On the intrinsic geometry of a unit vector field *   *   Comment. Mat. Univ. Carolinae 43, 2 (2002), 299-317.

Yampolsky A.

Abstract
We study the geometrical properties of a unit vector field on a Riemannian 2-manifold, considering the field as a local imbedding of the manifold into its tangent sphere bundle with the Sasaki metric. For the case of constant curvature K   , we give a description of the totally geodesic unit vector fields for K = 0   and K = 1   and prove a non-existence result for K ≠ 0 , 1   . We also found a family ξ ω   of vector fields on the hyperbolic 2-plane L 2   of curvature − c 2   which generate foliations on T 1 L 2   with leaves of constant intrinsic curvature − c 2   and of constant extrinsic curvature − c 2 4   .
Keywords: Sasaki metric, vector field, sectional curvature, totally geodesic submanifolds.
AMS subject class: Primary 53B25, 53C42; Secondary 46E25, 20C20 Introduction A unit vector field ξ   on a Riemannian manifold M   is called holonomic if ξ   is a field of normals of some family of regular hypersurfaces in M   and non-holonomic otherwise. The geometry of non-holonomic unit vector fields has been developed by A.Voss at the end of the 19-th century. The foundations of this theory can be found in [1] . Recently, the geometry of a unit vector field has been considered from another point of view. Namely, let T 1 M   be the unit tangent sphere bundle of M   endowed with the Sasaki metric [9] . If ξ   is a unit vector field on M   , then one may consider ξ   as a mapping ξ : M → T 1 M   so that the image ξ ( M )   is a submanifold in T 1 M   with the metric induced from T 1 M   . So, one may apply the methods from the study of the geometry of submanifolds to determine geometrical characteristics of a unit vector field. For example, the unit vector field ξ   is said to be minimal if ξ ( M )   is of minimal volume with respect to the induced metric [6] . A number of examples of locally minimal vector unit fields has been found (see [2, 3, 7] ). On the other hand, using the geometry of submanifolds, we may find the Riemannian, Ricci or scalar curvature of a unit vector field using the second fundamental form of the submanifold ξ ( M ) ∈ T 1 M   found in [11] . In this paper we apply this approach to the simplest case when the base space is 2-dimensional and hence the submanifold ξ ( M ) ∈ T 1 M   is a hypersurface.
Aknowlegement. The author expresses his thanks to E.Boeckx for valuable remarks and the referee for careful reading and corrections.

1 The results

Let ξ   be a given unit vector field. Denote by e 0   a unit vector field such that ∇ e 0 ξ = 0   . Denote by e 1   a unit vector field, orthogonal to e 0   , such that ∇ e 1 ξ = λ η ,   where η   is a unit vector field, orthogonal to ξ   . The function λ   is a signed singular value of a linear operator ∇ ξ : T M → ξ ⊥   (acting as ( ∇ ξ ) X = ∇ X ξ   ).
Set ∇ ξ ξ = k η , ∇ η η = κ ξ .   The functions k   and κ   are the signed geodesic curvatures of the integral curves of the fields ξ   and η   respectively. We prove that λ 2 = k 2 + κ 2   .
Denote the signed geodesic curvatures of the integral curves of the fields e 0   and e 1   as μ   and σ   respectively. Then ∇ e 0 e 0 = μ e 1 , ∇ e 1 e 1 = σ e 0 .   The rientations of the frames ( ξ , η )   and ( e 0 , e 1 )   are independent. Set s = 1   if the orientations are coherent and s = 0   otherwise.
The following result ( Lemma  3.2 ) is a basic tool for the study.
Let M   be a 2-dimensional Riemannian manifold of Gaussian curvature K   .
The second fundamental form Ω   of the submanifold ξ ( M ) ⊂ T 1 M   is given by Ω = [ − μ λ 1 + λ 2 ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 e 1 ( λ 1 + λ 2 ) ] .   Using the formula for the sectional curvature of T 1 M n   , we find an expression for the Gaussian curvature of ξ ( M 2 )   ( Lemma  3.4 ).
The Gaussian curvature K ξ   of a hypersurface ξ ( M ) ∈ T 1 M   is given by
K ξ = K 2 4 + K ( 1 − K ) 1 + λ 2 + ( − 1 ) s + 1 λ 1 + λ 2 e 0 ( K ) +
1 2 μ e 1 ( 1 1 + λ 2 ) − ( ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ) 2 ,
where K   is the Gaussian curvature of M   .
As applications of these Lemmas, we prove the following theorems.
Theorem  3.1  Let M 2   be a Riemannian manifold of constant Gaussian curvature K   . A unit vector field ξ   generating a totally geodesic submanifold in T 1 M 2   exists if and only if K = 0   or K = 1   . Moreover,
  • (a) if K = 0   , then ξ   is either a parallel vector field or moving along a family of parallel geodesics with constant angle speed. Geometrically, ξ ( M 2 )   is either M 2   imbedded isometrically into M 2 × S 1   as a factor or a (helical) flat submanifold in M 2 × S 1   ;
  • (b) if K = 1   , then ξ   is a vector field on a standard sphere S 2   which is parallel along the meridians and moving along the parallels with a unit angle speed. Geometrically, ξ ( M 2 )   is a part of totally geodesic R P 2   locally isometric to sphere S 2   of radius 2 in T 1 S 2 ≈ i s o m R P 3 .  
Theorem  3.3  Let M 2   be a space of constant Gaussian curvature K   . Suppose that ξ   is a unit geodesic vector field on M 2   . Then ξ ( M 2 )   has constant Gaussian curvature in one of the following cases:
  • (a) K = − c 2 < 0   and ξ   is a normal vector field for the family of horocycles on the hyperbolic 2-plane L 2   of curvature − c 2   . In this case, K ξ = − c 2   and therefore ξ ( M 2 )   is locally isometric the base space;
  • (b) K = 0   and ξ   is a parallel vector field on M 2   . In this case K ξ = 0   and ξ ( M 2 )   is also locally isometric to the base space;
  • (c) K = 1   and ξ   is any (local) geodesic vector field on the standard sphere S 2   . In this case, K ξ = 0   .
Theorem  3.4  Let L 2   be a hyperbolic 2-plane of constant curvature − c 2   .
Then T 1 L 2   admits a hyperfoliation with leaves of constant intrinsic curvature − c 2   and of constant extrinsic curvature − c 2 4   . The leaves are generated by unit vector fields making a constant angle with a pencil of parallel geodesics on L 2 .  

2 Basic definitions and preliminary results

Let ( M , g )   be an ( n + 1 )   – dimensional Riemannian manifold with metric g   .
Let ∇   denote the Levi-Civita connection on M   . Then ∇ X ξ   is always orthogonal to ξ   and hence, ( ∇ ξ ) X = d e f ∇ X ξ : T p M → ξ p ⊥   is a linear operator at each p ∈ M   . We define an adjoint operator ( ∇ ξ ) * X : ξ p ⊥ → T p M   by 〈 ( ∇ ξ ) * X , Y 〉 g = 〈 X , ∇ Y ξ 〉 g .   Then there is an orthonormal frame e 0 , e 1 , . . . , e n   in T p M   and an orthonormal frame f 1 , . . . , f n   in ξ p ⊥   such that
( ∇ ξ ) e 0 = 0 , ( ∇ ξ ) e α = λ α f α , ( ∇ ξ ) * f α = λ α e α , α = 1 , . . . , n , (1)
where λ 1 , λ 2 , . . . λ n   are real-valued functions.
Definition 2.1 The orthonormal frames satisfying ( 1 ) are called singular frames for the linear operator ( ∇ ξ )   and the real valued functions λ 1 , λ 2 , . . . λ n   are called the (signed) singular values of the operator ∇ ξ   with respect to the singular frame.
Remark that the sign of the singular value is definied up to the directions of the vectors of the singular frame.
For each X ~ ∈ T ( p , ξ ) T M   there is a decomposition X ~ = X 1 h + X 2 v   where ( ⋅ ) h   and ( ⋅ ) v   are the horizontal and vertical lifts of vectors X 1   and X 2   from T p M   to T ( p , ξ ) T M   . The Sasaki metric is defined by the scalar product of the form < < X ~ , Y ~ > > = < X 1 , Y 1 > + < X 2 , Y 2 > ,   where < ⋅ , ⋅ >   means the scalar product with respect to metric g   .
The following lemma has been proved in [11] .
Lemma 2.1 At each point ( p , ξ ) ∈ ξ ( M ) ⊂ T M   the vectors
{ e ~ 0 = e 0 h , e ~ α = 1 1 + λ α 2 ( e α h + λ α f α v ) , α = 1 , . . . , n , (2)
form an orthonormal frame in the tangent space of ξ ( M )   and the vectors
n ~ σ | = 1 1 + λ σ 2 ( − λ σ e σ h + f σ v ) , σ = 1 , . . . , n , (3)
form an orthonormal frame in the normal space of ξ ( M )   .
Let R ( X , Y ) ξ = [ ∇ X , ∇ Y ] ξ − ∇ [ X , Y ] ξ   be the curvature tensor of M   . Introduce the following notation
r ( X , Y ) ξ = ∇ X ∇ Y ξ − ∇ ∇ X Y ξ . (4)
Then, evidently, R ( X , Y ) ξ = r ( X , Y ) ξ − r ( Y , X ) ξ .   The following Lemma has also been proved in [11] .
Lemma 2.2 The components of second fundamental form of ξ ( M ) ⊂ T 1 M   with respect to the frame ( 3 ) are given by
Ω ~ σ | 00 = 1 1 + λ σ 2 < r ( e 0 , e 0 ) ξ , f σ > ,
Ω ~ σ | α 0 = 1 2 1 ( 1 + λ σ 2 ) ( 1 + λ α 2 ) [ < r ( e α , e 0 ) ξ + r ( e 0 , e α ) ξ , f σ > +
λ σ λ α < R ( e σ , e 0 ) ξ , f α > ] ,
Ω ~ σ | α β = 1 2 1 ( 1 + λ σ 2 ) ( 1 + λ α 2 ) ( 1 + λ β 2 ) [ < r ( e α , e β ) ξ + r ( e β , e α ) ξ , f σ >
+ λ α λ σ < R ( e σ , e β ) ξ , f α > + λ β λ σ < R ( e σ , e α ) ξ , f β > ] ,
where { e 0 , e 1 , . . . , e n ; f 1 , . . . , f n }   is a singular frame of ( ∇ ξ )   and λ 1 , . . . , λ n   are the corresponding singular values.
Let ∇ ~   and ∇   be the Levi-Civita connections of the Sasaki metric of T M   and the metric of M   respectively. The Kowalski formulas [8] give the covariant derivatives of combinations of lifts of vector fields.
Lemma 2.3 (O.Kowalski) Let X   and Y   be vector fields on M   . Then at each point ( p , ξ ) ∈ T M   we have
∇ ~ X h Y h = ( ∇ X Y ) h − 1 2 ( R ( X , Y ) ξ ) v ,
∇ ~ X h Y v = 1 2 ( R ( ξ , Y ) X ) h + ( ∇ X Y ) v ,
∇ ~ X v Y h = 1 2 ( R ( ξ , X ) Y ) h ,
∇ ~ X v Y v = 0 ,
where R   is the Riemannian curvature tensor of ( M , g )   .
This basic result allows to find the curvature tensor of T M   (see [8] ) and the curvature tensor of T 1 M   (see [4] ). As a corollary, it is not too hard to find an expression for the sectional curvature of T 1 M   . It is well-known that ξ v   is a unit normal for T 1 M   as a hypersurface in T M   . Thus, X ~ = X 1 h + X 2 v   is tangent to T 1 M   if and only if < X 2 , ξ > = 0   .
Let X ~ = X 1 h + X 2 v   and Y ~ = Y 1 h + Y 2 v   , where X 2 , Y 2 ∈ ξ ⊥   , form an orthonormal base of a 2-plane π ~ ⊂ T ( p , ξ ) T 1 M   . Then we have [5] :
K ~ ( π ~ ) = < R ( X 1 , Y 1 ) Y 1 , X 1 > − 3 4 ∥ R ( X 1 , Y 1 ) ξ ∥ 2 + 1 4 ∥ R ( ξ , Y 2 ) X 1 + R ( ξ , X 2 ) Y 1 ∥ 2 + ∥ X 2 ∥ 2 ∥ Y 2 ∥ 2 − < X 2 , Y 2 > 2 + 3 < R ( X 1 , Y 1 ) Y 2 , X 2 > − < R ( ξ , X 2 ) X 1 , R ( ξ , Y 2 ) Y 1 > + < ( ∇ X 1 R ) ( ξ , Y 2 ) Y 1 , X 1 > + < ( ∇ Y 1 R ) ( ξ , X 2 ) X 1 , Y 1 > . (5)
Combining the results of Lemma  2.1 , Lemma  2.2 and ( 5 ), we can write an expression for the sectional curvature of ξ ( M )   .
Lemma 2.4 Let X ~   and Y ~   be an ortonormal vectors which span a 2-plane π ~   tangent to ξ ( M ) ⊂ T 1 M   . Denote by K ξ ( π ~ )   the sectional curvature ξ ( M )   with respect to metric, induced by Sasaki metric of T 1 M   . Then
K ξ ( π ~ ) = K ~ ( π ~ ) + ∑ σ ( Ω σ | ( X ~ , X ~ ) Ω σ | ( Y ~ , Y ~ ) − Ω σ | 2 ( X ~ , Y ~ ) ) , (6)
where K ~ ( π ~ )   is the sectional curvature of T 1 M   given by ( 5 ), Ω | σ   are the components of the second fundamental form of ξ ( M )   given by Lemma  2.2 and the vectors are given with respect to the frame ( 2 ).

3 The 2-dimensional case

Let M   be a 2-dimensional Riemannian manifold. The following proposition gives useful information about the relation between the singular values of the ( ∇ ξ )   operator, geometric characteristics of the integral curves of singular frame and the Gaussian curvature of the manifold.
Lemma 3.1 Let ξ   be a given smooth unit vector field on M 2   . Denote by e 0   a unit vector field on M 2   such that ∇ e 0 ξ = 0 .   Let η   and e 1   be the unit vector fields on M 2   such that ( ξ , η )   and ( e 0 , e 1 )   form two orthonormal frames on M 2   .
Denote by λ   a signed singular value of the operator ( ∇ ξ )   . Then we have ∇ e 1 ξ = λ η ,   and the following relations hold:
  • (a) if k = < ∇ ξ ξ , η >   is a signed geodesic curvature of a ξ   -curve and κ = < ∇ η η , ξ >   is a signed geodesic curvature of a η   -curve, then λ 2 = k 2 + κ 2 ;  
  • (b) if K   is the Gaussian curvature of M 2   , then ( − 1 ) s K = e 0 ( λ ) − λ σ ,   where σ = < ∇ e 1 e 1 , e 0 >   is a signed geodesic curvature of a e 1   -curve and s = { 1 if the frames ( ξ , η ) and ( e 0 , e 1 ) have the same orientation, 0 if the frames ( ξ , η ) and ( e 0 , e 1 ) have an opposite orientation .  
Proof. (a) If ( ξ , η )   is an orthonormal frame on M 2   , then
∇ ξ ξ = k η , ∇ ξ η = − k ξ , ∇ η ξ = − κ η , ∇ η η = κ ξ . (7)
Geometrically, the functions k   and κ   are the signed geodesic curvatures of ξ   and η   -curves respectively.
In a similar way we get
∇ e 0 e 0 = μ e 1 , ∇ e 0 e 1 = − μ e 0 , ∇ e 1 e 0 = − σ e 1 , ∇ e 1 e 1 = σ e 0 , (8)
where μ   and σ   are the signed geodesic curvatures of the e 0   and e 1   -curves respectively.
Let ω   be an angle function between ξ   and e 0   . Then we have two possible decompositions:
Or(+) { e 0 = cos ω ξ + sin ω η , e 1 = − sin ω ξ + cos ω η , Or(–) { e 0 = cos ω ξ + sin ω η , e 1 = sin ω ξ − cos ω η .
In the case O r ( + )   we have
∇ e 0 ξ = ( k cos ω − κ sin ω ) η ,
∇ e 1 ξ = − ( k sin ω + κ cos ω ) η ,
and due to the choice of e 0   and e 1   we see that { k cos ω − κ sin ω = 0 , k sin ω + κ cos ω = − λ .   So, for the case of O r ( + )   k = − λ sin ω , κ = − λ cos ω .   In a similar way, for the case of O r ( − )   k = λ sin ω , κ = λ cos ω .   In both cases λ 2 = k 2 + κ 2 .   (b) Due to the choice of the frames,
< R ( e 0 , e 1 ) ξ , η > = < ∇ e 0 ∇ e 1 ξ − ∇ e 1 ∇ e 0 ξ − ∇ ∇ e 0 e 1 − ∇ e 1 e 0 ξ , η > =
< ∇ e 0 ( λ η ) − ∇ − μ e 0 + σ e 1 ξ , η > = e 0 ( λ ) − λ σ .
On the other hand,
< R ( e 0 , e 1 ) ξ , η > = { − K for the case of O r ( + ) , + K for the case of O r ( − ) . (9)
Set s = 1   for the case O r ( + )   and s = 0   for the case O r ( − )   . Combining the results, we get ( − 1 ) s K = e 0 ( λ ) − λ σ ,   which completes the proof.
The result of Lemma  2.2 can also be simplified in the following way.
Lemma 3.2 Let M   be a 2-dimensional Riemannian manifold of Gaussian curvature K   . In terms of Lemma  3.1 the second fundamental form of the submanifold ξ ( M ) ⊂ T 1 M   can be presented in two equivalent forms:
  • (i) Ω = [ − μ λ 1 + λ 2 ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 e 1 ( λ 1 + λ 2 ) ] ,  
  • (ii) Ω = [ − μ λ 1 + λ 2 1 2 ( σ λ + 1 − λ 2 1 + λ 2 e 0 ( λ ) ) 1 2 ( σ λ + 1 − λ 2 1 + λ 2 e 0 ( λ ) ) e 1 ( λ 1 + λ 2 ) ] .  
Proof. At each point ( p , ξ ) ∈ ξ ( M )   the vectors { e ~ 0 = e 0 h , e ~ 1 = 1 1 + λ 2 ( e 1 h + λ η v )   form an orthonormal frame in the tangent space of ξ ( M )   and n ~ = 1 1 + λ 2 ( − λ e 1 h + η v ) ,   is a unit normal for ξ ( M ) ⊂ T 1 M   .
Thus we see that in a 2-dimensional case the components of Ω   take the form
Ω 00 = 1 1 + λ 2 < r ( e 0 , e 0 ) ξ , η > , Ω 11 = 1 ( 1 + λ 2 ) 3 / 2 < r ( e 1 , e 1 ) ξ , η > ,
Ω 01 = 1 2 1 1 + λ 2 [ < r ( e 1 , e 0 ) ξ + r ( e 0 , e 1 ) ξ , η > + λ 2 < R ( e 1 , e 0 ) ξ , η > ] .
Keeping in mind ( 4 ), ( 8 ) and ( 9 ), we see that
< r ( e 0 , e 0 ) ξ , η > = − μ λ , < r ( e 0 , e 1 ) ξ , η > = e 0 ( λ ) ,
< r ( e 1 , e 0 ) ξ , η > = σ λ , < r ( e 1 , e 1 ) ξ , η > = e 1 ( λ ) ,
< R ( e 0 , e 1 ) ξ , η > = ( − 1 ) s K .
So we have
Ω 00 = − μ λ 1 + λ 2 , Ω 11 = e 1 ( λ ) ( 1 + λ 2 ) 3 / 2 = e 1 ( λ 1 + λ 2 ) ,
Ω 01 = 1 2 ( 1 + λ 2 ) ( e 0 ( λ ) + λ σ − λ 2 ( − 1 ) s K ) = { ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 1 2 ( σ λ + 1 − λ 2 1 + λ 2 e 0 ( λ ) ) ,
where Lemma  3.1 (b) has been applied in two ways.

3.1 Totally geodesic vector fields

The main goal of this section is to prove the following theorem.
Theorem 3.1 Let M 2   be a Riemannian manifold of constant Gaussian curvature K   . A unit vector field ξ   generating a totally geodesic submanifold in T 1 M 2   exists if and only if K = 0   or K = 1   . Moreover,
  • a) if K = 0   , then ξ   is either a parallel vector field or is moving along a family of parallel geodesics with constant angle speed. Geometrically, ξ ( M 2 )   is either M 2   imbedded isometrically into M 2 × S 1   as a factor or a (helical) flat submanifold in M 2 × S 1   ;
  • b) if K = 1   , then ξ   is a vector field on a sphere S 2   which is parallel along the meridians and moving along the parallels with a unit angle speed.
    Geometrically, ξ ( M 2 )   is a part of totally geodesic R P 2   locally isometric to sphere S 2   of radius 2 in T 1 S 2 ≈ i s o m R P 3   .
The proof will be divided into a series of separate propositions.
Proposition 3.1 Let M 2   be a Riemannian manifold. Let D   be a domain in M 2   endowed with a semi-geodesic coordinate system such that d s 2 = d u 2 + f 2 d v 2 ,   where f ( u , v )   is some non-vanishing function. Denote by ( e 0 , e 1 )   an orthonormal frame in D   and specify e 0 = ∂ u , e 1 = f − 1 ∂ v   . If ξ   is a unit vector field in D   parallel along u   -geodesics, then ξ   can be written given as ξ = cos ω e 0 + sin ω e 1 ,   where ω = ω ( v )   is an angle function and (a) a singular frame for ξ   may be chosen as { e 0 , e 1 , η = − sin ω e 0 + cos ω e 1 } ;   (b) a singular value for ξ   in this case is λ = e 1 ( ω ) − σ ,   where σ   is a signed geodesic curvature of the e 1   -curves.
Proof. Indeed, if ξ   is parallel along u   -geodesics, then evidently the angle function ω   between ξ   and the u   -curves does not depend on u   . So this function has the form ω = ω ( v )   and ξ = cos ω e 0 + sin ω e 1   . Moreover, since
∇ e 0 e 0 = 0 , ∇ e 0 e 1 = 0 ,
∇ e 1 e 0 = f u f e 1 , ∇ e 1 e 1 = − f u f e 0 ,
we see that σ = − f u f   and ∇ e 1 ξ = ( e 1 ( ω ) − σ ) η ,   where η = − sin ω e 0 + cos ω e 1   .
Therefore, λ = e 1 ( ω ) − σ   and the proof is complete.
Proposition 3.2 Let M 2   be a Riemannian manifold of constant negative curvature K = − r − 2 < 0   . Then there is no totally geodesic unit vector field on M 2   .
Proof. Suppose ξ   is totally geodesic unit vector field on M 2   . Set Ω ≡ 0   in Lemma  3.2 . Then λ μ ≡ 0   . If λ ≡ 0   in some domain D ⊂ M 2   , then ξ   is parallel in this domain and hence M 2   is flat in D   , which contradicts the hypothesis. Suppose that μ ≡ 0   at least in some domain D ⊂ M 2   . This means that e 0   -curves are geodesics in D   and the field ξ   is parallel along them. Choose a family of e 0   -curves and the orthogonal trajectories as a local coordinate net in D   . Then the first fundamental form of M 2   takes the form d s 2 = d u 2 + f 2 d v 2 ,   where f ( u , v )   is some function. Since M 2   is of constant curvature K = − 1 r 2   , the function f   satisfies the equation f u u − 1 r 2 f = 0 .   The general solution of this equation is f ( u , v ) = A ( v ) cosh ( u / r ) + B ( v ) sinh ( u / r ) .   There are two possible cases:
( i ) A 2 ( v ) ≡ B 2 ( v ) over the whole domain D ;
( i i ) A 2 ( v ) ≠ B 2 ( v ) in some subdomain D ′ ⊂ D .
Case (i). In this case, in dependence of the signs of A ( v )   and B ( v ) ,   f ( u , v ) = A ( v ) e u / r or f ( u , v ) = A ( v ) e − u / r .   Consider the first case (the second case can be reduced to the first one after the parameter change u → − u   ). Making an evident v   -parameter change, we reduce the metric to the form d s 2 = d u 2 + r 2 e 2 u / r d v 2 .   Applying Proposition  3.1 for f = r e u / r   , we get λ = 1 r ( ω ′ e − u / r + 1 ) .   Setting Ω 11 ≡ 0   , we see that e 1 ( λ ) ≡ 0   . Hence ω ′ ′ = 0   , i.e., ω = a v + b   . Therefore, λ = 1 r ( a e − u / r + 1 ) .   Considering Ω 01 ≡ 0   (with s = 1   because of O r ( + )   -case), we get − 1 2 r 2 + 1 r e 0 ( a e − u / r + 1 ) 1 + 1 r 2 ( e − u / r a + 1 ) 2 = − ( 1 r 2 + 1 ) ( a e − u / r + 1 ) 2 − a 2 e − 2 u / r 2 r 2 [ 1 + 1 r 2 ( a e − u / r + 1 ) 2 ] ≢ 0 ,   and hence, this case is not possible.
Case (ii). Choose a subdomain D ′ ⊂ D   such that A 2 ( v ) < B 2 ( v )   or A 2 ( v ) > B 2 ( v )   over D ′   . Then the function f   may be presented respectively in two forms:
(a) f ( u , v ) = B 2 − A 2 sinh ( u / r + θ ) or
(b) f ( u , v ) = A 2 − B 2 cosh ( u / r + θ ) ,
where θ ( v )   is some function.
Consider the case (a). After a v   -parameter change, the metric in D ′   takes the form d s 2 = d u 2 + r 2 sinh 2 ( u / r + θ ) d v 2 .   Applying Proposition  3.1 for f = r sinh ( u / r + θ )   , we get λ = ω ′ r sinh ( u / r + θ ) + 1 r coth ( u / r + θ ) .   Considering Ω 11 ≡ 0   , we have e 1 ( λ ) ≡ 0   which implies the identity ω ′ ′ sinh ( u / r + θ ) − ω ′ θ ′ cosh ( u / r + θ ) − θ ′ ≡ 0 .   From this we get ω ′ ′ = 0 , θ ′ = 0   and hence { θ = c o n s t , ω = a v + b   ( a , b = c o n s t   ).
After a parameter change we reduce the metric to the form d s 2 = d u 2 + r 2 sinh 2 ( u / r ) d v 2   Applying Proposition  3.1 for f = r sinh ( u / r )   , we get λ = a + cosh ( u / r ) r sinh ( u / r ) .   The substitution into Ω 01   gives − 1 2 ( 1 r 2 + 1 ) [ a + cosh ( u / r ) ] 2 − a 2 + 1 r 2 sinh 2 ( u / r ) + [ a + cosh ( u / r ) ] 2 ≢ 0 ,   which completes the proof for the polar case.
The Cartesian case consideration gives ω = a v + b , λ = a + sinh ( u / r ) r cosh ( u / r )   and Ω 01 = − 1 2 ( 1 r 2 + 1 ) [ a + sinh ( u / r ) ] 2 − a 2 − 1 r 2 cosh 2 ( u / r ) + [ a + sinh ( u / r ) ] 2 ≢ 0 ,   which completes the proof.
Proposition 3.3 Let M 2   be a Riemannian manifold of constant positive curvature K = r − 2 > 0   . Then a totally geodesic unit vector field ξ   on M 2   exists if r = 1   and ξ   is parallel along the meridians of M 2   locally isometric to S 2   and moves along the parallels with a unit angle speed. Geometrically, ξ ( M 2 )   is a part of totally geodesic R P 2   locally isometric to sphere S 2   of radius 2 in T 1 S 2 ≈ i s o m R P 3   .
Proof. Suppose ξ   is totally geodesic unit vector field on M 2   . The same arguments as in Proposition  3.2 lead to the case μ ≡ 0   at least in some domain D ⊂ M 2   . So, choose again a family of e 0   -curves and the orthogonal trajectories as a local coordinate net in D   . Then the first fundamental form of M 2   can be expressed as d s 2 = d u 2 + f 2 d v 2 ,   where f ( u , v )   is some function. Since M 2   is of constant curvature K = r − 2   , the function f   satisfies the equation f u u + 1 r 2 f = 0 .   The general solution of this equation f ( u , v ) = A ( v ) cos ( u / r ) + B ( v ) sin ( u / r )   may be presented in two forms:
(a) f ( u , v ) = A 2 + B 2 sin ( u / r + θ ) or
(b) f ( u , v ) = A 2 + B 2 cos ( u / r + θ ) ,
where θ ( v )   is some function.
Consider first, the case (a). After v   -parameter change, the metric in D   takes the form d s 2 = d u 2 + r 2 sin 2 ( u / r + θ ) d v 2 .   Applying Proposition  3.1 for f = r sin ( u / r + θ )   , we get λ = ω ′ r sin ( u / r + θ ) + 1 r cot ( u / r + θ ) .   Setting Ω 11 ≡ 0   , we find e 1 ( λ ) ≡ 0   which implies the identity ω ′ ′ sin ( u / r + θ ) − ω ′ θ ′ cos ( u / r + θ ) + θ ′ ≡ 0 .   From this ω ′ ′ = 0 , θ ′ = 0   and we have again { θ = c o n s t , ω = a v + b   a , b = c o n s t   .
After a suitable u   -parameter change, we reduce the metric to the form d s 2 = d u 2 + r 2 sin 2 ( u / r ) d v 2   Applying Proposition  3.1 for f = r sin ( u / r )   , we get λ = a + cos ( u / r ) r sin ( u / r ) .   Substitution into Ω 01   gives 1 2 ( 1 r 2 − 1 ) [ a + cos ( u / r ) ] 2 + a 2 − 1 r 2 sin 2 ( u / r ) + [ a + cos ( u / r ) ] 2 ≡ 0 ,   which is possible only if r = 1   and | a | = 1   . So, we obtain to the standard sphere metric d s 2 = d u 2 + sin 2 u d v 2   and ( after the ± v + b → v   parameter change ) the unit vector field ξ = { cos v , sin v sin u } .   This vector field is parallel along the meridians of S 2   and moves helically along the parallels of S 2   with unit angle speed.
For the case (b) one can find ω = a v + b , λ = a − sin ( u / r ) r cos ( u / r )   and Ω 01 = 1 2 ( 1 r 2 − 1 ) [ a − sin ( u / r ) ] 2 + a 2 − 1 r 2 cos 2 ( u / r ) + [ a − sin ( u / r ) ] 2 ≡ 0 ,   which gives r = 1   and | a | = 1   as a result. Thus, we have a metric d s 2 = d u 2 + cos 2 u d v 2   and a vector field ξ = { cos v , sin v cos u } .   It is easy to see that the results of cases (a) and (b) are geometrically equivalent.
Introduce the local coordinates ( u , v , ω )   on T 1 S 2   , where ω   is the angle between arbitrary unit vector ξ   and the coordinate vector field X 1 = { 1 , 0 }   .
The first fundamental form of T 1 S 2   with respect to these coordinates is [10]  d s ~ 2 = d u 2 + d v 2 + 2 cos u d v d ω + d ω 2 .   The local parameterization of the submanifold ξ ( S 2 )   , generated by the given field, is ω = v   and the induced metric on ξ ( S 2 )   is d s ~ 2 = d u 2 + 2 ( 1 + cos u ) d v 2 = d u 2 + 4 cos 2 u / 2 d v 2 .   Thus, ξ ( S 2 )   is locally isometric to sphere S 2   of radius 2. Since T 1 S 2 ≈ i s o m R P 3   and there are no other totally geodesic submanifolds in R P 3   except R P 2   , we see that ξ ( S 2 )   is a part of R P 2   . So the proof is complete.
Proposition 3.4 Let M 2   be a Riemannian manifold of constant zero curvature K = 0   . Then a totally geodesic unit vector field ξ   on M 2   is either parallel or moves along the family of parallel geodesics with constant angle speed. Geometrically, ξ ( M 2 )   is either E 2   imbedded isometrically into E 2 × S 1   as a factor or a helical flat submanifold in E 2 × S 1   .
Proof. Suppose ξ   is totally geodesic unit vector field on M 2   . Set Ω ≡ 0   in Lemma  3.2 . Then λ μ ≡ 0   . If λ ≡ 0   over some domain D ⊂ M 2   , then ξ   is parallel in this domain.
Suppose λ ≢ 0   in a domain D ⊂ M 2   . Then μ ≡ 0   on at least a subdomain D ′ ⊂ D   . This means that the e 0   -curves are geodesics in D ′   and the field ξ   is parallel along them. Choose a family of e 0   -curves and the orthogonal trajectories as a local coordinate net in D ′   . Then the first fundamental form of M 2   takes the form d s 2 = d u 2 + f 2 d v 2   and since M 2   is of zero curvature, f   satisfies the equation f u u = 0 .   A general solution of this equation is f ( u , v ) = A ( v ) u + B ( v ) .   There are two possible cases:
(a) A ( v ) ≠ 0 in some subdomain D ′ ′ ⊂ D ′ ;
(b) A ( v ) ≡ 0 over the whole domain D ′ .
Case(a). The function f   may be presented over D ′ ′   in the form f ( u , v ) = A ( v ) ( u + θ ) ,   where θ ( v ) = B ( v ) / A ( v )   . After a v   -parameter change, the metric in D ′ ′   takes the form d s 2 = d u 2 + ( u + θ ) 2 d v 2 .   Applying Proposition  3.1 for f = u + θ   , we get λ = ω ′ + 1 u + θ .   Setting Ω 11 ≡ 0   , we obtain the identity ω ′ ′ ( u + θ ) − ( ω ′ + 1 ) θ ′ ≡ 0 .   From this we get { ω ′ ′ = 0 ω ′ = − 1   or { ω ′ ′ = 0 θ ′ = 0   . In the first case, λ = 0   and the field ξ   is parallel again. In the second case { θ = c o n s t , ω = a v + b   a , b = c o n s t   .
Making a parameter change, we reduce the metric to the form d s 2 = d u 2 + u 2 d v 2   Applying Proposition  3.1 with f ( u , v ) = u   , we get λ = a + 1 u .   The substitution into Ω 01   gives the condition − a + 1 u 2 + ( a + 1 ) 2 = 0   which is possible only if a = − 1   . But this means that again λ = 0   and hence ξ   is a parallel vector field.
Case (b). After a v   -parameter change, the metric takes the form d s 2 = d u 2 + d v 2 .   Applying Proposition  3.1 for f ≡ 1   , we get λ = ω ′ .   Setting Ω 11 ≡ 0   , we find ω ′ ′ ≡ 0   . This means that ω = a v + b   and ξ   is either parallel along the u   -lines ( a = 0 )   or moves along the u   -lines helically with constant angle speed.
Let ( u , v , ω )   be standard coordinates in E 2 × S 1   . Then the first fundamental form of E 2 × S 1   is d s ~ 2 = d u 2 + d v 2 + d ω 2 .   If a = 0   , then with respect to these coordinates the local parameterization of ξ ( E 2 )   is ω = c o n s t   and ξ ( E 2 )   is nothing else but E 2   isometrically imbedded into E 2 × S 1   . If a ≠ 0   , then the local parameterization of ξ ( E 2 )   is ω = a v + b   and the induced metric is d s ~ 2 = d u 2 + ( 1 + a 2 ) d v 2   which is flat. The imbedding is helical in the sense that this submanifold meets each flat element of the cylinder p : E 2 × S 1 → S 1   under constant angle φ = arccos 1 1 + a 2   . So the proof is complete.

3.2 The curvature

The main goal of this section is to obtain an explicit formula for the Gaussian curvature of ξ ( M 2 )   and apply it to some specific cases. The first step is the following lemma.
Lemma 3.3 Let ξ   be a unit vector field on a 2-dimensional Riemannian manifold of Gaussian curvature K   . In terms of Lemma  3.1 , the sectional curvature K T 1 M ( ξ )   of T 1 M   along 2-planes tangent to ξ ( M )   is given by K T 1 M ( ξ ) = K 2 4 + K ( 1 − K ) 1 + λ 2 + ( − 1 ) s + 1 λ 1 + λ 2 e 0 ( K ) .  
Proof. Let π ~   be a 2-plane tangent to ξ ( M )   . Then X ~ = e 0 h   and Y ~ = 1 1 + λ 2 ( e 1 h + λ η v )   form an orthonormal basis of π ~   . So we may apply ( 5 ) setting X 1 = e 0   , X 2 = 0   , Y 1 = 1 1 + λ 2 e 1   , Y 2 = λ 1 + λ 2 η   .
We get
< R ( X 1 , Y 1 ) Y 1 , X 1 > = 1 1 + λ 2 < R ( e 0 , e 1 ) e 1 , e 0 > = 1 1 + λ 2 K ,
∥ R ( X 1 , Y 1 ) ξ ∥ 2 = 1 1 + λ 2 ∥ R ( e 0 , e 1 ) ξ ∥ 2 = 1 1 + λ 2 K 2 ,
∥ R ( ξ , Y 2 ) X 1 ∥ 2 = λ 2 1 + λ 2 ∥ R ( ξ , η ) e 0 ∥ 2 = λ 2 1 + λ 2 K 2 ,
< ( ∇ X 1 R ) ( ξ , Y 2 ) Y 1 , X 1 > = λ 1 + λ 2 < ( ∇ e 0 R ) ( ξ , η ) e 1 , e 0 > = − ( − 1 ) s λ 1 + λ 2 e 0 ( K ) ,
where K   is the Gaussian curvature of M   . Applying directly ( 5 ) we obtain
K T 1 M ( ξ ) = 1 1 + λ 2 ( K − 3 4 K 2 + λ 2 K 2 4 + ( − 1 ) s + 1 λ e 0 ( K ) )
= 1 1 + λ 2 ( K ( 1 − K ) + ( 1 + λ 2 ) K 2 4 + ( − 1 ) s + 1 λ e 0 ( K ) )
= K 2 4 + K ( 1 − K ) 1 + λ 2 + ( − 1 ) s + 1 λ 1 + λ 2 e 0 ( K ) .
Now we have the following.
Lemma 3.4 Let ξ   be a unit vector field on a 2-dimensional Riemannian manifold M   . In terms of Lemma  3.1 , the Gaussian curvature K ξ   of the hypersurface ξ ( M ) ∈ T 1 M   is given by
K ξ = K 2 4 + K ( 1 − K ) 1 + λ 2 + ( − 1 ) s + 1 λ 1 + λ 2 e 0 ( K ) +
1 2 μ e 1 ( 1 1 + λ 2 ) − ( ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ) 2 ,
where K   is the Gaussian curvature of M   .
Proof. In our case, one can easily reduce the formula ( 6 ) to the form K ξ = K T 1 M ( ξ ) + det Ω .   Applying Lemma  3.2 , we see that
det Ω = − μ λ 1 + λ 2 e 1 ( λ 1 + λ 2 ) − ( ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ) 2 =
− 1 2 μ e 1 ( λ 2 1 + λ 2 ) − ( ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ) 2 =
1 2 μ e 1 ( 1 1 + λ 2 ) − ( ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ) 2 .
Combining this result with Lemma  3.3 , we get what was claimed.
As an application of Lemma  3.4 we prove the following theorems.
Theorem 3.2 Let M 2   be a 2-dimensional Riemannian manifold of Gaussian curvature K   . Suppose that ξ   is a unit geodesic vector field on M 2   . Then the submanifold ξ ( M 2 ) ⊂ T 1 M 2   has non-positive extrinsic curvature.
Proof. By definition, the extrinsic curvature of a submanifold is the difference between the sectional curvature of the submanifold and the sectional curvature of ambient space along the planes, tangent to the submanifold. In our case , this is det Ω   .
If ξ   is a geodesic vector field, then we may choose e 0 = ξ   and then μ = k = 0   .
Therefore, for the extrinsic curvature we get − ( ( − 1 ) s + 1 K 2 + e 0 ( λ ) 1 + λ 2 ) 2 ≤ 0 .  
Theorem 3.3 Let M 2   be a space of constant Gaussian curvature K   . Suppose that ξ   is a unit geodesic vector field on M 2   . Then ξ ( M 2 )   has constant Gaussian curvature in one of the following cases:
  • (a) K = − c 2 < 0   and ξ   is a normal vector field for the family of horocycles on the hyperbolic 2-plane L 2   . In this case K ξ = − c 2   and therefore ξ ( L 2 )   is locally isometric to L 2   ;
  • (b) K = 0   and ξ   is a parallel vector field on M 2   . In this case K ξ = 0   and ξ ( M 2 )   is also locally isometric to M 2   ;
  • (c) K = 1   and ξ   is any (local) geodesic vector field on the standard sphere S 2   . In this case K ξ = 0   .
Proof. Since ξ   is geodesic, we may set e 0 = ξ   , e 1 = η , s = 1   . Taking into account ( 7 ) and ( 8 ), we see that λ = − κ = − σ   . Lemma  3.1 (b) gives − K = − e 0 ( σ ) + σ 2   .
So the result of Lemma  3.4 takes the form
K ξ = K 2 4 + K ( 1 − K ) 1 + σ 2 − ( K 2 − e 0 ( σ ) 1 + σ 2 ) 2 =
K 2 4 + K ( 1 − K ) 1 + σ 2 − ( K 2 − K + σ 2 1 + σ 2 ) 2 =
K ( 1 − K ) 1 + σ 2 + K ( K + σ 2 ) 1 + σ 2 − ( K + σ 2 1 + σ 2 ) 2 =
K − ( K + σ 2 1 + σ 2 ) 2 .
Suppose that K ξ   is constant. Then the following cases should be considered:
(a) σ = c o n s t ≠ 0   . This means that the orthogonal trajectories of the field ξ   consist of curves of constant curvature. With respect to this natural coordinate system, the metric of M 2   takes the form d s 2 = d u 2 + f 2 d v 2 .   Set σ = − c   . Then the function f   should satisfy the equation f u f = c   the general solution of which is f ( u , v ) = A ( v ) e c u   . After v   -parameter change we obtain metric of the form d s 2 = d u 2 + e 2 c u d v 2 .   So, the manifold M 2   is locally isometric to the hyperbolic 2-plane L 2   of curvature − c 2   and the field ξ   is a geodesic field of (internal or external) normals to the family of horocycles. (b) σ = 0   . Then evidently ξ   is a parallel vector field and therefore the manifold M 2   is locally Euclidean which implies K ξ   =0.
(c) σ   is not constant. Then K ξ   is constant if K = 1   only. So, M 2   is contained in a standard sphere S 2   and the curvature of ξ ( S 2 )   does not depend on σ   . Thus, the field ξ   is any (local) geodesic vector field. Evidently, K ξ = 0   for this case.
The case ( a )   of the Theorem  3.3 has an interesting generalization of the following kind.
Theorem 3.4 Let L 2   be a hyperbolic 2-plane of curvature − c 2   .
Then T 1 L 2   admits a hyperfoliation with leaves of constant intrinsic curvature − c 2   and of constant extrinsic curvature − c 2 4   . The leaves are generated by unit vector fields making a constant angle with a pencil of parallel geodesics on L 2 .  
Proof. Consider L 2   with metric d s 2 = d u 2 + e 2 c u d v 2   and a family of vector fields ξ ω = cos ω X 1 + sin ω X 2 ( ω = c o n s t ) ,   where X 1 = { 1 , 0 } , X 2 = { 0 , e − c u }   are the unit vector fields.
Since ∇ X 1 ξ ω = 0   , we may set e 0 = X 1 , e 1 = X 2   end therefore we have σ = − c , λ = c .   Then, setting K = − c 2   and λ = c   in Lemma  3.4 , we get K ξ = − c 2 .   The extrinsic curvature of ξ ( L 2 )   is also constant since det Ω = − 1 4 c 2 .   Now fix a point P ∞   at infinity boundary of L 2   and draw a pencil of parallel geodesics from P ∞   through each point of L 2   . Define a family of submanifolds ξ ω ( L 2 )   for this pencil. Evidently, through each point ( p , ζ ) ∈ T 1 L 2   there passes only one submanifold of this family. Thus, a family of submanifods ξ ω   form a hyperfoliation on T 1 L 2   of constant intrinsic curvature − c 2   and constant extrinsic curvature − c 2 4   .
Geometrically, ξ ω ( L 2 )   is a family of coordinate hypersurfaces ω = c o n s t   in T 1 L 2   . Indeed, let ( u , v , ω )   form a natural local coordinate system on T 1 L 2   .
Then the metric of T 1 L 2   has the form d s 2 = d u 2 + 2 e 2 u d v 2 + 2 d v d ω + d ω 2 .   With respect to these coordinates, the coordinate hypersurface ω = c o n s t   is nothing else but ξ ω ( L 2 )   and the induced metric is d s 2 = d u 2 + 2 e 2 c u d v 2 .   Evidently, its Gaussian curvature is constant and equal to − c 2   .
References

  1. Aminov Yu. The geometry of vector fields.//Gordon&Breach Publ., 2000.
  2. Boeckx E., Vanhecke L. Harmonic and minimal radial vector fields.// Acta Math. Hungar. 90(2001), 317-331.
  3. Boeckx E., Vanhecke L. Harmonic and minimal vector fields on tangent and unit tangent bundles.// Differential Geom. Appl.13 (2000), 77-93.
  4. Boeckx E., Vanhecke L. Characteristic reflections on unit tangent sphere bundle.// Houston J. Math. 23(1997), 427-448.
  5. Borisenko A., Yampolsky A. The sectional curvature of the Sasaki metric of T 1 M n   .//Ukr. Geom. Sb. 30 (1987), 10-17. (Engl. transl.: J. Sov. Math. 51(1990), No 5, 2503-2508).
  6. Gluck H., Ziller W. On the volume of a unit vector field on the three-sphere. // Comm. Math. Helv. 61 (1986), 177-192.
  7. González-Dávila J.C., Vanhecke L. Examples of minimal unit vector fields.// Ann Global Anal. Geom. 18 (2000), 385-404.
  8. Kowalski O. Curvature of the induced Riemannian metric on the tangent bundle of a Riemannian manifold.// J. Reine Angew. Math. 250(1971), 124-129.
  9. Sasaki S. On the differential geometry of tangent bundles of Riemannian manifolds. // Tôhoku Math. J. 10 (1958), 338-354.
  10. Klingenberg W, Sasaki S. Tangent sphere bundle of a 2– sphere.// Tôhoku Math. J. 27 (1975), 45-57.
  11. Yampolsky A. On the mean curvature of a unit vector field.// Math. Publ. Debrecen, 2002, to appear.

Department of Geometry, Faculty of Mechanics and Mathematics, Kharkiv National University, Svobody Sq. 4, 61077, Kharkiv, Ukraine. e-mail: yamp@univer.kharkov.ua