 ω, f22 = L2 ω, f12 = −βLω, (1.7)
Throughout the paper we frequently use equation (1.6) to (1.9) without quoting them.
§2. An h−Vector
Let bi (x, y) be components of a covariant vector in the Finsler space (M n , L). It is called an
h−vector if there exists a scalar function ρ such that
∂bi
= ρhij , (2.1)
∂y j
∂2L
hij = gij − li lj = L .
∂y i ∂y j
∂˙j ∂˙k bi = (∂˙k ρ)hij + ρL−1 {L2 ∂˙i ∂˙j ∂˙k L + hij lk },
(n − 2)[∂˙k ρ + ρL−1 lk ] = 0,
∂ρ j
∂y j y = −ρ.
We shall frequently use equation (2.2) without quoting it in the next article.
To find the relation between fundamental quantities of (M n , L) and (M n , L), we use the fol-
lowing results
∂˙i β = bi , ∂˙i L = li , ∂˙j li = L−1 hij . (3.1)
l i = f1 li + f2 b i , (3.2)
fp
hij = hij + f L2 wmi mj , (3.3)
L
where
β
p = f1 + Lf2 ρ, mi = bi − li .
L
The quantities corresponding to (M n , L) will be denoted by putting bar on the top of those
quantities.
From (3.2) and (3.3) we get the following relations between metric tensors of (M n , L) and
(M n , L)
fp
g ij = gij − L−1 {β(f1 f2 − f βLω) + Lρf f2 }li lj
L
+(f L2 ω + f22 )bi bj + (f1 f2 − f βLω)(li bj + lj bi ). (3.4)
The contravariant components of the metric tensor of (M n , L) will be obtained from (3.4)
as follows:
L ij Lv L4 ω i j L2 u
gij = g + 3 li lj − b b − 2 (li bj + lj bi ), (3.5)
fp f pt f pt f pt
where, we put bi = g ij bj , li = g ij lj , b2 = g ij bi bj and
u = f1 f2 − f βLω + Lρf22 ,
v = (f1 f2 − f βLω)(f β + △f2 L2 ) + Lρf2 {f (f + L2 ρf2 )
+L2 △(f22 + f L2 ω)}
and
β2
t = f1 + L3 ω△ + Lf2 ρ, △ = b2 − . (3.6)
L2
50 O.P.Pandey and H.S.Shukla
f
(a) ∂˙i f = li + f2 mi
L
(b) ∂˙i f1 = −βLωmi
(c) ∂˙i f2 = L2 ωmi
(d) ∂˙i p = −Lω(β − ρL2 )mi
3ω
(e) ∂˙i ω = − li + ω2 mi
L
(f) ˙ 2
∂i b = −2C..i + 2ρmi
2
(g) ∂˙i △ = −2C..i − 2 (β − ρL2 )mi , (3.7)
L
β i
where mi = g ij mj = bi − Ll .
i
To find C jk = g ih C jhk we use (3.5), (3.9), (3.10) and get
i i q sL3 L
C jk = Cjk + (hjk mi + hij mk + hik mj ) + mj mk mi − C.jk ni
2f p 2f p ft
4
Lq△ 2Lq + L △s
− 2 hjk ni − mj mk n i , (3.11)
2f pt 2f 2 pt
To find the v-curvature tensor of (M n , L) with respect to Cartan’s connection, we use the
following:
h
Cij hhk = Cijk , hik hkj = hij , hij ni = f L2 ωmj . (3.13)
and (M n , L):
fp
S hijk = Shijk + dhj dik − dhk dij + Ehk Eij − Ehj Eik , (3.15)
L
where
dij = P C.ij − Qhij + Rmi mj , (3.16)
The tangent vector space Mxn to M n at every point x is considered as the Riemannian n-
space (Mxn , gx ) with the Riemannian metric gx = gij (x, y)dy i dy j . Then the components of the
Cartan’s tensor are the Christoffel symbols associated with gx :
1 ih ˙
i
Cjk = g (∂k gjh + ∂˙j ghk − ∂˙h gjk ).
2
i
Thus Cjk defines the components of the Riemannian connection on Mxn and v-covariant deriva-
tive, say
Xi |j = ∂˙j Xi − Xh Cij
h
i
is the covariant derivative of covariant vector Xi with respect to Riemannian connection Cjk on
n n
Mx . It is observed that the v-curvature tensor Shijk of (M , L) is the Riemannian Christoffel
curvature tensor of the Riemannian space (M n , gx ) at a point x. The space (M n , gx ) equipped
with such a Riemannian connection is called the tangent Riemannian n-space [2].
It is well known [1] that any Riemannian n-space V n can be imbedded isometrically in a
Euclidean space of dimension n(n+1)2 . If n + r is the lowest dimension of the Euclidean space
n
in which V is imbedded isometrically, then the integer r is called the imbedding class number
of V n . The fundamental theorem of isometric imbedding ([1] page 190) is that the tangent
Riemannian n-space (Mxn , gx ) is locally imbedded isometrically in a Euclidean (n + r)−space if
and only if there exist r−number ǫP = ±1, r−symmetric tensors H(P )ij and r(r−1) 2 covariant
vector fields H(P,Q)i = −H(Q,P )i ; P, Q = 1, 2, · · · , r, satisfying the Gauss equations
X
Shijk = ǫP {H(P )hj H(P )ik − H(P )ij H(P )hk }, (4.1)
P
The numbers ǫP = ±1 are the indicators of unit normal vector NP to M n and H(P )ij are
the second fundamental tensors of M n with respect to the normals NP . Thus if gx is assumed
to be positive definite, there exists a Cartesian coordinate system (z i , up ) of the enveloping
Euclidean space E n+r such that ds2 in E n+r is expressed as
X X
ds2 = (dz i )2 + ǫp (dup )2 .
i p
r+2
X
S hijk = ǫλ {H (λ)hj H (λ)ik − H (λ)hk H (λ)ij },
λ=1
X
(b) H (r+1)ij kk − H (r+1)ik kj = ǫQ {H (Q)ij H (Q,r+1)k − H (Q)ik H (Q,r+1)j }
Q
X
(c) H (r+2)ij kk − H (r+2)ik kj = ǫQ {H (Q)ij H (Q,r+2)k − H (Q)ik H (Q,r+2)j }
Q
where ki denotes v-covariant derivative in (M n , L), i.e. covariant derivative in tangent Rieman-
i
nian n-space (Mxn , g x ) with respect to its Christoffel symbols C jk . Thus
h
Xi kj = ∂˙j Xi − Xh C ij .
To prove these equations we note that for any symmetric tensor Xij satisfying Xij li =
j
Xij l = 0, we have from (3.11),
q
Xij kk − Xik kj = Xij |k − Xik |j − (hik X.j − hij X.k )
2f t
L3 ω q
+ (C.ik X.j − C.ij X.k ) − (Xij mk − Xik mj )
t 2f p
L3 (2qω − sp)
+ (X.j mk − X.k mj ) mi . (5.4)
2f pt
q
q
fp
˙ fp √q
Since L
= ∂k L = 2 f Lp
mk , applying formula (5.4) for H(P )ij , we get
k
r ! r ! r
fp
fp
fp
H
− H
= {H(P )ij |k − H(P )ik |j }
L (P )ij k L (P )ik j L
r r
q fp L3 ω f p
− {hik H(P ).j − hij H(P ).k } + {C.ik H(p).j − C.ij H(p).k }
2f t L t L
√
L2 L(2qω − sp)
+ √ {H(P ).j mk − H(P ).k mj }mi . (5.6)
2t f p
q
q
fp
fp
Substituting the values of L H(P )ij
− L H (P )ik
from (5.6) and the values
k j
of dij from (3.16) in (5.5) we find that equation (5.5) is identically satisfied due to equation
54 O.P.Pandey and H.S.Shukla
(4.2).
Applying formula (5.4) for dij and substituting the values of d.j from (5.12) and dij from
(3.16), we get
√
Lq f ωp
dij kk − dik k=
j dij |k − dik |j − (hik C..j − hij C..k )
2f t3/2
L4 ω(2qω − sp)
+ √ (C..j mk − C..k mj )mi
2 f ωp.t3/2
√
L4 ω f ωp
+ (C.ik C..j − C.ij C..k )
t3/2
L4 ω△(3qω − sp)
+ √ (C.ik mj − C.ij mk )
2 f ωp.t3/2
Lq△(3qω − sp)
− √ (hik mj − hij mk ). (5.13)
4f f ωp.t3/2
Since,
√
L4 ω f ωp Lf p{pω2 + 3Lω 2 (β − ρL2 )}
∂˙k P = 3/2
C ..k + √
t 2 f ωp.t3/2
Lqω
+ √ mk ,
2 f ωpt
Lpqω pq
∂˙k Q = √ 3/2
C..k − 3
√ lk
2 f ωp.t 2L f ωpt
(β − ρL2 )(qω + sp) pq(qω + f pω2 )
− √ mk − √ mk
2L f ωpt 4L2 (f ωp)3/2 t
pq{3ω(β − ρL2 ) − L2 △ω2 }
+ √ mk (5.15)
4L f ωp t3/2
and
Substituting the value of dij |k − dik |j from (5.16) in (5.13), then value of dij kk − dik kj
thus obtained in (5.7), and using equations (4.1) and (3.17), it follows that equation (5.7) holds
identically.
sp − 2qω
Eij kk − Eik kj = √ (dij mk − dik mj ). (5.17)
2f ω pt
pq + L3 △(sp − qω)
E.j = √ mj . (5.18)
2L2 p f ω
56 O.P.Pandey and H.S.Shukla
Applying formula (5.4) for Eij and substituting the value of E.j from (5.18) and the value
of Eij from (3.17), we get
qL△(sp − 2qω)
Eij kk − Eik kj = Eij |k − Eik |j + √ (hij mk − hik mj )
4f pt f ω
Lω{pq + L3 △(sp − qω)}
+ √ (C.ik mj − C.ij mk ). (5.19)
2pt f ω
Now,
q (β − ρL2 )s q(f ω2 + f2 ω)
(∂˙k S) = − √ lk − √ + mk (5.21)
2L3 f ω 2L f ω 4L2 (f ω)3/2
and
sp − qω
(∂˙k T ) = − √ lk + term containing mk ,
2p f ω
L(sp − qω)
Eij |k − Eik |j = √ (C.ij mk − C.ik mj )
2p f ω
q(sp − 2qω)
− 2 (hij mk − hik mj ). (5.22)
4L p(f ω)3/2
Substituting the value of Eij |k − Eik |j from (5.22) in (5.19), then the value of Eij kk − Eik kj
thus obtained in (5.17), and then using (3.16) in the right-hand side of (5.17), we find that the
equation (5.17) holds identically.
This completes the proof of Codazzi equations of (Mxn , gx ). The Ricci Kühne equations of
(Mxn , g x ) consist of the following four equations
X
(a) H (P,Q)i kj − H (P,Q)j ki + ǫR {H (R,P )i H (R,Q)j
R
−H (R,P )j H (R,Q)i } + ǫr+1 {H (r+1,P )i H (r+1,Q)j
−H (r+1,P )j H (r+1,Q)i } + ǫr+2 {H (r+2,P )i H (r+2,Q)j
−H (r+2,P )j H (r+2,Q)i } + ghk {H (P )hi H (Q)kj
−H (P )hj H (Q)ki } = 0, P, Q = 1, 2, · · · , r (5.23)
β−Change of Finsler Metric by h-Vector and Imbedding Classes of Their Tangent Spaces 57
X
(b) H (P,r+1)i kj − H (P,r+1)j ki + ǫR {H (R,P )i H (R,r+1)j − H (R,P )j H (R,r+1)i }
R
+ ǫr+2 {H (r+2,P )i H (r+2,r+1)j − H (r+2,P )j H (r+2,r+1)i }
+ ghk {H (P )hi H (r+1)kj − H (P )hj H (r+1)ki } = 0, P = 1, 2, · · · , r
X
(c) H (P,r+2)i kj − H (P,r+2)j ki + ǫR {H (R,P )i H (R,r+2)j − H (R,P )j H (R,r+2)i }
R
+ ǫr+1 {H (r+1,P )i H (r+1,r+2)j − H (r+1,P )j H (r+1,r+2)i }
+ ghk {H (P )hi H (r+2)kj − H (P )hj H (r+2)ki } = 0, P = 1, 2, · · · , r
X
(d) H (r+1,r+2)i kj − H (r+1,r+2)j ki + ǫR {H (R,r+1)i H (R,r+2)j − H (R,r+1)j
R
×H (R,r+2)i } + ghk {H (r+1)hi H (r+2)kj − H (r+1)hj H (r+2)ki } = 0.
fp
g hk {H(P )hi H(Q)kj − H(P )hj H(Q)ki } = g hk {H(P )hi H(Q)kj
L
L3 ω
−H(P )hj H(Q)ki } − {H(P ).i H(Q).j − H(P ).j H(Q).i }.
t
Also, we have H(P,Q)i kj − H(P,Q)j ki = H(P,Q)i |j − H(P,Q)j |i . Hence equation (5.24) is
satisfied identically by virtue of (4.3).
H(P ).i kj − H(P ).j ki = H(P ).i |j − H(P ).j |i = {H(P )hi |j − H(P )hj |i }bh
−g hk {H(P )hi C.kj − H(P )hj C.ki } (5.26)
58 O.P.Pandey and H.S.Shukla
√ ! √ !
L Lω
L Lω
√
= ∂˙j √
t j t
4
√ √
L ω Lω L Lω
= C..j + {pω2 + 3Lω 2 (β − ρL2 )} mj (5.27)
t3/2 2ωt3/2
and
r s
hk fp L hk
g {H(P )hi dkj − H(P )hj dki } = g ×
L fp
√
L3 ω L
{H(P )hi dkj − H(P )hj dki } − √ {H(P ).i d.j − H(P ).j d.i }. (5.28)
t fp
After using (3.16) and (5.12) the equation (5.28) may be written as
r √
hk fp L Lω hk
g {H(P )hi dkj − H(P )hj dki } = √ g ×
L t
√
L4 ω Lω
{H(P )hi C.kj − H(P )hj C.ki } − {H(P ).i C..j − H(P ).j C..i }
√ t3/2
L Lω
− [pω2 + 3Lω 2 (β − ρL2 )]{H(P ).i mj − H(P ).j mi }. (5.29)
2ωt3/2
Since Ekj lk = 0 = Ejk lk , from (3.5), we find that the value of ghk {H(P )hi Ekj − H(P )hj Eki }
is s √
L hk L3 ω L
.g {H(P )hi Ekj − H(P )hj Eki } − √ {H(P ).i E.j − H(P ).j E.i },
fp t fp
sp−2qω
where N = √ .
2f ω pt
β−Change of Finsler Metric by h-Vector and Imbedding Classes of Their Tangent Spaces 59
Since dhi lh = 0, Ekj lk = 0, from (3.5), we find that the value of ghk {dhi Ekj − dhj Eki } is
L hk L4 ω
g {dhi Ekj − dhj Eki } − {d.i E.j − d.j E.i },
fp f pt
L3 (2qω − sp)
− √ {C..i mj − C..j mi }.
2f p.t3/2
Also,
we have
L3 (sp − 2qω)
(N mi )kj − (N mj )ki = √ (C..j mi − C..i mj ). (5.32)
2f p.t3/2
Hence equation (5.31) is satisfied identically. Therefore Ricci Kühne equations of (Mxn , gx )
given in (5.23) are satisfied.
Hence the Theorem A given in introduction is satisfied for the β−change (1.3) of Finsler
metric given by h−vector. 2
References