1. Introduction
In the generalization from the theory of submanifolds in Riemannian to the theory of submanifolds in semi-Riemannian manifolds, the induced metric on submanifolds may be degenerate (lightlike). Therefore, there is a natural existence of lightlike submanifolds and for which the local and global geometry is completely different than nondegenerate case. In lightlike case, the standard text book definitions do not make sense, and one fails to use the theory of nondegenerate geometry in the usual way. The primary difference between the lightlike submanifolds and nondegenerate submanifolds is that in the first case, the normal vector bundle intersects with the tangent bundle. Thus, the study of lightlike submanifolds becomes more difficult and different from the study of nondegenerate submanifolds. Moreover, the geometry of lightlike submanifolds is used in mathematical physics, in particular, in general relativity since lightlike submanifolds produce models of different types of horizons (event horizons, Cauchy’s horizons, and Kruskal’s horizons). The universe can be represented as a four-dimensional submanifold embedded in a (4 + n)-dimensional spacetime manifold. Lightlike hypersurfaces are also studied in the theory of electromagnetism [1]. Thus, large number of applications but limited information available motivated us to do research on this subject matter. Kupeli [2] and Duggal and Bejancu [1] developed the general theory of degenerate (lightlike) submanifolds. They constructed a transversal vector bundle of lightlike submanifold and investigated various properties of these manifolds.
In the study of Riemannian geometry, Chen and Yano [
3] introduced the notion of a
Riemannian manifold of a quasiconstant curvature as a Riemannian manifold
with the curvature tensor
satisfying the condition
(1.1)
for any vector fields
X,
Y,
Z, and
W on
, where
α,
β are scalar functions and
θ is a 1-form defined by
(1.2)
where
ζ is a unit vector field on
which called the
curvature vector field. It is well known that if the curvature tensor
is of the form (
1.1), then the manifold is conformally flat. If
β = 0, then the manifold reduces to a space of constant curvature.
A nonflat Riemannian manifold of dimension
n(>2) is defined to be a quasi-Einstein manifold [
4] if its Ricci tensor satisfies the condition
(1.3)
where a, b are scalar functions such that b ≠ 0, and ϕ is a nonvanishing 1-form such that for any vector field X, where U is a unit vector field. If b = 0, then the manifold reduces to an Einstein manifold. It can be easily seen that every Riemannian manifold of quasiconstant curvature is a quasi-Einstein manifold.
The subject of this paper is to study the geometry of lightlike submanifolds of a semi-Riemannian manifold of quasiconstant curvature. We prove two characterization theorems for such a lightlike submanifold (M, g, S(TM), S(TM⊥)) as follows.
Theorem 1.1. Let M be an r-lightlike submanifold of a semi-Riemannian manifold of quasiconstant curvature. If the curvature vector field ζ of is tangent to M and S(TM) is totally geodesic in M, then one has the following results:
- (1)
if S(TM⊥) is a Killing distribution, then the functions α and β, defined by (1.1), vanish identically. Furthermore, , M, and the leaf M* of S(TM) are flat manifolds;
- (2)
if S(TM⊥) is a conformal Killing distribution, then the function β vanishes identically. Furthermore, and M* are space of constant curvatures, and M is an Einstein manifold such that Ric = (r/(m − r))g, where r is the induced scalar curvature of M.
Theorem 1.2. Let M be an irrotational r-lightlike submanifold of a semi-Riemannian manifold of quasiconstant curvature. If ζ is tangent to M, S(TM) is totally umbilical in M, and S(TM⊥) is a conformal Killing distribution with a nonconstant conformal factor, then the function β vanishes identically. Moreover, and M* are space of constant curvatures, and M is a totally umbilical Einstein manifold such that Ric = (c/(m − r))g, where c is the scalar quantity of M.
2. Lightlike Submanifolds
Let (
M,
g) be an
m-dimensional lightlike submanifold of an (
m +
n)-dimensional semi-Riemannian manifold
. We follow Duggal and Bejancu [
1] for notations and results used in this paper. The radical distribution Rad(
TM) =
TM∩
TM⊥ is a vector subbundle of the tangent bundle
TM and the normal bundle
TM⊥, of rank
r (1 ≤
r ≤ min{
m,
n}). Then, in general, there exist two complementary nondegenerate distributions
S(
TM) and
S(
TM⊥) of Rad(
TM) in
TM and
TM⊥, respectively, called the
screen and
coscreen distribution on
M, and we have the following decompositions:
(2.1)
where the symbol ⊕
orth denotes the orthogonal direct sum. We denote such a lightlike submanifold by
M = (
M,
g,
S(
TM),
S(
TM⊥)). Let tr(
TM) and ltr(
TM) be complementary (but not orthogonal) vector bundles to
TM in
and
TM⊥ in
S(
TM)
⊥, respectively, and let {
Ni} be a lightlike basis of
consisting of smooth sections of
, where
𝒰 is a coordinate neighborhood of
M, such that
(2.2)
where {
ξ1, …,
ξr} is a lightlike basis of Γ(Rad(
TM)). Then,
(2.3)
We say that a lightlike submanifold (
M,
g,
S(
TM),
S(
TM⊥)) of
is
(1) r-lightlike submanifold if 1 ≤ r < min{m, n},
(2) coisotropic submanifold if 1 ≤ r = n < m,
(3) isotropic submanifold if 1 ≤ r = m < n,
(4) totally lightlike submanifold if 1 ≤ r = m = n.
The above three classes (2)~(4) are particular cases of the class (1) as follows: S(TM⊥) = {0}, S(TM) = {0}, and S(TM) = S(TM⊥) = {0}, respectively.
Example 2.1. Consider in the 1-lightlike submanifold M given by equations
(2.4)
then we have
TM = span{
U1,
U2} and
TM⊥ = {
H1,
H2}, where we set
(2.5)
It follows that Rad(
TM) is a distribution on
M of rank 1 spanned by
ξ =
H1. Choose
S(
TM) and
S(
TM⊥) spanned by
U2 and
H2 where are timelike and spacelike, respectively. Finally, the lightlike transversal vector bundle
(2.6)
and the transversal vector bundle
(2.7)
are obtained.
Let be the Levi-Civita connection of and P the projection morphism of Γ(TM) on Γ(S(TM)) with respect to the decomposition (2.1). For an r-lightlike submanifold, the local Gauss-Weingartan formulas are given by
(2.8)
(2.9)
(2.10)
(2.11)
(2.12)
for any
X,
Y ∈ Γ(
TM), where ∇ and ∇
* are induced linear connections on
TM and
S(
TM), respectively, the bilinear forms
and
on
M are called the
local lightlike second fundamental form and
local screen second fundamental form on
TM, respectively, and
is called the
local radical second fundamental form on
S(
TM).
, and
are linear operators on Γ(
TM), and
τij,
ρiα,
ϕαi, and
θαβ are 1-forms on
TM.
Since is torsion-free, ∇ is also torsion-free and both and are symmetric. From the fact that , we know that are independent of the choice of a screen distribution. Note that , and ρiα depend on the section ξ ∈ Γ(Rad(TM)|𝒰). Indeed, take , then we have [5].
The induced connection ∇ on TM is not metric and satisfies
(2.13)
where
ηi is the 1-form such that
(2.14)
But the connection ∇
* on
S(
TM) is metric. The above three local second fundamental forms of
M and
S(
TM) are related to their shape operators by
(2.15)
(2.16)
(2.17)
(2.18)
(2.19)
and
ϵβθαβ = −
ϵαθβα, where
X,
Y ∈ Γ(
TM). From (
2.19), we know that the operators
are shape operators related to
for each
i, called the
radical shape operators on
S(
TM). From (
2.16), we know that the operators
are Γ(
S(
TM)) valued. Replace
Y by
ξj in (
2.15), then we have
for all
X ∈ Γ(
TM). It follows that
(2.20)
Also, replace
X by
ξj in (
2.15) and use (
2.20), then we have
(2.21)
Thus
ξi is an eigenvector field of
corresponding to the eigenvalue 0. For an
r-lightlike submanifold, replace
Y by
ξi in (
2.17), then we have
(2.22)
From (2.15)~(2.18), we show that the operators and are not self-adjoint on Γ(TM) but self-adjoint on Γ(S(TM)).
Theorem 2.2. Let (M, g, S(TM), S(TM⊥)) be an r-lightlike submanifold of a semi-Riemannian manifold , then the following assertions are equivalent:
- (i)
are self-adjoint on Γ(TM) with respect to g, for all i,
- (ii)
satisfy for all X ∈ Γ(TM), i and j,
- (iii)
for all i and j, that is, the image of Rad(TM) with respect to for each i is a trivial vector bundle,
- (iv)
for all X, Y ∈ Γ(TM) and i, that is, is a shape operator on M, related by the second fundamental form .
Proof. From (2.15) and the fact that are symmetric, we have
(2.23)
(i)⇔(ii). If for all X ∈ Γ(TM), i and j, then we have for all X, Y ∈ Γ(TM), that is, are self-adjoint on Γ(TM) with respect to g. Conversely, if are self-adjoint on Γ(TM) with respect to g, then we have
(2.24)
for all
X,
Y ∈ Γ(
TM). Replace
Y by
ξj in this equation and use the second equation of (
2.20), then we have
for all
X ∈ Γ(
TM),
i and
j.
(ii)⇔(iii). Since S(TM) is nondegenerate, from the first equation of (2.21), we have , for all i and j.
(ii)⇔(iv). From (2.16), we have for any X, Y ∈ Γ(TM) and for all i and j.
Corollary 2.3. Let (M, g, S(TM), S(TM⊥)) be a 1-lightlike submanifold of a semi-Riemannian manifold , then the operators are self-adjoint on Γ(TM) with respect to g.
Definition 2.4. An r-lightlike submanifold (M, g, S(TM), S(TM⊥)) of a semi-Riemannian manifold is said to be irrotational if for any X ∈ Γ(TM) and i.
For an r-lightlike submanifold M of , the above definition is equivalent to and for any X ∈ Γ(TM). In this case, are self-adjoint on Γ(TM) with respect to g, for all i.
We need the following Gauss-Codazzi equations (for a full set of these equations see [
1, chapter 5]) for M and
S(
TM). Denote by
, and
R* the curvature tensors of the Levi-Civita connection
of
, the induced connection ∇ of
M, and the induced connection ∇
* on
S(
TM), respectively:
(2.25)
(2.26)
(2.27)
(2.28)
(2.29)
(2.30)
The Ricci tensor of
is given by
(2.31)
for any
X,
. Let
=
m +
n. Locally,
is given by
(2.32)
where {
E1, …,
Em+n} is an orthonormal frame field of
. If
and
(2.33)
then
is an
Einstein manifold. If
, any
is Einstein, but
in (
2.33) is not necessarily constant. The
scalar curvature is defined by
(2.34)
Putting (
2.33) in (
2.34) implies that
is Einstein if and only if
(2.35)
3. The Tangential Curvature Vector Field
Let
R(0,2) denote the induced Ricci tensor of type (0,2) on
M, given by
(3.1)
Consider an induced quasiorthonormal frame field
(3.2)
where {
Ni,
Wα} is a basis of Γ(tr(
TM)|
𝒰) on a coordinate neighborhood
𝒰 of
M such that
Ni ∈ Γ(ltr(
TM)|
𝒰) and
Wα ∈ Γ(
S(
TM⊥)|
𝒰). By using (
2.29) and (
3.1), we obtain the following local expression for the Ricci tensor:
(3.3)
(3.4)
Substituting (
2.25) and (
2.27) in (
3.3) and using (
2.15)
~(
2.18) and (
3.4), we obtain
(3.5)
for any X, Y ∈ Γ(TM). This shows that R(0,2) is not symmetric. A tensor field R(0,2) of M, given by (3.1), is called its induced Ricci tensor if it is symmetric. From now and in the sequel, a symmetric R(0,2) tensor will be denoted by Ric.
Using (
2.28), (
3.5), and the first Bianchi identity, we obtain
(3.6)
From this equation and (
2.28), we have
(3.7)
Theorem 3.1 (see[5]). Let M be a lightlike submanifold of a semi-Riemannian manifold , then the tensor field R(0,2) is a symmetric Ricci tensor Ric if and only if each 1-form tr(τij) is closed, that is, d(tr(τij)) = 0, on any 𝒰 ⊂ M.
Note 1. Suppose that the tensor R(0,2) is symmetric Ricci tensor Ric, then the 1-form tr(τij) is closed by Theorem 3.1. Thus, there exist a smooth function f on 𝒰 such that tr(τij) = df. Consequently, we get tr(τij)(X) = X(f). If we take , it follows that . Setting Δ = exp(f) in this equation, we get for any X ∈ Γ(TM|𝒰). We call the pair on 𝒰 such that the corresponding 1-form tr(τij) vanishes the canonical null pair of M.
For the rest of this paper, let
M be a lightlike submanifold of a semi-Riemannian manifold
of quasiconstant curvature. We may assume that the curvature vector field
ζ of
is a unit spacelike tangent vector field of
M and
>4,
(3.8)
(3.9)
(3.10)
for all
X,
Y ∈ Γ(
TM). Substituting (
3.8)
~(
3.10) into (
3.5), we have
(3.11)
Definition 3.2. We say that the screen distribution S(TM) of M is totally umbilical [1] in M if, on any coordinate neighborhood 𝒰 ⊂ M, there is a smooth function γi such that for any X ∈ Γ(TM), or equivalently,
(3.12)
In case
γi = 0 on
𝒰, we say that
S(
TM) is
totally geodesic in
M.
A vector field
X on
is said to be a
conformal Killing vector field [
6] if
for any smooth function
δ, where
denotes the Lie derivative with respect to
X, that is,
(3.13)
In particular, if
δ = 0, then
X is called a
Killing vector field [
7]. A distribution
𝒢 on
is called a
conformal Killing (resp.,
Killing)
distribution on
if each vector field belonging to
𝒢 is a conformal Killing (resp., Killing) vector field on
. If the coscreen distribution
S(
TM⊥) is a Killing distribution, using (
2.10) and (
2.17), we have
(3.14)
Therefore, since
are symmetric, we obtain
(3.15)
Theorem 3.3. Let M be an r-lightlike submanifold of a semi-Riemannian manifold , then the coscreen distribution S(TM⊥) is a conformal Killing (resp., Killing) distribution if and only if there exists a smooth function δα such that
(3.16)
Theorem 3.4. Let M be an irrotational r-lightlike submanifold of a semi-Riemannian manifold of quasiconstant curvature. If the curvature vector field ζ of is tangent to M, S(TM) is totally umbilical in M, and S(TM⊥) is a conformal Killing distribution, then the tensor field R(0,2) is an induced symmetric Ricci tensor of M.
Proof. From (2.17)~(2.20), (2.22), (3.16), and (3.11), we have
(3.17)
(3.18)
Using (
3.17), we show that
R(0,2) is symmetric.
4. Proof of Theorem
As
, we get
by (
2.30). From (
2.27) and (
3.16), we have
(4.1)
By Theorems
3.1 and
3.4, we get
dτ = 0 on
TM. Thus, we have
due to (
2.28). From the above results, we deduce the following equation:
(4.2)
Taking
X =
ξi and
Z =
X to (
4.2) and then comparing with (
3.9), we have
(4.3)
Case 1. If S(TM⊥) is a Killing distribution, that is, δα = 0, then we have
(4.4)
Substituting (
4.3) into (
1.1) and using (
2.25) and the facts
and
due to (
1.1), we have
(4.5)
Thus,
M is a space of constant curvature −
α. Taking
X =
Y =
ζ to (
4.3), we have
β = −
α. Substituting (
4.3) into (
3.18) with
δα =
γi = 0, we have
(4.6)
On the other hand, substituting (
4.5) and
g(
R(
ξi,
Y)
X,
Ni) = 0 into (
3.4), we have
(4.7)
From the last two equations, we get
α = 0 as
m > 1. Thus,
β = 0, and
and
M are flat manifolds by (
1.1) and (
4.5). From this result and (
2.29), we show that
M* is also flat.
Case 2. If S(TM⊥) is a conformal Killing distribution, assume that β ≠ 0. Taking X = Y = ζ to (4.3), we have . From this and (4.3), we show that
(4.8)
Substituting (
4.8) into (
1.1) and using (
2.25) with
and (
3.16), we have
(4.9)
for all
X,
Y,
Z,
W ∈ Γ(
TM). Substituting (
4.8) into (
3.18) with
γi = 0, we have
(4.10)
by the fact that
. On the other hand, from (
2.27), (
3.9), and (
4.3), we have
g(
R(
ξi,
Y)
X,
Ni) = 0. Substituting this result and (
4.9) into (
3.4), we have
(4.11)
The last two equations imply
β = 0 as
m −
r > 1. It is a contradiction. Thus,
β = 0 and
is a space of constant curvature
α. From (
2.29) and (
4.9), we show that
M* is a space of constant curvature
. But
M is not a space of constant curvature by (
3.17)
3. Let
, then the last two equations reduce to
(4.12)
Thus
M is an Einstein manifold. The scalar quantity
r of
M [
8], obtained from
R(0,2) by the method of (
2.34), is given by
(4.13)
Since
M is an Einstein manifold satisfying (
4.12), we obtain
(4.14)
Thus, we have
(4.15)
which provides a geometric interpretation of half lightlike Einstein submanifold (the same as in Riemannian case) as we have shown that the constant
κ =
r/(
m −
r).
5. Proof of Theorem
Assume that
ζ is tangent to
M,
S(
TM) is totally umbilical, and
S(
TM⊥) is a conformal Killing vector field. Using (
1.1), (
2.26) reduces to
(5.1)
for all
X,
Y,
Z ∈ Γ(
TM). Replacing
W by
N to (
1.1), we have
(5.2)
for all
X,
Y,
Z ∈ Γ(
TM) and where
ei =
θ(
Ni). Applying ∇
X to (
3.12) and using (
2.13), we have
(5.3)
for all
X,
Y,
Z ∈ Γ(
TM). Substituting this equation into (
2.30), we obtain
(5.4)
Substituting this equation and (
5.2) into (
2.27) and using
θ(
ξi) = 0, we obtain
(5.5)
Replacing
Y by
ξi to this and using (
2.20)
1 and the fact
θ(
ξi) = 0, we have
(5.6)
for all
X,
Y ∈ Γ(
TM). Differentiating (
3.16) and using (
5.1), we have
(5.7)
Replacing
Y by
ξi in the last equation and using (
2.20)
1, we obtain
(5.8)
As the conformal factor
δα is nonconstant, we show that
δα −
ϵαρiα(
ξi) ≠ 0. Thus, we have
(5.9)
where
. From (
3.17)
1 and (
5.9), we show that the second fundamental form tensor
h, given by
, satisfies
(5.10)
Thus,
M is totally umbilical [
5]. Substituting (
5.9) into (
5.6), we have
(5.11)
for all
X,
Y ∈ Γ(
TM). Taking
X =
Y =
ζ to this equation, we have
(5.12)
Assume that
β ≠ 0, then we have
(5.13)
Substituting (
5.13) into (
1.1) and using (
2.25), (
3.12), (
3.17)
1, and (
5.9), we have
(5.14)
for all
X,
Y,
Z,
W ∈ Γ(
TM). Substituting (
5.9) and (
5.13) into (
3.18), we have
(5.15)
On the other hand, substituting (
5.14) and the fact that
(5.16)
into (
3.4), we have
(5.17)
Comparing (
5.15) and (
5.17), we obtain (
m − 1)
β = 0. As
m > 1, we have
β = 0, which is a contradiction. Thus, we have
β = 0. Consequently, by (
1.1), (
2.29), and (
5.14), we show that
and
M* are spaces of constant curvatures
α and
, respectively. Let
(5.18)
Thus,
M is an Einstein manifold. The scalar quantity
c of
M is given by
(5.20)
Example 5.1. Let (M, g) be a lightlike hypersurface of an indefinite Kenmotsu manifold equipped with a screen distribution S(TM), then there exist an almost contact metric structure on , where J is a (1,1)-type tensor field, ζ is a vector field, ϑ is a 1-form, and is the semi-Riemannian metric on such that
(5.22)
for any vector fields
X,
Y on
, where
is the Levi-Civita connection of
. Using the local second fundamental forms
B and
C of
M and
S(
TM), respectively, and the projection morphism
P of
M on
S(
TM), the curvature tensors
, and
R* of the connections
, ∇, and ∇
* on
, and
S(
TM), respectively, are given by (see [
9])
(5.23)
for any
X,
Y,
Z,
W ∈ Γ(
TM). In case the ambient manifold
is a space form
of constant
J-holomorphic sectional curvature
c,
is given by (see [
10])
(5.24)
Assume that
M is almost screen conformal, that is,
(5.25)
where
φ is a nonvanishing function on a neighborhood
𝒰 in
M, and
ζ is tangent to
M, then, by the method in Section 2 of [
9], we have
(5.26)
where
ρ is a nonvanishing function on a neighborhood
𝒰. Then the leaf
M* of
S(
TM) is a semi-Riemannian manifold of quasiconstant curvature such that
α = −1 + 2
φρ2,
β = −2
φρ2, and
θ =
ϑ in (
1.1).
Acknowledgment
The authors are thankful to the referee for making various constructive suggestions and corrections towards improving the final version of this paper.