Harmonic Vector Fields: Variational Principles and Differential Geometry

Table of Contents

Cover image

Front Matter

Copyright

Preface

Chapter One. Geometry of the Tangent Bundle

1.1. The Tangent Bundle

1.2. Connections and Horizontal Vector Fields

1.3. The Dombrowski Map and the Sasaki Metric

1.4. The Tangent Sphere Bundle

1.5. The Tangent Sphere Bundle over a Torus

Chapter Two. Harmonic Vector Fields

2.1. Vector Fields as Isometric Immersions

2.2. The Energy of a Vector Field

2.3. Vector Fields Which Are Harmonic Maps

2.4. The Tension of a Vector Field

2.5. Variations through Vector Fields

2.6. Unit Vector Fields

2.7. The Second Variation of the Energy Function

2.8. Unboundedness of the Energy Functional

2.9. The Dirichlet Problem

2.10. Conformal Change of Metric on the Torus

2.11. Sobolev Spaces of Vector Fields

Chapter Three. Harmonicity and Stability

3.1. Hopf Vector Fields on Spheres

3.2. The Energy of Unit Killing Fields in Dimension 3

3.3. Instability of Hopf Vector Fields

3.4. Existence of Minima in Dimension > 3

3.5. Brito's Functional

3.6. The Brito Energy of the Reeb Vector

3.7. Vector Fields with Singularities

3.8. Normal Vector Fields on Principal Orbits

3.9. Riemannian Tori

Chapter Four. Harmonicity and Contact Metric Structures

4.1. H-Contact Manifolds

4.2. Three-Dimensional H-Contact Manifolds

4.3. Stability of the Reeb Vector Field

4.4. Harmonic Almost Contact Structures

4.5. Reeb Vector Fields on Real Hypersurfaces

4.6. Harmonicity and Stability of the Geodesic Flow

Chapter Five. Harmonicity with Respect to g-Natural Metrics

5.1. g-Natural Metrics

5.2. Naturally Harmonic Vector Fields

5.3. Vector Fields Which Are Naturally Harmonic Maps

5.4. Geodesic Flow with Respect to g-Natural Metrics

Chapter Six. The Energy of Sections

6.1. The Horizontal Bundle

6.2. The Sasaki Metric

6.3. The Sphere Bundle U(E)

6.4. The Energy of Cross Sections

6.5. Unit Sections

6.6. Harmonic Sections in Normal Bundles

6.7. The Energy of Oriented Distributions

6.8. Examples of Harmonic Distributions

6.9. The Chacon-Naveira Energy

Chapter Seven. Harmonic Vector Fields in CR Geometry

7.1. The Canonical Metric

7.2. Bundles of Hyperquadrics in (T(M), J, Gs)

7.3. Harmonic Vector Fields from C(M)

7.4. Boundary Values of Bergman-Harmonic Maps

7.5. Pseudoharmonic Maps

7.6. The Pseudohermitian Biegung

7.7. The Second Variation Formula

Chapter Eight. Lorentz Geometry and Harmonic Vector Fields

8.1. A Few Notions of Lorentz Geometry

8.2. Energy Functionals and Tension Fields

8.3. The Spacelike Energy

8.4. The Second Variation of the Spacelike Energy

8.5. Conformal Vector Fields

Appendix A. Twisted Cohomologies

Appendix B. The Stokes Theorem on Complete Manifolds

Appendix C. Complex Monge-Ampère Equations

Appendix D. Exceptional Orbits of Highest Dimension

Appendix E. Reilly's Formula

References

Index

Front Matter

Harmonic Vector Fields

Harmonic Vector Fields

Variational Principles and Differential Geometry

Sorin Dragomir

Università degli Studi della Basilicata

Domenico Perrone

Università del Salento

Amsterdam • Boston • Heidelberg • London • New York • Oxford • Paris • San Diego • San Francisco • Singapore • Sydney • Tokyo

Copyright

Elsevier

225 Wyman Street, Waltham, MA 02451, USA

The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, UK

© 2012 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher's permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

Library of Congress Cataloging-in-Publication Data

Dragomir, Sorin, 1955–

Harmonic vector fields : variational principles and differential geometry / Sorin Dragomir and Domenico Perrone.

p. cm.

Includes bibliographical references and index.

ISBN 978-0-12-415826-9 (alk. paper)

1. Vector fields. 2. Geometry, Differential. I. Perrone, Domenico, 1949– II. Title.

QA613.619.D73 2011

514’.72–dc23

2011030072

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

For information on all Elsevier publications visit our website at www.elsevierdirect.com

Printed in the United States of America

11 12 13 14 9 8 7 6 5 4 3 2 1

Preface

The main object of study in this monograph are harmonic vector fields on Riemannian manifolds. Let ( M, g) be a real n-dimensional Riemannian manifold and Sn−1 → S( M) → M its tangent sphere bundle. It is a classical fact that S( M(the Sasaki metric) naturally associated to the Riemannian metric g on the base manifold (cf. e.g., D.E. Blair, [42]). Therefore any smooth unit tangent vector field X : M → S( M) may be looked at as a smooth map among the Riemannian manifolds ( M, g) and ( S( M) so that one may consider (by assuming that M is compact and orientable) the ordinary Dirichlet energy

(P.1)

familiar in the theory of harmonic maps (cf. e.g., H. Urakawa, [292]). A harmonic vector field is then a critical point X ∈ Γ ∞ ( S( M)) of E : Γ ∞ ( S( M)) → ℝ i.e., for any smooth 1-parameter variation { X t} | t| < ϵ of X (with X0 = X) through unit tangent vector fields X t ∈ Γ ∞ ( S( M)), | t| < ϵ, one has { dE( X t)/ dt} t=0 = 0. Any harmonic vector field is a smooth solution to the nonlinear elliptic PDE system

(P.2)

These are precisely the Euler-Lagrange equations associated to the constrained variational principle δ E( X) = 0 and g( X, X) = 1 ([309], [316]). Here Δ g is a second order elliptic operator acting on vector fields, e.g., if X is a C² vector field on M then locally

with respect to a local orthonormal frame { E i : 1 ≤ i ≤ n} of T( M). One may think of the covariant derivative ∇ X as a section in the vector bundle T*( M) ⊗ T( M) → M . If ∇* is the formal adjoint of ∇ i.e., (∇* φ, X) = ( φ, ∇ X) for any φ ∈ Ω ⁰( T*( M) ⊗ T( M, then Δ g = ∇*∇ and although (P.2) is nonlinear, an obvious notion of weak solution to (P.2) may be introduced. Here we made use of the L. A systematic study of weak solutions (e.g., existence and local properties) to (P.2) (the harmonic vector field by a standard mix of functional analysis and calculus of variations (e.g., the Eberlein-Smulian theorem on the characterization of reflexive Banach spaces, Kondrakov's compact embedding theorem, and existence theorems for minimizers of lower semicontinuous functionals). Here

are described in Section 2.11 of Chapter 2. Other important exceptions are Theorem 3.43 (due to E. Boeckx & L. Vanhecke, [51]) and Theorem 3.44 (due to G.~Nunes & J. Ripoll, [224]). Both results furnish examples of weak solutions to the harmonic vector fields system (cf. Definition 2.52 in Chapter 2) within interesting geometric contexts [E. Boeckx & L. Vanhecke's result is that radial vector fields on harmonic manifolds are weakly harmonic while G. Nunes & J. Ripoll's result is that normal vector fields to the principal orbits of a cohomogeneity one action (of a compact Lie group of isometries of a compact orientable Riemannian manifold) are weakly harmonic].

Harmonic vector fields aren't harmonic maps unless the additional curvature condition

(P.3)

is satisfied. The resulting theory of harmonic vector fields is similar in many respects to the more consolidated theory of harmonic maps yet presents new and intriguing aspects captured in a rapidly growing specific literature cf. e.g., E. Boeckx & L. Vanhecke, given by

This is the total bending functional, a measure of the failure of X ∈ Γ ∞ ( S( M)) to be parallel. A closer look at the properties of the Sasaki metric G s on S( M) shows however that the Dirichlet and total bending functionals are related

(P.4)

so that the theories in [316] and [309] are identical. The relation (P.4) may be used to show that the search for vector fields which are critical points of $E : Γ ∞ ( S( M)) → ℝ, rather than critical points of E : C∞ ( M, T( M, is the only appropriate choice. Indeed the only smooth vector fields which are critical points of E : C∞ ( M, T( M) are the parallel vector fields.

The authors' interest in the theory of harmonic vector fields arose in relationship to the study of the geometry of contact Riemannian manifolds (cf. [237], [238], [239], [240], [241], [242], [243], [244], [245], [246] and [247]) and of nonlinear subelliptic systems of variational origin appearing in the theory of Hörmander systems of vector fields (cf. e.g., J. Jost & C-J. Xu, [180]) and CR geometry (cf. e.g., E. Barletta et al., [25]).

The exposition of the material collected in this book is organized as follows. Chapter 1 is devoted to the basic geometric properties of the tangent bundle over a Riemannian manifold. A description of the tangent sphere bundle S¹ → S( T²) → T² over a torus T² and a classification of its smooth sections (up to homotopy) complete Chapter 1 (and prepare several instances where the general theory may be applied, cf. Section 2.6 and Section 2.10 in Chapter 2 and Section 3.9 in Chapter 3 of this book).

Chapter 2 presents the basic theoretic material while the remaining chapters deal mainly with applications and generalizations. From a technical point of view, the main achievement of Chapter 2 is perhaps the explicit expression (2.21) of the tension field

(P.5)

of a C∞ vector field X : M → T( M). Of course $X$ is thought of as a map of the Riemannian manifolds ( M, g) and ( T( M), G s) where G s is the Sasaki metric on T( Mis the first fundamental form of ι : S( M) ↪ T( M)). The proof of (P.5) is a rather involved calculation exploiting the relationship among the Levi-Civita connections of ( T( M), G s) and ( M, g) (related through the formalism of vertical and horizontal lifting and including curvature calculations, cf. e.g., [42], p. 139–141). A consequence of (P.5) (that is (2.27) in Theorem 2.19) may then be used to characterize unit vector fields which are harmonic maps (as the smooth unit vector fields satisfying (P.2) and (P.3). Given a unit tangent vector field X ∈ Γ ∞ ( S( M)) and a smooth 1-parameter variation X t ∈ Γ ∞ ( S( M)) of X, one derives (cf. (2.32) and (2.36) in Chapter 2) the first and second variation formulae

(P.6)

(P.7)

where V = { dU t/ dt} t=0. Related to (P.7) one discusses stability results for critical points of E : Γ ∞ ( S( M)) → ℝ. The Dirichlet problem

(P.8)

(P.9)

is considered in Section 2.9. Here Ω is a smoothly bounded strictly pseudoconvex domain in ℂ n (with n ≥ 2) endowed with the Bergman metric g. We report on a recent result by E. Barletta, [22], dealing with Cto (P.8) and (P.9). The existence problem is open so far.

of the metric on the base manifold. As argued in Section 2.10 the study is confined to the 2-dimensional case (and if M is 2-dimensional, compact, orientable and admits globally defined nowhere zero vector fields then M must be the torus T²) because the term (div( X) X − ∇ XX) u in the identity

(cf. (2.110) in Chapter 2) may be calculated in terms of the Gaussian curvature of M in dimension n = 2, a calculation which appears to admit no obvious analog in higher dimension. The main result is G. Wiegmink's (cf. op. cit..

Chapter 2 ends up with the construction of Sobolev type spaces of vector fields (appropriate for the study of weak solutions to (P.2).

The harmonicity and stability of Hopf and Killing vector fields is discussed in Chapter 3. There we introduce (following F. Brito, [71]) the functional

(P.10)

where H X . This is referred to as Brito's functional. Brito's functional is an attempt to avoid the difficulties arising from the fact that Hopf vector fields on a sphere S² m+1 are (by a result of C.M. Wood, [316]) unstable critical points of E : Γ ∞ ( S( S² m. Section 3.8 furnishes a detailed proof of the result by G. Nunes & J. Ripoll (cf. [224]) mentioned earlier in the preface. Section 3.9 of Chapter 3 reports on a beautiful result by G. Wiegmink, [309], giving a complete description of harmonic vector fields on a Riemannian torus.

Harmonicity and stability of special vector fields appearing on contact Riemannian manifolds (such as the Reeb vector field underlying a contact Riemannian structure) are studied in Chapter 4. One of the main notions in Chapter 4 is that of an H- contact manifold (a contact metric manifold whose Reeb vector is a harmonic vector field). In a long series of papers one of the authors of this monograph has emphasized (cf. [237], [238], [239], [240], [241], [242], [243], [244], [245], [246] and [247]) that H-contact metric manifolds possess special features and may be quite explicitly described, especially in real dimension 3.

For each Riemannian manifold ( M, g), its Sasaki metric G s belongs to a large family of Riemannian metrics on T( M), the family of Riemannian g(cf. [6]). In Chapter 5, we endow T( M) with an arbitrary Riemannian g-natural metric G and study smooth vector fields V on M thought of as maps of ( M, g) into ( T( M), G) (cf. [2]) i.e., we look at harmonicity of V as a map of ( M, g) into ( S( Mis the metric induced by G on S( M) (cf. [3], [250]). One decomposes the tension field into vertical and horizontal components and derives two equations describing the harmonicity of V : ( M, g) → ( S( M. In particular the equation (P.2) is invariant under a 4-parameter deformation of the Sasaki metric. The remaining equation is a natural generalization of (P.3).

Many of the results in Chapter 2, Chapter 3 and Chapter 4 admit versions holding for sections in Riemannan vector bundles (cf. e.g., J.J. Konderak, [194]). The main findings in this direction are presented in Chapter 6.

Chapter 7 is devoted to generalizations of the notion of a harmonic vector field within CR and pseudohermitian geometry (cf. e.g., G. Tomasini et al., [110]). Section 7.1, Section 7.2 and Section 7.3 are an attempt (based on the results in D. Perrone et al., [107]) to relate harmonicity of vector fields to the geometry of the Fefferman metric (a Lorentz metric on the total space of the canonical circle bundle over a strictly pseudoconvex CR manifold). Section 7.4, Section 7.5, Section 7.6 and Section 7.7 rely on results by Y. Kamishima et al., [103] and generalize harmonic vector fields in the spirit of the work by J. Jost & C-J. Xu, [180] and E. Barletta et al., [25] (dealing with generalizations of harmonic maps within the theory of Hörmander systems of vector fields and CR geometry). The treatment in Chapter 7 leads naturally to nonlinear subelliptic systems of variational origin and exhibits a nontrivial link among the differential geometry of harmonic vector fields and the analysis of subelliptic partial differential equations.

In Chapter 8 we discuss, within the framework of Lorentz geometry, a version of the energy functional (the spacelike energy) due to O. Gil-Medrano & A. Hurtado, [130], defined on reference frames (unit timelike vector fields) and the corresponding critical points (the spatially harmonic vector fields). These are ordinary harmonic vector fields when geodesic. The treatment is tentative, as the study of harmonic vector fields on a Lorentz manifold is still in its infancy. Nevertheless the subject looks promising, especially in its potential applications to the general relativity theory.

The Authors,

Potenza-Lecce, June 9, 2011.

Chapter One. Geometry of the Tangent Bundle

Contents

1.1 The Tangent Bundle 2

1.2 Connections and Horizontal Vector Fields 4

1.3 The Dombrowski Map and the Sasaki Metric 6

1.3.1 Preliminaries on Local Calculations 9

1.3.2 Isotropic Almost Complex Structures 11

1.3.3 Invariant Isotropic Complex Structures 20

1.4 The Tangent Sphere Bundle 26

1.5 The Tangent Sphere Bundle over a Torus 29

also carries a natural almost complex structure J is an almost Kähler manifold. The almost complex structure J(discovered by P. Dombrowski, 1961) is rarely integrable (in fact only when the base Riemannian manifold is locally Euclidean) yet J appears to be but one of the many isotropic contains a large subfamily of complex structures (among which the invariant is related to the properties of the twisted .

Keywords: Tangent bundle; nonlinear connection; Dombrowski map; Sasaki metric; isotropic almost complex structure; twisted Dolbeau complex; Bruschlinsky group; Riemannian torus.

The scope of this chapter is to briefly review the basic facts in the geometry of the tangent bundle T( M) over a Riemannian manifold ( M, g), such as nonlinear connections, the Dombrowski map and the Sasaki metric Gs. Remarkably T( M) also carries a natural almost complex structure J (arising from g) compatible to Gs and such that ( T( M), J, Gs) is an almost Kähler manifold. The almost complex structure J (discovered by P. Dombrowski, [99]) is rarely integrable (in fact only when the base Riemannian manifold is locally Euclidean) yet J appears to be but one of the many isotropic almost complex structures Jδ, σ built by R.M. Aguilar, [11]. On the other hand the existence of an integrable isotropic almost complex structure only requires that ( M, g) has constant sectional curvature and, if this is the case, the family Jδ, σ contains a large subfamily of complex structures (among which the invariant ones may be completely determined, cf. Theorem 1.20). When an almost complex structure Jδ, σ is non integrable the geometry of ( T( M), Jδ, σ) is related to the properties of the twisted Dolbeau complex (a description of which is given in Appendix A of this book). Further information on the geometry of T( M) (over a semi-Riemannian manifold M) is furnished in Chapter 7. Chapter 1 also contains the calculation (due to G. Wiegmink, [309]) of the Bruschlinsky group of a torus T² endowed with an arbitrary Riemannian metric leading to the classification up to homotopy of the unit tangent vector fields on T². For the classical aspects of the geometry of the tangent bundle over a Riemannian manifold (the Sasaki metric, the almost contact metric structure of the unit tangent bundle, etc.) the reader may also consult the books by K. Yano & S. Ishihara, [324] and D.E. Blair, [42].

1.1. The Tangent Bundle

Let M be a real n-dimensional C∞ manifold and π : T( M) → M is a local coordinate system on M be the naturally induced local coordinates on T( M) i.e.,

Hence T( M) is a real 2 nmanifold. We set

(defined on U) and x). Let

is an embedding of M is the vertical space at vis a vertical vector. The assignment

distribution of rank n on T( M.

Definition 1.1

is called the vertical distribution on T( M) is a vertical vector field on T( M).

.

is

such that the diagram

and p .

Definition 1.2

be a tangent vector field on M, is called the natural lift of X.

.

By a customary language abuse, one may identify the tangent bundle T( Mboth have rank n be the curve given by

be the map given by

be a local coordinate system on M . Then

In particular, γv .

Definition 1.3

is called the vertical lift.

the vertical vector field γ s is the vertical lift of s. Note that in general s may fail to be the natural lift of a vector field on M. The vertical lift of X.

1.2. Connections and Horizontal Vector Fields

Let ∇ be a linear connection on Mon M be the local connection coefficients i.e.,

Any linear connection ∇ on M be a local coordinate system on M be the induced local coordinates on T( M). We set by definition

(1.1)

It may be easily checked that the definition doesn't depend upon the choice of local coordinates on M and that .

Definition 1.4

on T( M) is called a nonlinear connection on T( M) if

(1.2)

is the vertical distribution.

on T( M) is also referred to as a horizontal distribution on T( Mis the corresponding horizontal bundle. By (1.2) any horizontal distribution on T( M) has rank n.

given by

Locally

Moreover the following sequence of vector bundles and vector bundle morphisms

on T( M) the restriction of L .

Definition 1.5

be a nonlinear connection on T( Mis called the horizontal lift ).

be a local coordinate system on M be a nonlinear connection on T( M). Then

. Applying L are the local coefficients .

Definition 1.6

defined by

is called the Liouville vector.

by setting

is any smooth extension of A to T( Mvector field on T( M.

Definition 1.7

is said to be regular is a nonlinear connection on T( M).

Proposition 1.8

For any linear connection ∇ on M the induced connection in is regular.

Proof

Let A be a tangent vector field on T( Mif and only if

(1.3)

is locally the span of

(1.4)

distribution of rank n on T( M). Therefore to check .

An inspection of (1.4) also shows that

Corollary 1.9

The local coefficients of the nonlinear connection on T( M) determined by a linear connection ∇ on M are given by .

be a nonlinear connection on T( Mis a horizontal vectoris a horizontal vector field on T( M). For any section s is the horizontal lift of s. Given a vector field X on M its horizontal lift of X.

1.3. The Dombrowski Map and the Sasaki Metric

Definition 1.10

be a nonlinear connection on T( Mdefined by

is called the Dombrowski map .

be a nonlinear connection on T( M. It is an easy consequence of definitions that the sequence of vector bundles and vector bundle morphisms

is exact. Locally

.

Let g be a Riemannian metric on Mbe a local coordinate system on M . Let Xi . We set

.

Definition 1.11

be a nonlinear connection on T( M) and g a Riemannian metric on M. The Riemannian metric Gs on T( M) defined by

(1.5)

is called the Sasaki metric on T( M.

Of course the Sasaki metric Gs may be thought of as induced by the given Riemannian metric g on M is but the lifting tool (from M to T( M)).

be a nonlinear connection on T( M-tensor field on T( M) defined by

(1.6)

It is then immediate that

Proposition 1.12

Let ( M, g) be a Riemannian manifold. For any nonlinear connection on T( M) the synthetic object is an almost Hermitian manifold.

Proof

as a consequence of (1.6), where I , so that J is an almost complex structure on T( M.

It may be easily shown that

Proposition 1.13 P. Dombrowski, [99]

J is integrable if and only if ( M, g) is locally Euclidean.

Our convention for the sign of the curvature tensor field R is

.

Proposition 1.14

Let ( M, g) be a Riemannian manifold and a regular connection in . Let be the nonlinear connection on T( M) associated to and the corresponding horizontal lift. Let Gsbe the Sasaki metric associated to the pair . Then the Levi-Civita connection D of is expressed by

for any .

For a proof of Proposition 1.14 one may see [42], p. 139–140.

given by (1.6) was introduced by P. Dombrowski, [99]. It is but one of a larger family of isotropic almost complex structures Jδ, σ due to R.M. Aguilar, [11]. Any isotropic almost complex structure determines an almost Kähler metric whose Kähler 2-form is the pullback to T( M. Moreover isotropic complex structures exist precisely when M has constant sectional curvature (cf. Theorem 1 in [11], p. 431). The remainder of this section is devoted to a brief presentation of the results in [11], which are believed (as we argue later on) to offer the possibility of further development for the harmonic vector field theory.

1.3.1. Preliminaries on Local Calculations

Let ( M, gbe a local orthonormal frame of T( Mbe the dual coframe on U is the corresponding dual coframe. We set

be the 1-form on T( M) defined by

Then

(1.7)

are given by

It may be easily shown that

(1.8)

. Then

(1.9)

be the connection 1-forms of ∇ (the Levi-Civita connection of ( M, gthe curvature 1-forms i.e.,

Then (by the structure equations for ∇)

(1.10)

(1.11)

Next we establish the following

Lemma 1.15 R.M. Aguilar, [11]

With the notations above

(1.12)

on .

Proof

Let us take the exterior derivative of (1.7)

and