Totally geodesic submanifolds in the tangent bundle of a Riemannian 2-manifold.

Alexander YAMPOLSKY

November 27, 2006

Abstract
We give a full description of totally geodesic submanifolds in the tangent bundle of a Riemannian 2-manifold of constant curvature and present a new class of a cylinder-type totally geodesic submanifolds in the general case.
Keywords: Sasaki metric, totally geodesic submanifolds in the tangent bundle.
AMS subject class: Primary 53B25; Secondary 53B20.
Introduction Let ( M n , g )   be a Riemannian manifold with metric g   and T M n   its tangent bundle. S. Sasaki [7] introduced on T M n   a natural Riemannian metric T g   . With respect to this metric, all the fibers are totally geodesic and intrinsically flat submanifolds. Probably M.-S. Liu [5] was the first who noticed that the base manifold embedded into T M n   by the zero section is totally geodesic, as well. Soon afterwards, Sato K. [9] described geodesics (the totally geodesic submanifolds of dimension 1) in the tangent bundle over space forms. The next step was made by P.Walczack [10] who tried to find a non-zero section ξ : M n → T M n   such that the image ξ ( M n )   is a totally geodesic submanifold. He proved that if ξ   is of constant length and ξ ( M n )   is totally geodesic, then ξ   is a parallel vector field. As a consequence, the base manifold should be reducible. The irreducible case stays out of considerations up to now. A general conjecture stated by A.Borisenko claims that, in irreducible case, the zero vector field is the unique one which generates a totally geodesic submanifold ξ ( M n )   or, equivalently, the base manifold is the unique totally geodesic submanifold of dimension n   in T M n   transversal to fibers. A dimensional restriction is essential. M.T.K. Abbassi and the author [1] treated the case of fiber transversal submanifolds in T M n   of dimension l < n   and have found some examples of totally geodesic submanifolds of this type. Earlier this problem had been considered in [11] .
It is also worthwhile to mention that in the case of tangent sphere bundle the situation is different. Sasaki S. [8] described geodesics in the tangent sphere bundle over space forms and Nagy P. [6] described geodesics in the tangent sphere bundle over symmetric spaces. The author has given a full description of totally geodesic vector fields on 2-dimensional manifolds of constant curvature [12] and an example of a totally geodesic unit vector field on positively/negatively curved manifolds of non-constant curvature [13] . A full description of 2-manifolds which admit a totally geodesic unit vector field was given in [14] .
In this paper we consider a more general problem concerning the description of all possible totally geodesic submanifolds in the tangent bundle of Riemannian 2-manifold with a sign-preserving curvature. For the spaces of constant curvature this problem was posed by A.Borisenko in [2] .
In Section  2 we prove the following theorems.
Theorem 1. Let M 2   be Riemannian manifold of constant curvature K ≠ 0   . Suppose that F ~ 2 ⊂ T M 2   is a totally geodesic submanifold. Then locally F ~ 2   is one of the following submanifolds:
  • (a) a single fiber T q M 2   ;
  • (b) a cylinder-type surface based on a geodesic γ   in M 2   with elements generated by a parallel unit vector field along γ   ;
  • (c) the base manifold embedded into T M 2   by zero vector field.
Remark that the item (b) of Theorem 1 is a consequence of more general result.
Theorem 2 Let M 2   be a Riemannian manifold of sign-preserving curvature.
Suppose that F ~ 2 ⊂ T M 2   is a totally geodesic submanifold having non-transversal intersection with the fibers. Then locally F ~ 2   is a cylinder-type surface based on a geodesic γ   in M 2   with elements generated by a parallel unit vector field along γ   .
Moreover, a general Riemannian manifold M n   admits this class of totally geodesic surfaces in T M n   (see Proposition  2.4 ).
In Section  3 we prove the following general result.
Theorem 3. Let M 2   be a Riemannian manifold with sign-preserving curvature. Then T M 2   does not admit a totally geodesic 3-manifold even locally.
Acknowledgement. The author expresses his thanks to professor E. Boeckx (Leuven, Belgium) for useful remarks in discussing the results.

1 Necessary facts about the Sasaki metric

Let ( M n , g )   be an n   -dimensional Riemannian manifold with metric g   .
Denote by < ⋅ , ⋅ >   the scalar product with respect to g   . The Sasaki metric on T M n   is defined by the following scalar product: if X ~ , Y ~   are tangent vector fields on T M n   , then
< < X ~ , Y ~ > > : = < π * X ~ , π * Y ~ > + < K X ~ , K Y ~ > , (1)
where π * : T T M n → T M n   is the differential of the projection π : T M n → M n   and K : T T M n → T M n   is the connection map [3] . The local representations for π *   and K   are the following ones.
Let ( x 1 , . . . , x n )   be a local coordinate system on M n   . Denote by ∂ i : = ∂ / ∂ x i   the natural tangent coordinate frame. Then, at each point q ∈ M n   , any tangent vector ξ   can be decomposed as ξ = ξ i ∂ i | q   . The set of parameters { x 1 , . . . , x n ; ξ 1 , . . . , ξ n }   forms the natural induced coordinate system in T M n   , i.e. for a point Q = ( q , ξ ) ∈ T M n   , with q ∈ M n , ξ ∈ T q M n   , we have q = ( x 1 , . . . , x n ) , ξ = ξ i ∂ i | q   . The natural frame in T Q T M n   is formed by ∂ ~ i : = ∂ ∂ x i | Q , ∂ ~ n + i : = ∂ ∂ ξ i | Q   and for any X ~ ∈ T Q T M n   we have the decomposition X ~ = X ~ i ∂ ~ i + X ~ n + i ∂ ~ n + i .   Now locally, the horizontal and vertical projections of X ~   are given by
π * X ~ | Q = X ~ i ∂ i | q , K X ~ | Q = ( X ~ n + i + Γ j k i ( q ) ξ j X ~ k ) ∂ i | q , (2)
where Γ j k i   are the Christoffel symbols of the metric g   .
The inverse operations are called lifts . If X = X i ∂ i   is a vector field on M n   then the vector fields on T M   given by
X h = X i ∂ ~ i − Γ j k i ξ j X k ∂ ~ n + i ,
X v = X i ∂ ~ n + i
are called the horizontal and vertical lifts of X   respectively. Remark that for any vector field X   on M n   it holds
π * X h = X , K X h = 0 , π * X v = 0 , K X v = X . (3)
There is a natural decomposition T Q ( T M n ) = ℋ Q ( T M n ) ⊕ V Q ( T M n ) ,   where ℋ Q ( T M n ) = ker K   is called the horizontal distribution and V Q ( T M n ) = ker π *   is called the vertical distribution on T M n   . With respect to the Sasaki metric, these distributions are mutually orthogonal. The vertical distribution is integrable and the fibers are precisely its integral submanifolds.
The horizontal distribution is never integrable except the case of a flat base manifold.
For any vector fields X , Y   on M n   , the covariant derivatives of various combinations of lifts to the point Q = ( q , ξ ) ∈ T M n   can be found by the formulas [4] 
∇ ~ X h Y h | Q = ( ∇ X Y | q ) h − 1 2 ( R q ( X , Y ) ξ ) v , ∇ ~ X v Y h | Q = 1 2 ( R q ( ξ , X ) Y ) h , ∇ ~ X h Y v | Q = ( ∇ X Y | q ) v + 1 2 ( R q ( ξ , Y ) X ) h , ∇ ~ X v Y v | Q = 0 , (4)
where ∇   and R   are the Levi-Civita connection and the curvature tensor of M n   respectively.
Remark. The formulas  4 are applicable to the lifts of vector fields only.
A formal application to a general field on tangent bundle may lead to wrong result. For example,
∇ ~ X v ( ξ i ( ∂ i ) h ) = X v ( ξ i ) ∂ i h + ξ i ∇ ~ X v ∂ i h
= X i ∂ i h + ξ i 1 2 ( R ( ξ , X ) ∂ i ) h = X h + 1 2 ( R ( ξ , X ) ξ ) h
and we have an additional term in the formulas. We will use this rule in our calculations without special comments.

2 Local description of 2-dimensional totally geodesic submanifolds in T M 2  

In this section we prove Theorem 1. The proof is given in a series of subsections. Namely, in subsection  2.1 we prove the item (a), in subsection  2.2 we prove the item (c) and finally, in subsection  2.3 we prove Theorem 2 and therefore, the item (b) of Theorem 1.

2.1 Preliminary considerations.

Let F ~ 2   be a submanifold in T M 2   . Let ( x 1 , x 2 ; ξ 1 , ξ 2 )   be a local chart on T M 2   . Then locally F ~ 2   can be given by mapping f   of the form f : { x 1 = x 1 ( u 1 , u 2 ) , x 2 = x 2 ( u 1 , u 2 ) ,
ξ 1 = ξ 1 ( u 1 , u 2 ) ,
ξ 2 = ξ 1 ( u 1 , u 2 ) ,
where u 1 , u 2   are the local parameters on F ~ 2   . The Jacobian matrix f *   of the mapping f   is of the form f * = ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ∂ ξ 2 ∂ u 1 ∂ ξ 2 ∂ u 2 ) .   Since r a n k f * = 2   , we have three geometrically different possibilities to achieve the rank, namely
( a ) det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ) ≠ 0 ; ( b ) det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ) ≠ 0 ;
( c ) det ( ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ∂ ξ 2 ∂ u 1 ∂ ξ 2 ∂ u 2 ) ≠ 0 .
Without loss of generality we can consider these possibilities in a way that (b) excludes (a), and (c) excludes (a) and (b) restricting the considerations to a smaller neighbourhood or even to an open and dense subset.
Case (a). In this case one can locally parameterize the submanifold under consideration as f : { x 1 = u 1 , x 2 = u 1 ,
ξ 1 = ξ 1 ( u 1 , u 2 ) ,
ξ 2 = ξ 2 ( u 1 , u 2 ) ,
and we can consider the submanifold F ~ 2   as an image of the vector field ξ ( u 1 , u 2 )   on the base manifold. Denote F ~ 2   in this case by ξ ( M 2 )   . We analyze this case in subsection  2.2 .
Case (b). In this case one can parameterize the submanifold F 2   as f : { x 1 = u 1 , x 2 = x 2 ( u 1 , u 2 ) ,
ξ 1 = u 2 ,
ξ 2 = ξ 2 ( u 1 , u 2 ) .
Taking into account that we exclude the case (a) in considerations of the case (b), we should set det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ) = det ( 1 0 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ) = ∂ x 2 ∂ u 2 = 0 .   Therefore, x 2 ( u 1 , u 2 )   does not depend on u 2   and the local representation takes the form f : { x 1 = u 1 , x 2 = x 2 ( u 1 ) ,
ξ 1 = u 2 ,
ξ 2 = ξ 2 ( u 1 , u 2 ) .
Remark that π ( F ~ 2 ) = ( u 1 , x 2 ( u 1 )   is a regular curve on M 2   . If we denote this projection by γ ( s )   parameterized by the arc-length parameter and set u 2 : = t   , the local parametrization of F ~ 2   takes the form
γ ( s ) : { x 1 = x 1 ( s ) , x 2 = x 2 ( s ) , ξ ( t , s ) : { ξ 1 = t , ξ 2 = ξ 2 ( t , s ) (5)
We can interpret this kind of submanifolds in T M 2   as a one-parametric family of smooth vector fields over a regular curve on the base manifold.
We will refer to this kind of submanifolds as ruled submanifolds in T M 2   and analyze their totally geodesic property in subsection  2.3 .
Case (c). It this case a local parametrization of F ~ 2   can be given as f : { x 1 = x 1 ( u 1 , u 2 ) , x 2 = x 2 ( u 1 , u 2 ) ,
ξ 1 = u 1 ,
ξ 2 = u 2 .
Taking into account that we exclude the case (b) considering the case (c), we should suppose det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ) = det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 1 0 ) = − ∂ x 1 ∂ u 2 = 0 .   det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ ξ 2 ∂ u 1 ∂ ξ 2 ∂ u 2 ) = det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 0 1 ) = ∂ x 1 ∂ u 1 = 0 .   Thus, we conclude x 1 = c o n s t   . In the same way, we get x 2 = c o n s t   .
Therefore, a submanifold of this kind is nothing else but the fiber, which is evidently totally geodesic and there is nothing to prove.

2.2 Totally geodesic vector fields

In [1] the author has found the conditions on a vector field to generate a totally geodesic submanifold in the tangent bundle. Namely, let ξ   be a vector field on M n   . The submanifold ξ ( M n )   is totally geodesic in T M n   if and only if for any vector fields X , Y   on M n   the following equation holds
r ( X , Y ) ξ + r ( Y , X ) ξ − ∇ h ξ ( X , Y ) ξ = 0 , (6)
where r ( X , Y ) ξ = ∇ X ∇ Y ξ − ∇ ∇ X Y ξ   is ”half” the Riemannian curvature tensor and h ξ ( X , Y ) = R ( ξ , ∇ X ξ ) Y + R ( ξ , ∇ Y ξ ) X   .
It is natural to rewrite this equations in terms of ρ   and e ξ   where e ξ   is a unit vector field and ρ   is the length function of ξ   .
Lemma 2.1 Let ξ = ρ e ξ   be a vector field on a Riemannian manifold M n   . Then ξ ( M n )   is totally geodesic in T M n   if and only if for any vector field X   the following equations hold
{ H e s s ρ ( X , X ) − ρ 2 ( R ( e ξ , ∇ X e ξ ) X ) ( ρ ) − ρ | ∇ X e ξ | 2 = 0 , ρ 3 ∇ R ( e ξ , ∇ X e ξ ) X e ξ − 2 X ( ρ ) ∇ X e ξ − ρ ( r ( X , X ) e ξ + | ∇ X e ξ | 2 e ξ ) = 0 , (7)
where H e s s ρ ( X , X )   is the Hessian of the function ρ   .
Proof. Indeed, the equation ( 6 ) is equivalent to
r ( X , X ) ξ = ∇ R ( ξ , ∇ X ξ ) X ξ , (8)
where X   is an arbitrary vector field. Setting ξ = ρ e ξ   , where e ξ   is a unit vector field, we have
r ( X , X ) ξ = ∇ X ∇ X ( ρ e ξ ) − ∇ ∇ X X ( ρ e ξ ) =
∇ X ( X ( ρ ) e ξ + ρ ∇ X e ξ ) − ( ∇ X X ) ( ρ ) e ξ − ρ ∇ ∇ X X e ξ =
( X ( X ( ρ ) ) − ( ∇ X X ) ( ρ ) ) e ξ + 2 X ( ρ ) ∇ X e ξ + ρ r ( X , X ) e ξ
and ∇ R ( ξ , ∇ X ξ ) X ξ = ρ 2 ( R ( e ξ , ∇ X e ξ ) X ) ( ρ ) e ξ + ρ 3 ∇ R ( e ξ , ∇ X e ξ ) X e ξ .   If we remark that X ( X ( ρ ) ) − ( ∇ X X ) ( ρ ) = d e f H e s s ρ ( X , X )   and for a unit vector field e ξ   < r ( X , X ) e ξ , e ξ > = − | ∇ X e ξ | 2 ,   then we can easily decompose the equation ( 8 ) into components, parallel to and orthogonal to e ξ   , which gives the equations ( 7 ).
Corollary 2.1 Suppose that M n   admits a totally geodesic vector field ξ = ρ e ξ   . Then (a) the function ρ   has no strong maximums; (b) there is a bivector field e 0 ∧ ∇ e 0 e 0   such that e ξ   is parallel along it.
Particulary, if n = 2   then either M 2   is flat or e 0   is a geodesic vector field and ρ   is linear with respect to the natural parameter along each e 0   geodesic line. Moreover, the field ξ   makes a constant angle with each e 0   geodesic line.
Proof. Indeed, for any unit vector field η   consider the linear mapping ∇ Z η | q : T q M n → η q ⊥   , where η q ⊥   is an orthogonal complement to η   in T q M n   . For dimensional reasons it follows that the kernel of this mapping is not empty. In other words, there exists a (unit) vector field e 0   such that ∇ e 0 η = 0   .
Let e 0   be a unit vector field such that ∇ e 0 e ξ = 0   . Then from  7  ( ) 1   we conclude H e s s ρ ( e 0 , e 0 ) = 0   at each point of M n   . Therefore, the Hessian of ρ   can not be positively definite.
Moreover, from  7  ( ) 2   we see that r ( e 0 , e 0 ) e ξ = 0   , which gives ∇ e 0 ∇ e 0 e ξ − ∇ ∇ e 0 e 0 e ξ = − ∇ ∇ e 0 e 0 e ξ = 0   . Setting Z = e 0 ∧ ∇ e 0 e 0   , we get ∇ Z e ξ = 0 .   Suppose now that n = 2   . If Z ≠ 0   then e ξ   is a parallel vector field on M 2   which means that M 2   is flat. If Z = 0   then evidently e 0   is a geodesic vector field. Since in this case H e s s ρ ( e 0 , e 0 ) = e 0 ( e 0 ( ρ ) ) = 0   , we conclude that ρ   is linear with respect to the natural parameter along each e 0   geodesic line.
As concerns the angle function < e 0 , e ξ >   , we have e 0 < e 0 , e ξ > = < ∇ e 0 e 0 , e ξ > + < e 0 , ∇ e 0 e ξ > = 0 .  
Taking into account the Corollary  2.1 , introduce on M 2   a semi-geodesic coordinate system ( u , v )   such that e ξ   is parallel along u   -geodesics. Let
d s 2 = d u 2 + b 2 ( u , v ) d v 2 (9)
be the first fundamental form of M 2   with respect to this coordinate system.
Denote by ∂ 1   and ∂ 2   the corresponding coordinate vector fields. Then the following equations should be satisfied: ∇ ∂ 1 e ξ = 0 , ∂ 1 2 ( ρ ) = 0 .   Introduce the unit vector fields e 1 = ∂ 1 , e 2 = 1 b ∂ 2 .   Then the following rules of covariant derivation are valid
∇ e 1 e 1 = 0 , ∇ e 1 e 2 = 0 , ∇ e 2 e 1 = − k e 2 ∇ e 2 e 2 = k e 1 , (10)
where k   is a (signed) geodesic curvature of v   -curves. Remark that k = − ∂ 1 b b .   With respect to chosen coordinate system, the field ξ   can be expressed as
ξ = ρ ( cos ω e 1 + sin ω e 2 ) , (11)
where ω = ω ( u , v )   is an angle function, i.e.
e ξ = cos ω e 1 + sin ω e 2 .   Introduce a unit vector field ν ξ   by ν ξ = − sin ω e 1 + cos ω e 2 .   Then we can easily find
∇ e 1 e ξ = ∂ 1 ω ν ξ ,
∇ e 2 e ξ = ( e 2 ( ω ) − k ) ν ξ .
Since e ξ   is parallel along u   -curves, we conclude that ∂ 1 ω = 0   , so that ω = ω ( v )   .
Now the problem can be formulated as On a Riemannian 2-manifold with the metric ( 9 ), find a vector field of the form ( 11 ) with
∂ 1 2 ρ = 0 and ω = ω ( v ) (12)
satisfying the equation ( 8 ).
Lemma 2.2 Let M 2   be a Riemannian 2-manifold with the metric ( 9 ) and ξ   be a local vector field on M 2   satisfying ( 12 ). Then ξ   is totally geodesic if and only if
∇ e 2 ∇ e 2 ξ − ( k + c K ) ∇ e 1 ξ = 0 , ∇ e 1 ∇ e 2 ξ + ∇ e 2 ∇ e 1 ξ + ( k + c K ) ∇ e 2 ξ = 0 , (13)
or in a scalar form
{ e 2 ( e 2 ( ρ ) ) − ( k + c K ) e 1 ( ρ ) = ρ λ 2 , e 2 ( c ) = 0 , { 2 e 1 ( e 2 ( ρ ) ) + c K e 2 ( ρ ) = 0 , e 1 ( c ) + c ( k + c K ) = 0 (14)
where λ : = < ∇ e 2 e ξ , ν ξ > = e 2 ( ω ) − k   , c : = ρ 2 λ = ± | ξ ∧ ∇ e 2 ξ |   and K   is the Gaussian curvature of M 2   .
Proof. Indeed,
∇ e 1 ξ = e 1 ( ρ ) e ξ ,
∇ e 2 ξ = e 2 ( ρ ) e ξ + ρ λ ν ξ .
So, taking into account ( 10 ) and ( 12 ), we have
r ( e 1 , e 1 ) ξ = ∇ e 1 ∇ e 1 ξ − ∇ ∇ e 1 e 1 ξ = e 1 ( e 1 ( ρ ) ) e ξ = ∂ 1 2 ρ e ξ = 0 ,
r ( e 1 , e 2 ) ξ = ∇ e 1 ∇ e 2 ξ − ∇ ∇ e 1 e 2 ξ = ∇ e 1 ∇ e 2 ξ ,
r ( e 2 , e 1 ) ξ = ∇ e 2 ∇ e 1 ξ − ∇ ∇ e 2 e 1 ξ = ∇ e 2 ∇ e 1 ξ + k ∇ e 2 ξ ,
r ( e 2 , e 2 ) ξ = ∇ e 2 ∇ e 2 ξ − ∇ ∇ e 2 e 2 ξ = ∇ e 2 ∇ e 2 ξ − k ∇ e 1 ξ .
As concerns the right-hand side of ( 8 ), we have
R ( ξ , ∇ e 1 ξ ) e 1 = 0 , R ( ξ , ∇ e 1 ξ ) e 2 = 0 ,
R ( ξ , ∇ e 2 ξ ) e 1 = ρ 2 λ R ( e ξ , ν ξ ) e 1 = − ρ 2 λ K e 2 ,
R ( ξ , ∇ e 2 ξ ) e 2 = ρ 2 λ R ( e ξ , ν ξ ) e 2 = ρ 2 λ K e 1 .
Therefore, setting X = e 1   in ( 8 ), we obtain an identity. Setting X = e 2   , we have ∇ e 2 ∇ e 2 ξ − k ∇ e 1 ξ = ρ 2 λ K ∇ e 1 ξ .   Setting X = e 1 + e 2   , we obtain r ( e 1 , e 2 ) ξ + r ( e 2 , e 1 ) ξ = − ρ 2 λ K ∇ e 2 ξ ,   which can be reduced to ∇ e 1 ∇ e 2 ξ + ∇ e 2 ∇ e 1 ξ + k ∇ e 2 ξ = − ρ 2 λ K ∇ e 2 ξ .   It remains to mention that | ξ ∧ ∇ e 2 ξ | 2 = | ξ | 2 | ∇ e 2 ξ | 2 − < ξ , ∇ e 2 ξ > 2 = ρ 2 ( e 2 ( ρ ) 2 + ρ 2 λ 2 ) − ( e 2 ( ρ ) ρ ) 2 = ρ 4 λ 2 .   So, if we set c = ρ 2 λ   , we evidently obtain ( 13 ).
Moreover, continuing calculations, we see that
∇ e 2 ∇ e 2 ξ = [ e 2 ( e 2 ( ρ ) ) − ρ λ 2 ] e ξ + [ e 2 ( ρ ) λ + e 2 ( ρ λ ) ] ν ξ =
[ e 2 ( e 2 ( ρ ) ) − ρ λ 2 ] e ξ + 1 ρ e 2 ( c ) ν ξ ,
∇ e 1 ∇ e 2 ξ + ∇ e 2 ∇ e 1 ξ = [ e 2 ( e 1 ( ρ ) ) + e 1 ( e 2 ( ρ ) ) ] e ξ + [ e 1 ( ρ ) λ + e 1 ( ρ λ ) ] ν ξ =
[ e 2 ( e 1 ( ρ ) ) + e 1 ( e 2 ( ρ ) ) ] e ξ + 1 ρ e 1 ( c ) ν ξ .
Taking into account that e 1 ( e 2 ( ρ ) ) − e 2 ( e 1 ( ρ ) ) = k e 2 ( ρ )   , the equations ( 13 ) can be written as
[ e 2 ( e 2 ( ρ ) ) − ρ λ 2 ] e ξ + 1 ρ e 2 ( c ) ν ξ − ( k + c K ) e 1 ( ρ ) e ξ = 0
[ 2 e 1 ( e 2 ( ρ ) ) − k e 2 ( ρ ) ] e ξ + 1 ρ e 1 ( c ) ν ξ + ( k + c K ) [ e 2 ( ρ ) e ξ + ρ λ ν ξ ] = 0
and after evident simplifications we obtain the equations ( 14 ).
Proposition 2.1 Let M 2   be a Riemannian manifold of constant curvature.
Suppose ξ   is a non-zero local vector field on M 2   such that ξ ( M 2 )   is totally geodesic in T M 2   . Then M 2   is flat.
Proof. Let M 2   a Riemannian manifold of constant curvature K ≠ 0   . Then the function b   in ( 9 ) should satisfy the equation − ∂ 11 b b = K .   The general solution of this equation can be expressed in 3 forms:
  • (a) b ( u , v ) = A ( v ) cos ( u / r + θ ( v ) )   or b ( u , v ) = A ( v ) sin ( u / r + θ ( v ) )   for K = 1 / r 2 > 0   ;
  • (b) b ( u , v ) = A ( v ) cosh ( u / r + θ ( v ) )   or b ( u , v ) = A ( v ) sinh ( u / r + θ ( v ) )   for K = − 1 / r 2 < 0   ;
  • (c) b ( u , v ) = A ( v ) e u / r   for K = − 1 / r 2 < 0   ;
Evidently, we may set A ( v ) ≡ 1   (making a v   -parameter change) in each of these cases.
The equation  14  ( ) 2   means that c   does not depend on v   . Since K   is constant, the equation  14  ( ) 4   implies e 2 ( k ) = 0 .   If we remark that k = − ∂ 1 b b   then one can easily find θ ( v ) = c o n s t   in cases ( a )   and ( b )   .
After a u   -parameter change, the function b   takes one of the forms
  • (a) b ( u , v ) = cos ( u / r )   or b ( u , v ) = sin ( u / r )   for K = 1 / r 2 > 0   ;
  • (b) b ( u , v ) = cosh ( u / r )   or b ( u , v ) = sinh ( u / r )   for K = − 1 / r 2 < 0   ;
  • (c) b ( u , v ) = e u / r   for K = − 1 / r 2 < 0   ;
From the equation  14  ( ) 4   we find c K = − e 1 ( c ) c − k = − e 1 ( c ) c + e 1 ( b ) b = e 1 ( ln b / c ) .   Suppose first that e 2 ( ρ ) ≠ 0   . Multiplying  14  ( ) 3   by e 2 ( ρ )   we can easily solve this equation with respect to e 2 ( ρ )   by a chain of simple transformations:
2 e 2 ( ρ ) ⋅ e 1 ( e 2 ( ρ ) ) + e 1 ( ln b / c ) ⋅ [ e 2 ( ρ ) 2 ] = 0 ,
e 1 [ e 2 ( ρ ) 2 ] + e 1 ( ln b / c ) ⋅ [ e 2 ( ρ ) 2 ] = 0 ,
e 1 [ e 2 ( ρ ) 2 ] e 2 ( ρ ) 2 + e 1 ( ln b / c ) = 0 ,
e 1 [ ln e 2 ( ρ ) 2 ] + e 1 ( ln b / c ) = 0 ,
e 1 ( ln [ e 2 ( ρ ) 2 b / c ] ) = 0
and therefore, e 2 ( ρ ) 2 b / c = h ( v ) 2   or ∂ 2 ρ = h ( v ) c b .   Since ρ   is linear with respect to the u   -parameter, say ρ = a 1 ( v ) u + a 2 ( v )   , then ∂ 2 ρ = a 1 ′ u + a 2 ′   and therefore c b   is also linear with respect to u   , namely c b = m 1 ( v ) u + m 2 ( v ) = a 1 ′ h u + a 2 ′ h   . But the functions c   and b   do not depend on v   . Therefore m 1   and m 2   are constants, so a 1 = m 1 ∫ h ( v ) d v , a 2 = m 2 ∫ h ( v ) d v   . Thus c b = m 1 u + m 2 .   Now the function c   takes the form c ( u ) = ( m 1 u + m 2 ) 2 b   and therefore e 1 ( c ) = 2 m 1 ( m 1 u + m 2 ) b − ( m 1 u + m 2 ) 2 ∂ 1 b b 2 .   Substitution into  14  ( ) 4   gives 2 m 1 ( m 1 u + m 2 ) b − 2 ( m 1 u + m 2 ) 2 ∂ 1 b b 2 + ( m 1 u + m 2 ) 4 b 2 K = 0   or ( m 1 u + m 2 ) b 2 [ 2 m 1 b − 2 ( m 1 u + m 2 ) ∂ 1 b + ( m 1 u + m 2 ) 3 K ] = 0 .   The expression in brackets is an algebraic one and can not be identically zero if K ≠ 0   . Therefore m 1 = m 2 = 0   and hence ρ 2 λ : = c = 0   . But this identity implies λ = 0   or ρ = 0   . If λ = 0   then e ξ   is a parallel unit vector field and therefore, M 2   is flat and we come to a contradiction. Therefore ρ = 0   .
Remark. If K = 0   , we can not conclude that c = 0   . In this case the expression in brackets can be identically zero for m 1 = 0   and b = c o n s t   .
And we have c = m 2 = c o n s t   .
Suppose now that e 2 ( ρ ) = 0   . Then ρ = a 1 u + a 2 ,   where a 1 , a 2   are constants and we obtain the following system
− ( k + c K ) ∂ 1 ρ = ρ λ 2 , ∂ 2 c = 0 , ∂ 1 c + c ( k + c K ) = 0 . (15)
If ∂ 1 ρ = 0   then immediately ρ = 0   or λ = 0   . The identity λ = 0   implies K = 0   as above. Therefore, ρ = 0   .
Suppose ∂ 1 ρ ≠ 0   or equivalently a 1 ≠ 0   . Then from  15  ( ) 1   we get
( k + c K ) = − ρ λ 2 a 1 (16)
Since c = ρ 2 λ   , from  15  ( ) 2   we see that ∂ 2 λ = 0   or ∂ 2 [ ∂ 2 ω + ∂ 1 b b ] = 0   . Since b   does not depend on v   , we have ∂ 22 ω = 0   or equivalently ∂ 2 ω = α = c o n s t   .
Thus, λ = α + ∂ 1 b b   .
Now we can find ∂ 1 c   in two ways. First, from  15  ( ) 3   using ( 16 ) and keeping in mind that c = ρ 2 λ   :
∂ 1 c = c ρ λ 2 a 1 = ρ 3 λ 3 a 1   Second, directly:
∂ 1 c = 2 ρ ∂ 1 ρ λ + ρ 2 ∂ 1 λ .   It is easy to see that ∂ 1 λ = k λ − K   and hence we get ∂ 1 c = 2 a 1 ρ λ + ρ 2 ( k λ − K ) .   Equalizing, we have 2 a 1 ρ λ + ρ 2 ( k λ − K ) − ρ 3 λ 3 a 1 = 0   or ρ a 1 [ 2 a 1 2 λ + a 1 ρ ( k λ − K ) − ρ 2 λ 3 ] = 0 .   The expression in brackets is an algebraic one and can not be identically zero for K ≠ 0   . Since ρ ≠ 0   , we obtain a contradiction.
Remark. We do not obtain a contradiction if K = 0   , since we have another solution λ = 0   which gives ∂ 1 b + α = 0   and hence b = − α u + m   .
We have achieved the result by putting a restriction on the geometry of the base manifold. Putting a restriction on the vector field we are able to achieve a similar result. Recall that a totally geodesic vector field necessarily makes a constant angle with some family of geodesics on the base manifold ( see Corollary  2.1 ). It is not parallel along this family and this fact is essential for its totally geodesic property. Namely,
Proposition 2.2 Let M 2   be a Riemannian manifold. Suppose ξ   is a non-zero local vector field on M 2   which is parallel along some family of geodesics of M 2   . If ξ ( M 2 )   is totally geodesic in T M 2   then M 2   is flat.
Remark. Geometrically, this assertion means that if ξ ( M 2 )   is not transversal to the horizontal distribution on T M 2   then ξ ( M 2 )   is never totally geodesic in T M 2   except when M 2   is flat.
Proof. Let M 2   be a non-flat Riemannian manifold and suppose that the hypothesis of the theorem is fulfilled. Then, choosing a coordinate system as in Lemma  2.2 , we have ∇ e 1 ξ = 0   and we can reduce ( 13 ) to
∇ e 2 ∇ e 2 ξ = 0 , ∇ e 1 ∇ e 2 ξ + ( k + c K ) ∇ e 2 ξ = 0 . (17)
Now make a simple computation.
R ( e 2 , e 1 ) ∇ e 2 ξ = ∇ e 2 ∇ e 1 ∇ e 2 ξ − ∇ e 1 ∇ e 2 ∇ e 2 ξ − ∇ [ e 2 , e 1 ] ∇ e 2 ξ =
∇ e 2 ∇ e 1 ∇ e 2 ξ − k ∇ e 2 ∇ e 2 ξ = ∇ e 2 ∇ e 1 ∇ e 2 ξ .
On the other hand, differentiating  17  ( ) 2   , we find ∇ e 2 ∇ e 1 ∇ e 2 ξ = − e 2 ( k + c K ) ∇ e 2 ξ .   So we have R ( e 2 , e 1 ) ∇ e 2 ξ = − e 2 ( k + c K ) ∇ e 2 ξ .   Therefore, either ∇ e 2 ξ = 0   or e 2 ( k + c K ) = 0   . If we accept the first case we see that ξ   is a parallel vector field on M 2   and we get a contradiction.
If we accept the second case, we obtain R ( e 2 , e 1 ) ∇ e 2 ξ = 0 ,   which means that ∇ e 2 ξ   belongs to a kernel of the curvature operator of M 2   .
In dimension 2 this means that M 2   is flat or, equivalently, ξ   is a parallel vector field and we obtain a contradiction, as well.

2.3 Ruled totally geodesic submanifolds in T M 2  

Proposition 2.3 Let M 2   be a Riemannian manifold of sign-preserving curvature. Consider a ruled submanifold F ~ 2   in T M 2   given locally by γ ( s ) : { x 1 = x 1 ( s ) , x 2 = x 2 ( s ) , ξ ( t , s ) : { ξ 1 = t , ξ 2 = ξ 2 ( t , s ) .   Then F ~ 2   is totally geodesic in T M 2   if γ ( s )   is a geodesic in M 2   , ξ ( t , s ) = t ρ ( s ) e ( s ) ,   where e ( s )   is a unit vector field which is parallel along γ   and ρ ( s )   is an arbitrary smooth function.
Remark. Geometrically, F ~ 2   is a cylinder-type surface based on geodesic γ ( s )   with elements directed by a unit vector field e ( s )   parallel along γ ( s )   .
Proof. Fixing s = s 0   , we see that F 2   meets the fiber over x 1 ( s 0 ) , x 2 ( s 0 )   by a curve ξ ( t , s 0 )   . If F 2   is supposed to be totally geodesic, then this curve is a straight line on the fiber. Therefore, the family ξ ( t , s )   should be of the form ξ ( t , s ) : { ξ 1 = t , ξ 2 = α ( s ) t + β ( s ) .   Introduce two vector fields given along γ ( s )   by
a = ∂ 1 + α ( s ) ∂ 2 , b = β ( s ) ∂ 2 . (18)
Then we can represent ξ ( t , s )   as ξ ( t , s ) = a ( s ) t + b ( s ) .   Denote by τ   and ν   the vectors of the Frenet frame of the curve γ ( s )   .
Denote also by ( ′   ) the covariant derivative of vector fields with respect to the arc-length parameter on γ ( s )   . Then { τ ′ = k ν , ν ′ = − k τ .   Denote by ∂ ~ 1 , ∂ ~ 2   the s   and t   coordinate vector fields on F 2   respectively.
A simple calculation yields ∂ ~ 1 = τ h + ( ξ ′ ) v , ∂ ~ 2 = a v .   One of the unit normal vector fields can be found immediately, namely N ~ 1 = ν h   . Consider the conditions on F 2   to be totally geodesic with respect to the normal vector field N ~ 1   . Using formulas ( 4 ), ∇ ~ ∂ ~ 1 N ~ 1 = ∇ ~ τ h + ( ξ ′ ) v ν h = − k τ h − 1 2 [ R ( τ , ν ) ξ ] v + 1 2 [ R ( ξ , ξ ′ ) ν ] h   Therefore, < < ∇ ~ ∂ ~ 1 N ~ 1 , ∂ ~ 2 > > = − 1 2 < R ( τ , ν ) ξ , a > = − 1 2 < R ( τ , ν ) b , a > = 0 .   Since M 2   is supposed to be non-flat, it follows b ∧ a = 0   . From ( 18 ) we conclude b = 0   . Thus, ξ ( t , s ) = a ( s ) t   . Moreover,
< < ∇ ~ ∂ ~ 1 N ~ 1 , ∂ ~ 1 > > = − k − 1 2 < R ( τ , ν ) ξ , ξ ′ > + 1 2 < R ( ξ , ξ ′ ) ν , τ > =
− k + < R ( ξ , ξ ′ ) ν , τ > = − k + t 2 < R ( a , a ′ ) ν , τ > = 0
identically with respect to parameter t   . Therefore, k = 0   and a ∧ a ′ = 0   .
Thus, γ ( s )   is a geodesic line on M 2   . In addition, ( a ∧ a ′ = 0 ) ∼ ( a ′ = λ a )   .
Set a = ρ ( s ) e ( s )   , where ρ = | a ( s ) |   . Then ( a ′ = λ a ) ∼ ( ρ ′ e + ρ e ′ = λ ρ e )   , which means that e ′ = 0   . From this we conclude ξ ( t , s ) = t ρ ( s ) e ( s ) ,   where ρ ( s )   is arbitrary function and e ( s )   is a unit vector field, parallel along γ ( s )   . Therefore, ∂ ~ 1 = τ h + t ρ ′ e v , ∂ ~ 2 = ρ e v   and we can find another unit normal vector field N ~ 2 = ( e ⊥ ) v   , where e ⊥ ( s )   is a unit vector field also parallel along γ ( s )   and orthogonal to e ( s )   . For this vector field we have
∇ ~ ∂ ~ 1 N ~ 2 = ∇ ~ τ h + ( ξ ′ ) v ( e ⊥ ) v = [ ( e ⊥ ) ′ ] v + 1 2 [ R ( ξ , e ⊥ ) τ ] h = 1 2 t ρ [ R ( e , e ⊥ ) τ ] h ,
∇ ~ ∂ ~ 2 N ~ 2 = 0
Evidently, < < ∇ ~ ∂ ~ i N ~ 2 , ∂ ~ k > > = 0   for all i , k = 1 , 2   . Thus, the submanifold is totally geodesic.
The converse statement is true in general.
Proposition 2.4 Let M n   be a Riemannian manifold. Consider a cylinder type surface F ~ 2 ⊂ T M n   parameterized as { γ ( s ) , t ρ ( s ) e ( s ) } ,   where γ ( s )   is a geodesic in M n   , e ( s )   is a unit vector field, parallel along γ   and ρ ( s )   is an arbitrary smooth function. Then F ~ 2   is totally geodesic in T M n   and intrinsically flat.
Proof. Indeed, the tangent basis of F ~ 2   is consisted of ∂ ~ 1 = γ ′ h + t ρ ′ e v , ∂ ~ 2 = ρ e v .   By formulas ( 4 ),
∇ ~ ∂ ~ 1 ∂ ~ 1 = ( ∇ γ ′ γ ′ ) h + 1 2 [ R ( γ ′ , γ ′ ) ξ ] v = 0 ,
∇ ~ ∂ ~ 1 ∂ ~ 2 = ( ∇ γ ′ ρ e ) v + 1 2 [ R ( ξ , ρ e ) γ ′ ] h = ρ ′ e v ∼ ∂ ~ 2 ,
∇ ~ ∂ ~ 2 ∂ ~ 1 = 1 2 [ R ( ξ , ρ e ) γ ′ ] h = 1 2 [ R ( t ρ e , ρ e ) γ ′ ] h = 0
∇ ~ ∂ ~ 2 ∂ ~ 2 = ρ e v ( ρ ) e v = 0 .
It is easy to find the Gaussian curvature of this submanifold, since it is equal to the sectional curvature of T M 2   along the ∂ ~ 1 ∧ ∂ ~ 2   plane. Using the curvature tensor expressions [4] , we find G a u s s ( F ~ 2 ) = < < R ~ ( τ h , e v ) e v , τ h > > = 1 4 | R ( ξ , e ) τ | 2 = 0 .  

3 Local description of 3-dimensional totally geodesic submanifolds in T M 2  

Theorem 3.1 Let M 2   be Riemannian manifold with Gaussian curvature K   . A totally geodesic submanifold F ~ 3 ⊂ T M 2   locally is either a) a 3-plane in T M 2 = E 4   if K = 0   , or b) a restriction of the tangent bundle to a geodesic γ ∈ M 2   such that K | γ = 0   if K ≢ 0   . If M 2   does not contain such a geodesic, then T M 2   does not admit 3-dimensional totally geodesic submanifolds.
Proof. Let F ~ 3   be a submanifold it T M 2   . Let ( x 1 , x 2 ; ξ 1 , ξ 2 )   be a local chart on T M 2   . Then locally F ~ 3   can be given mapping f   of the form f : { x 1 = x 1 ( u 1 , u 2 , u 3 ) , x 2 = x 2 ( u 1 , u 2 , u 3 ) , ξ 1 = ξ 1 ( u 1 , u 2 , u 3 ) , ξ 2 = ξ 2 ( u 1 , u 2 , u 3 ) ,   where u 1 , u 2 , u 3   are the local parameters on F ~ 3   . The Jacobian matrix f *   of the mapping f   is of the form f * = ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 1 ∂ u 3 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ∂ x 2 ∂ u 3 ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ∂ ξ 1 ∂ u 3 ∂ ξ 2 ∂ u 1 ∂ ξ 2 ∂ u 2 ∂ ξ 2 ∂ u 3 ) .   Since r a n k f * = 3   , we have two geometrically different possibilities to achieve the rank, namely ( a ) det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 1 ∂ u 3 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ∂ x 2 ∂ u 3 ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ∂ ξ 1 ∂ u 3 ) ≠ 0 ; ( b ) det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 1 ∂ u 3 ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ∂ ξ 1 ∂ u 3 ∂ ξ 2 ∂ u 1 ∂ ξ 2 ∂ u 2 ∂ ξ 2 ∂ u 3 ) ≠ 0 .   Without loss of generality we can consider this possibilities in such a way that (b) excludes (a).
Consider the case (a). In this case we can locally parameterize the submanifold F 3   as f : { x 1 = u 1 , x 2 = u 2 , ξ 1 = u 3 , ξ 2 = ξ 2 ( u 1 , u 2 , u 3 ) .   By hypothesis, the submanifold F ~ 3   is totally geodesic in T M 2   . Therefore, it intersects each fiber of T M 2   by a vertical geodesic, i.e. by a straight line.
Fix u 0 = ( u 0 1 , u 0 2 )   . Then the parametric equation of F ~ 3 ∩ T u 0 M 2   with respect to fiber parameters is { ξ 1 = u 3 , ξ 2 = ξ 2 ( u 0 1 , u 0 2 , u 3 ) .   On the other hand, this equation should be the equation of a straight line and hence { ξ 1 = u 3 , ξ 2 = α ( u 0 1 , u 0 2 ) u 3 + β ( u 0 1 , u 0 2 ) ,   where α ( u ) = α ( u 1 , u 2 )   and β ( u ) = β ( u 1 , u 2 )   some smooth functions on M 2   . From this viewpoint, after setting u 3 = t   the submanifold under consideration can be locally represented as a one-parametric family of smooth vector fields ξ t   on M 2   of the form ξ t ( u ) = t ∂ 1 + ( α ( u ) t + β ( u ) ) ∂ 2   with respect to the coordinate frame ∂ 1 = ∂ / ∂ u 1 , ∂ 2 = ∂ / ∂ u 2   .
Introduce the vector fields
a ( u ) = ∂ 1 + α ( u ) ∂ 2 , b ( u ) = β ( u ) ∂ 2 . (19)
Then ξ t   can be expressed as ξ t ( u ) = t a ( u ) + b ( u ) .   It is natural to denote by ξ t ( M 2 )   a submanifold F ~ 3 ⊂ T M 2   of this kind.
Denote by ∂ ~ i ( i = 1 , . . . , 3 )   the coordinate vector fields of ξ t ( M 2 )   . Then
∂ ~ 1 = { 1 , 0 , 0 , t ∂ 1 α + ∂ 1 β } ,
∂ ~ 2 = { 0 , 1 , 0 , t ∂ 2 α + ∂ 2 β } ,
∂ ~ 3 = { 0 , 0 , 1 , α } .
A direct calculation shows that these fields can be represented as
∂ ~ 1 = ∂ 1 h + t ( ∇ ∂ 1 a ) v + ( ∇ ∂ 1 b ) v ,
∂ ~ 2 = ∂ 2 h + t ( ∇ ∂ 2 a ) v + ( ∇ ∂ 2 b ) v ,
∂ ~ 3 = a v .
Denote by N ~   a normal vector field of ξ t ( M 2 )   . Then N ~ = ( a ⊥ ) v + Z t h ,   where < a ⊥ , a > = 0   and the field Z t = Z t 1 ∂ 1 + Z t 2 ∂ 2   can be found easily from the equations
< < ∂ ~ i , N ~ > > = < Z t , ∂ i > + t < ∇ ∂ i a , a ⊥ > + < ∇ ∂ i b , a ⊥ > = 0 ( i = 1 , 2 )
Using the formulas ( 4 ), one can find
∇ ~ ∂ ~ i a v = ∇ ~ ∂ i h + t ( ∇ ∂ i a ) v + ( ∇ ∂ i b ) v a v =
( ∇ ∂ i a ) v + 1 2 [ R ( ξ t , a ) ∂ i ] h = ( ∇ ∂ i a ) v + 1 2 [ R ( b , a ) ∂ i ] h .
If the submanifold ξ t ( M 2 )   is totally geodesic, then the following equations should be satisfied identically < < ∇ ~ ∂ ~ i ∂ ~ 3 , N ~ > > = < ∇ ∂ i a , a ⊥ > + 1 2 < R ( b , a ) ∂ i , Z t > = 0   with respect to the parameter t   . To simplify the further calculations, suppose that the coordinate system on M 2   is the orthogonal one, so that < ∂ 1 , ∂ 2 > = 0   and R ( b , a ) ∂ 2 = g 11 K | b ∧ a | ∂ 1 , R ( b , a ) ∂ 1 = − g 22 K | b ∧ a | ∂ 2 ,   where K   is the Gaussian curvature of M 2   and g 11 , g 22   are the contravariant metric coefficients. Then we have
< R ( b , a ) ∂ 1 , Z t > = − g 22 K | b ∧ a | < Z t , ∂ 2 > =
g 22 K | b ∧ a | ( t < ∇ ∂ 2 a , a ⊥ > + < ∇ ∂ 2 b , a ⊥ > ) ,
< R ( b , a ) ∂ 2 , Z t > = g 11 K | b ∧ a | < Z t , ∂ 1 > =
− g 11 K | b ∧ a | ( t < ∇ ∂ 1 a , a ⊥ > + < ∇ ∂ 1 b , a ⊥ > ) .
Thus we get the system { g 22 K | b ∧ a | < ∇ ∂ 2 a , a ⊥ > t + < ∇ ∂ 1 a , a ⊥ > + g 22 K | b ∧ a | < ∇ ∂ 2 b , a ⊥ > = 0 , g 11 K | b ∧ a | < ∇ ∂ 1 a , a ⊥ > t − < ∇ ∂ 2 a , a ⊥ > + g 11 K | b ∧ a | < ∇ ∂ 1 b , a ⊥ > = 0 ,   which should be satisfied identically with respect to t   . As a consequence, we have 3 cases:
  • (i) K = 0 , { < ∇ ∂ 1 a , a ⊥ > = 0 , < ∇ ∂ 2 a , a ⊥ > = 0 ;  
  • (ii) K ≠ 0   , | b ∧ a | = 0 , { < ∇ ∂ 1 a , a ⊥ > = 0 , < ∇ ∂ 2 a , a ⊥ > = 0 ;  
  • (iii) K ≠ 0   , | b ∧ a | ≠ 0   , { < ∇ ∂ 1 a , a ⊥ > = 0 , < ∇ ∂ 2 a , a ⊥ > = 0 , { < ∇ ∂ 1 b , a ⊥ > = 0 , < ∇ ∂ 2 b , a ⊥ > = 0 ;  
Case (i). In this case the base manifold is flat and we can choose a Cartesian coordinate system, so that the covariant derivation becomes a usual one and we have { ∇ ∂ i a = { 0 , ∂ i α } ( i = 1 , 2 ) a ⊥ = { − α , 1 }   From < ∇ ∂ i a , a ⊥ > = 0   it follows that α = c o n s t   , i.e. a   is a parallel vector field. Moreover, in this case
∂ ~ 1 = { 1 , 0 , 0 , ∂ 1 β } = ∂ 1 h + ( ∂ 1 b ) v ,
∂ ~ 2 = { 0 , 1 , 0 , ∂ 2 β } = ∂ 1 h + ( ∂ 1 b ) v ,
∂ ~ 3 = { 0 , 0 , 1 , α } ,
N ~ = { − ∂ 1 β , − ∂ 2 β , − α , 1 } .
Now we can find ∇ ~ ∂ ~ i ∂ ~ k = ( ∇ ∂ i ∂ k b ) v = { 0 , 0 , 0 , ∂ i k β }   and the conditions < < ∇ ~ ∂ ~ i ∂ ~ k , N ~ > > = 0   imply ∂ i k β = 0   . Thus, β = m 1 u 1 + m 2 u 2 + m 0   , where m 1 , m 2 , m 0   are arbitrary constants. As a consequence, the submanifold ξ t ( M 2 )   is described by parametric equations of the form { x 1 = u 1 , x 2 = u 2 , ξ 1 = t , ξ 2 = α t + m 1 u 1 + m 2 u 2 + m 0   and we have a hyperplane in T M 2 = E 4   .
Case(ii). Keeping in mind ( 19 ), the condition b ∧ a = 0   implies b = 0   .
The conditions { < ∇ ∂ 1 a , a ⊥ > = 0 , < ∇ ∂ 2 a , a ⊥ > = 0   imply ∇ ∂ 1 a = λ 1 ( u ) a , ∇ ∂ 2 a = λ 2 ( u ) a   . As a consequence, we have
ξ t = t a
∂ ~ 1 = ∂ 1 h + t ( ∇ ∂ 1 a ) v = ∂ 1 h + t λ 1 a v ,
∂ ~ 2 = ∂ 2 h + t ( ∇ ∂ 2 a ) v = ∂ 2 h + t λ 2 a v ,
∂ ~ 3 = a v ,
N ~ = ( a ⊥ ) v .
Using formulas ( 4 ),
∇ ~ ∂ ~ i ∂ ~ k = ∇ ~ ∂ i h + t λ i a v ( ∂ k h + t λ k a v ) =
∇ ~ ∂ i h ∂ k h + t λ i ∇ ~ a v ∂ k h + ∇ ~ ∂ i h ( t λ k a v ) + t 2 λ i λ k ∇ ~ a v a v =
( ∇ ∂ i ∂ k ) h − 1 2 [ R ( ∂ i , ∂ k ) ξ t ] v + t λ i 1 2 [ R ( ξ t , a ) ∂ k ] h +
t ∂ i ( λ k ) a v + t λ k ( ∇ ∂ i a ) v + t λ k 1 2 [ R ( ξ t , a ) ∂ i ] h =
( ∇ ∂ i ∂ k ) h − t 1 2 [ R ( ∂ i , ∂ k ) a ] v + t ∂ i ( λ k ) a v + t λ k λ i a v .
Evidently, for i ≠ k   < < ∇ ~ ∂ ~ i ∂ ~ k , N ~ > > = − t 1 2 < R ( ∂ i , ∂ k ) a , a ⊥ > ≠ 0 ,   since M 2   is non-flat and a ≠ 0   . Contradiction. Case (iii). The conditions imply ∇ i a = λ i ( u ) a , ∇ i b = μ i ( u ) a ( i = 1 , 2 )   and we have
ξ t = t a + b
∂ ~ 1 = ∂ 1 h + ( t λ 1 + μ 1 ) a v ,
∂ ~ 2 = ∂ 1 h + ( t λ 2 + μ 2 ) a v ,
∂ ~ 3 = a v ,
N ~ = ( a ⊥ ) v .
A calculation as above leads to the identity
< < ∇ ~ ∂ ~ i ∂ ~ k , N ~ > > = − 1 2 < R ( ∂ i , ∂ k ) ξ t , a ⊥ > = − t 1 2 < R ( ∂ i , ∂ k ) a , a ⊥ > − 1 2 < R ( ∂ i , ∂ k ) b , a ⊥ > = 0  
which can be true if and only if { < R ( ∂ i , ∂ k ) a , a ⊥ > = 0 , < R ( ∂ i , ∂ k ) b , a ⊥ > = 0 .   The first condition contradicts K ≠ 0   .
Consider the case (b). In this case the submanifold F ~ 3   can be locally parametrized by { x 1 = u 1 , x 2 = x 2 ( u 1 , u 2 , u 3 ) ξ 1 = u 2 , ξ 2 = u 3 .   Since we exclude the case (a), we should suppose det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 1 ∂ u 3 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ∂ x 2 ∂ u 3 ∂ ξ 1 ∂ u 1 ∂ ξ 1 ∂ u 2 ∂ ξ 1 ∂ u 3 ) = det ( 1 0 0 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ∂ x 2 ∂ u 3 0 1 0 ) = − ∂ x 2 ∂ u 3 = 0 ;   det ( ∂ x 1 ∂ u 1 ∂ x 1 ∂ u 2 ∂ x 1 ∂ u 3 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ∂ x 2 ∂ u 3 ∂ ξ 2 ∂ u 1 ∂ ξ 2 ∂ u 2 ∂ ξ 2 ∂ u 3 ) = det ( 1 0 0 ∂ x 2 ∂ u 1 ∂ x 2 ∂ u 2 ∂ x 2 ∂ u 3 0 0 1 ) = ∂ x 2 ∂ u 2 = 0 ;   Therefore, in this case we have a submanifold, which can be parametrized by { x 1 = x 1 ( s ) , x 2 = x 2 ( s ) ξ 1 = u 2 , ξ 2 = u 3 ,   where s   is a natural parameter of the regular curve γ ( s ) = { x 1 ( s ) , x 2 ( s ) }   on M 2   . Geometrically, a submanifold of this class is nothing else but the restriction of T M 2   to the curve γ ( s )   . Denote by τ   and ν   the Frenet frame of γ ( s )   . It is easy to verify that ∂ ~ 1 = τ h , ∂ ~ 2 = ∂ 1 v , ∂ ~ 3 = ∂ 2 v , N ~ = ν h .   By formulas ( 4 ), for i = 1 , 2   < < ∇ ~ ∂ ~ 1 + i N ~ , ∂ ~ 1 > > = < < ∇ ~ ∂ i v ν h , τ h > > = 1 2 < R ( ξ , ∂ i ) ν , τ > = 1 2 < R ( τ , ν ) ∂ i , ξ > = 0   for arbitrary ξ   . Evidently, M 2   must be flat along γ ( s )   .
References

  1. Abbassi M.T.K., Yampolsky A. Totally geodesic submanifolds transverse to fibers of the tangent bundle of a Riemannian manifold with Sasaki metric, Math. Publ. Debrecen 64/1-2 (2004), 129–154.
  2. Borisenko A., Yampolsky A. Riemannian greometry of bundles, Uspechi Math. Nauk 46/6 (1991), 51-95 (Russian). Engl. transl. Russian Math Surveys, 46/6 (1991), 55–106.
  3. Dombrowski P. On the geometry of tangent bundle, J. Reine Angew. Math. 210/1-2 (1962), 73–88.
  4. Kowalski O. Curvature of the induced Riemannian metric on the tangent bundle of a Riemannian manifold, J. Reine Angew. Math. 250 (1971), 124–129.
  5. Liu M.-S. Affine maps of tangent bundles with Sasaki metric, Tensor N.S. 28 (1974), 34–42.
  6. Nagy P. On the tangent sphere bundlre of a Riemannian 2-manifold, Tohôku Math. J. 29 (1977), 203–208.
  7. Sasaki S. On the differential geometry of tangent bundles of Riemannian manifolds, Tohôku Math. J. 10 (1958), 338–354.
  8. Sasaki S. Geodesics on the tangent sphere bundle over space forms, J. Reine Angew. Math 288 (1976), 106–120
  9. Sato K. Geodesics on the tangent bundle over space forms, Tensor N.S. 32 (1978), 5–10.
  10. Walczak P. On totally geodesic submanifolds of the tangent bundles with Sasaki metric, Bull. Acad Pol. Sci., ser Sci. Math. 28/3-4 (1980), 161–165.
  11. Yampolsky A. On totally geodesic vector fields on a submanifold, Matematicheskaya fizika, analiz, geometriya 1/3-4 (1994), 540–545 (Russian).
  12. Yampolsky A. On the intrinsic geometry of a unit vector field, Comment. Math. Univ. Carolinae 43/2 (2002), 299–317.
  13. Yampolsky A. On extrinsic geometry of unit normal vector field of Riemannian hyperfoliation, Math. Publ. Debrecen 63/4 (2003), 555 – 567.
  14. Yampolsky A. Full description of totally geodesic unit vector fields on 2-dimensional Riemannian manifolds, Matematicheskaya fizika, analiz, geometriya 11/3 (2004), 355–365.