Mathematics 241A

Introduction to Global Analysis

John Douglas Moore

Department of Mathematics
University of California
Santa Barbara, CA, USA 93106
e-mail: [email protected]

Fall, 2010
Preface

These are slightly revised lecture notes for the graduate course, Topics in Ge-
ometry, given at UCSB during the fall quarter of 2009. We intend to include
additional topics later. It might be helpful to start with an overview of the
subject presented.
Morse theory might have developed in three main stages, although as events
transpired, the three stages were actually intertwined.
The first stage should have been finite-dimensional Morse theory, which re-
lates critical points of proper nonnegative functions on finite-dimensional man-
ifolds to the topology of these manifolds. Indeed, the foundations for this were
laid in Marston Morse’s first landmark article on Morse theory [55], but he
quickly turned his attention to problems from the calculus of variations, which
ultimately became part of the infinite-dimensional theory. In subsequent devel-
opments, finite-dimensional Morse theory became a primary tool for studying
the topology of finite-dimensional manifolds and has had many successes, in-
cluding the celebrated h-cobordism theorem of Smale [51], which settled the
generalized Poincaré conjecture in dimensions greater than five: any compact
manifold of dimension at least five which is homotopy equivalent to a sphere
must be homeomorphic to a sphere. Modern expositions of finite-dimensional
Morse theory often construct a chain complex from the free abelian group gener-
ated by the critical points, the boundary being defined by orbits of the gradient
flow which connect the critical points. The homology of this chain complex,
called Morse homology, is isomorphic to the usual integer homology of the man-
ifold; see [74] for example.
What might have been the second stage—the Morse theory of geodesics—
formed the core of what Morse [56] called “the calculus of variations in the
large.” Morse was interested in studying the critical points of the length func-
tion or action function on the space of paths joining two points in a Riemannian
manifold, the critical points being geodesics. His idea was to approximate the
infinite-dimensional space of paths by a finite-dimensional manifold of very high
dimension, and then apply finite-dimensional Morse theory to this approxima-
tion. As explained in Milnor’s classical book on Morse theory [50], this approach
produced many striking results in the theory of geodesics in Riemannian geome-
try, such as the theorem of Serre that any two points on a compact Riemannian
manifold can be joined by infinitely many geodesics. It also provided the first
proof of the Bott periodicity theorem from homotopy theory. One might regard

i
the Morse theory of geodesics as an application of topology to the study of
ordinary differential equations, in particular, to those equations which like the
equation for geodesics, arise from classical mechanics. Thus Morse theory arises
from the very core of “applied mathematics.”
Palais and Smale were able to provide an elegant reformulation of the Morse
theory of geodesics in the language of infinite-dimensional Hilbert manifolds
[62]. They showed that the action function on the infinite-dimensional manifold
of paths satisfies “Condition C,” a condition replacing “proper” in the finite-
dimensional theory, which they showed is sufficient for the development of Morse
theory in infinite dimensions. This became the standard approach to the Morse
theory of geodesics during the last several decades of the twentieth century.
Future historians will most likely regard the third stage of Morse theory
as encompassing many strands, but our viewpoint is to focus on techniques
for applying Morse theory to nonlinear elliptic partial differential equations
coming from the calculus of variations in which the domain is a two-dimensional
compact surface. Morse himself hoped to apply the ideas of his theory to a
central case—the partial differential equations for minimal surfaces in Euclidean
space. The first steps in this direction were taken by Morse and Tompkins, as
well as Shiffman, who established the theorem that if a simple closed curve in
Euclidean space R3 bounds two stable minimal disks, it bounds a third, which is
not stable. This provided a version of the so-called “mountain pass lemma” for
minimal disks in Euclidean space. The results of Morse, Tompkins and Shiffman
suggested that Morse inequalities should hold for minimal surfaces in Euclidean
space with boundary constrained to lie on a given Jordan curve, and indeed,
such inequalities were later established under appropriate hypotheses [41].
But the most natural extension of the Morse theory of geodesics to the realm
of partial differential equations would be a Morse theory of two-dimensional
minimal surfaces in a more general curved ambient Riemannian manifold M ,
instead of the ambient Euclidean space used in the classical theory of minimal
surfaces. The generalization to a completely general ambient space requires
new techniques. Unfortunately, if Σ is a connected compact surface, it becomes
somewhat awkward to extend the finite-dimensional approximation procedure—
so effective in the theory of geodesics—to the space of mappings Map(Σ, M )
from Σ to M . One might hope that a better approach would be based upon the
theory of infinite-dimensional manifolds, as developed by Palais and Smale, but
a serious problem is encountered: the standard energy function

E : Map(Σ, M ) → R,

used in the theory of harmonic maps and parametrized minimal surfaces, fails to
satisfy the Condition C which Palais and Smale had used so effectively in their
theory, when Map(Σ, M ) is completed with respect to a norm strong enough to
lie within the space of continuous functions.
To get around this difficulty, Sacks and Uhlenbeck introduced the α-energy
[68], [69], a perturbation of the usual energy which does satisfy Condition C
when Map(Σ, M ) is completed with respect to a Banach space norm which

ii
is both weak enough to satisfy condition C and strong enough to have the
same homotopy type as the space of continuous maps from Σ to M . In simple
terms, we could say that the α-energy lies within “Sobolev range.” The α-
energy approaches the usual energy (plus a constant) as the parameter α in
the perturbation goes to one, and we can therefore say that the usual energy
is “on the border of Sobolev range.” Using the α-energy, Sacks and Uhlenbeck
were able to establish many striking results in the theory of minimal surfaces in
Riemannian manifolds, including the fact that any compact simply connected
Riemannian manifold contains at least one nonconstant minimal two-sphere,
which parallels the classical theorem of Fet and Lyusternik stating that any
compact Riemannian manifold contains at least one smooth closed geodesic. But
they also discovered the phenomenon of “bubbling” as α → 1, which prevents
Morse inequalitites from holding for compact parametrized minimal surfaces in
complete generality.
A somewhat different approach to existence of parametrized minimal sur-
faces in Riemannian manifolds was developed at about the same time by Schoen
and Yau [72], using Morrey’s solution to the Plateau problem in a Riemannian
manifold and arguments based upon a “replacement procedure.” Their approach
yielded striking theorems relating positive scalar curvature to topology of three-
manifolds, and was a step toward the first proof of the positive mass theorem of
general relativity. The Schoen-Yau replacement procedure can also be used to
obtain many of the existence results of Sacks and Uhlenbeck, and an alternate
treatment of many of their theorems is provided by Jost [39], as well as by other
authors, who use yet different techniques, including heat flow.
However, the approach via Sacks-Uhlenbeck perturbation seems to provide
the strongest link with the global analysis approach to nonlinear partial differ-
ential equations, and the clearest insight into bubbling, the phenomenon men-
tioned above which is observed as the perturbation is turned off. Bubbling
appears to be an essential component of any complete critical point theory of
minimal surfaces within Riemannian manifolds, and this phenomenon also ap-
pears in the study of other nonlinear partial differential equations, such as the
Yang-Mills equation on four-dimensional manifolds.
If one were to replace the Riemannian metric on the ambient manifold by a
conformal equivalence class of such metrics, the theory of two-dimensional min-
imal surfaces would become part of a broader context—nonlinear partial differ-
ential equations which are conformally invariant and lie on the border of Sobolev
range. These equations include the Yang-Mills equations on four-dimensional
manifolds from the standard model for particle physics, the anti-self-dual equa-
tions used so effectively by Donaldson, the Seiberg-Witten equations, and Gro-
mov’s equations for pseudoholomorphic curves. A technology has been devel-
oped for studying these equations. First one needs to develop a transversality
theory using Smale’s generalization of Sard’s theorem from finite-dimensional
differential topology. This generally shows that in generic situtations solutions
to the nonlinear partial differential equation form a finite-dimensional submani-
fold of an infinite-dimensional function space. The tangent space to this subman-
ifold is studied via the linearization of the nonlinear partial differential equation

iii
at a given solution; often, the dimension of the tangent space is obtained by
application of the Atiyah-Singer index theorem, which reduces to the Riemann-
Roch theorem in the case of parametrized minimal surfaces. Next one develops
a suitable compactness theorem. Finally, one uses topological and geometric
methods to derive important geometrical conclusions (for example, existence of
minimal surfaces) or to construct differential topological invariants of manifolds
(as in Seiberg-Witten theory).
As mentioned before, for minimal surfaces, bubbling implies that the most
obvious extension of the Morse inequalities to minimal surfaces in Riemannian
manifolds cannot hold, but also provides the framework for analyzing how the
Morse inequalities fail.
Not only does Sacks-Uhlenbeck bubbling interfere with the Morse inequali-
ties, but when minimizing area one must allow for variations not only over the
space of functions but over the conformal structure on the surface, an element
of Teichmüller space, and a sequence of harmonic maps may degenerate as the
conformal structure moves to the boundary. Moreover, branched coverings of
a given minimal surfaces count as new critical points within the space of func-
tions, although they are not geometrically distinct from the covered surface.
Finally, the energy is invariant under the action of the mapping class group, so
the energy descends to a function on the quotient space, a space which has a
more complicated topology than that of Map(Σ, M ) when the genus of Σ is at
least two. One might suspect that a procedure for constructing minimal surfaces
that can fail in several different ways is too flawed to be of much use. However,
we argue that a different perspective is more productive—since the minimax
procedure for a given homology class does not always yield interesting minimal
surfaces, one should divide homology classes into various categories depending
upon which of the possible difficulties will likely arise.
It is our purpose here to provide some of the foundations for such a theory,
a theory which promises important applications similar to those found in the
Morse theory of geodesics.
It has long been known that that the extension of Morse theory to infinite-
dimensional manifolds is not really necessary for the study of geodesics. Indeed,
Bott expresses it this way in his beautiful survey article on Morse theory [9]
in 1982: ”I know of no aspect of the geodesic question where [the infinite-
dimensional approach] is essential; however it clearly has some aesthetic advan-
tages, and points the way for situations where finite dimensional approximations
are not possible...” One could imagine constructing a finite-dimensional approx-
imation suitable for studying the α-energy when α > 1, but it would be far more
awkward than the infinite-dimensional manifold of maps, and an analysis of how
it breaks down as α → 1 seems to require a study of infinite-dimensional mani-
fold pieces. Thus, for any partial Morse theory of minimal surfaces, in contrast
to closed geodesics, calculus on infinite-dimensional manifolds appears to play
an essential conceptual and simplifying role.
We assume the reader is familiar with the basics of finite-dimensional dif-
ferential geometry, including geodesics, curvature, the tubular neighborhood
theorem, and the second variation formula for geodesics. We will also assume

iv
some familiarity with basic complex analysis, the foundations of Banach and
Hilbert space theory, and the willingness to accept results from the linear the-
ory of elliptic partial differential operators, in particular the theory of Fredholm
operators on Sobolev spaces. All of these topics are treated very well in highly
accessible sources, to which we can refer at the appropriate time.
We begin with an overview of global analysis on infinite-dimensional man-
ifolds of maps. The second chapter reviews the theory of geodesics on Rie-
mannian manifolds which owes much to the pioneering work of Bott, Gromoll,
Klingenberg and Meyer. We then turn to minimal surfaces, providing a brief
introduction to the key theorems of Sacks, Uhlenbeck, Meeks, Schoen and Yau
which helped elucidate the topology of three-dimensional manifolds. A final
chapter is planned that will cover the bumpy metric theorem of [52] and some
of its immediate applications. Starred sections can be omitted without loss of
continuity.

Doug Moore

Santa Barbara, CA, December, 2010

v
Contents

1 Infinite-dimensional manifolds 1
1.1 A global setting for nonlinear DE’s . . . . . . . . . . . . . . . . . 1
1.2 Infinite-dimensional calculus . . . . . . . . . . . . . . . . . . . . . 2
1.3 Infinite-dimensional manifolds . . . . . . . . . . . . . . . . . . . . 15
1.4 The basic mapping spaces . . . . . . . . . . . . . . . . . . . . . . 23
1.5 Homotopy type of the space of maps . . . . . . . . . . . . . . . . 29
1.6 The α-Lemma* . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.7 The tangent and cotangent bundles . . . . . . . . . . . . . . . . . 36
1.8 Differential forms . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
1.9 Riemannian and Finsler metrics . . . . . . . . . . . . . . . . . . . 43
1.10 Vector fields and ODE’s . . . . . . . . . . . . . . . . . . . . . . . 47
1.11 Condition C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
1.12 Topological constraints give critical points . . . . . . . . . . . . . 53
1.13 de Rham cohomology . . . . . . . . . . . . . . . . . . . . . . . . 56

2 Morse Theory of Geodesics 63

2.1 Geodesics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.2 Condition C for the action . . . . . . . . . . . . . . . . . . . . . . 66
2.3 Existence of smooth closed geodesics . . . . . . . . . . . . . . . . 71
2.4 Second variation . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.5 Morse nondegenerate critical points . . . . . . . . . . . . . . . . . 78
2.6 The Sard-Smale Theorem . . . . . . . . . . . . . . . . . . . . . . 82
2.7 Existence of Morse functions . . . . . . . . . . . . . . . . . . . . 85
2.8 Bumpy metrics for smooth closed geodesics* . . . . . . . . . . . . 89
2.9 Adding handles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
2.10 Morse inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . 100
2.11 The Morse-Witten complex . . . . . . . . . . . . . . . . . . . . . 104

3 Harmonic and minimal surfaces 110

3.1 The energy of a smooth map . . . . . . . . . . . . . . . . . . . . 110
3.2 Minimal surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
3.3 Minimal surfaces of higher genus . . . . . . . . . . . . . . . . . . 122
3.4 The Bochner Lemma . . . . . . . . . . . . . . . . . . . . . . . . . 129
3.5 The α-energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

vi
 3.6 Regularity of (α, ω)-harmonic maps . . . . . . . . . . . . . . . . . 138
3.7 Morse theory for the perturbed energy . . . . . . . . . . . . . . . 141
3.8 Local control of energy density . . . . . . . . . . . . . . . . . . . 147
3.9 Bubbling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
3.10 Existence of minimizing spheres . . . . . . . . . . . . . . . . . . . 156
3.11 Existence of minimal tori . . . . . . . . . . . . . . . . . . . . . . 163
3.12 Higher genus minimal surfaces* . . . . . . . . . . . . . . . . . . . 166
3.13 Complex form of second variation . . . . . . . . . . . . . . . . . . 168
3.14 Isotropic curvature . . . . . . . . . . . . . . . . . . . . . . . . . . 170

Bibliography 176

vii
Chapter 1

Infinite-dimensional
manifolds

1.1 A global setting for nonlinear DE’s

Linear differential equations are often fruitfully studied via techniques from lin-
ear functional analysis, including the theory of Banach and Hilbert spaces. In
contrast, the proper setting for an important class of nonlinear partial differen-
tial equations is a nonlinear version of functional analysis, which is based upon
infinite-dimensional manifolds modeled on Banach and Hilbert spaces. The the-
ory of such manifolds was developed by Eells, Palais and Smale among others in
the 1950’s and 1960’s, and has proven to be extremely useful for understanding
many of the nonlinear differential equations which arise in geometry, such as
1. the equation for geodesics in a Riemannian manifold,

2. the equation for harmonic maps from surfaces into a Riemannian manifold,
or for minimal surfaces in a Riemannian manifold,
3. the equations for pseudoholomorphic curves in a symplectic manifold,
4. and the Seiberg-Witten equations.

In all of these examples, the solutions can be expressed as critical points of

a real-valued function (often called the action or the energy) defined on an
infinite-dimensional manifold, such as the function space manifolds described
in the following pages. In favorable cases, a gradient (or pseudogradient) of
the action or energy can then be used to locate critical points (solutions to the
differential equations) via what is often called the “method of steepest descent.”
For this procedure to work, the topology on the function space must be
strong enough for the action or energy to be differentiable, yet weak enough to
force convergence of a sequence which is tending toward a minimum (or to a
minimax solution for a given constraint). The two conflicting conditions often

1
select a unique acceptable topology for the space of functions. In the most
favorable circumstances, the topology is strong enough so that it lies within
the space of continuous functions, a space which has been studied extensively
by topologists. The function space is then often homotopy equivalent, that is
equivalent in the sense of homotopy theory, to the space of continuous functions.
This makes some existence questions within the theory of nonlinear differential
equations accessible by topological methods.
In the case of the Seiberg-Witten equations, the logic is reversed. Instead of
topology shedding light on existence of solutions to partial differential equations,
it is the space of solutions to the Seiberg-Witten equations that enable one to
distinguish between different smooth structures on smooth four-manifolds (as
explained in [54]). This illustrates that at a very fundamental level, topology
and nonlinear partial differential equations are closely related, and underscores
the importance of developing a global theory of partial differential equations
based upon the theory of infinite-dimensional manifolds.

1.2 Infinite-dimensional calculus

Our first topic is the theory of infinite-dimensional manifolds. We refer to the
excellent presentations of Lang [43] or of Abraham, Marsden and Ratiu [2] for
further elaboration of topics only briefly introduced in the following pages.
It was pointed out by Smale, Abraham, Lang and others in the 1960’s that
several variable calculus could be developed not just in finite-dimensional Eu-
clidean spaces, but also with very little extra work within the context of infinite-
dimensional Hilbert and Banach spaces. Most of the proofs of theorems are
straightforward modifications of the proofs in Rn , so we will go very rapidly
over this basic material.
Definition. A pre-Hilbert space is a real vector space E together with a function
h·, ·i : E × E → R which satisfies the following axioms:
1. hx, yi = hy, xi, for x, y ∈ E.

2. hax, yi = ahx, yi, for a ∈ R and x, y ∈ E.

3. hx + y, zi = hx, zi + hy, zi, for x, y, z ∈ E.
4. hx, xi ≥ 0, for x ∈ E, equality holding only when x = 0.

These axioms simply state that hx, yi is a positive-definite symmetric bilinear

form on E.
p such a pre-Hilbert space, we can define a map k · k : E → R by
Given
kxk = hx, xi. We can regard E as a metric space by defining the distance
between elements x and y of E to be d(x, y) = kx − yk.
Definition. A Hilbert space is a pre-Hilbert space which is complete in terms
of the metric d.

2
An example of a finite-dimensional Hilbert spaces is Rn with inner product h·, ·i
defined by
h(x1 , . . . , xn ), (y1 , . . . yn )i = x1 y1 + · · · + xn yn .
An example of an infinite dimensional Hilbert space is the space L2 ([0, 1], R)
studied in basic analysis courses. To construct it, one starts with defining an
inner product h·, ·i on the space
C ∞ ([0, 1], R) = { C ∞ functions f : [0, 1] → R }
by Z 1
hφ, ψi = φ(t)ψ(t)dt.
0
This inner product satisfies the first four of the above axioms but not the last
one. We let L2 ([0, 1], R) denote the equivalence classes of Cauchy sequences
from C ∞ ([0, 1], R), two Cauchy sequences {φi } and {ψi } being equivalent if for
each  > 0 there is a positive integer N such that
i, j > N ⇒ kφi − ψi k < . (1.1)
Equivalence classes of sequences form a vector space, and we define an inner
product h·, ·i on L2 ([0, 1], R) by
h{φi }, {ψi }i = lim hφi , ψi i.
i,i→∞

We say that L2 ([0, 1], R) is the completion of C ∞ ([0, 1], R) with respect to h·, ·i.
The process we have described is virtually the same as that often used to con-
struct the real numbers from the rationals.
Definition. A pre-Banach space is a vector space E together with a function
k · k : E → R which satisfies the following axioms:
1. kaxk = |a|kxk, when a ∈ R and x ∈ E.
2. kx + yk ≤ kxk + kyk, when x, y ∈ E.
3. kxk ≥ 0, for x ∈ E
4. kxk = 0 only if x = 0.
A function k · k : E → R which satisfies the first three axioms is called a
seminorm on E. If, in addition, it satisfies the fourth axiom it is called a norm.
As in the case of Hilbert spaces, we can make E into a metric space by
defining the distance between elements x and y of E to be d(x, y) = kx − yk.
Definition. A Banach space is a pre-Banach space which is complete in terms
of the metric d.

p every Hilbert space is a Banach space with norm k · k defined by

Of course,
kxk = hx, xi. Let
C 0 ([0, 1], R) = { continuous functions f : [0, 1] → R },

3
and define

k · k : C 0 ([0, 1], R) → R by kf k = sup{|f (t)| : t ∈ [0, 1]}.

Then k · k makes C 0 ([0, 1], R) into a Banach space. More generally, we can
consider the space

C k ([0, 1], R) = { functions f : [0, 1] → R which

have continuous derivatives up to order k },

a Banach space with respect to the norm

( k )
X
k (i)
k · kk : C ([0, 1], R) → R defined by kf kk = sup |f (t)| : t ∈ [0, 1] ,
i=0
(1.2)
where f (i) (t) denotes the derivative of f of order i.
When 1 ≤ p < ∞ and p 6= 2, the spaces Lp ([0, 1], R) studied in basic
analysis courses are Banach spaces which are not Hilbert spaces. To construct
these spaces, one starts with defining a function k · k on the space C ∞ ([0, 1], R)
of C ∞ functions f : [0, 1] → R by
Z 1 1/p
p
kφk = |φ(t)| dt ,
0

which is shown to be a norm by means of the Minkowski inequality. This agrees

with the norm previously defined on L2 ([0, 1], R) when p = 2. Just as in the
construction of L2 ([0, 1], R), one can use this norm to define Cauchy sequences,
and let Lp ([0, 1], R) be the set of equivalence classes of Cauchy sequences, where
two Cauchy sequences {φi } and {ψi } are equivalent if for each  > 0 there is a
positive integer N such that (1.1) holds. The basic properties of Lp spaces are
treated in standard references on functional analysis; thus for the Hölder and
Minkowski inequalities, for example, one can refer to Theorem III.1 of [65].
Each Banach space E has a metric

d:E×E →R defined by d(e1 , e2 ) = ke1 − e2 k,

and we say that a subset U of E is open if

p∈U ⇒ B (p) = {q ∈ E : d(p.q) < } ⊂ U,

for some  > 0. A map T : E1 → E2 between Banach spaces is continuous if

T −1 (U ) is open for each open U ⊂ E2 , or equivalently, if there is a constant
c > 0 such that
kT (e1 )k2 ≤ cke1 k1 , for all e1 ∈ E1 ,
where k · k1 and k · k2 are the norms on E1 and E2 respectively.

4
 We let L(E1 , E2 ) be the space of continuous linear maps T : E1 → E2 , a
Banach space in its own right under the norm

kT k = sup{kT (e1 )k : e1 ∈ E1 , ke1 k = 1}.

It is easily shown that L(E1 , E2 ) is the same as the space of linear maps from
E1 to E2 , which are bounded in terms of this norm. In particular, we can define
the dual of a Banach space E to be E ? = L(E, R). We say that a Banach space
is reflexive if (E ? )? is isomorphic to E. It is proven in analysis courses that
Lp ([0, 1], R) is reflexive when 1 < p < ∞ but L1 ([0, 1], R) and C 0 ([0, 1], R) are
not.
Banach spaces and continuous linear maps form a category, as do Hilbert
spaces and continuous linear maps. The subject functional analysis is con-
cerned with properties of the categories of Hilbert spaces, Banach spaces and
more general spaces of functions, and is one of the major tools in studying linear
partial differential equations. Key theorems from the theory of Banach spaces
include the Open Mapping Theorem, the Hahn-Banach Theorem and the Uni-
form Boundedness Theorem. These theorems are discussed in [65], [66] and [67];
we will need to use the statements of these theorems and their consequences.
The Open Mapping Theorem states that if T : E1 → E2 is a continuous
surjective map between Banach spaces, it takes open sets to open sets. Thus if T
is a continuous bijection, its inverse is continuous. The Hahn-Banach Theorem
implies that if e is a nonzero element of a Banach space E, then there is a
continuous linear function λ : E → R such that λ(e) 6= 0. A bilinear map
B : E1 × E2 → F is said to be continuous if there is a constant c > 0 such that

kB(e1 , e2 )k ≤ cke1 k1 ke2 k1 , for all e1 ∈ E1 and all e2 ∈ E2 .

One of the consequences of the Uniform Boundedness Theorem is that such a

Bilinear map is continuous is and only if

B(·, e2 ) : E1 → F and B(e1 , ·) : E1 → F

are continuous for each e1 ∈ E1 and each e2 ∈ E2 . The three theorems are
almost trivial to prove for finite-dimensional Banach spaces, but the proofs are
more subtle for infinite-dimensional Banach spaces.
We now turn to the question of how to develop differential calculus for func-
tions defined on Banach spaces. It is actually the topology, or the equivalence
class of norms on the Banach space, that is important for calculus, two norms
k · k1 and k · k2 on a linear space E being equivalent if there is a constant c > 1
such that
1
kxk1 < kxk2 < ckxk1 , for x ∈ E.
c
Lang [43] calls a vector space E a Banachable space if it is endowed with an
equivalence class of Banach space norms. However, most other authors do not
use this term, and we will simply use the simpler term Banach space, it being
understood however, that we may sometimes pass to an equivalent norm in the

5
middle of an argument, when it is the underlying vector space with its topology,
the “topological vector space”—not the norm itself—that is important.
Definition. Suppose that E1 and E2 are Banach spaces, and that U is an open
subset of E1 . A continuous map f : U → E2 is said to be differentiable at the
point x0 ∈ U if there exists a continuous linear map T : E1 → E2 such that

kf (x0 + h) − f (x0 ) − T (h)k

lim = 0,
khk→0 khk

where k · k denotes both the Banach space norms on E1 and E2 . We will call T
the derivative of f at x0 and write Df (x0 ) for T . Note that the derivative can
also be calculated from the formula
f (x0 + th) − f (x)
Df (x0 )h = lim .
t→0 t

Just as in ordinary calculus, the derivative Df (x0 ) determines the linearization

of f near x0 , which is the affine function

f˜(x) = f (x0 ) + Df (x0 )(x − x0 )

which most closely approximates f near x0 .

If f is differentiable at every x ∈ U , a derivative Df (x) is defined for each
x ∈ U and thus we have a set-theoretic map Df : U → L(E1 , E2 ). If this map
Df is continuous, we can also ask whether it is differentiable at x0 ∈ U . This
will be the case if there is a continuous linear map T : E1 → L(E1 , E2 ) such
that
kDf (x0 + h) − Df (x0 ) − T (h)k
lim = 0,
khk→0 khk
in which case we write D2 f (x0 ) for T and call D2 f (x0 ) the second derivative of
f at x0 . Note that

D2 f (x0 ) ∈ L(E1 , L(E1 , E2 )) = L2 (E1 , E2 )

= {continuous bilinear maps T : E1 × E1 → E2 },

which can also be made into a Banach space in an obvious way.

We say that a function f : U → E2 , where U is an open subset of E1 , is
1. C 0 if it is continuous.
2. C 1 if it is continuous and differentiable at every x ∈ U , and Df : U →
L(E1 , E2 ) is continuous.

3. C k for k ≥ 2 if it is C 1 and Df : U → L(E1 , E2 ) is C k−1 .

4. C ∞ or smooth if it is C k for every nonnegative integer k.

6
 With the above definition of differentiation, many of the arguments in several
variable calculus can be transported without difficulty to the Banach space
setting, as carried out in detail in [43] or [2]. For example, the Leibniz rule for
differentiating a product carries over immediately to the infinite-dimensional
setting:.
Proposition 1.2.1. Suppose that B : F1 × F2 → G is a continuous bilinear
map between Banach spaces, that U is an open subsets of a Banach space E
and
f : U1 −→ F1 , f2 : U2 −→ F2
are C 1 maps. Then e 7→ g(e) = B(f1 (e), f2 (e)): is a C 1 map, and

Dg(x0 )h = B(Df1 (x0 )h, f2 (e)) + B(f1 (x0 )Df2 (x0 )h).

Sketch of Proof: To simplify notation, we write

g(x) = B(f1 (x), f2 (x)) = f1 (x) · f2 (x).

Then

g(x + h) − g(x) = (f1 (x + h) − f1 (x))f2 (x + h) + f1 (x)(f2 (x + h) − f2 (x)),

and hence
g(x + h) − g(x) f1 (x + h) − f1 (x) f2 (x + h) − f2 (x)
= f2 (x + h) + f1 (x) .
khk khk khk

The Proposition follows by taking the limit as khk → 0.

The following lemma is called the chain rule.
Proposition 1.2.2. If U1 and U2 are open subsets of Banach spaces E1 and
E2 and
f : U1 −→ U2 , g : U2 −→ E3
are C 1 maps, then so is g ◦ f : U1 → E3 , and

D(g ◦ f )(x0 ) = Dg(f (x0 ))Df (x0 ), for x0 ∈ U1 .

Sketch of Proof: We have

f (x0 + h) = f (x0 ) + Df (x0 )h + o(h),

where the symbol o(h) stands for an element in E2 such that o(h)/khk → 0 as
khk → 0. Similarly,

g(f (x0 ) + k) = g(f (x0 )) + Dg(f (x0 ))(k) + o(k).

7
Setting k = Df (x0 )h + o(h) yields

g(f (x0 + h)) = g(f (x0 )) + Dg(f (x0 ))(Df (x0 )h + o(h)) + o(k).

One checks without difficulty using continuity of g that an o(k) expression,

where k is a bounded linear function of h, is also o(h), and hence

g(f (x0 + h)) = g(f (x0 )) + Dg(f (x0 ))Df (x0 )h + o(h),

which gives the desired conclusion.

By induction, one immediately shows that the composition of C k maps is C k
and the composition of C ∞ maps is C ∞ .
Example 1.2.3. We suppose that the domain of the function is the Banach
space E = Lp (S 1 , RN ), the completion of the space C ∞ (S 1 , RN ) of smooth
RN -valued functions on S 1 with respect to the Lp -norm
Z 1/p
p
kφkLp = |φ(t)| dt , for φ ∈ Lp (S 1 , RN ).
S1

Here S 1 is regarded as the quotient of the interval [0, 1] obtained by identifying

the points 0 and 1, and possesses the standard measure dt with respect to which
S 1 has measure one. A useful tool for dealing with functions in the Lp spaces
is the Hölder inequality which states: if φ ∈ Lp , ψ ∈ Lq and
1 1 1
+ = where p, q, r ≥ 1, then φψ ∈ Lr and kφψkLr ≤ kφkLp kψkLq .
p q r
Using this inequality and the chain rule, it is not difficult to show that when
p ≥ 2, the function
Z
f : E −→ R defined by f (φ) = (1 + |φ(t)|2 )p/2 dt
S1

is C 2 and that its first and second derivatives are given by the formulae
Z
Df (φ)(ψ) = p (1 + |φ(t)|2 )p/2−1 φ(t) · ψ(t)dt
S1

and
Z
(D2 f )(φ)(ψ1 , ψ2 ) = p (1 + |φ(t)|2 )p/2−1 ψ1 (t) · ψ2 (t)dt
S1
Z
+ p(p − 2) (1 + |φ(t)|2 )p/2−2 (φ(t) · ψ1 (t))(φ(t) · ψ2 (t))dt.
S1

Indeed, to carry this out, we apply the chain rule to f = h ◦ g, where h is

integration over S 1 and

g : E → L1 (S 1 , R) by g(φ) = (1 + |φ(t)|2 )p/2 .

8
On the other hand, if f were C 3 , one can verify that g would also be C 3 , with
third derivative given by the formula

(D3 g)(ψ1 , ψ2 , ψ3 ) = (3p/2)(p/2 − 1)(1 + |φ|2 )p/2−1 φψ1 ψ2 ψ3

+ (3p/2)(p/2 − 1)(p/2 − 2)(1 + |φ|2 )p/2−2 φ3 ψ1 ψ2 ψ3 ,

and if p < 3, for any smooth choice of φ, we could choose ψ1 , ψ2 and ψ3 in Lp

such that the product ψ1 ψ2 ψ3 is not in L1 . This implies that f cannot be C 3
when 2 < p < 3.
Another familiar theorem from several variable calculus in finite dimension
is the “equality of mixed partials.” To state the infinite-dimensional version, we
let E and F be Banach spaces, U an open subset of E. Suppose that f : U → F
is a C 2 map. Then for x0 ∈ U ,

D2 f (x0 ) ∈ L(E, L(E, F )) = L2 (E, F )

= {continuous bilinear maps T : E × E → F }.

Of course, a very important case is the one where F = R, the base field of real
numbers.
Proposition 1.2.4. D2 f (x0 ) is symmetric; in other words,

D2 f (x0 )(h, k) = D2 f (x0 )(k, h), for h, k ∈ E.

Sketch of proof: First note that by the Hahn-Banach theorem, it suffices to

show that if λ : F → R is any continuous linear functional, then D2 (λ◦f )(x0 ) =
λ ◦ D2 f (x0 ) is symmetric, because if D2 f (x0 ) is not symmetric, the same will
be true of λ ◦ D2 f (x0 ), for some linear function λ. This reduces the proof to
the case where F = R. Next note that via the chain rule,
Z 1
f (x + h) − f (x) = (Df )(x + th)hdt,
0

and by iteration,
Z 1
f (x + h + k) − f (x + k) − f (x + h) + f (x) = (Df )(x + th + k)hdt
0
Z 1 Z 1
= (D(Df )(x + th + sk)(k))hdsdt.
0 0

Interchanging h and k yields

Z 1
f (x + h + k) − f (x + h) − f (x + k) + f (x) = (Df )(x + tk + h)kdt
0
Z 1 Z 1
= (D(Df )(x + sk + th)(h))kdsdt.
0 0

9
Since the left-hand sides of the last two expressions are equal, so are the right-
hand sides, and hence
Z 1Z 1
[(D(Df )(x + th + sk)(k))h − (D(Df )(x + th + sk)(h))k]dsdt = 0.
0 0

Since D(Df ) is a continuous function, this can only happen if D2 f (x)(k, h) =

D2 f (x)(h, k), for all x, h and k, which is exactly what we needed to prove.
More generally, if f : U → F is a C k -map, then

Dk f (x0 ) = D(Dk−1 f )(x0 ) ∈ L(E, Lk−1 (E, F )) = Lk (E, F ),

and by an induction based on the previous lemma, we see that in fact

Dk f (x0 ) ∈ Lks (E, F ) = {T ∈ Lk (E, F ) : T is symmetric}.

By symmetric we mean that

T (hσ(1) , . . . , hσ(k) ) = T (h1 , . . . , hk ),

for all permutations σ in the symmetric group Sk on k letters.

It is often useful to have an explicit formula for the higher derivatives of
a composition. The following such formula comes from §3 of [1]. If p and k
are positive integers with k ≤ p and (i1 , . . . , ik ) is a k-tuple of positive integers
such that i1 + · · · + ik = p and i1 ≤ i2 ≤ · · · ik , we define integers σkp (i1 , . . . , ik )
recursively by requiring that σ11 (1) = 1 and
k
X
σkp (i1 , . . . , ik ) = δi11 σk−1
p−1
(i2 , . . . , ik ) + σkp−1 (i1 , . . . , ij + 1, . . . ik ),
j=1

where δi11 is 1 if i1 = 1 and 0 otherwise.

Proposition 1.2.5. Suppose that U and V are open subsets of Banach spaces
E and F respectively, and that f : U → V and g : V → G are C p maps, where
G is a third Banach space. Then
p
X
Dp (g ◦ f ) = (Dk g ◦ f )Pkp (f ),
k=1

where X
Pkp (f ) = σkp (i1 , . . . , ik )(Di1 f, . . . , Dik f ),
the sum being taken over all p-tuples of positive integers such that i1 +· · ·+ik = p
and i1 ≤ i2 ≤ · · · ≤ ik .
The proof of Proposition 1.2.5 is by induction on p starting with the case p = 1,
which is an immediate application of the chain rule. For p ≥ 2, one obtains the

10
formula for σkp and the expression for Dp (g ◦ f ) by applying the chain rule and
the Leibniz rule for differentiating a product.
In order to put Taylor’s theorem in the Banach space setting, we need to
define the integral of a continuous map γ : [0, 1] → E into a Banach space E.
The definition is particularly easy if E is a reflexive Banach space; in this case,
we just set Z 1 Z 1
γ(t)dt = e, where λ(e) = λ ◦ γ(t)dt.
0 0

A definition of the integral for general Banach spaces can be found in [43].
Let U be an open subset of a Banach space E. Following [2], we define a
thickening of U to be an open subset U e ⊂ E × E such that

1. U × {0} ⊂ U
e.

2. (x, h) ∈ U
e ⇒ x + th ∈ U , for t ∈ [0, 1].

3. (x, h) ∈ U
e ⇒ x ∈ U.

Proposition 1.2.6. (Taylor’s Theorem.) If a map f : U → F is of class C r ,

there exist continuous maps

φk : U → Lks (E, F ), for 1 ≤ k ≤ r e → Lr (E, F ),

and R : U s

where Ũ is a thickening of U , such that R(x, 0) = 0 and

1 1
f (x + h) = f (x) + φ1 (x)h + φ2 (x)(h, h) + · · · + φr (x)(h, · · · , h)
2! r!
+ R(x, h)(h, · · · , h).

Here φk (x) = (Dk f )(x), for 1 ≤ k ≤ r.

To see this, we first reduce to the case where F = R by applying the Hahn-
Banach Theorem, and then establish via induction,
Z 1
f (x + h) = f (x) + (Df )(x + th)hdt
0
Z 1
= f (x) + (Df )(x)h + [(Df )(x + th) − (Df )(x)]hdt
0
Z 1 Z 1
= f (x) + (Df )(x)h + [(D2 f )(x + sth)](h, h)tdsdt
0 0
1
= f (x) + (Df )(x)h + (D2 f )(x)(h, h)
2!
Z 1Z 1
+ [(D2 f )(x + sth) − D2 f (x)](h, h)tdsdt
0 0

11
Continuing in the same manner, we find that

1
f (x + h) = f (x) + (Df )(x)h + (D2 f )(x)(h, h)
2!
1
+ · · · + (Dk f )(x)(h, · · · , h) + R(x, h)(h, · · · , h),
k!
where R(x, h) ∈ Lks (E, F ) depends continuously on x and h and R(x, 0) = 0.
We have seen that many of the basic results of differential calculus of several
variables extend with little change to the infinite-dimensional context. The
following theorem is somewhat deeper:
Theorem 1.2.7. (Inverse Function Theorem.) If U1 and U2 are open
subsets of Banach spaces E1 and E2 with x0 ∈ U1 , and f : U1 → U2 is a C ∞ map
such that Df (x0 ) ∈ L(E1 , E2 ) is invertible, then there are open neighborhoods
V1 of x0 and V2 of f (x0 ), and a C ∞ map g : V2 → V1 , such that

f ◦ g = idV2 and g ◦ f = idV1 .

Moreover, Dg(f (x)) = [Df (x)]−1 , for x ∈ V1 .

Sketch of proof: We can assume without loss of generality that x0 = 0 ∈ U1
and f (0) = 0 ∈ U2 . We can assume, moreover, that E1 = E2 and Df (0)
is the identity map by replacing f by Df (0)−1 ◦ f . Define h : U1 → E1 by
h(x) = x − f (x). Then Dh(0) = 0, and by continuity of Dh there exists δ > 0
such that
1
kxk ≤ δ ⇒ x ∈ U1 and kdh(x)k < .
2
If kxk, kyk < δ, then it follows from the chain rule that
Z 1 
 d 
kh(x) − h(y)k =  (h(tx + (1 − t)y))dt
0 dt
Z 1 
 
=  Dh(tx + (1 − t)y)(x − y)dt
0
Z 1 
1
≤ kDh(tx + (1 − t)y)kdt kx − yk < kx − yk.
0 2

More generally, if kyk < δ/2, and we define the map hy by hy (x) = h(x) + y,
then
δ 1
kxk ≤ δ ⇒ khy (x)k ≤ kyk + kh(x)k < + kxk ≤ δ,
2 2
so hy takes the closed ball of radius δ to itself and

1
khy (x) − hy (x0 )k = kh(x) − h(x0 )k < kx − x0 k,
2

12
so hy is a contraction. Thus by the well-known Contraction Lemma, given y
with kyk < δ/2, there is a unique fixed point x of hy ; that is, there is a unique
x such that kxk ≤ δ and

hy (x) = x ⇒ x − f (x) + y = x ⇒ f (p) = q.

Let

V2 = {y ∈ E1 : kyk < δ/2} and V1 = {x ∈ E1 : kxk < δ, f (x) ∈ V2 }

and define

g : V2 → V1 by g(y) = x ∈ V1 ⇔ f (x) = y.

Then g is a set-theoretic inverse to f : V1 → V2 .

To show that g is continuous, it suffices to show that |x−x0 | ≤ 2|f (x)−f (x0 )|.
But

|x − x0 | ≤ |(x − f (x)) − (x0 − f (x0 ))| + |f (x) − f (x0 )|

1
≤ |h(x) − h(x0 )| + |f (x) − f (x0 )| ≤ |x − x0 | + |f (x) − f (x0 )|,
2
which clearly implies the desired result.
To see that that g is C 1 , we note first that if x0 ∈ V1 ,

f (x) − f (x0 ) = Df (x0 )(x − x0 ) + o(|x − x0 |).

If y0 = f (x0 ) and y = f (x), we can rewrite this equation as

y − y0 = Df (x0 )(g(y) − g(y0 )) + o(|x − x0 |),

or g(y) − g(y0 ) = [Df (x0 )]−1 (y − y0 + o(|x − x0 |)).

The continuity argument shows that [Df (x0 )]−1 (o(|x − x0 |)) is o(|y − y0 |), so g
is C 1 with derivative Dg(y) = (Df )−1 (g(y)).
Finally, one uses “bootstrapping”:

g ∈ C 1 ⇒ (Df )−1 ◦ g ∈ C 1 ⇒ Dg ∈ C 1 ⇒ g ∈ C 2 ⇒ · · ·

to conclude that g is C ∞ , and the theorem is proven. Later, we will see that
the technique of bootstrapping used here is a fundamental tool for the study of
nonlinear PDE’s.
Before stating an important corollary of the Inverse Function Theorem, we point
out one of the difficulties in dealing with Banach spaces. A closed linear subspace
E1 of a Banach space E is a Banach space in its own right, but there may not
exist a complementary closed subspace E2 such that E is linearly homeomorphic
to E1 ⊕ E2 . We say that a subspace E1 of a Banach space E is split if there does
exist such a complement E2 . For example, any finite-dimensional subspace of a

13
Banach space is split as is any closed subspace of finite codimension. Moreover,
any closed subspace of a Hilbert space is split, because the inner product can
be used to define the orthogonal complement.
Corollary 1.2.8. If U is an open subset of the Banach space E with x ∈ U ,
and f : U → F is a C ∞ map such that Df (x) ∈ L(E, F ) is surjective with
split kernel, then there exists an open subset V ⊂ U and a diffeomorphism
φ : V1 × V2 → V , where V1 and V2 are open subsets of Banach spaces E1 and E2
with E = E1 ⊕ E2 and E2 ∼ = F , such that f ◦ φ is the projection on the second
factor.
Sketch of proof: We can assume without loss of generality that x = 0 and f (0) =
0. Let E1 be the kernel of Df (0) and since it splits, let E2 be a complement
such that E = E1 ⊕ E2 . Note that Df (p) establishes an isomorphism from E2
to F . Let
g : E = E1 ⊕ E2 → E = E1 ⊕ E2 by g(x1 , x2 ) = (x1 , f (x1 , x2 )).
One easily checks that Dg(0) is invertible. Now apply the inverse function
theorem to construct a smooth map
φ : V1 × V2 → V ⊂ U such that g ◦ φ = id,
the identity map. The projection on E2 is just f ◦ φ.
Remark. Notably absent from our examples of Banach spaces is the space

C ∞ ([0, 1], R) = { functions f : [0, 1] → R which

have continuous derivatives of all orders },
which one suspects should have importance for the theory of nonlinear partial
differential equations. Unfortunately, the natural topology to use on this space
is not defined by a single norm or seminorm, but by a countable collection of
norms {k · kk : k ∈ Z, k ≥ 0}, where k · kk is defined by (1.2).
Definition. A Fréchet space is a vector space E together with a countable
collection of seminorms {k · kk : k ∈ Z, k ≥ 0} satisfying the following axioms:
1. kxkk = 0 for all k only if x = 0.
2. Suppose that {xi } is a sequence of elements from E. If for each  > 0,
there is a N such that i, j ≥ N implies that kxi − xj kk <  for all k, then
there is an element x ∈ E such that kxi − xkk converges to zero for all k.

Of course every Banach space is a Fréchet space, but C ∞ ([0, 1], R) with the
collection of norms defined above is a Fréchet space which is not a Banach
space. Given a Fréchet space E with seminorms {k · kk : k ∈ Z, k ≥ 0}, we can
define a distance function
∞
X 1
d : E × E → R by d(x, y) = kx − ykk ,
2k
k=0

14
which makes E into a metric space. Thus we can talk of continuous maps
f : E1 → E2 from the Fréchet space E1 to the Fréchet space E2 , and we have a
category consisting of Fréchet spaces and continuous linear maps.
Definition. Suppose that E1 and E2 are Fréchet spaces, and that U is an
open subset of E1 . A continuous map f : U → E2 is said to be continuously
differentiable on U if there exists a continuous map Df : U × E1 → E2 such
that
f (x + ty) − f (x)
Df (x)y = lim ,
t→0 t
where t ranges throughout R − {0}, it being understood that the limit on the
right-hand side exists for all x ∈ U and all y ∈ E1 .
It is proven in Hamilton’s survey article [32], Part I, 3.2.3 and 3.2.5 that if
f : U → E2 is continuously differentiable, the map y 7→ Df (x)y is linear. Thus
if E1 and E2 are Banach spaces the above definition agrees with the definition
previously given.
We could develop much of the infinite-dimensional calculus and the theory
of infinite-dimensional manifolds in the category of Fréchet spaces, and in fact
this is carried out in [32], but the Inverse Function Theorem does not hold for
Fréchet spaces. Moreover, in the existence theory for solutions to nonlinear
partial differential equations, it is often convenient to first prove existence in
a given Banach space and then prove regularity using bootstrapping, just as
we did in the proof of the Inverse Function Theorem. This technique seems
particularly well-adapted to Banach spaces. For these reasons, we prefer to
think of C ∞ ([0, 1], R) as the intersection of a “chain” of Banach spaces,

· · · ⊆ C k+1 ([0, 1], R) ⊆ C k ([0, 1], R) ⊆ C k−1 ([0, 1], R) ⊆ · · · ⊆ C 0 ([0, 1], R).

For proving theorems, we will usually work in the category of Banach spaces
so that we can use theorems like the Inverse Function Theorem. However,
the statements of theorems are sometimes more elegant when phrased in the
category of Fréchet spaces.

1.3 Infinite-dimensional manifolds

Definition. Let E be a Banach space. A connected smooth manifold mod-

eled on E is a connected Hausdorff space M together with a collection A =
{(Uα , φα ) : α ∈ A}, where each Uα is an open subset of M and each

φα : Uα −→ φα (Uα ) ⊂ E

is a homeomorphism such that

S
1. {Uα : α ∈ A} = M.
2. φβ ◦ φ−1 ∞
α : φα (Uα ∩ Uβ ) → φβ (Uα ∩ Uβ ) is C , for all α, β ∈ A.

15
A nonconnected Hausdorff space is called a smooth manifold or a Banach mani-
fold if each component is a connected smooth manifold modeled on some Banach
space. (There is no harm in allowing different components to be modeled on
different Banach spaces.) The smooth manifold is called a Hilbert manifold if
each component is modeled on a Hilbert space.
We say that A = {(Uα , φα ) : α ∈ A} is the atlas defining the smooth structure
on M, and each (Uα , φα ) is one of the charts in the atlas.
Remark. We could define Fréchet manifolds by simply replacing the phrase
“Banach space” by “Fréchet space” in the above definition.
Let M and N be smooth manifolds modeled on Banach spaces E and F re-
spectively. Suppose that M and N have atlases A = {(Uα , φα ) : α ∈ A} and
B = {(Vβ , ψβ ) : β ∈ B}. Then a continuous map F : M → N is said to be
smooth if ψβ ◦ F ◦ φ−1 ∞
α is C , where defined, for α ∈ A and β ∈ B. It follows
from the chain rule that the composition of smooth maps is smooth. In this way
we obtain a category whose objects are smooth manifolds modeled on Banach
spaces and whose morphisms are smooth maps between such manifolds.
As in the case of finite-dimensional manifolds, a diffeomorphism is a smooth
map between manifolds with smooth inverse. We will often identify two smooth
manifolds if there is a diffeomorphism from one to the other. Later we will
construct invariants (such as de Rham cohomology) that will often enable us to
determine that two infinite-dimensional manifolds cannot be diffeomorphic.
Of course, the simplest example of a smooth manifold modeled on E is an
open subset U of E in which the atlas is {(U, idU )}. However, the examples of
most interest to us will be function spaces.
Example. Suppose that M n is a complete Riemannian manifold of finite di-
mension n, which we can assume is isometrically imbedded in Euclidean space
RN by the celebrated Nash imbedding theorem [57]. Suppose that Σ is a com-
pact smooth manifold of finite dimension m. Then
C 0 (Σ, RN ) = {continuous maps f : Σ → RN }
is a Banach space, and we claim that the subspace
C 0 (Σ, M ) = {continuous maps f : Σ → M } ⊆ C 0 (Σ, RN )
is an infinite-dimensional Banach manifold.
To construct the charts on C 0 (Σ, M ) we use the exponential map of M .
Suppose that f : Σ → M is a smooth (C ∞ ) map, and consider the Banach
space
C 0 (f ∗ T M ) = {continuous sections of f ∗ T M },
with the C 0 -norm
kXk = sup{|X(p)| : p ∈ Σ},
where the | · | on the right is length as defined by the Riemannian metric on M .
For  > 0, we set
Vf, = {X ∈ C 0 (f ∗ T M ) : kXk < },

16
and we define

Uf, = {g ∈ C 0 (Σ, M ) : dM (g(p), f (p)) <  for all p ∈ Σ},

where dM : M × M → R is the distance function on M defined by the Rieman-

nian metric. Then we define ψf, : Vf, → Uf, by

ψf, (X)(p) = expf (p) (X(p)).

By the the proof of the standard tubular neighborhood theorem from finite-
dimensional Riemannian geometry, one concludes that when  > 0 is sufficiently
−1
small, ψf, is a bijection, and we set φf, = ψf, . We set

A = {(Uf, , φf, ) : f : Σ → M is a C ∞ map and  > 0 is

small enough that ψf, is a homeomorphism } .

Since smooth maps f : Σ → M are dense in C 0 (Σ, M ), the union of the elements
of A cover C 0 (Σ, M ).
Hence to verify that C 0 (Σ, M ) is a smooth manifold, we need only check
that φf2 ,2 ◦ (φf1 ,1 )−1 is smooth where defined, when

(Uf1 ,1 , φf1 ,1 ), (Uf2 ,2 , φf2 ,2 ) ∈ A.

Let U be the open subset of the total space of f1∗ T M on which (expf2 (p) )−1 ◦
expf1 (p) is defined. We can assume that U ∩ Tf1 (p) M is convex with compact
closure for each p ∈ Σ, and define

g : U → (total space of f2∗ T M )

by
g(p, v) = (p, (expf2 (p) )−1 ◦ expf1 (p) (v)), for p ∈ Σ, v ∈ Tf1 (p) M .
We can think of g as a family of smooth maps

p 7→ g̃p : (Tf1 (p) M ∩ U ) → Tf2 (p) M,

and since Σ is compact and U has compact closure in each fiber, all derivatives
Dk (g̃p ) are bounded. Using the open neighborhood U of the zero-section, we
define
Ũ = {X ∈ C 0 (f1∗ T M ) : X(Σ) ⊆ U }
and define ωg : Ũ → C 0 (f2∗ T M ) by ωg (X) = g ◦ X.
To finish the proof that C 0 (Σ, M ) is a smooth manifold, it suffices to show
that ωg is smooth. We can formalize this statement in a theorem, known as the
ω-lemma:
Lemma 1.3.1. Suppose that E1 and E2 are finite-dimensional vector bundles
over the compact smooth manifold Σ and that U is a bounded open neighbor-
hood of the zero section of E1 whose restriction to each fiber of E1 is convex.

17
If g : U → (total space of E2 ) is a smooth map which takes the fiber of E1 over
p to the fiber of E2 over p (for each p ∈ Σ), and

Ũ = {σ ∈ C 0 (E1 ) : σ(Σ) ⊆ U },

then the map ωg : Ũ → C 0 (E2 ), defined by ωg (σ) = g(σ), is smooth.

The first step is to show that ωg is continuous; this is straightforward and we
can safely leave it to the reader.
Suppose now that σ, η, σ + η ∈ Ũ . It follows from Taylor’s theorem that for
each p ∈ Σ,

g̃p ((σ + η)(p)) = g̃p (σ(p)) + Dg̃p (σ(p))η(p) + R(σ, η)(p)(η(p)),

where
Z 1 
R(σ, η)(p)(η(p)) = [Dg̃p ((σ + tη)(p)) − Dg̃p (σ(p))]dt η(p). (1.3)
0

We can write this as

ωg (σ + η) = ωg (σ) + T (σ)(η) + R(σ, η)η,

where
T (σ)(η)(p) = Dg̃p (σ(p))η(p),
and R(σ, η)η is the remainder term given by (1.3). Note that

kT (σ)(η)k = sup |Dg̃p (σ(p))η(p)| ≤ sup |Dg̃p (σ(p))|kηk,

p∈Σ p∈Σ

so T (σ) is a continuous linear map from a neighborhood of 0 in C 0 (E1 ) to

C 0 (E2 ). Thus T extends to a linear map from C 0 (E1 ) to C 0 (E2 ). We next
estimate the error term R(σ, η)η; since
Z 1 
 [Dg̃p ((σ + tη)(p)) − Dg̃p (σ(p))]dt ≤ |D2 g̃p (σ(p))||η(p)|,
 
 
0

we conclude that

kR(σ, η)ηk ≤ sup |D2 g̃p (σ(p))kηk2 = o(η).

p∈Σ

This implies that ωg has the derivative Dωg (σ) = T (σ) at σ. We have defined
a map
Dωg : Ũ −→ L(C 0 (E1 ), C 0 (E2 )), (1.4)
and it is relatively straightforward to show that Dωg is continuous, showing
that ωg is C 1 .

18
 We would like to extend this argument to higher derivatives, and for this we
need to factor the derivative given by (1.4) as follows: Recalling that g : U →
(total space of E2 ), we define a corresponding map

f : U → (total space of L(E1 , E2 ))

Dg by setting Dg(p)
f = Dg̃p .

(We can regard Dgf as a “partial derivative” of g in which the point p ∈ Σ is

held fixed.) We can then define
0
f : Ũ → C (Σ, L(E1 , E2 ))
ωDg by ωDg
f (σ)(p) = Dg̃p (σ(p)).

Using the fact that C 0 (Σ, R) is a Banach algebra, we can show that the map

A : C 0 (Σ, L(E1 , E2 )) → L(C 0 (Σ, E1 ), C 0 (Σ, E2 ))

defined by A(T )(σ)(p) = T (p) · σ(p),

is smooth, thus providing the desired factorization,

Dωg = A · ωDg
f.

1
The argument presented in the preceding paragraph now shows that ωDg f is C ,
from which we conclude that Dωg is C 1 , and hence ωg is C 2 .
1 3
Next, we show that ωDg 2 g is C , which implies that ωg is C , and so forth.

By induction, we establish that ωg is C k for all k ∈ N, and hence ωg is C ∞ .

The above lemma has the following consequence:
Theorem 1.3.2. If Σ and M are smooth manifolds with Σ compact, then
C 0 (Σ, M ) is a smooth manifold modeled on the Banach spaces

C 0 (f ∗ T M ) = {continuous sections of f ∗ T M },

where f : Σ → M is a smooth map. Moreover, if g : M → N is a C ∞ map,

then the map

ωg : C 0 (Σ, M ) → C 0 (Σ, N ) defined by ωg (f ) = g ◦ f,

is smooth.
To prove the moreover part of the theorem, we need to show that if f1 : Σ → M
and f2 : Σ → N are smooth, then

φf2 ,2 ◦ ωg ◦ φ−1

f1 ,1 is smooth where defined.

The proof of this is a straightforward application of Lemma 1.2.1.

A modification of the previous example is often quite useful. Let Σ be a compact
smooth manifold of finite dimension m with boundary ∂Σ, M n a complete

19
Riemannian manifold of finite dimension n and f0 : ∂Σ → M a fixed smooth
mapping. We claim that

C00 (Σ, M ) = {continuous maps f : Σ → M : f |∂Σ = f0 }.

is a Banach manifold. The component containing f is modeled on the Banach

space

C00 (f ∗ T M ) = {continuous sections of f ∗ T M

which vanish on the boundary of M }.

The proof is a straightforward modification of the proof of Theorem 1.2.2.

Unfortunately, the manifolds C 0 (Σ, M ) are not sufficient for constructing a
global theory of partial differential equations. We need to be able to differenti-
ate elements in our function spaces. Thus we need to start off with a somewhat
stronger Banach space than the space C 0 (Σ, R) of continuous real-valued func-
tions on Σ.
Thus for k ∈ N, we are led to consider the space C k (Σ, R) of real-valued
functions on Σ which have continuous derivatives up to order k, a Banach space
with respect to the norm

kf kC k = sup{kf k(p) + kDf k(p) + · · · + kDk f k(p) : p ∈ Σ}. (1.5)

In fact, it is easily checked that if f, g ∈ C k (Σ, R), then

kf gkC k ≤ kf kC k kgkC k , (1.6)

so C k (Σ, R) is in fact a Banach algebra.

More generally, we can consider the space C k (Σ, RN ) of RN -valued functions
on Σ which have continuous derivatives up to order k, which is also a Banach
space with norm defined by (1.5). The Banach algebra condition (1.6) ensures
that we can define a continuous multiplication

µ : C k (Σ, L(RN , RM )) × C k (Σ, RN ) −→ C k (Σ, RM )

by simply multiplying in the range.

If M is an n-dimensional Riemannian manifold which isometrically imbedded
in Rn , we let

C k (Σ, M ) = {f ∈ C k (Σ, RN ) : f (p) ∈ M for all p ∈ Σ }.

We claim that C k (Σ, M ) is a smooth manifold.

The construction of the atlas is just like the construction for C 0 (Σ, M ). For
 > 0, we set
Vf, = {X ∈ C k (f ∗ T M ) : kXkC 0 < },
an open subset of C k (f ∗ T M ), and we define

Uf, = {g ∈ C 0 (Σ, M ) : dM (g(p), f (p)) <  for all p ∈ Σ}.

20
As before, we define ψf, : Vf, → Uf, by

ψf, (X)(p) = expf (p) (X(p)).

−1
When  > 0 is sufficiently small, ψf, is a bijection, and we set φf, = ψf, . As
k
smooth atlas for C (Σ, M ), we take

A = {(Uf, , φf, ) : f : Σ → M is a C ∞ map and  > 0 is

small enough that ψf, is a homeomorphism } .

Just as before, since smooth maps f : Σ → M are dense in C k (Σ, M ), the

union of the elements of A cover C k (Σ, M ). To finish the proof that C k (Σ, M )
is an infinite-dimensional smooth manifold, we need to verify once again that
φf2 ,2 ◦ (φf1 ,1 )−1 is smooth where defined, when

(Uf1 ,1 , φf1 ,1 ), (Uf2 ,2 , φf2 ,2 ) ∈ A.

But this follows from a corresponding ω-lemma:

Lemma 1.3.3. Suppose that E1 and E2 are finite-dimensional vector bundles
over the compact smooth manifold Σ and that U is a bounded open neighbor-
hood of the zero section of E1 whose restriction to each fiber of E1 is convex.
If g : U → (total space of E2 ) is a smooth map which takes the fiber of E1 over
p to the fiber of E2 over p (for each p ∈ Σ), and

Ũ = {σ ∈ C k (E1 ) : σ(Σ) ⊆ U },

then the map ωg : Ũ → C k (E2 ), defined by ωg (σ) = g(σ), is smooth.

The proof is virtually identical to the proof given for Lemma 1.2.1. The proof
extends to C k maps because the space C k (Σ, R) has two key properties:
1. It is a Banach algebra, and
2. there is be a continuous inclusion from C k (Σ, R) into the Banach algebra
C 0 (Σ, R) of continuous functions.

Thus just as before, we can construct an important family of Banach manifolds:

Theorem 1.3.4. If Σ and M are smooth manifolds with Σ compact, then for
each k ∈ N, C k (Σ, M ) is a smooth manifold modeled on the Banach spaces

C k (f ∗ T M ) = {C k sections of f ∗ T M },

whenever f : Σ → M is a smooth map. Moreover, if g : M → N is a C ∞ map,

then the map

ωg : C k (Σ, M ) → C k (Σ, N ) defined by ωg (f ) = g ◦ f,

is smooth.

21
To summarize, for each pair (Σ, M ) of finite-dimensional smooth manifolds,
with Σ compact, we have a chain of Banach manifolds,
· · · ⊆ C k+1 (Σ, M ) ⊆ C k (Σ, M ) ⊆ C k−1 (Σ, M ) ⊆ · · · ⊆ C 0 (Σ, M ).
The intersection of these manifolds is the space C ∞ (Σ, M ) of C ∞ maps from Σ
to M , which could be made into a Fréchet manifold, but we will not enter into
the details of that here.
We can now try to formulate calculus of variations problems in terms of the
infinite-dimensional manifolds that we have constructed. Thus, for example, we
can define the action function
Z
1
1 1
J : C (S , M ) −→ R by J(γ) = |γ 0 (t)|2 dt,
2 S1
and check without much difficulty that J is a smooth map. As we learned
in elementary differential geometry courses, the “critical points” for J are the
smooth closed geodesics on M .
Suppose that Σ is a compact two-dimensional Riemann surface. Thus we
can imagine that Σ has a Riemannian metric, but we forget about the metric
except for its conformal equivalence class, which we denote by ω. Suppose that
{(Uα , (xα , yα )) : α ∈ A}
is an atlas of isothermal coordinate charts on Σ, and let {ψα : α ∈ A} be
a partition of unity subordinate to {Uα : α ∈ A}. We can then define the
Dirichlet integral
"  #
 ∂f 2  ∂f 2
Z X  
1 1 1
Eω : C (S , M ) −→ R by Eω (f ) = ψα   +
  dxα dyα .
2 Σ ∂xα  ∂yα 
α∈A

Once again it is relatively easy to check that Eω is a smooth map on the infinite-
dimensional manifold C 1 (Σ, M ). Later we will see that the “critical points” for
Eω are harmonic maps.
More generally, suppose that Σ is an m-dimensional Riemannian manifold
with Riemannian metric expressed in local coordinates (1 , . . . xm ) on Σ by
m
X
h= ηab dxa dxb .
a,b=1

If f : Σ → M ⊆ RN is a smooth map and (η ab ) denotes the matrix inverse to

(ηab ), we set
m
X ∂X ∂X p
|df |2 = η ab · and dA = det(ηab )dx1 · · · dxm .
∂xa ∂xb
a,b=1

We can then define the Dirichlet integral

Z
1 1
E : C (Σ, M ) −→ R by E(f ) = |df |2 dA,
2 Σ

22
which is once again a smooth real-valued function on the infinite-dimensional
manifold C 1 (Σ, M ). In the case where the domain is two-dimensional, choice
of isothermal parameters leads to exactly the same integrand as before, so this
generalizes the previous energy functions to higher dimensional domains.

1.4 The basic mapping spaces

For the study of geodesics, harmonic and minimal surfaces and pseudoholomor-
phic curves, as well as other nonlinear partial differential equations, we need a
collection of function spaces with weak enough topologies that it is relatively
easy to prove convergence of a sequence which is tending toward an infimum
of energy on a given component. The infinite-dimensional manifolds that have
proven to be most useful in this regard are those modeled on Sobolev spaces.
In this section, we describe the simplest of these spaces.
If Σ is a compact Riemannian manifold, we can define an inner product (·, ·)
on the space C ∞ (Σ, R)) of smooth real-valued maps by
Z
(f, g) = (f g + hDf, Dgi)dA,
Σ

where the inner product h·, ·i on the right is the usual inner product in the
cotangent space defined by the Riemannian metric on Σ, and dA denotes the
area or volume form on Σ. The inner product (·, ·) makes space C ∞ (Σ, R))
of smooth functions into a pre-Hilbert space. Any pre-Hilbert space has a
Hilbert space completion, the set of equivalence classes of Cauchy sequences, as
described at the beginning of §1.2. The Hilbert space completion in our case is
denoted by L21 (Σ, R)), and is called the Sobolev space of L21 -functions on Σ.
A second important Sobolev space generalizes the Lp spaces studied in real
analysis when 1 < p < ∞. We start by defining a norm k · kLp1 on C ∞ (Σ, R)) by
 p Z
kf kLp1 = (|f |p + |Df |p )dA,
Σ

where |Df | is the length calculated with respect to the Riemannian metric
on Σ. This norm makes C ∞ (Σ, R)) into a pre-Banach space. As before, we
can construct the Banach space completion. This Banach space completion is
denoted by Lp1 (Σ, R)) and is called the Sobolev space of Lp1 -functions on Σ. Of
course, when p = 2 this reduces to the previous example.
We can also define versions of these Sobolev spaces for higher numbers of
derivatives. Thus we can define a norm k · kLpk on C ∞ (Σ, R)) by
 p Z
kf kLpk = (|f |p + |Df |p + · · · + |Df k |p )dA,
Σ
and construct the completion with respect to this norm, which is denoted
Lpk (Σ, R). The resulting space is a Banach space, and a Hilbert space when
p = 2. We thus obtain a chain of Banach spaces,
· · · ⊆ Lpk (Σ, R) ⊂ · · · ⊆ Lp1 (Σ, R) ⊆ Lp (Σ, R),

23
and the intersection of all the spaces in the chain is just the space C ∞ (Σ, R)) of
smooth functions. These spaces are essential for the modern theory of partial
differential equations and they are compared by means of the Sobolev Lemma,
which in general form states

Lpk (Σ, R) ⊆ C l (Σ, R) for p(k − l) > dim(Σ), (1.7)

where ⊆ indicates continuous inclusion. A complement to the Sobolev Lemma

states that for pk > dim(Σ), Lpk (Σ, R) is a Banach algebra in an appropri-
ate norm defining the “Banachable” structure. Additional information on the
Sobolev spaces is found in standard references, such as Evans [19].
For the case where Σ = S 1 , where S 1 is the unit interval [0, 1] with endpoints
identified, the Sobolev Lemma is relatively easy to establish, and we do that
here:
Lemma 1.4.1. There is a continuous linear injection i : L21 (S 1 , R) → C 0 (S 1 , R)
which extends the inclusion C ∞ (S 1 , R) ⊂ C 0 (S 1 , R).
Proof: We begin with the sequence of inequalities:
Z τ Z
|f (t)| ≤ |f (τ )| + |f 0 (u)|du ≤ |f (τ )| + |f 0 (u)|du.
t S1

Averaging over τ and using the Cauchy-Schwarz inequality yields

Z Z 
0
|f (t)| ≤ |f (u)|du + |f (u)|du
S1 S1
Z 1/2
≤ [|f (u)|2 + |f 0 (u)|2 ]du = (f, f )1/2 ,
S1

where (·, ·) denotes the L21 inner product. Taking the supremum over all t yields

kf kC 0 ≤ kf kL21 .

Thus a Cauchy sequence with respect to the L21 inner product gets taken under
the inclusion C ∞ (S 1 , R) ⊂ C 0 (S 1 , R) to a Cauchy sequence with respect to the
C 0 -norm. By definition, an element of L21 is an equivalence class of Cauchy
sequences, and the map i is defined by sending this equivalence class to the
limit of the C 0 Cauchy sequence. It is immediate that i is injective.
Lemma 1.4.2. L21 (S 1 , R) is a Banach algebra; multiplication of functions is a
continuous bilinear map

L21 (S 1 , R) × L21 (S 1 , R) −→ L21 (S 1 , R).

24
Proof: It follows from the Cauchy-Schwarz inequality that
Z Z
0 2
kf gk2L2 = (f g, f g) = 2
[(f g) + [(f g) ] ]dt = [(f g)2 + (f 0 g + f g 0 )2 ]dt
1
S1 S1
Z
= [(f g)2 + (f 0 )2 g 2 + 2f gf 0 g 0 + f 2 (g 0 )2 ]dt
S1
Z Z
0 2 2 2 0 2
2
(f 0 g 0 )dt
 
≤ (f g) + (f ) g + f (g ) dt + 2kf gkC 0
S1 S1
≤ kf k2C 0 kgk2L2 + kgk2C 0 kf k2L2 + kf k2C 0 kgk2L2 + 2kf gkC 0 kf kL21 kgkL21 .
1 1 1

0
Since the C norm is less than the L21 , we find that

kf gk2L2 ≤ kf k2L2 kgk2L2 ,

1 1 1

finishing the proof of the Lemma.

It follows from this Lemma that the multiplication map

L21 (S 1 , Hom(Rm , Rn )) × L21 (S 1 , Rm ) −→ L21 (S 1 , Rn )

is continuous.
Suppose now that M is a complete connected finite-dimensional Rieman-
nian manifold isometrically imbedded as a proper submanifold of an ambient
Euclidean space RN . We let

L21 (S 1 , M ) = {γ ∈ L21 (S 1 , RN ) : γ(t) ∈ M for all t ∈ S 1 },

which is a closed subspace of L21 (S 1 , RN ) by Lemma 1.4.1.

We claim that L21 (S 1 , M ) is an infinite-dimensional smooth manifold, the
proof being just like the proof for C 0 (S 1 , M ). If γ : S 1 → M is a smooth curve,
we let

Vγ, = {X ∈ L21 (S 1 , γ ∗ T M ) : the L21 norm of X is < },

Uγ, = {λ ∈ L21 (S 1 , M ) : dM (λ(t), γ(t)) <  for all t ∈ S 1 },

and if  > 0 is sufficiently small, we define

−1
ψγ, : Vγ, → Uγ, by ψγ, (X)(p) = expγ(t) (X(t)), φγ, = ψγ, .

We then set

A = (Uγ, , φγ, ) : γ : S 1 → M is a C ∞ map and  > 0 is



small enough that ψγ, is a homeomorphism } .

and prove that it is a smooth atlas by the same argument used to establish
Lemma 1.3.1. Thus we obtain:

25
Theorem 1.4.3. If M is a smooth manifold, then L21 (S 1 , M ) is a smooth
manifold modeled on the Banach spaces L21 (γ ∗ T M ) for γ : S 1 → M a smooth
map. Moreover, if g : M → N is a C ∞ map, then the map
ωg : L21 (S 1 , M ) → L21 (S 1 , N ) defined by ωg (γ) = g ◦ γ,
is also C ∞ .
We are also interested in Lp1 -maps from a compact oriented surface Σ. It turns
out that these are Hölder continuous in accordance with the following definition.
Definition. If Σ is a metric space, a map f : Σ → R is said to be Hölder
continuous with Hölder exponent γ ∈ (0, 1] if there is a constant C > 0 such
that
|f (p) − f (q)| ≤ Cd(p, q)γ for all p, q ∈ Σ.
We let C 0,γ (Σ, R) be the space of all functions f : Σ → R which are Hölder
continuous. If f ∈ C 0,γ (Σ, R), we let
 
f (p) − f (q)
[f ]C = sup
0,γ : p, q ∈ Σ, p 6= q .
(d(p, q)γ

Lemma 1.4.4. If Σ is a compact oriented surface and p > 2, there is a con-

tinuous linear injection i : Lp1 (Σ, R) → C 0,γ (Σ, R), where γ = 1 − 2/p, which
extends the inclusion C ∞ (Σ, R) ⊂ C 0,γ (Σ, R). Moreover, there is a constant
C > 0 such that
[f ]C 0,γ ≤ Ckf kLp1 .

A complete proof of this is given in Evans [19], §5.6.2. We only prove the
weaker result that Lp1 (Σ, R) ⊆ C 0,γ (Σ, R). We begin by assuming that Σ is
the torus with flat Riemannian metric expressed in terms of suitable conformal
coordinates as ds2 = dx2 + dy 2 . We consider a smooth function f (x, y) on the
disk D(p, r0 ) of radius r0 about p ∈ Σ defined in terms of Euclidean coordinates
centered at p by x2 + y 2 ≤ r02 , then after shifting to polar coordinates (r, θ)
defined by
x = r cos θ, y = r sin θ,
we see that
Z s Z s
∂f
|f (s, θ) − f (p)| = (r, θ)dr ≤ |Df |(r, θ)dr,
0 ∂r 0

and hence
Z 2π 2π s
|Df |
Z Z Z
|f (s, θ) − f (p)|dθ ≤ |Df |(r, θ)drdθ ≤ dxdy.
0 0 0 D(p,r0 ) r
Thus
2π r0 Z r0  "Z #
|Df |
Z Z
|f (s, θ) − f (p)|sdsdθ ≤ rdr dxdy
0 0 0 D(p,r0 ) r

26
and hence
r2 |Df |
Z Z
|f (x, y) − f (p)|dxdy ≤ 0 dxdy.
D(p,r0 ) 2 D(p,r0 ) r

It follows from the Hölder inequality that

"Z #1/p "Z #(p−1)/p
r02 |Df | r2
Z
dxdy
dxdy ≤ 0 p
|Df | dxdy
2 D(p,r0 ) r 2 D(p,r0 ) D(p,r0 ) rp/(p−1)

while direct integration yields

Z Z
dxdy
= rp/(1−p) dxdy
D(p,r0 ) rp/(p−1) D(p,r0 )
2π r0
2π(p − 1) (p−2)/(p−1)
Z Z
= r1/(1−p) drdθ = r0 .
0 0 p−2

Thus
(p−1)/p
r02

2π(p − 1)
Z
(p−2)/p
|f (x, y) − f (p)|dxdy ≤ r0 kDf kLp . (1.8)
D(p,r0 ) 2 p−2

It follows from (1.8) and the Hölder inequality that

Z Z
πr02 |f (0)| ≤ |f (x, y) − f (p)|dxdy + |f (x, y)|dxdy
D(p,r0 ) D(p,r0 )

≤ (constant)kDf kLp + (constant)kf kL1

(p−1)/p
≤ (constant)kDf kLp + (constant)kf kLp (area of D(p, r0 ))
≤ (constant)kf kLp1 ,

which quickly yields the desired result when Σ is the flat torus.
If Σ is a more general Riemann surface, we can give Σ a Riemannian metric of
constant curvature of constant curvature and total volume one. Choose r0 > 0
less than the injectivity radius of this metric. A modification of the above
argument can then be applied to a normal coordinate ball of radius r0 showing
that if p ∈ Σ, then

|f (p)| ≤ (constant)kf kLp1 , and hence kf kC 0 ≤ (constant)kf kLp1 .

(Note that changing the Riemannian metric on Σ merely replaces the Lp1 -norm
by an equivalent norm, so adopting the constant curvature metric imposes no
restriction.) Thus if a sequence {fi } of smooth functions on Σ converges to a
limit in Lp1 -norm, {fi } converges also in C 0 norm to a unique limit function
f∞ ∈ C 0 . Thus any element of Lp1 (Σ, R) can be identified with a continuous
function and the Lemma is proven.

27
Remark 1.4.5. The previous Sobolev Lemma for Lp1 maps from a surface
begins to fail as p approaches 2 from above. The reason is that the highest
order term in the L21 -norm is conformally invariant and hence invariant under
dilations. In the case where D is the unit disk centered at the origin in R2 this
highest order term is Z
|Df |2 dxdy.
D
In contrast, the highest order term in the Lp1 -norm is not invariant under dila-
tions. Given a smooth map f : D → RN which takes the boundary ∂D to a
point, we can define a dilated map f : D → RN , where D is the ball of radius
 > 0 centered at the origin in R2 , by
x y 1   x y 
f (x, y) = f , . Then |Df (x, y)| = Df , .

    
and if x̃ = x/ and ỹ = y/ denote the coordinates corresponding to x and y on
the unit disk D1 ,
Z  (p−2) Z
p 1
|Df | dxdy = |Df |2α dx̃dỹ.
D  D1

Thus as  → 0, the Lp1 -norm of f approaches infinity as long as p > 2. In

particular, a bound on the L21 -norm does not imply a bound on the C 0 -norm.
Lemma 1.4.6. If Σ is a compact oriented surface and p > 2, Lp1 (Σ, R) satisfies
kf gkLp1 ≤ 2kf kLp1 kgkLp1 .
Thus after passing to an equivalent norm, we can show that Lp1 (Σ, R) is a Banach
algebra.
Sketch of Proof: By the previous Lemma,
kf gkLp ≤ kf kC 0 kgkLp + kgkC 0 kf kLp ≤ kf kLp1 kgkLp + +kgkLp1 kf kLp
and
kD(f g)kLp ≤ kf kC 0 kDgkLp +kgkC 0 kDf kLp ≤ kf kLp1 kDgkLp ++kgkLp1 kDf kLp .
Adding these two inequalities yields the statement of the Lemma.
If Σ is a compact surface and p > 2, we let
Lp1 (Σ, M ) = {f ∈ Lp1 (Σ, RN ) : f (p) ∈ M for all p ∈ Σ},
a closed subspace of Lp1 (Σ, RN ) by Lemma 1.4.3.
If Σ is a compact surface and p > 2, we claim that Lp1 (Σ, M ) is an infinite-
dimensional smooth manifold. In this case, when f : Σ → M is a smooth curve,
we let

Vγ, = {X ∈ Lp1 (Σ∗ T M ) : the Lp1 norm of X is < },

Uγ, = {g ∈ Lp1 (Σ, M ) : dM (f (p), g(p)) <  for all p ∈ Σ},

28
and if  > 0 is sufficiently small, we define
−1
ψγ, : Vγ, → Uγ, by ψγ, (X)(p) = expf (p) (X(p)), φγ, = ψγ, .

We then set

A = {(Uγ, , φγ, ) : f : Σ → M is a C ∞ map and  > 0 is

small enough that ψγ, is a homeomorphism } .

and once again prove that it is a smooth atlas by the same argument used to
establish Lemma 1.3.1. Thus we obtain:
Theorem 1.4.7. If Σ is a compact smooth surface and M is a smooth manifold,
then for p > 2, Lp1 (Σ, M ) is a smooth manifold modeled on the Banach spaces
Lp1 (f ∗ T M ) for f : Σ → M a smooth map. Moreover, if g : M → N is a C ∞
map, then the map

ωg : Lp1 (Σ, M ) → Lp1 (Σ, N ) defined by ωg (γ) = g ◦ γ,

is also C ∞ .
In exactly the same way, we could show that if Σ is an m-dimensional smooth
manifold and p > m, then Lp1 (Σ, M ) is an infinite-dimensional smooth manifold.

1.5 Homotopy type of the space of maps

A continuous map f : X → Y between topological spaces is said to be a homo-
topy equivalence if there is a continuous map g : Y → X such that f ◦ g and
g ◦ f are both homotopic to the identity. The following theorem was proven
quite early in the theory of manifolds of maps; see Eells [16] for the appropriate
references.
Theorem 1.5.1. Let M be a compact connected Riemannian manifold. Then
the inclusions

C k (S 1 , M ) ⊂ L21 (S 1 , M ) and C k (S 1 , M ) ⊂ C 0 (S 1 , M )

are homotopy equivalences when k ≥ 1.

The point of this Theorem is that from the point of view of homotopy theory,
L21 (S 1 , M ) is essentially the same as the space of continuous maps C 0 (S 1 , M )
with the compact open topology. This latter space has been extensively studied
by topologists and much is known about its homotopy and homology groups, as
we will see later.
The proof of the Theorem is an application of the theory of “smoothing opera-
tors.”

29
 For preparation, we suppose that φ : R → R is a smooth map which vanishes
outside [−1, 1]. Suppose, moreover, that
Z
φ≥0 and φ = 1.
R

For  > 0, let φ (t) = (1/)φ(t/), so that

Z
supp(φ ) ⊂ [−, ] and φ = 1.
R

If γ ∈ C 0 (S 1 , RN ), we can regard γ as an element of C 0 (R, RN ) such that

γ(t + 1) = γ(t) for all t, and we define φ ∗ γ ∈ C 0 (R, RN ) by
Z Z
(φ ∗ γ)(t) = φ (t − τ )γ(τ )dτ = φ (s)γ(t − s)ds.
R R

∞
It is immediately checked that φ ∗ γ is C and

dk dk
Z  k 
d
k
(φ  ∗ γ)(t) = k
(φ  ) ∗ γ = φ (t − τ )γ(τ )dτ.
dt dt R dtk

Moreover, (φ ∗ γ)(t + 1) = (φ ∗ γ)(t), and hence φ ∗ γ ∈ C ∞ (S 1 , RN ). We can

thus define smoothing operators

S : C 0 (S 1 , RN ) → C k (S 1 , RN ), S : L21 (S 1 , RN ) → C k (S 1 , RN )

by S (γ) = φ ∗ γ. It is not difficult to show that the maps

S : C 0 (S 1 , RN ) → C k (S 1 , RN ), S : L21 (S 1 , RN ) → C k (S 1 , RN )

are continuous.
Proof of Theorem 1.5.1: Recall that we regard M as a submanifold of RN .
Choose δ > 0 so small that the exponential map

exp : N M → RN , defined by exp(v) = p + v, for v ∈ Tp M ,

maps N Mδ = {v ∈ N M : |v| < δ} diffeomorphically onto

M (δ) = {p ∈ RN : d(p, M ) < δ}.

Then the “nearest point projection” map r : M (δ) → M , defined by r(p+v) = p

for p ∈ M , is a strong deformation retraction from M (δ) to M . To see this, we
define
h : M (δ) × [0, 1] → M (δ) by h(p + v, t) = p + (1 − t)v,
and check that
1. h(q, 0) = q, for q ∈ M (δ),

30
 2. h(q, 1) = r(q) ∈ M , for q ∈ M (δ), and
3. h(p, t) = p, for p ∈ M .
We have a similar strong deformation retraction on the function space level.
The ω-Lemma gives us a smooth map

ωh : L21 (S 1 , M (δ) × [0, 1]) → L21 (S 1 , M (δ)) defined by ωh (γ) = h ◦ γ.

We define

j : L21 (S 1 , M (δ)) × [0, 1] → L21 (S 1 , M (δ) × [0, 1]) by j(γ, τ )(t) = (γ(t), τ )

and let H = ωh ◦ j. Then

1. H(γ, 0) = γ, for γ ∈ L21 (S 1 , M (δ)),
2. H(γ, 1) = r ◦ γ ∈ L21 (S 1 , M ), for γ ∈ L21 (S 1 , M (δ)), and
3. H(γ, t) = γ, for γ ∈ L21 (S 1 , M ).
We can therefore define a strong deformation retraction

R : L21 (S 1 , M (δ)) −→ L21 (S 1 , M )

by R(γ) = ωr (γ) = H(γ, 1). In a similar fashion, we can define a strong

deformation retraction

R : C k (S 1 , M (δ)) −→ C 0 (S 1 , M ),

whenever k ≥ 0.
Let εk = 2−k and let

C 0 (S 1 , M )εk = {γ ∈ C 0 (S 1 , M ) : γ maps the closed interval

[(m − 1)2−k , (m + 1)2−k ] into the open ball
B(γ(m2−k ); δ) for each integer m such that 0 ≤ m ≤ 2k },

an open set in the compact-open topology. They key point of this set is that
when |s − t| < 2−k then the straight line from γ(s) to γ(t) in RN lies entirely
within M (δ) hence S ? γ lies within M (δ) when  ≤ k . Note that
∞
[
C 0 (S 1 , M ) = C 0 (S 1 , M )εk , L21 (S 1 , M )εk+1 ⊂ L21 (S 1 , M )εk ,
k=1

and we therefore say that C 0 (S 1 , M ) is a monotone union of the subspaces

C 0 (S 1 , M )εk . Similarly, we let

L21 (S 1 , M )εk = L21 ((S 1 , M ) ∩ C 0 (S 1 , M )εk ,

C k (S 1 , M )εk = C k ((S 1 , M ) ∩ C 0 (S 1 , M )εk , when k ≥ 1,

31
thereby expressing L21 (S 1 , M ) and C k (S 1 , M ) as monotone unions for k ≥ 1.
By analogous formulae, we define

C 0 (S 1 , M (δ))εk , L21 (S 1 , M (δ))εk and C k (S 1 , M (δ))εk ,

for k ≥ 1. We can then define smoothing operators

Sεk : L21 (S 1 , M )εk −→ C k (S 1 , M (δ))εk ,

since
γ ∈ L21 (S 1 , M )εk ⇒ Sεk ? γ ∈ C k (S 1 , M (δ))εk .
We define s to the composition of

Sεk : L21 (S 1 , M )εk → C k (S 1 , M (δ))εk and

R : C k (S 1 , M (δ))εk → C k (S 1 , M )εk .

We claim that if i : C k (S 1 , M )εk ⊂ L21 (S 1 , M )εk is the inclusion, then

s◦i and i ◦ s

are homotopic to the identity. This is easy to verify. To get the homotopy from
s ◦ i to the identity, we simply define

H1 : C k (S 1 , M )εk × [0, 1] → C k (S 1 , M )εk

by H1 (γ, t) = R ◦ (tSεk + (1 − t)id)) ◦ i(γ).

Similarly, to get the homotopy from i ◦ s to the identity, we define

H2 : L21 (S 1 , M )εk × [0, 1] → L21 (S 1 , M )εk

by H2 (γ, t) = i ◦ R ◦ (tSεk + (1 − t)id)(γ),

This shows that for each k ∈ N, the inclusion

C k (S 1 , M )εk ⊂ L21 (S 1 , M )εk

is a homotopy equivalence.
To finish the proof, we must take an appropriate limit as k → ∞. Suppose
that the metrizable space X is a monotone union of open subsets, by which we
mean that we have a sequence of spaces
[
U1 ⊂ U2 ⊂ U3 ⊂ · · · such that X = {Uk : k ∈ N}.

Suppose, moreover, that we let

X ? = (U1 × [1, 2]) ∪ (U2 × [2, 3]) ∪ (U3 × [3.4]) ∪ · · ·

topologized as a subset of X × R. We say that X is the homotopy direct limit

of the subspaces {Uk : k ∈ N} if the projection p : X ? → X on the first factor

32
is a homotopy equivalence. If the subsets Uk are open, then the open cover
{Uk : k ∈ N} has a C 0 subordinate partition of unity {ψk : k ∈ N}. In this case,
the map
∞
!
X
?
f :X→X defined by f (x) = x, kψk (x)
k=1

is a homotopy inverse to p, showing that X is a homotopy direct limit in this

case. Using this argument, one easily verifies that C k (S 1 , M ) is a homotopy
direct limit of its subspaces C k (S 1 , M )εk and L21 (S 1 , M ) is a homotopy direct
limit of L21 (S 1 , M )εk .
Now we apply the following Lemma, which is just Theorem A from the
Appendix to Milnor’s book on Morse theory [50]:
Lemma 1.5.2. Suppose that X is the homotopy direct limit of {Uk : k ∈ N}
and that Y is the homotopy direct limit of {Vk : k ∈ N}. If f : X → Y is
a continuous map such that f (Uk ) ⊆ Vk and the restriction of f to Uk is a
homotopy equivalence from Uk to Vk , then f itself is a homotopy equivalence.
We refer the reader to Milnor for the proof of this Lemma. It implies that the
inclusion C k (S 1 , M ) ⊂ L21 (S 1 , M ) is a homotopy equivalence when k ≥ 1. In
a similar manner, one verifies that the inclusion C k (S 1 , M ) ⊂ C 0 (S 1 , M ) is a
homotopy equivalence when k ≥ 1.
Theorem 1.5.3. Let M be a compact connected Riemannian manifold, Σ a
compact connected Riemann surface, p > 2. Then the inclusions

C k (Σ, M ) ⊂ Lp1 (Σ, M ), C k (Σ, M ) ⊂ C 0 (Σ, M )

are homotopy equivalences.

The proof is essentially the same as for the previous theorem, with 2 replaced by
p, except for the definition of smoothing operators defined on Σ. The construc-
tion of such operators is a standard technique in the theory of partial differential
equations. We describe only the case Σ = T 2 here, where T 2 = C/Λ, Λ being
a lattice in C; the construction in this case is particularly transparent. (In the
general case, the ideas are the same, but one constructs the smoothing operators
by piecing together using a partition of unity on Σ.)
Note that an element f ∈ Lp1 (T 2 , RN ) can be regarded as a map f : C → RN
such that f (z + λ) = f (z) for λ ∈ Λ.
Suppose that φ : C → [0, ∞) is a smooth map which vanishes outside D =
{z ∈ C : |z| ≤ 1} such that Z
φ dxdy = 1,
C

where (x, y) are the standard coordinates on C. Let φ (z) = (1/2 )φ(z/), so
that Z
supp(φ ) ⊂ {z ∈ C : |z| ≤ } and φ dxdy = 1.
C

33
If f : C → M comes from an element f ∈ Lp1 (T 2 , RN ), we define φ ∗ f ∈
C ∞ (C, RN ) by Z
(φ ∗ γ)(z) = φ (z − w)γ(w)dw.
C
It is immediately checked that (φ ∗ f )(z + λ) = (φ ∗ f )(z) for λ ∈ Λ, so (φ ∗ f )
can be identified with an element of C ∞ (T 2 , RN ).
Thus we can define smoothing operators

S : C 0 (T 2 , RN ) → C k (T 2 , RN ), S : Lp1 (T 2 , RN ) → C k (T 2 , RN )

by S (f ) = φ ∗ f . The proof of the Theorem for maps from Σ = T 2 now

continues in exactly the same way as for maps from S 1 .
Remark 1.5.4. It is interesting to consider Sobolev spaces of maps which do
not lie in “Sobolev range.” Thus we could consider

W1p (Σ, M ) = {f ∈ Lp1 (Σ, M ) : f (p) ∈ M for almost all p ∈ Σ },

p
H1,S (Σ, M ) = (Closure of C ∞ (Σ, M ) in Lp1 (Σ, M )),
p
for p ≤ dim Σ. Although H1,S (Σ, M ) ⊆ W1p (Σ, M ) the inclusion is often strict,
and neither space is in general homotopically equivalent to the space C 0 (Σ, M )
of continuous maps. These Sobolev spaces have been extensively studied by
Hang and Lin [33], among others, and applications to the theory of harmonic
maps are described in the review article of Brezis [11]. When the dimension of
Σ is at least three, harmonic maps from Σ to M are vastly more complicated
than geodesics and harmonic surfaces; for example, they need not be smooth.

1.6 The α-Lemma*

We have seen that there is a covariant functor from finite-dimensional smooth
manifolds and smooth maps to infinite-dimensional smooth manifolds and smooth
maps,

M 7→ C k (Σ, M ), g:M →N 7→ ωg : C k (Σ, M ) → C k (Σ, N ),

where ωg (f ) = g ◦ f . One might hope that when M is a fixed smooth manifold,

there is a similar contravariant functor from compact manifolds Σ to infinite-
dimensional smooth manifolds C k (Σ, M ) in which

f : Σ1 → Σ2 7→ αf : C k (Σ2 , M ) → C k (Σ1 , M ),

where αf (g) = g ◦ f . However, it turns out that the maps αf are no longer C ∞
smooth, but only C k :
α-Lemma 1.6.1. If M is a fixed smooth manifold, a smooth map g : Σ1 → Σ2
between smooth compact manifolds induces a C k map

αg : C k (Σ2 , M ) → C k (Σ1 , M ),

34
for each k.
A proof of this lemma can be found in [1].
We could try to put the α- and ω-Lemmas into a single theorem. This is partially
accomplished by the following theorem, which we will not prove; it is stated in
the survey article [16], which also includes numerous references:
Theorem 1.6.2. If S, M and N are smooth finite-dimensional manifolds, then

Φ : C k (S, M ) × C k+s (M, N ) → C k (S, N ), Φ(f, g) = g ◦ f,

is C s .
Remark. The loss of derivatives in the statement of Theorem 1.6.2 has far-
reaching implications. For example, suppose that we want to construct examples
of infinite-dimensional Banach Lie groups, which are defined just like ordinary
Lie groups, except that of being finite-dimensional manifolds, they are infinite-
dimensional Banach manifolds. We could start with a finite-dimensional Lie
group G with identity e, smooth multiplication map

µ : G × G → G, µ(σ, τ ) = σ · τ

and smooth inverse map

ν : G → G, ν(σ) = σ −1 .

We could then define the corresponding loop group L21 (S 1 , G) with identity the
constant map to e, smooth multiplication

ωµ : L21 (S 1 , G) × L21 (S 1 , G) → L21 (S 1 , G), ωµ (f, g) = f · g,

where the dot on the right denotes multiplication within G, and smooth inverse

ων : L21 (S 1 , G) → L21 (S 1 , G), ων (f )(p) = (f (p))−1 .

It is not difficult to check that L21 (S 1 , G) is in fact an infinite-dimensional Banach

Lie group. A rich theory of these loop groups has been developed, although they
are often modeled on Fréchet rather than Banach spaces.
On the other hand, although the space of C k diffeomorphisms is a per-
fectly well-behaved topological group under composition, this composition fails
to be smooth because of the loss of derivatives implicit in the statement of the
Theorem 1.6.2. There seems to be no simple way to make the group of C k
diffeomorphisms into an infinite-dimensional Lie group modeled on a Banach
space. This fact seems to interfere with potential applications of global analysis
techniques to important nonlinear systems of PDE’s, such as those governing
incompressible fluids.
One consequence of Theorem 1.6.2 is a smoothness result for the evaluation map

ev : C k (Σ, M ) × M → M defined by ev(f, p) = f (p). (1.9)

35
Lemma 1.6.3.The map ev : C k (Σ, M ) × Σ → M , defined by (1.9).
For a direct proof we refer to [2], page 99.
We can extend Theorem 1.6.2 and many of its consequences to the Sobolev man-
ifolds Lpj (Σ2 , M ) when p and k are large enough that Lpj (Σ2 , M ) ⊂ C k (Σ2 , M )
in accordance with (1.7). For example, one consequence is:
Lemma 1.6.4. If Lpj (Σ, M ) ⊂ C k (Σ, M ), the map

ev : Lpj (Σ, M ) × Σ → M defined by ev(f, p) = f (p)

is C k .
Note that it follows from this Lemma that

ev : L2k (S 1 , M ) × Σ → M is C k−1 ,

while if Σ is a surface,

ev : Lpk (Σ, M ) × Σ → M is C k−1 ,

when p > 2.

1.7 The tangent and cotangent bundles

Many constructions from the theory of finite-dimensional manifolds can be gen-
eralized to infinite-dimensional Banach or Hilbert manifolds. These include
tensors of various ranks, vector fields and differential equations, connections,
Riemannian metrics on Hilbert manifolds, Finsler metrics on Banach mani-
folds, differential forms and de Rham cohomology. Many of these constructions
are carried out in great detail in the Lang’s book [43] on infinite-dimensional
manifolds. We provide a brief summary here.
We first extend familiar definitions of tangent and cotangent bundles to the
infinite-dimensional context. Let M be an infinite-dimensional smooth manifold
modeled on a Banach space E with smooth atlas {Uα , φα ) : α ∈ A}. Consider
the collection of triples (α, p, v), where α ∈ A, p ∈ Uα and v ∈ E. On this
collection of triples we define an equivalence relation ∼ by

(α, p, v) ∼ (β, q, w) ⇔ p = q and w = D(φβ ◦ φ−1

α )(φα (p))v.

The set of equivalence classes is called the tangent bundle of M and is denoted
by T M.
Let [α, p, v] denote the equivalence class of (α, p, v) and define

π : T M −→ M by π([α, p, v]) = p.

Let Ũα = {[α, p, v]; p ∈ Uα , v ∈ E}, and define

φ̃α : Ũα −→ E ⊕ E by φ̃α ([α, p, v]) = (φα (p), v).

36
Then {(Ũα , φ̃α ) : α ∈ A} is a smooth atlas on T M making T M into a smooth
manifold modeled on the Banach space E ⊕ E. If p ∈ M, we let Tp M = π −1 (p),
the fiber of the tangent bundle over p, and call Tp M the tangent space to M
at p.
Just as in the finite-dimensional case, elements of Tp M are called tangent
vectors. If γ : (a, b) → M is a smooth curve, t ∈ (a, b) and γ(t) ∈ Uα , we define

γ 0 (t) ∈ Tp M by γ 0 (t) = [α, γ(t), D(φα ◦ γ)(t) · 1],

a tangent vector called the velocity vector to γ at t.

If F : M → N is a smooth map between manifolds with atlases {(Uα , φα ) :
α ∈ A} and {(Vβ , ψβ ) : β ∈ B} and p ∈ M, we can define the differential
(F∗ )p : Tp M → TF (p) N by

(F∗ )p ([α, p, v]) = (β, F (p), (D(ψβ ◦ F ◦ φ−1

α )(φα (p))(v)],

where p ∈ Uα and F (p) ∈ Vβ . Note that if γ : (a, b) → M is a C 1 curve,

(F∗ )p (γ 0 (t)) = (F ◦ γ)0 (t), for t ∈ (a, b).

The differentials fit together to form a map of tangent bundles F∗ : T M → T N .

In a very similar way, we can also describe the cotangent bundle of M. We
consider a similar collection of triples (α, p, v ∗ ), where α ∈ A, p ∈ Uα and
v ∗ ∈ E ∗ , where E ∗ is the Banach space dual to E. This time we choose the
equivalence relation

(α, p, v ∗ ) ∼ (β, q, w∗ ) ⇔ p = q and v ∗ = [D(φβ ◦ φ−1 ∗ ∗

α )(φα (p))] w ,

where (·)∗ denotes transpose map defined by

[D(φβ ◦ φ−1 ∗ ∗
(v) = w∗ (D(φβ ◦ φ−1


α )(φα (p))] w α )(φα (p))(v).

The cotangent bundle T ∗ M is the set of equivalence classes and is a smooth

manifold modeled on the Banach space E⊕E ∗ . Once again, we have a projection

π : T M −→ M defined by π([α, p, v]) = p.

If p ∈ M, the fiber Tp∗ M of the cotangent bundle over p is called the cotangent
space to M at p. A smooth map F : M → N induces a map in the opposite
direction

(F ∗ )p : TF∗ (p) N → Tp M by (F ∗ )p (w∗ )(v) = w∗ ((F∗ )p (v)).

In terms of local coordinates, this can be written

(F ∗ )p ([β, F (p), w∗ ]) = [α, p, (D(ψβ ◦ F ◦ φ−1 ∗ ∗

α )(φα (p)) (w )]. (1.10)

We can also define the k-th tensor power and the k-th exterior power of the
cotangent bundle. To do this, we start with the Banach space Lk (E, R) of maps

T : E × E × · · · E(k times) −→ R

37
which are linear in each variable, the so-called space of k-linear maps, or its
subspace of alternating k-linear maps Lka (E, R). An element T ∈ Lk (E, R) is
said to be alternating if

T (hσ(1) , . . . , hσ(k) ) = (sgn(σ))T (h1 , . . . , hk ), for all σ ∈ Sk ,

where Sk denotes the symmetric group on k letters and sgn(σ) denotes the sign
of the permutation σ ∈ Sk . A continuous linear map Φ : E → F between
Banach spaces induces continuous linear maps

Φ∗ : Lk (F, R) → Lk (E, R), Φ∗ : Lka (F, R) → Lka (E, R)

by means of the formula

(Φ∗ T )(v1 , . . . vk ) = T (Φ(v1 ), . . . , Φ(vk )).

To define the k-th tensor power of the cotangent bundle, we start with triples
(α, p, T ), where α ∈ A, p ∈ Uα and T ∈ Lk (E, R) and the equivalence relation

(α, p, Tα ) ∼ (β, q, Tβ ) ⇔ p = q and Tα = Φ∗ Tβ ,

where Φ = D(φβ ◦ φ−1α )(φα (p)). The k-th tensor power of the cotangent bundle
⊗k T ∗ M is the set of equivalence classes. The k-th exterior power is defined
the same way, except that T is taken to lie in Lka (E, R). The fibers ⊗k Tp∗ M
and Λk Tp∗ M are called the k-th tensor power and the exterior power of the
cotangent space to M at p.
In the case where M = L21 (S 1 , M ), M being oriented, the tangent bundle
has another description, namely

T L21 (S 1 , M ) = L21 (S 1 , T M ).

To see this, recall how we constructed the atlas on L21 (S 1 , M ). If γ is a

smooth element of L21 (S 1 , M ), we let

Uγ, = {λ ∈ L21 (S 1 , M ) : dM (λ(t), γ(t)) <  for all t ∈ S 1 },

Vγ, = {X ∈ L21 (S 1 , Rn )) : hX(t), X(t)i <  for all t ∈ S 1 },

and define

ψγ, : Vγ, −→ Uγ, by (ψγ, (X))(t) = expγ(t) (X(t)).

For  sufficiently small, ψγ, is a bijection with inverse φγ, : Uγ, → Vγ, , and

{(Uγ, , φγ, ) : γ is smooth and  is sufficiently small}

is a smooth atlas for L21 (S 1 , M ).

Now for each smooth γ, we can construct a lift γ̃ ∈ L21 (S 1 , T M ) by setting

γ̃(t) = 0γ(t) ∈ Tγ(t) M,

38
and construct a corresponding chart on L21 (S 1 , T M ). The remarkable fact is
that we can choose the chart to be valid over all of

Ũγ, = Ω−1 2 1
π (Uγ, ) = {X ∈ L1 (S , T M ) : π ◦ X ∈ Uγ, }.

Indeed, we can set

Ṽγ, = Vγ, × { L21 -sections of γ ∗ T M }

and define ψ̃γ, : Ṽγ, → Ũγ, by

ψ̃γ, (X, Y ) = (expγ(t) X(t), (d(expγ(t) )X(t) Y (t)).

−1
Finally, define φ̃γ, : Ũγ, → Ṽγ, by φ̃γ, = ψ̃γ, . Then

(φ̃γ1 , ◦ φ̃−1 −1 −1
γ2 , )(X, Y ) = ((φγ1 , ◦ φγ2 , )(X), D(φγ1 , ◦ φγ2 , )(X)(Y )).

Thus the charts transform exactly the way they should for the tangent bundle.
In a quite similar fashion, we can show that if Σ is a compact Riemann
surface and p > 2,
T Lp1 (Σ, M ) = Lp1 (Σ, T M ).
It is important to observe that just as the imbedding i : M → RN induces an
imbedding ωi : L21 (S 1 , M ) → L21 (S 1 , RN ), so the imbedding i : T M → T RN =
R2N induces an imbedding

ωi : T L21 (S 1 , T M ) = L21 (S 1 , T M ) −→ L21 (S 1 , T RN ) = L21 (S 1 , R2N ),

allowing us to realize T L21 (S 1 , T M ) as a subspace of a Banach space. Similarly,

T Lp1 (Σ, M ) can be regarded as a subspace of a Banach space.
Note that the Hilbert space inner product allows us to identify the model
space L21 (S 1 , Rn ) for M = L21 (S 1 , M ) with its dual. Using this fact, it is not
difficult to verify that

T ∗ L21 (S 1 , M ) = L21 (S 1 , T ∗ M ), ⊗k T ∗ L21 (S 1 , M ) = L21 (S 1 , ⊗k T ∗ M ),

and
Λk T ∗ L21 (S 1 , M ) = L21 (S 1 , Λk T ∗ M ).
It is actually the “sections” of the tangent bundle and the exterior powers
of the cotangent bundle that will be of most importance for us; these are called
vector fields and differential forms, respectively.
Definition. A smooth vector field on M is a smooth map

X : M −→ T M such that π ◦ X = idM .

Note that if X : M → T M is a smooth vector field and f : M → R is a smooth

function, we can define the vector field f X : M → T M by (f X )(p) = f (p)X (p).

39
This makes the space of smooth vector fields into a module over the ring of
smooth real-valued functions.
Example. If M is a finite-dimensional Riemannian manifold and X : M → T M
is a smooth vector field on M , then

ωX : L21 (S 1 , M ) −→ L21 (S 1 , T M ) = T L21 (S 1 , M ) and

ωX : Lp1 (Σ, M ) −→ Lp1 (Σ, T M ) = T L21 (Σ, M )

are smooth vector fields on L21 (S 1 , M ) and Lp1 (Σ, M ).

1.8 Differential forms

For calculations on smooth manifolds, differential forms are often more conve-
nient to use than general tensor fields. We now describe how some of the familiar
operations on differential forms extend to infinte-dimensional manifolds.
Definition. A smooth covariant tensor field of rank k on M is a smooth map

φ : M −→ ⊗k T ∗ M such that π ◦ φ = idM .

A smooth differential form of degree k on M (or a smooth k-form) is a smooth

map
φ : M −→ Λk T ∗ M such that π ◦ φ = idM .

As in the case of vector fields, we can multiply covariant tensor fields or differ-
ential forms by functions.
An important example of differential one-form occurs when f : M → R is a
smooth function. Then for the coordinate chart (Uα , φα ), we have

D(f ◦ φ−1 ∗
α ) : Uα → L(E, R) = E , (1.11)

where E is the model space of M. The differential of f is the smooth one-form

df such that
df (p) = [α, p, D(f ◦ φ−1
α )(p)], for p ∈ Uα .
It is readily verified that the local representatives transform as they should
under change of coordinates. Moreover it is easily checked that the differentials
of functions at a point p generate the cotangent space. The Leibniz rule for
differentiation implies that d(f g) = gdf + f dg.
Definition. A point p ∈ M is a critical point for the real-valued function
f : M → R if df (p) = 0.
An important example to keep in mind is the real-valued function
Z
1
2 1
J : L1 (S , M ) → M, J(γ) = J(γ) = |γ 0 (t)|2 dt,
2 S1

40
when M is a Riemannian manifold. In this case, regularity theory will show
that a critical point in this case is actually a C ∞ map, hence a smooth closed
geodesic in M .
Using the notion of differential of a function, we can define the directional
derivative of a function f in the direction of X by

X (f )(p) = df (p)(X (p)),

the right hand side being the dual pairing between cotangent and tangent spaces.
Lemma 1.8.1. Let X and Y be smooth vector fields on the Banach manifold
M. Then there is a unique vector field [X , Y] on M which satisfies the equation

([X , Y](f ))(p) = (X Y(f ))(p) − (YX (f ))(p).

Sketch of proof: It suffices to show that the above formula is equivalent to an

expression for [X , Y] in terms of a local coordinate chart (Uα , φα ). Suppose that

X̃ , Ỹ : Uα → E are defined by X (p) = [p, α, X̃ (p)], Y(p) = [p, α, Ỹ(p)].

Using the chain rule, one can check that the Lie bracket must then be given by
h i
[X , Y](p) = p, α, DỸ(p)X̃ (p) − DX̃ (p)Ỹ(p) .

The vector field [X , Y] is known as the Lie bracket of X and Y; it is easily

verified that it satisfies the identity:

[f X , gY] = f g[X , Y] + f X (g)Y − gY(f )X .

We next note that differential forms are “functorial.” If F : M → N is a

smooth map and φ is a smooth differential form of degree k on N , we can define
a differential form F ∗ φ on M by

[F ∗ φ](p) = Fp∗ (φ(F(p));

it follows from (1.10) that F ∗ φ is smooth.

If X1 , . . . , Xk are smooth vector fields on M and φ is a smooth k-form on M,
then the smooth function

φ(X1 , . . . , Xk ) : M −→ R

is defined fiberwise via the continuous (k + 1)-linear map

Lka (E, R) × E × · · · × E −→ R,

where E is the model space for M.

41
Definition. If φ is a smooth k-form on M and ω is a smooth l-form on M, the
wedge product of φ and ω is the (k + l)-form on M defined by

k!l! X
(φ ∧ ω)(X1 , . . . , Xk+l ) = sgn(σ)(Xσ(1) , . . . , Xσ(k+l) ).
(k + l)!
σ∈Sk+l

Here Sk+l is the symmetric group on k + l letters and sgn(σ) is the sign of the
permutation σ ∈ Sk+l .
We remind the reader that some authors prefer to define the wedge product
using the factor
1 k!l!
instead of .
(k + l)! (k + l)!
With either convention, the wedge product is bilinear and associative, just as in
the case of finite-dimensional manifolds, but not commutative. If φ is a k-form
and ω an l-form,
φ ∧ ω = (−1)kl ω ∧ φ.

Definition. If φ is a smooth k-form on M and X is a smooth vector field, the

interior product ιX φ is the smooth (k − 1)-form on M defined by the formula

ιX φ(X2 , . . . , Xk ) = φ(X , X2 , . . . , Xk ),

whenever X1 , . . . , Xk are smooth vector fields on M. (It is readily checked that

there is a unique such differential form.) It is easily checked that

ιX (φ ∧ ψ) = (ιX φ) ∧ ψ + (−1)deg(φ) φ ∧ ιX ψ.

Finally, the exterior derivative d is the collection of R-linear maps from

k-forms to (k + 1)-forms which satisfy the following axioms, familiar from finite-
dimensional differential topology:
1. If ω is a k-form, the value dω(p) depends only on ω and its derivatives at
p.
2. If f is a smooth real-valued function regarded as a differential 0-form, d(f )
is the differential of f defined before.

3. d ◦ d = 0.
4. If ω is a k-form and φ is an l-form, then

d(ω ∧ φ) = (dω) ∧ φ + (−1)k ω ∧ (dφ).

5. If F : N → M is a smooth map, F ∗ ◦ d = d ◦ F ∗ on differential forms.

42
Just as in the finite-dimensional case, one can prove:
Theorem 1.8.2. There is a unique of linear maps of real vector spaces,

d : {differential k-forms} −→ {differential (k + 1)-forms},

which satisfy the five above axioms. Moreover, these linear maps satisfy the
explicit formula
X  
dω(X0 , . . . , Xk ) = (−1)i Xi ω(X0 , . . . , Xbi , . . . , Xk )
X
+ (−1)i+j ω([Xi , Xj ], X0 , . . . , Xbi , . . . , Xbj , . . . , Xk ), (1.12)
i<j

where the hats denote elements which are left out.

We sketch the proof under the assumption that the corresponding theorem for
finite-dimensional manifolds has been established. Using the fifth axiom, we
can reduce the proof of uniqueness to the case where the manifold is an open
subset of the model space. Since (1.12) is linear over functions, it suffices to
establish the formula in the case where X0 , . . . Xk are constant in terms of
the local chart, in which case all brackets [Xi , Xj ] vanish. Thus it suffices to
prove uniqueness when the vector fields X0 , . . . Xk are tangent to a (k + 1)-
dimensional affine subspace of the model space, and in this case (1.12) follows
from the corresponding formula on this (k + 1)-dimensional subspace, a finite-
dimensional submanifold.
To prove the local existence, one merely defines the exterior derivatives by
the formula (1.12) and check that they satisfy all of the axioms. It is simplest
to verify the first four axioms by showing that if any of these axioms were to
fail, it would have to fail already on finite-dimensional manifolds.
Finally, one notes that if operators satisfying the five axioms are defined and
unique for any open subset of the model space, they are defined and unique for
any set Uα in a smooth atlas {Uα : α ∈ A} for M. These operators must agree
on overlaps Uα ∩ Uβ , for α, β ∈ A, and hence must fit together to form well-
defined operators on M, and the resulting operators must be unique because
their restrictions to each Uα are unique.
Differential forms of degree k should be thought of as integrands for inte-
gration over compact oriented k-dimensional submanifolds. From this point
of view, it is evident that they should be determined by their restrictions to
finite-dimensional manifolds, where the above axioms for exterior derivative are
familiar.

1.9 Riemannian and Finsler metrics

To construct critical points of functions such as the action or energy, we need
to develop the “method of steepest descent” within the context of infinite-
dimensional manifolds. And to find the direction of steepest descent, we are

43
led to seek a notion of gradient, which in the finite-dimensional context de-
pends upon a Riemannian metric. Thus it is important to consider how to
extend the notion of Riemannian metric to infinite-dimensional manifolds. It is
to be expected that a somewhat stronger theory is possible for Hilbert manifolds
than for Banach manifolds. Indeed, we will see that there is no fully satisfactory
notion of Riemannian metric or gradient on Banach manifolds. We will need to
make do with weaker notions of Finsler metrics and pseudogradients.
Suppose, therefore that M is a Hilbert manifold modeled on the Hilbert
space E. If (Uα , φα ) is a smooth chart on M and Ũα is the subset of T M
projecting to Uα , define

εα : Ũα −→ E by εα ([α, p, v]) = v.

The Hilbert space inner product (·, ·) pulls back via εα to Tp M: we let

(·, ·)α : Tp M × Tp M → R by (v, v)α = (εα (v), εα (w)).

Definition. A Riemannian metric on a Hilbert manifold M is a function which

assigns to each p ∈ M an inner product

h·, ·ip : Tp M × Tp M −→ R

such that:
1. There is some constant cp > 0 such that

1
(v, v)α < hv, vip < cp (v, v)α , for all v ∈ Tp M.
cp

(Thus the topology induced by the Riemannian metric on Tp M agrees

with the model space topology.)
2. h·, ·ip varies smoothly with p; in other words, p 7→ h·, ·ip is a smooth
covariant tensor field of rank two.

Example 1.9.1. Suppose that we have a proper isometric imbedding of the

smooth Riemannian manifold M into RN . Then the Hilbert manifold L21 (S 1 , M )
can be given a very natural Riemannian metric hh·, ·ii as follows: if X, Y ∈
Tγ L21 (S 1 , M ), we can regard X and Y as maps X, Y : S 1 → RN such that
X(t) ∈ Tγ(t) M for each t ∈ S 1 . We set
Z
hX, Y iγ = [X(t) · Y (t) + X 0 (t) · Y 0 (t)]dt.
S1

This can be regarded as the pullback of the “flat” Riemannian metric on the
Hilbert space L21 (S 1 , RN ), and it is smooth because the pullback of a smooth
covariant tensor field via a smooth map is smooth.

44
Definition. If M is a Hilbert manifold with Riemannian metric p 7→ h·, ·ip , the
gradient of a C 1 function f : M → R is the vector field grad(f ) defined by

hgrad(f )(p), vip = dfp (v), for all v ∈ Tp M.

The idea behind the method of steepest descent for finding critical points of a
nonnegative function f is to follow flowlines for the vector field −grad(f ); in
favorable cases, these flowlines will converge to a critical point for f .
In the case of Banach manifolds, the metrics best suited to our applications
are not Riemannian, but Finsler. If (Uα , φα ) is a smooth chart on a Banach
manifold M, Ũα is the subset of T M projecting to Uα and

εα : Ũα −→ E by εα ([α, p, v]) = v

as before, the Banach space norm k · k on E pulls back via εα to Tp M: we let

k · kα : Tp M → R by kvkα = kεα (v)k.

Definition. A Finsler metric on a Banach manifold M is a function which

assigns to each p ∈ M a norm

k · kp : Tp M −→ R

such that
1. There is some constant cp > 0 such that
1
kvkα < kvkp < cp kvkα , for all v ∈ Tp M.
cp

(Thus the Finsler norm on Tp M is equivalent to the Banach space norm

for the model space.)
2. k · kp varies continuously with p.

Note that any Riemannian metric on a Hilbert manifold determines a Finsler

metric: simply define
q
k · kp : Tp M −→ R by kvkp = hv, vip .

Even in this special case, however, the norm k·kp is only continuous, not smooth
as a function on T M.
The Riemannian metric on a Hilbert manifold establishes a norm-preserving
vector bundle isomorphism from Tp M to Tp∗ M. We do not have such an iso-
morphism in the case of a Finsler metric on a Banach manifold, but the norm
k · kp on Tp M induces a dual norm (which we also denote by k · kp for simplicity)
on Tp∗ M:
kφkp = sup{|φ(v)| : v ∈ Tp M and kvkp = 1}.

45
Example 1.9.2. Suppose that we have a proper isometric imbedding of the
smooth Riemannian manifold M into RN . Given a compact Riemann surface Σ
and a real number p > 2, the Banach manifold Lp1 (Σ, M ) can be given a Finsler
metric as follows: if X ∈ Tf (Lp1 (Σ, M ), we can regard X as a map X : Σ → RN
such that X(p) ∈ Tf (p) M for each p ∈ Σ. We then let kXkf be the Lp1 -norm
of X as a mapping into Euclidean space. The Finsler metric f 7→ k · kf can
be regarded as the pullback of the “flat” Finsler metric on the Banach space
Lp1 (Σ, RN ).
If M is a connected Banach manifold with Finsler metric k·k and γ : [0, 1] → M
is a C 1 curve, we can define its length L(γ) by
Z 1
L(γ) = kγ 0 (t)kdt,
0

where integration along a path can be defined as in [43], §1.4. We can then
define a distance function d : M × M → R by

d(p, q) = inf L(γ) : γ : [0, 1] → M is a C 1 path, γ(0) = p, γ(1) = q . (1.13)

Proposition 1.9.3. Given a Finsler metric on a regular Banach manifold, the

distance function d defined aboveis a metric in the metric space sense, and the
metric topology agrees with the manifold topology.
We apologize to the reader for not giving a complete proof of this Proposition,
referring instead to [61] (see the appendix to §2).
However we will describe a proof of the Proposition for Examples 1.8.1 and
1.8.2. Note that it is quite easily verified that the distance function d satisfies

d(p, q) = d(q, p), d(p, r) ≤ d(p, q) + d(q, r), and d(p, p) = 0.

That only leaves the property d(p, q) = 0 ⇒ p = q.

Lemma 1.9.4. In each of our two key examples, L21 (S 1 , M ) and Lp1 (Σ, M ) with
p > 2, d is a metric and the metric topology agrees with the manifold topology.
Moreover, L21 (S 1 , M ) and Lp1 (Σ, M ) are complete as metric spaces.
Proof: Let us consider Lp1 (Σ, M ). If f, g ∈ Lp1 (Σ, M ) and d(f, g) = 0, then
there exist arbitrarily short paths connecting f and g in Lp1 (Σ, M ). But a path
connecting f and g in Lp1 (Σ, M ) also connects f and g in the ambient Banach
space E = Lp1 (Σ, RN ), so there are arbitrarily short curves connecting f and
g in the ambient Banach space. However, by a straightforward modification of
the finite-dimensional argument, it is easily verified that if γ : [0, 1] → E is a
C 1 -map into a Banach space, then
Z 1
kγ 0 (t)kdt ≥ kγ(0) − γ(1)k,
0

46
so two distinct points in E cannot be joined by curves of arbitrarily small length.
Since the map ωi : Lp1 (Σ, M ) → Lp1 (Σ, RN ) induced by the inclusion i : M →
N
R is an imbedding, it is now easy to verify that the metric topology agrees with
the manifold topology. Finally, since ωi : Lp1 (Σ, M ) → Lp1 (Σ, RN ) is distance-
decreasing, a Cauchy sequence {fi } in Lp1 (Σ, M ) is also a Cauchy sequence in
Lp1 (Σ, RN ), and must therefore converge. Since Lp1 (Σ, M ) is a closed subset of
Lp1 (Σ, RN ), we see that Lp1 (Σ, M ) must be complete as a metric space.
The case of L21 (S 1 , M ) is treated in the same way.

1.10 Vector fields and ODE’s

It is well-known that the global qualitative theory of systems of ordinary dif-
ferential equations is best formulated within the language of vector fields on
finite-dimensional manifolds. This theory, including the fundamental existence
and uniqueness theorem for systems of ordinary differential equations, can be
extended to infinite-dimensional manifolds. A detailed exposition of this exten-
sion is presented in Chapter IV of [43].
Definition. A C 1 curve γ : (a, b) → M is called an integral curve for the vector
field X if
X (γ(t)) = γ 0 (t), for t ∈ (a, b), (1.14)
where γ 0 (t) is the velocity vector to γ at t.
Just as in the finite-dimensional case, a fundamental existence and uniqueness
theorem states that given a smooth vector field X on M, there is a unique
integral curve for X which passes through any point of M:
Theorem 1.10.1. (Existence and Uniqueness Theorem for Ordinary
Differential Equations.) Suppose that X is a C 1 vector field on M and
p ∈ M. Then there is an open neighborhood U of p, an  > 0 and a C 1 map

φ : (−, ) × U −→ M

such that if φt (q) = φ(t, q) for t ∈ (−, ) and q ∈ U , then

1. each curve t 7→ φt (q) is an integral curve for X ,
2. any integral curve for X which passes through U is of the form t 7→ φt (q)
for some q ∈ U ,
3. φ0 is the inclusion U ⊂ M, and
4. φt ◦ φs = φt+s , whenever, both sides are defined.
We will call (−, ) × U a local flow box for X .
Idea of proof (following Chapter IV of Lang [43]): The proof is based upon the
Contraction Lemma, just like the proof of the Inverse Function Theorem.

47
 We can replace the differential equation (1.14) by its local coordinate repre-
sentation, and consider the initial value problem

γ 0 (t) = f (γ(t)), γ(0) = q, (1.15)

for γ : (a, b) → V and f : V → E, where V is a suitable open subset of a Banach

space E. We can assume that

kf k ≤ K and kDf k ≤ L

on V . Integrating both sides of (1.15) yields the equivalent integral equation

Z t
γ(t) = q + f (γ(u))du. (1.16)
0

We can assume that the closed ball B2δ (p) of radius δ about p is contained in V
and suppose that q ∈ Bδ (p), the open ball of radius δ about p. Let I = [−, ],
where  > 0 will be chosen later, and let
n o
X = γ : I → U : γ is continuous, γ(0) = q and γ(I) ⊂ B2δ (p) .

We can make X into a complete metric space by defining the distance function
d by
d(γ1 , γ2 ) = sup{kγ1 (t) − γ2 (t)k : t ∈ I}.
If γ ∈ X, we set Z t
T (γ)(t) = q + f (γ(u))du.
0
We choose  so that  < δ/K, and hence

kT (γ)(t) − qk ≤ K ≤ δ,

so T (γ) ∈ X. Finally, we note that

d (T (γ1 ), T (γ2 )) = sup {kT (γ1 )(t) − T (γ2 )(t)k : t ∈ I}

≤  sup{kf (γ1 (t)) − f (γ2 (t))k : t ∈ I}
≤ L sup{kγ1 (t)γ2 (t)k : t ∈ I} = Ld(γ1 , γ2 ).

Thus by choosing  so that L < 1, we can ensure that T : X → X will be

a contraction. Then, by the Contraction Lemma, T has a unique fixed point
γq ∈ X, which must be a solution to the integral equation (1.16). This fixed
point γq is C 1 and is the unique solution to the initial value problem (1.15).
Thus we can define

φ : (−, ) × Bδ (p) → V by φ(t, q) = γq (t).

It is relatively easy to check that properties 2, 3 and 4 of the Theorem hold and
that φ is continuous. It is a little more challenging to check that φ is C 1 , and
for that we refer the reader to the excellent presentation in [43].

48
Once we have the Existence and Uniqueness Theorem, we can piece together
the locally defined maps to form a map
φ : V −→ M, where V is an open neighborhood of {0} × M in R × M.
We say that the maps {φt } defined by φt (q) = φ(t, q) form the one-parameter
group of local diffeomorphisms of M which corresponds to the vector field X .

1.11 Condition C
We want to apply the existence and uniqueness theorem from the preceding
section to find critical points of a C 2 real-valued map f : M → [0, ∞) via the
method of steepest descent, where M is an infinite-dimensional manifold. In
order to get this method to work, we need to assume that the function f assumes
a “compactness condition” introduced by Palais and Smale [62]:
Definition. Suppose that M is a Banach manifold with a complete Finsler
metric. (For example, M might be a Hilbert manifold with a complete Rieman-
nian metric.) Then a C 2 function f : M → [0, ∞) is said to satisfy condition C
if whenever {pi } is a sequence in M such that
1. f (pi ) is bounded and
2. kdf (pi )k is not bounded away from zero,
then {pi } possesses a subsequence which converges to a critical point for f .
It is easiest to utilize this compactness condition in the case where M is a
Hilbert manifold:
Theorem 1.11.1. Suppose that M is a Hilbert manifold with a complete
Riemannian metric h·, ·i. If f : M → [0, ∞) is a C 2 function which satisfies
condition C, X = −grad(f ) and {φt } is the local one-parameter group of dif-
feomorphisms corresponding to X , then
1. for each p ∈ M, φt (p) is defined for all t ≥ 0, and
2. there is a sequence ti → ∞ such that {φti (p)} converges to a critical point
for f .

The proof will be based upon a collection of inequalities. Suppose that 0 < t1 <
t2 . Then
Z t2
d
f (φt1 (p)) − f (φt2 (p)) = − f (φt (p))dt
t1 dt
Z t2 Z t2
=− df (φt (p))(X (φt (p))dt = hgrad(f )(φt (p)), X (φt (p)idt
t1 t1
Z t2 Z t2
= kgrad(f )(φt (p))k2 dt = kdf (φt (p))k2 dt. (1.17)
t1 t1

49
On the other hand, using the metric d on the Riemannian manifold M, we have
Z t2
d


d(φt1 (p), φt2 (p)) ≤ (φt (p)) dt
dt
t1
Z t2 Z t2
≤ kX (φt (p))kdt = kdf (φt (p))kdt. (1.18)
t1 t1

Now we use the Cauchy-Schwarz inequality to obtain

Z t2 2
2
d(φt1 (p), φt2 (p)) ≤ kdf (φt (p))kdt
t1
Z t2
≤ (t2 − t1 ) kdf (φt (p))k2 dt = (t2 − t1 )(f (φt1 (p)) − f (φt2 (p))). (1.19)
t1

Let t̄ = sup{t ∈ R : φt (p) is defined}, and let {ti } be a sequence of real

numbers < t̄ such that ti → t̄. It follows from (1.19) that {φti (p)} is a Cauchy
sequence in M. Since (M, d) is a complete metric space, φti (p) → q, for some
q ∈ M. But by the Existence and Uniqueness Theorem there is a flow box
(−, ) × U containing (0, q). This implies that the curve t 7→ φt (p) can be
extended beyond t̄, giving a contradiction. Thus we see that φt (p) is defined for
all t ≥ 0, and the first statement of the Theorem is proven.
Next, it follows from (1.17) that
Z ∞
kgrad(f )(φt (p))k2 dt < ∞,
0

and hence there must exist a sequence ti → ∞ such that

kdf (φti (p))k = kgrad(f )(φti (p))k → 0.

By Condition C, a subsequence of {φti (p)} converges to a critical point for f ,

finishing the proof of the Theorem.
Of course, we would like a version of the above Theorem to hold for the case
of a C 2 function f : M → R, where M is only a Banach manifold. However,
it is not possible to define Riemannian metrics on M in this case, so we need
a replacement for the notion of gradient, such as the following, similar to a
definition suggested by Palais [59]:
Definition. Suppose that f : M → R is a C 2 function on a Banach manifold
which has a Finsler metric and let U be an open subset of M. A C 1 vector field
X : U → T M is called a pseudogradient for f over U if there is a constant  > 0
such that for each p ∈ U ,
1. dfp (X (p)) ≥ kdfp k2 ,
2. kX (p)k ≤ (1/)kdfp k.

50
In both inequalities, kdfp k = sup{|dfp (v)| : v ∈ Tp M and kvk ≤ 1}, which is the
dual norm on the cotangent space to M at p.
Of course, in the case of a Hilbert manifold, X = grad(f ) satisfies both con-
ditions in the definition with  = 1; in other words, a gradient on a Hilbert
manifold is also a pseudogradient. The definition was set up so that the follow-
ing Theorem would be true:
Theorem 1.11.2. Suppose that M is a Banach manifold with a complete
Finsler metric k · k. Suppose that f : M → [0, ∞) is a C 2 function which
satisfies condition C. Let
K = {p ∈ M : df (p) = 0}
and let U = M − K. If X is a pseudogradient for f on U , and {φt } is the local
one-parameter group of diffeomorphisms corresponding to −X , then
1. for each p ∈ M, φt (p) is defined for all t ≥ 0, and
2. there is a sequence ti → ∞ such that {φti (p)} converges to a critical point
for f .

The proof is almost identical to the proof of Theorem 1.11.1 (except that we
call the vector field −X instead of X ). Assuming that p is not a critical point
for f and that 0 < t1 < t2 , we use the first condition in the definition of
pseudogradient to replace (1.17) by the inequality
Z t2
d
f (φt1 (p)) − f (φt2 (p)) = − f (φt (p))dt
t1 dt
Z t2 Z t2
= df (φt (p))(X (φt (p))dt ≥  kdf (φt (p))k2 dt. (1.20)
t1 t1

We use the second condition in the definition of pseudogradient to replace (1.18)

by
Z t2
1 t2
Z
d(φt1 (p), φt2 (p)) ≤ kX (φt (p))kdt ≤ kdf (φt (p))kdt. (1.21)
t1  t1
Then we use the Cauchy-Schwarz inequality exactly as before to obtain
Z t2 2
2 1
d(φt1 (p), φt2 (p)) ≤ 2 kdf (φt (p))kdt
 t1
t2 − t1 t2 t2 − t1
Z
≤ kdf (φt (p))k2 dt ≤ (f (φt1 (p)) − f (φt2 (p))). (1.22)
2 t1 3
We can now use (1.22) instead of (1.19) to show that φt (p) is defined for all
t ≥ 0. Finally, it follows from (1.20) that
Z ∞
kdf (φt (p))k2 dt < ∞,
0

51
which enables us to find a sequence ti → ∞ such that kdf (φti (p))k → 0, and by
condition C, a subsequence of {φti (p)} converges to a critical point for f .
The only problem now is to show that given a C 2 function f : M → [0, ∞)
on a Banach manifold, we can construct a corresponding pseudogradient on
U = M − K, where K is the set of critical points for f . The standard technique
for constructing a pseudogradient consists of constructing pseudogradients over
each open set of an open cover of U and then piecing these together using a
partition of unity.
Let M be a metrizable infinite-dimensional smooth manifold modeled on
a Banach space. According to a well-known theorem of Stone, M must be
paracompact. That means that every open cover of M must have an open
locally finite refinement.
Definition. Let U = {Uα : α ∈ A} be an open cover of M. A partition of
unity subordinate to U is a collection {ψα : α ∈ A} of continuous real-valued
functions on M such that
1. ψα : M → [0, 1],

2. the support of ψα is a closed subset of Uα ,

3. if p ∈ M, there is an open neighborhood V of p which intersects the
supports of only finitely many ψα , and
P
4. ψα = 1.

It is known (and proven in topology texts) that any open cover of a paracompact
Hausdorff space possesses a subordinate continuous partition of unity.
Moreover, as proven in Lang [43], C ∞ Hilbert manifolds possess C ∞ parti-
tions of unity. However, for Banach manifolds, we encounter a perhaps unex-
pected obstacle. Banach manifolds need not possess C ∞ partitions of unity. As
pointed out in [16], for example, to construct C k partitions of unity one needs
to be able to construct nontrivial real-valued C k functions on the model Banach
space E with bounded support.
Fortunately, in the case where Σ is a Riemann surface and p > 2 the Banach
manifold Lp1 (Σ, M ) does possess partitions of unity of class C 2 . To see why, we
notice that the function

f : Lp1 (Σ, R) −→ R defined by f (φ) = kφkp

is C 2 , by an argument similar to that given in Example 1.2.3. Let g : R → [0, 1]

be a smooth function such that
1. g(s) = 1 when |s| ≤ 1, and

2. g(s) = 0 when |s| ≥ 2.

52
Then the map
 
2f (φ)
f : Lp1 (Σ, Rn ) → [0, 1] defined by f (φ) = g

is a C 2 function which equals one on a small neighborhood of the origin and
has support contained in the set {φ ∈ Lp1 (Σ, Rn ) : kφk ≤ }. Using local
coordinates, we can transport this function to Lp1 (Σ, M ) thereby obtaining a C 2
function f : Lp1 (Σ, M ) → R which is one in a neighborhood of a given point p
and vanishes outside a given open neighborhood of p.
Using this function to start with, we can follow the familiar argument (such
as given in Lang [43], Chapter II, §3) to construct C 2 partitions of unity subor-
dinate to any open cover on the smooth manifold Lp1 (Σ, M ).
Lemma 1.11.3. If f : Lp1 (Σ, M ) → R is a C 2 function, where p ≥ 2, then f pos-
sesses a C 2 pseudogradient X which is tangent to every Lpk (Σ, M ) ⊂ Lp1 (Σ, M ),
for k ∈ N.
Proof: Suppose that 0 <  < 1. If p is not a√critical point for f , we can choose
a unit vector u ∈ Tp M such that |dfp (u)| > kdfp k; then
√
v = kdfp ku satisfies kvk ≤ kdfp k, dfp (v) ≥ kdfp k2 ,

the two conditions in the definition of pseudogradient at the point p. We can

extend v to a smooth vector field on some neighborhood of p which is a pseudo-
gradient; for example, we could choose it to be constant in terms of some smooth
chart. Thus we can construct a pseudogradient on an open neighborhood about
any point p which is not in the set K of critical points of F . If M admits C 2
partitions of unity, one can piece together a C 2 pseudogradient on M − K.
In the above construction, we can choose v to lie in the dense subspace
Lpk (Σ, M ) of Lp1 (Σ, M ) and the C 2 partition of unity on Lp1 (Σ, M ) can be chosen
so that it restricts to a C 2 partition of unity on Lpk (Σ, M ) for every k ≥ 1. When
this is done, the pseudogradient will be tangent to every Lpk (Σ, M ), finishing the
proof of the Lemma.

1.12 Topological constraints give critical points

Suppose now that M is a Banach manifold with a complete Finsler metric.

Suppose that f : M → [0, ∞) is a C 2 function and let

Ma = {p ∈ M : f (p) ≤ a}.

Definition. An ambient isotopy of M is a smooth map Ψ : M×[0, 1] → M such

that if ψt : M → M is defined by ψt (p) = Ψ(p, t), then ψt is a diffeomorphism
for each t ∈ [0, 1] and ψ0 = id.
Theorem 1.12.1. (Deformation Theorem) Suppose that M is a Banach
manifold with a complete Finsler metric. If f : M → R is a nonnegative C 2

53
function satisfying condition C and there are no critical points p for f such that
a ≤ f (p) ≤ b, then
1. Ma is a strong deformation retract of Mb .
2. there is a smooth ambient isotopy Ψ = {ψt : t ∈ [0, 1]} of M such that
ψ1 (Mb ) ⊂ Ma .

Proof: It follows from Condition C that there is an  > 0 such that there are no
critical points p for f such that a −  < f (p) < b + . Using a partition of unity,
we construct a smooth function η : M → [0, 1] such that
1. η ≡ 1 on {p ∈ M : a ≤ f (p) ≤ b},
2. η ≡ 0 outside {p ∈ M : a −  < f (p) < b + }.
Let Y be a pseudogradient for f on U = M − K, where K is the critical locus
of F and set X = −ηY.
Since f satisfies Condition C, there is a constant k > 0 such that kdf k ≥ k
on {p ∈ M : a ≤ f (p) ≤ b}. Indeed, if not, there would exist a sequence {pi } in
{p ∈ M : a ≤ f (p) ≤ b} such that kdf (pi )k → 0. By condition C, a subsequence
of {pi } would converge to a critical point for f in {p ∈ M : a ≤ f (p) ≤ b},
contradicting the hypothesis of the Theorem.
Theorem 1.10.2 shows that the one-parameter group {φt : t ∈ R} of local
diffeomorphisms determined by X is globally defined for all t ≥ 0. We claim
that if p ∈ Mb , then φt (p) ∈ Ma for t > (b − a)/(k). Indeed, if φt (p) ∈
/ Ma ,
it follows from (1.20) that
Z t
f (p) − f (φt (p)) ≥  kdf (φτ (p))kdτ ≥ kt,
0

and hence kt ≤ b − a, or equivalently, t ≤ (b − a)/(k). We now set ψt = φct ,

where c = 2(b − a)/(k), and define

Ψ : M × [0, 1] → M by Ψ(p, t) = ψt (p).

Then Ψ is an ambient isotopy such that ψ1 (Mb ) ⊂ Ma .

This proves the second assertion of the Theorem. For the first, we let τ (p)
be the first time t such that φt (p) ∈ Ma and define Φ : Mb × [0, 1] → Mb by
(
φt(τ (p)) , for p ∈ Mb − Ma ,
Φ(p, t) =
p, for p ∈ Ma .

Then Φ is a strong deformation retraction from Mb to Ma , finishing the proof

of the Theorem.
Definition. Let F be a family of subsets of M. We say that F is ambient
isotopy invariant if
A∈F ⇒ ψ1 (A) ∈ F

54
whenever {ψt : t ∈ [0, 1]} is an ambient isotopy of M.
Theorem 1.12.2. (Minimax Theorem) Suppose that M is a smooth man-
ifold with a complete Finsler metric. Suppose, moreover, that f : M → [0, ∞)
is a C 2 function satisfying condition C and F is a nonempty family of subsets
of M which is ambient isotopy invariant. Then

Minimax(f, F) = inf {sup{f (p) : p ∈ A} : A ∈ F}

is a critical value for f .

Proof: Let c = Minimax(f, F). If c is not a critical value for f , then Condition
C implies that there exists an  > 0 such that there are no critical points p ∈ M
with f (p) ∈ (c − , c + ). But by definition of Minimax(f, F), there exists an
A ∈ F such that A ⊂ Mc+ . Theorem 1.11.1 then gives a smooth isotopy
{ψt : t ∈ [0, 1]} such that

ψ1 (Mc+ ) ⊂ Mc− , so ψ1 (A) ⊂ Mc− .

But then ψ1 (A) ∈ F showing that Minimax(f, F) ≤ c − , a contradiction.

For example, we could let M0 be a component of M and let

F = {A ⊆ M : A ⊆ M0 }.

Then F is ambient isotopy invariant and the previous Theorem implies:

Corollary 1.12.3. Suppose that M is a smooth manifold with a complete
Finsler metric, and that f : M → [0, ∞) is a C 2 function satisfying condition
C. Then f assumes its minimum value on each component of M.
We can also use invariants from algebraic topology to construct critical points.
For example, suppose that M is simply connected and [α] is a nonzero element
in πk (M), the k-th homotopy group of M. In this case,

F[α] = { h(S k ) such that h : S k → M is a continuous map representing [α]}

is ambient isotopy invariant, and hence there is a minimax critical point corre-
sponding to the homotopy class [α]. Alternatively, suppose that x is a nonzero
element in Hk (M; R), the singular homology group of M of degree k with real
coefficients, and let

Fx = {h(A) such that A is a compact oriented manifold of dimension k

with fundamental class [µA ] and h : A → M is a
continuous map with h∗ ([µA ]) a nonzero multiple of x}.

Once again Fx is ambient isotopy invariant, and one obtains a minimax critical
point corresponding to the homology class x. For a third example, we suppose

55
that θ is a differential k-form on M such that dθ = 0, and let

Fθ = { h(A) such that A is a compact oriented manifold of dimension k

Z 
1 ∗
and h : A → M is a C -map such that h θ 6= 0 .
A

It follows from the Homotopy Lemma to be proven in the next section that Fθ
is ambient isotopy invariant, so once again we obtain a minimax critical point
corresponding to the differential form θ.
Theorem 1.12.4. Suppose that K is the set of critical points for f whose
critical value is Minimax(f, F) and that U is an open neighborhood of K within
M. If the elements in F are compact, then there is an element A ∈ F such that
for some  > 0,
φt (A) ⊂ Mc− ∪ U, for all t ≥ 0.

Proof: We follow an argument presented by Klingenberg [42]. First choose an

open subset V of M such that K ⊂ V ⊂ U and the distance from V to M−U is
a positive number δ. It follows from condition C that the norm of df is bounded
below on M − V —otherwise, a sequence of points in M − V would converge
to a critical point which would not lie in K. Moreover, any orbit of −X which
enters V can only leave U if it travels for a distance δ in U − V , but then it
follows from (1.21) that the value of f must decrease by at least δ. Thus if we
set  = δ/2, there will exist an element A ∈ F such that A ⊂ Mc+ , and when
T is sufficiently large, every orbit starting in A will have either passed below
level c −  or entered V at least once in time T . Hence A0 = φT (A) is an element
of F and φt (A0 ) ⊂ Mc− ∪ U for all t ≥ 0.

1.13 de Rham cohomology

Once one has C 2 partitions of unity on Banach manifolds, it is relatively straight-
forward to extend de Rham cohomology to Banach manifolds. Indeed, C 2 par-
titions of unity will enable us to piece together C 1 differential forms with C 1
exterior derivatives, elements of the vector space

Ωk (M) = {C 1 differential forms ω on M of degree k : dω ∈ C 1 }.

We make the Ωk (M)’s into a cochain complex in which the differential is the
exterior derivative
d : Ωk (M) −→ Ωk+1 (M).
We say that an element ω ∈ Ωk (M) is closed if dω = 0 and exact if ω = dθ
for some θ ∈ Ωk−1 (M). Since d ◦ d = 0, every exact k-form is closed. The
quotient space

k closed elements of Ωk (M)

HdR (M; R) = .
exact elements of Ωk (M)

56
is called the de Rham cohomology of M. If ω ∈ Ωk (M) is closed, we let [ω]
k
denote its cohomology class in HdR (M; R). In the terminology of algebraic
topology, the de Rham cohomology is the cohomology of the cochain complex

· · · → Ωk−1 (M) → Ωk (M) → Ωk+1 (M) → · · · . (1.23)

Note that de Rham cohomology has a cup product defined by

[ω] ∪ [φ] = [ω ∧ φ],

which makes the direct sum

∞
X
∗ k
HdR (M; R) = HdR (M; R)
k=0

into a graded commutative algebra over the ring of smooth real-valued functions
on M . Moreover, the cup product behaves well under smooth maps: If F : M →
N is a smooth map, the linear map F ∗ on differential forms induces a linear
map

F ∗ : HdR
k k
(N ; R) −→ HdR (M; R) such that F ∗ ([ω] ∪ [φ]) = F ∗ [ω] ∪ F ∗ [φ].

Moreover, the identity map on M induces the identity on de Rham cohomology

and if F : M → N and G : N → P are smooth maps, then (G ◦ F )∗ = F ∗ ◦ G∗ ,
so

M 7→ H k (M; R), (F : M → N ) 7→ (F ∗ : HdR

k k
(N ; R) → HdR (M; R))

is a contravariant functor from the category of smooth manifolds and smooth

maps to the category of real vector spaces and linear maps.
Lemma 1.13.1. (Poincaré Lemma.) If U is a convex open subset of a
Banach space E, or more generally any contractible open subset of E, then the
de Rham cohomology of U is trivial:
(
k ∼ R if k = 0,
HdR (U ; R) =
0 if k 6= 0.

One can modify the proof that is used in the finite-dimensional case. We only
sketch the key ideas—the reader should refer to Lang [43] for details. Since the
inclusion from a point into the convex set U is a homotopy equivalence, the
Poincaré Lemma is an immediate consequence of
Lemma 1.13.2. (Homotopy Lemma.) Smoothly homotopic maps F, G :
M → N induce the same map on cohomology,

F ∗ = G∗ : HdR
k k
(N ; R) −→ HdR (M; R).

57
On the other hand via functoriality, this follows from the special case of the
Homotopy Lemma for the inclusion maps
i0 , i1 : M −→ [0, 1] × M, i0 (p) = (0, p), i1 (p) = (1, p).
Indeed, if H : [0, 1] × M → N is a smooth homotopy from F to G, then by
definition of homotopy, F = H ◦ i0 and G = H ◦ i1 , so
i∗0 = i∗1 ⇒ F ∗ = i∗0 ◦ H ∗ = i∗1 ◦ H ∗ = G∗ .
This special case, however, can be established by integrating over the fiber
of the projection on the second factor [0, 1] × M → M. More precisely, let t
be the standard coordinate on [0, 1], T the vector field tangent to the fiber of
[0, 1] × M such that dt(T ) = 1. We then define integration over the fiber
Z 1
k k−1
π∗ : Ω ([0, 1] × M) → Ω (M) by π∗ (ω)(p) = (ιT ω)(t, p)dt.
0
∗ k
Here the interior product (ιT ω)(t, p) is an element of Λ T(t,p) ([0, 1] × M) and
the integration is possible because the exterior power at (t, p) is canonically iso-
∗
morphic to Λk T(0,p) ([0, 1] × M). The key to proving that i∗0 = i∗1 in cohomology
is the “cochain homotopy” formula
i∗1 ω − i∗0 ω = d(π∗ (ω)) + π∗ (dω).
This formula can be verified by using naturality to reduce the proof to the
finite-dimensional case, and then calculating in local coordinates just as in the
familiar finite-dimensional treatment found in [10]. Note that
dω = 0 ⇒ i∗1 ω − i∗0 ω = d(π∗ (ω)) ⇒ [i∗1 ω] = [i∗0 ω],
and hence on the cohomology level i∗0 = i∗1 . This finishes our sketch of the proof
of the Homotopy Lemma and the Poincaré Lemma.
Remark 1.13.3. If N is a submanifold of M with inclusion map i : N → M,
we let
Ωk (M, N ) = ker(i∗ : Ωk (M) → Ωk (N )),
and note that the exterior derivative makes this into the k-th cochain group of a
cochain complex Ω∗ (M, N ). The cohomology of this complex is called the rela-
k
tive de Rham cohomology of the pair (M, N ) and is denoted by HdR (M, N ; R).
The short exact sequence of cochain complexes
0 → Ω∗ (M, N ) → Ω∗ (M) → Ω∗ (N ) → 0. (1.24)
yields a long exact sequence via the “snake lemma” (Theorem 2.16 in [35]) from
algebraic topology:
k k k
· · · → HdR (M, N ; R) → HdR (M; R) → HdR (N ; R)
k+1 k+1
→ HdR (M, N ; R) → HdR (M; R) → · · · .

58
This is very useful for calculating de Rham cohomology.
For us, one of the primary uses of differential forms will be to calculate
cohomology of infinite-dimensional manifolds. It is important to realize that
the de Rham cohomology is the same as the singular or Čech cohomology with
real coefficients that is studied in algebraic topology:
Theorem 1.13.4. (de Rham Theorem.) Suppose that either M = Lpk (Σ, M )
where pk > dim(Σ) or M is finite-dimensional. Suppose, moreover that M
admits C 2 partitions of unity. Then the cohomology of the cochain complex
Ω∗ (M) is isomorphic to the Čech cohomology of M.
The proof (due to André Weil) is via the zig-zag construction described in the
excellent text on de Rham theory by Bott and Tu [10], which we follow closely.
In the case where M is finite-dimensional, we let U = {Uα : α ∈ A} be
a locally finite open cover of M by sets which are geodesically convex with
respect to a Riemannian metric on M. If M = Lpk (Σ, M ) where pk > dim(Σ),
we construct an open cover U = {Uα : α ∈ A} in which each open set Uα is of
the form
Uα = {g ∈ Lpk (Σ, M ) : kg − f kC 0 < δ}.
The Sobolev Lemma guarantees the existence of such an open cover. We choose
δ so small that if p, q ∈ M and d(p, q) < 2δ, then there is a unique minimizing
geodesic

γp,q : [0, 1] → M such that γp,q (0) = p, γp,q (1) = q,

and moreover this geodesic depends smoothly on p and q. (Then any two points
in a δ-ball about a given point can be connected by a unique such geodesic.) If
g1 and g2 are two elements of Uα , we can then define a path

Γg1 ,g2 : [0, 1] → Lpk (Σ, M ) by Γg1 ,g2 (t)(p) = γg1 (p),g2 (p) (t).

It is easily checked in either case that the intersection of any collection of el-
ements from U is contractible. Readers familiar with cohomology theory will
remember that the Čech cohomology of such an open covering is the same as
the Čech cohomology of M (and we do not need to take direct limits). Such an
open cover is often called a “good” cover or “Leray” cover.
We now construct a double complex K ∗,∗ in which the (p, q)-element is
p
K p,q = Č (U, Ωq ),

which is defined to be the space of functions ω which assign to each distinct

ordered (p + 1)-tuple (α0 , . . . , αp ) of indices such that Uα0 ∩ · · · ∩ Uαp 6= 0 an
element
ωα0 ···αp ∈ Ωq (Uα0 ∩ · · · ∩ Uαp )
in such a way that if the order of elements in a sequence is permuted, ωα0 ,...,αp
changes by the change of the permutation; thus

ωα0 α1 = −ωα1 α0 , ωαα = 0, and so forth.

59
We have two differentials on the double complex, the exterior derivative
p p
d : Č (U, Ωq ) → Č (U, Ωq+1 ) defined by (dω)α0 ···αp = dωα0 ···αp ,

and the Čech differential

p p+1
δ : Č (U, Ωq ) → Č (U, Ωq )

defined by
p+1
X
(δω)α0 ···αp+1 = (−1)i ωα0 ···α̂i ···αp+1 ,
i=0

the forms on the right being restricted to the intersection.

The first differential is exact except when q = 0 by the Poincaré Lemma,
while in the case q = 0 we find that
p p p
[Kernel of d : Č (U, Ω0 ) → Č (U, Ω1 )] = Č (U; R),

the space of Čech cocycles for the covering U on M. The Čech cohomology of
the cover U is by definition the cohomology of the cochain complex
p−1 p p+1
· · · → Č (U; R) → Č (U; R) → Č (U; R) → · · · (1.25)
p
and is denoted by Ȟ (M; R).
The second differential is exact except when p = 0 and it is at this point
that the C 2 partition of unity {ψα : α ∈ A} subordinate to U is used. Indeed,
p
given a δ-cocycle ω ∈ Č (U, Ωq ), we set
X p−1
τα0 ···αp−1 = ψα ωαα0 ···αp−1 ∈ Č (U, Ωq ),
α

noting that since ψα is C 2 we stay in the class of C 1 forms with C 1 exterior

derivatives. Then
X
(δτ )α0 ···αp = (−1)i ψα ωαα0 ···α̂i ···αp
i,α

and it follows from the fact that δω = 0 that

p
X
(−1)i ωαα0 ···α̂i ···αp = ωα0 ···αp .
i=1

Hence !
X
(δτ )α0 ···αp = ψα ωα0 ···αp = ωα0 ···αp ,
α

establishing exactness. When p = 0, we find that

0 1
Kernel of δ : Č (U, Ωq ) → Č (U, Ωq ) = Ωq (M),

60
the space of smooth q-forms on M.
We can summarize the previous discussion by stating that the rows and
columns in the following commutative diagram are exact:

· · ·
↑ ↑ ↑
0 1 2
0 → Ω2 (M) → Č (U, Ω2 ) → Č (U, Ω2 ) → Č (U, Ω2 ) → · · ·
↑ ↑ ↑
0 1 2
0 → Ω (M) → Č (U, Ω ) → Č (U, Ω ) → Č (U, Ω1 ) → · · ·
1 1 1

↑ ↑ ↑
0 1 2
0 → Ω0 (M) → Č (U, Ω0 ) → Č (U, Ω0 ) → Č (U, Ω0 ) → · · ·
↑ ↑ ↑
0 1 2
Č (U, R) Č (U, R) Č (U, R)
↑ ↑ ↑
0 0 0

The remainder of the proof uses this diagram. Given a de Rham cohomol-
p
ogy class [ω] ∈ HdR (M; R) with p-form representative ω we construct a cor-
p
responding cohomology class s([ω]) in the Čech cohomology Ȟ (M; R) as fol-
0
lows: The differential form defines an element ω 0p ∈ Č (U, Ωp ) by simply re-
stricting ω to the sets in the cover. It is readily checked that ω 0p is closed
with respect to the total differential D = δ + (−1)p d on the double com-
∗
plex K ∗,∗ = Č (U, Ω∗ ). Using the Poincaré Lemma, we construct an element
0
ω 0,p−1 ∈ Č (U, Ωp−1 ) such that dω 0,p−1 = ω 0p . Let ω 1,p−1 = δω 0,p−1 and ob-
serve that dω 1,p−1 = 0 and ω 1,p−1 is cohomologous to ω 0p with respect to D.
1
Using the Poincaré Lemma again, we construct an element ω 1,p−2 ∈ Č (U, Ωp−2 )
1,p−2 1,p−1 2,p−2 1,p−2 2,p−2
such that dω =ω . Let ω = δω and note that ω is co-
homologous to ω 0p with respect to D. Continue in this fashion until we reach
p
a D-cocycle ω p0 ∈ Č (U, Ω0 ) which is cohomologous to ω 0p . Since dω p0 = 0,
p0
each function ωα0 ···αp is constant, and thus ω p0 determines a Čech cocycle s(ω)
whose cohomology class is s([ω]).
By the usual diagram chasing, the cohomology class obtained is independent
of choices made. Moreover, reversing the zig-zag construction described in the
preceding paragraph yields an inverse to s. This finishes our sketch of the proof
of de Rham’s theorem; for more details, one can consult [10], Chapter 2.
Remark 1.13.5. The proof shows that the cohomologies of the two cochain
complexes (1.23) and (1.25) are isomorphic. It follows that the cohomology of
the cochain complex (1.25) is independent of the choice of good cover. On the
other hand, in the case where M has C ∞ partitions of unity the argument can
be repeated with C ∞ differential forms to show that the de Rham cohomology
is the same whether calculated with C ∞ forms of C 1 forms with C 1 exterior
derivatives.
Remark 1.13.6. This remark assumes some familiarity with singular cohomol-

61
ogy, as treated in Chapter 3 of [35]. One could replace the double complex the
occurs in the proof by
Ksp,q = Csp (U, Ωq ),
which is defined to be the space of functions ω which assign to each distinct
ordered (p + 1)-tuple (α0 , . . . , αp ) of indices such that Uα0 ∩ · · · ∩ Uαp 6= 0 an
element sα0 ···αp in the space of singular cochains within Uα0 ∩ · · · ∩ Uαp with
coefficients in R. The above proof can then be modified to give an isomorphism
from singular cohomology to the Čech cohomology of M. The argument can
also be modified so that it applies to relative cohomology. Thus readers familiar
with standard cohomology theory can rest assured that de Rham cohomology
gives exactly the same results as the cohomology they have studied in algebraic
topology courses.
Remark 1.13.7. Suppose we take an arbitrary cover of M , not necessarily a
good cover. Then the rows in the above diagram are still exact, even though
the columns are not. For example, we can take two open sets U and V such
that M = U ∪ V . Then the above diagram collapses to a short exact sequence
of de Rham complexes

0 → Ω∗ (M ) → Ω∗ (U ) ⊕ Ω∗ (V ) → Ω∗ (U ∩ V ) → 0.

By the “snake lemma” from algebraic topology, we get a long exact sequence

k k k k
· · · → HdR (M ; R) → HdR (U ; R) ⊕ HdR (V ; R) → HdR (U ∩ V ; R)
k+1 k+1 k+1
→ HdR (M ; R) → HdR (U ; R) ⊕ HdR (V ; R) → · · ·

which is called the Mayer-Vietoris sequence. The Mayer-Vietoris sequence,

together with the Homotopy Lemma, is very helpful in computing de Rham
cohomology.

62
Chapter 2

Morse Theory of Geodesics

2.1 Geodesics
Our next goal is to explain how critical point theory on infinite-dimensional
manifolds can be used to used to produce periodic solutions to a class of im-
portant nonlinear ordinary differential equations, the equations for geodesics in
Riemannian manifolds.
The geodesic equation is a generalization of the simplest second-order lin-
ear ordinary differential equation—the equation of a particle moving with zero
acceleration in Euclidean space. This asks for a vector-valued function

γ : (a, b) −→ RN such that γ 00 (t) = 0,

and its solutions are the constant speed straight lines. The simplest way to
make this differential equation nonlinear is to consider a proper submanifold M
of RN with the induced Riemannian metric, and ask for a function

γ : (a, b) −→ M ⊂ RN such that (γ 00 (t))T = 0,

where (·)T denotes projection into the tangent space of M . In simple terms,
we are asking for the curves which are as straight as possible subject to the
constraint that they lie within M . In the terminology of differential geometry,
the tangential projection of the ordinary derivative is known as the covariant
derivative, and one often writes

Dγ 0
(γ 00 (t))T = ∇γ 0 γ 0 (t) or (γ 00 (t))T = (t),
dt
where ∇ and D are two commonly used notations for the covariant derivative.
The smooth maps γ which satisfy the equation ∇γ 0 γ 0 (t) = 0 are called the
smooth geodesics in M .
We can put the geodesic equation into the more general context of simple
mechanical systems: We let (M, h·, ·i) be a Riemannian manifold, which we can

63
assume has a proper isometric imbedding ι : M n → RN into Euclidean space;
such an imbedding is provided by the Nash imbedding theorem. In addition, we
consider a smooth function φ : M → R, to be called the potential . The triple
(M, h·, ·i, φ) is called a simple mechanical system. For a simple mechanical
system, “Newton’s equation of motion” is

∇γ 0 γ 0 (t) = −(grad φ)(γ 0 (t)).

The left-hand side can be interpreted as the acceleration of a moving particle

of unit mass, while the right-hand side is the force (per unit mass) produced by
the potential φ.
As we mentioned before, the idea behind the calculus of variations is to
regard solutions to ordinary differential equations—such as the geodesic equa-
tion or Newton’s equation of motion—as critical points of functions defined on
infinite-dimensional manifolds. In the case of geodesics, the infinite-dimensional
manifold is

L21 ([0, 1], M ) = {γ ∈ L21 ([0, 1], RN ) : γ(t) ∈ M , for all t ∈ [0, 1] },

where M is a proper submanifold of RN , or one of the many useful subspaces

of L21 ([0, 1], M ). These include the free loop space

L21 (S 1 , M ) = {γ ∈ L21 ([0, 1], M ) : γ(0) = γ(1)},

the space of paths from p to q,

Ω(M, p, q) = {γ ∈ L21 ([0, 1], M ) : γ(0) = p, γ(1) = q},

where p and q are points of M , and the space of paths from S1 to S2 ,

Ω(M, S0 , S1 ) = {γ ∈ L21 ([0, 1], M ) : γ(0) ∈ S0 , γ(1) ∈ S1 },

when S0 and S1 are compact imbedded submanifolds of M .

The first step towards formulating the equations of geodesics within critical
point theory is to define the Euclidean action

1 1 0
Z
JRN : L21 ([0, 1], RN ) → R by JRN (γ) = γ (t) · γ 0 (t)dt,
2 0
the dot denoting the Euclidean dot product. The map JRN is clearly smooth,
being the restriction of a continuous bilinear map
Z
1
(γ, λ) 7→ γ 0 (t) · λ0 (t)dt
2 S1
to the diagonal.
Recall from §1.4 that the isometric imbedding ι : M → RN induces by
composition a map

ωι : L21 ([0, 1], M ) → L21 ([0, 1], RN )

64
which is also a smooth imbedding. Moreover, the Hilbert space structure on
the spaces L21 ([0, 1], RN ) induces Riemannian metrics on L21 ([0, 1], M ) and its
subspaces L21 (S 1 , M ) and Ω(M, p, q). We define

JM : L21 ([0, 1], M ) → R by JM = JRN ◦ ωι ,

a map which is clearly smooth, being the composition of smooth maps. In

addition, if φ : M → R is a smooth function, the map
Z 1
γ ∈ L21 ([0, 1], M ) 7→ φ(γ(t))dt
0

is also smooth, since it is the composition of

ωφ : L21 ([0, 1], M ) → L21 ([0, 1], R)

with integration, integration being a continuous linear map. Finally, we define

the action for the simple mechanical system (M, h·, ·i, φ) to be the function
Z 1 
2 1 0 0
JM,φ : L1 ([0, 1], M ) → R, JM,φ (γ) = hγ (t), γ (t)i − φ(γ(t)) dt.
0 2
By restriction, we also get smooth maps

JM,φ : L21 (S 1 , M ) → R, JM,φ : Ω(M, p, q) → R.

With these preparations out of the way, we can now state Hamilton’s prin-
ciple of least action: the motion of a simple mechanical system is described by
a critical point for the action function JM,φ on L21 (S 1 , M ) or Ω(M, p, q).
Focusing first on the case of free loops, the case needed for studying periodic
motion, we are led to ask: what is the differential

(dJM,φ )γ : Tγ (L21 (S 1 , M )) −→ R?

We will assume that γ is smooth and that V ∈ Tγ (L21 (S 1 , M )) lies in the space
of C ∞ sections of γ ∗ T M .
To calculate the differential, we suppose that α : S 1 × (−, ) → M is a
smooth one parameter family of curves with α(t, 0) = γ(t) and D2 α(t, 0) = V (t),
where D2 denote the partial derivative with respect to the second slot. Then

ωα : L21 (S 1 , S 1 × (−, )) −→ L21 (S 1 , M )

is smooth. Let

µ̄ : (−, ) → L21 (S 1 , S 1 × (−, )) by µ̄(τ )(t) = (t, τ ),

for t ∈ S 1 . Clearly, µ̄ is smooth and hence ᾱ = ωα ◦ µ̄ is a smooth curve in

L21 (S 1 , M ) with ᾱ0 (0) = X. A straightforward calculation now shows that
 Z
d
[hγ 0 (t), ∇γ 0 V i − dφ(γ(t))(V (t))]dt,

dJM,φ (γ)(V ) = (JM,φ ◦ ᾱ) =
dt τ =0 S 1

65
where ∇γ 0 V is the directional derivative of V in the direction of γ 0 projected
into the tangent space, otherwise known as the covariant derivative of V . Since
γ is assumed to be smooth, we can integrate by parts to obtain
Z
dJM,φ (γ)(V ) = − h∇γ 0 γ 0 (t) + (grad φ)(γ 0 (t)), V (t)idt, (2.1)
S1

the boundary terms cancelling by periodicity. Since X can be an arbitrary

smooth vector field along γ, a critical point for JM,φ must be a periodic solution
to Newton’s equation of motion,

∇γ 0 γ 0 (t) = −(grad φ)(γ 0 (t)). (2.2)

Thus Hamilton’s principle implies that the motion of a simple mechanical system
should be represented by solutions to Newton’s equation of motion. In the case
where φ = 0, a critical point for the action is simply a smooth closed geodesic.
In a quite similar fashion, one finds that critical points to JM,φ : Ω(M ; p, q) →
R are solutions γ to (2.2) such that γ(0) = p and γ(1) = q. On the other hand,
if one calculates the derivative on the larger space Ω(M, S0 , S1 ), the integration
by parts gives additional boundary terms and we must replace (2.1) by
Z
dJM,φ (γ)(V ) = − h∇γ 0 γ 0 (t) + (grad φ)(γ 0 (t)), V (t)idt
S1
+ hγ 0 (1), V (1)i − hγ 0 (0), V (0)i.

This more complicated formula must be used when considering critical points
for dJM,φ : Ω(M, S0 , S1 ) → R, and in this case V (0) and V (1) are constrained
to be tangent to S0 and S1 . The first variation formula implies that critical
points are solutions to (2.2) which are perpendicular to S0 and S1 .

2.2 Condition C for the action

In order to apply the method of steepest descent to calculus of the variations
problems described in the preceding section, we need the topology on the space
of maps to satisfy two conditions:
1. It must be strong enough so that JM,φ is continuous.
2. It must be weak enough so that suitable sequences {γi } (such as sequences
such that JM,φ (γi ) tends to a minimum or minimax value subject to some
constraint) will converge in the topology.
If we chose the topology to be too strong, this would make it difficult to es-
tablish convergence of {γi }. The two conflicting requirements single out L21 as
the appropriate space to use when studying critical points of the action JM,φ .
indeed, we next show that when M is compact, the action functions

JM , JM,φ : L21 (S 1 , M ) −→ R

66
satisfy condition C, thereby making available the minimax principle from §1.11.
Moreover, we will show that the critical points of J are actually C ∞ curves.
Theorem 2.2.1. If (M, h·, ·i) is a compact Riemannian manifold, the function
JM : L21 (S 1 , M ) → R, defined by
Z
1
JM (γ) = hγ 0 (t), γ 0 (t)idt,
2 S1

satisfies Condition C: if {γi } is a sequence in L21 (S 1 , M ) such that

JM (γi ) is bounded and kdJM (γi )k → 0, (2.3)

then it possesses a subsequence which converges to a critical point for JM .

Proof: We start by recalling the proof of the Sobolev Lemma from §1.4. Suppose
that {γi } is a sequence of elements of L21 (S 1 , M ) satisfying (2.3). We can regard
each γi as an element of the space L21 (S 1 , RN ). Then for t1 < t2 ,

t2 t2 1/2
√
Z Z
|γi (t1 ) − γi (t2 )| ≤ |γi0 (t)|dt ≤ t2 − t1 |γi0 (t)|2 dt
t1 t1
√ p
≤ t2 − t1 2JRN (γi ).

Since JM (γi ) is bounded, we see that {γi } is equicontinuous. Since γi takes

values in the compact submanifold M ⊂ RN , {γi } is also uniformly bounded.
It therefore follows from Arzela’s theorem or Ascoli’s theorem ([66], page 179)
that a subsequence of {γi } will converge uniformly to a continuous map γ∞ :
S 1 → M . For simplicity, we continue to denote the subsequence by {γi }. To
finish the proof, we need to show that

kdJM (γi )k → 0 ⇒ {γi } is a Cauchy sequence in L21 (S 1 , M ).

To understand |dJM (γ)k we need to be able to compare an element of

Tγ L21 (S 1 , RN ) with its projection into Tγ L21 (S 1 , M ). Recalling that M is a
submanifold of RN with inclusion i : M → RN , we let P denote the vector
bundle map
P : i∗ T (RN ) −→ T M
which projects onto the tangential component. By the ω-Lemma, we have a
smooth map
ωP : L21 (S 1 , i∗ T (RN )) −→ L21 (S 1 , T M )
which is also a vector bundle map.
Lemma 2.2.2. If h·, ·i denotes the Riemannian metric on either L21 (S 1 , RN ) or
L21 (S 1 , M ) and X ∈ L21 (S 1 , i∗ T (RN )) with ωπ (X) = γ, then

hωP (X), ωP (X)i ≤ [1 + CJM (γ)]hX, Xi,

where C is some constant which depends on M but is independent of γ.

67
Proof: We can regard P as a section of the bundle i∗ (End(RN )) over M and it
possesses a differential dP which is a section of T ∗ M ⊗ [i∗ (End(RN ))]. Let

C1 = sup{kdP (v)k : v is a unit-length element of T M },

a finite constant since M is compact. Then it follows from the Leibniz rule that

(ωP (X))(t) = Pγ(t) X(t) ⇒ (ωP (X))0 (t) = dP (γ 0 (t))(X(t)) + Pγ(t) X 0 (t).

Recall that
Z
k(ωP (X))0 (t)k2 + k(ωP (X))(t)k2 dt,
 
hωP (X), ωP (X)i =
S1

where k · k denotes the norm in RN , and hence

Z
hωP (X), ωP (X)i = [kdP (γ 0 (t))(X(t)) + Pγ(t) X 0 (t)k2 + kPγ(t) (X)(t)k2 ]dt
S1
Z
≤ [kdP (γ 0 (t))(X(t))k2 + kPγ(t) X 0 (t)k2 + kPγ(t) (X))(t)k2 ]dt
S 1
Z
≤ [C1 kγ 0 (t)k2 k(X(t)k2 + kX 0 (t)k2 + kX(t)k2 ]dt.
S1

Thus
Z
hωP (X), ωP (X)i ≤ C1 sup{k(X(t)k : t ∈ S } 2 1
kγ 0 (t)k2 dt + hX, Xi
S1
≤ (CJM (γ) + 1)hX, Xi,

for some positive constant C. This finishes the proof of the lemma.
Lemma 2.2.3. If X ∈ L21 (S 1 , i∗ T (RN )) and ωπ (X) = γ, where

ωπ : L21 (S 1 , i∗ T (RN )) → L21 (S 1 , M )

is the projection induced by π : i∗ T (RN ) → M , then

Z
dJM (γ)(ωP (X)) = [γ 0 (t) · X 0 (t) + α(γ 0 (t), γ 0 (t)) · X(t)]dt, (2.4)
S1

where α : T M × T M → N M is the second fundamental form of M in RN and

the dot products are taken in the ambient Euclidean space RN .
Proof: Note that it suffices to establish (2.4) in the case where γ and X are
smooth because both sides are continuous in the L21 -topology. It follows from
(2.1) with φ = 0 that
Z
dJM (γ)(ωP (X)) = − γ 00 (t) · Pγ(t) (X(t))dt.
S1

68
But since α(γ 0 (t), γ 0 (t)) is the normal component of γ 00 (t),

γ 00 (t) · X(t) = Pγ(t) (γ 00 (t)) · X(t) + α(γ 0 (t), γ 0 (t)) · X(t)

= γ 00 (t) · Pγ(t) (X(t)) + α(γ 0 (t), γ 0 (t)) · X(t),

and hence
Z
dJM (γ)(ωP (X)) = [−γ 00 (t) · X(t) + α(γ 0 (t), γ 0 (t)) · X(t)]dt.
S1

An integration by parts yields the claim.

Returning to the proof of the theorem, we let {γi } be a sequence satisfying (2.3)
which converges uniformly to a continuous map γ∞ : S 1 → M . Since JM (γi ) is
bounded, hγi , γi i = kγi k2 is bounded, and hence

kγi − γj k2 ≤ (kγi k + kγj k)2

is also bounded as i, j → ∞. Lemma 2.2.2 implies that k(ωP )γi (γi − γj )k is

bounded as well. Since kdJM (γi )k → 0, for every  > 0 there exists N ∈ N such
that
|dJM (γi )((ωP )γi (γi − γj )) − dJM (γj )((ωP )γj (γi − γj ))| <  (2.5)
for i, j > N . Here (ωP )γi (γi − γj ) lies in the tangent space to L21 (S 1 , M ) at γi .
It follows from (2.5) and the explicit formula (2.4) for dJM that for every  > 0
there exists N ∈ N such that
Z
[γi0 (t) · (γi0 (t) − γj0 (t)) + α(γi0 (t), γi0 (t)) · (γi (t) − γj (t))]dt
S 1
Z
− [γj0 (t) · (γi0 (t) − γj0 (t)) + α(γj0 (t), γj0 (t)) · (γi (t) − γj (t))]dt < ,
S1

for i, j > N . But

|α(γi0 (t), γi0 (t))| ≤ (constant)|γi0 (t)|2 ,
the constant is a bound for the norm of the second fundamental form of M and
hence depends only on the submanifold M . Since |γi (t) − γj (t)| < 1/n for i and
j sufficiently large,
Z 
0 0  ≤ (constant)JM (γi ) 1 → 0
 
 α(γ i (t), γi (t)) · (γ i (t) − γ j (t))dt
 1
S
 n
as i, j → ∞, a similar implication holding when i is replaced by j. Hence for
every  > 0 there exists N ∈ N such that
Z 
0 0 0 0
 

 (γi (t) − γj (t)) · (γi (t) − γj (t))dt < ,
S1

for i, j > N . This, together with the C 0 convergence of {γi } implies that {γi }
is a Cauchy sequence in L21 (S 1 , RN ). Since L21 (S 1 , RN ) is complete, {γi } has a

69
limit in L21 (S 1 , RN ), which must of course be γ∞ . Thus γi → γ∞ ∈ L21 (S 1 , M ),
and by continuity of dJM , γ∞ must be a critical point for JM . This concludes
the proof of Theorem 2.2.1.
Theorem 2.2.4. Suppose that (M, h·, ·i) is a compact Riemannian manifold,
φ : M → R is a smooth function and ψ : S 1 → RN is an L21 map. The real-valued
functions JM,φ and JM,ψ on L21 (S 1 , M ), defined by
Z
1
JM,φ (γ) = [hγ 0 (t), γ 0 (t)i − φ(γ(t))] dt and
2 S1
Z
1
JM,ψ (γ) = [hγ 0 (t), γ 0 (t)i + γ(t) · ψ(t)] dt,
2 S1
satisfy Condition C.
Sketch of proof: Suppose that {γi } is a sequence in L21 (S 1 , M ) such that JM,φ (γi )
is bounded and kdJM,φ (γi )k → 0. Then JM (γi ) is also bounded and just as
before, {γi } is uniformly bounded and equicontinuous and thus by the Arzela-
Ascoli theorem, possesses a subsequence which converges uniformly to a contin-
uous map γ∞ .
We can now mimic the preceding proof up to equation (2.5). At this point,
we need to account for an extra term in dJM,φ , namely
Z
− dφ(γi (t))(Pγi (t) (γi (t) − γj (t)))dt.
S1

However, this term goes to zero, since {γi (t)} is Cauchy in C 0 . Similarly, there
is an extra term in JM,ψ which goes to zero. With these minor changes, the
argument proceeds to the desired conclusion exactly as before.
For the statement of the next theorem, we suppose that M is a complete Rie-
mannian manifold which has a proper isometric immersion into some Euclidean
space RN .
Theorem 2.2.5. Suppose that (M, h·, ·i) is a compact Riemannian manifold
and φ : M → R is a smooth function. The real-valued functions

JM,φ : Ω(M, p, q) → R and JM,φ : Ω(M, S1 , S2 ) → R,

defined by Z 1
1
JM,φ (γ) = [hγ 0 (t), γ 0 (t)i − φ(γ(t))] dt,
2 0
satisfy condition C.
Proof: A straightforward modification of preceding arguments.
Theorem 2.2.6. if the path γ within L21 (S 1 , M ), Ω(M, p, q) or Ω(M, S1 , S2 ) is
a critical point for JM or JM,φ , then γ is C ∞ . Moreover, if ψ ∈ L2k (S 1 , RN ),
then any critical point for JM,ψ lies in L2k+2 (S 1 , M ).

70
Proof: We prove only the free loop space cases. If γ is a critical point for JM ,
it follows from (2.4) that
Z Z
0 0
γ (t) · X (t)dt = − α(γ 0 (t), γ 0 (t)) · X(t)dt,
S1 S1

for all X ∈ Tγ L21 (S 1 , RN ). Thus in the sense of distributions,

γ 00 (t) = α(γ 0 (t), γ 0 (t)). (2.6)

Note that
γ ∈ L21 ⇒ γ 0 ∈ L2 ⇒ α(γ 0 (t), γ 0 (t)) ∈ L1 .
Thus it follows from (2.6) and standard theorems of analysis that γ is C 1 . Now
we use the technique of “elliptic bootstrapping”:

γ ∈ C 1 ⇒ RHS of (2.6) is C 1 ⇒ γ ∈ C 2 ⇒ RHS of (2.6) is C 2 ⇒ · · · .

By induction, we see that γ is C ∞ and Theorem 2 is established for JM .

The proofs for JM,φ and JM,ψ are the same except that (2.6) is replaced by

γ 00 (t) = α(γ 0 (t), γ 0 (t)) − (grad φ)(γ 0 (t))

or γ 00 (t) = α(γ 0 (t), γ 0 (t)) + (ψ(t))T ,

where (·)T is the tangential component.

2.3 Existence of smooth closed geodesics

In studying closed geodesics on a Riemannian manifold (M, h·, ·i), we let M =
L21 (S 1 , M ), an infinite-dimensional Hilbert manifold with a complete Rieman-
nian metric, and define the action
Z
1
J : M −→ R by J(γ) = hγ 0 (t), γ 0 (t)idt,
2 S1
a function which satisfies condition C by Theorem 2.2.1. We have seen that
M = L21 (S 1 , M ) is homotopy equivalent to C 0 (S 1 , M ), the free loop space
studied by topologists.
Theorem 2.3.1. Let (M, h·, ·i) be a compact connected Riemannian manifold
and let [S 1 , M ] denote the set of free homotopy classes of continuous maps from
S 1 to M . If α ∈ [S 1 , M ] and

L21,α (S 1 , M ) = {γ ∈ L21 (S 1 , M ) : the free homotopy class of γ is α},

then J assumes its minimum on L21,α (S 1 , M ). If α 6= 0, this minimum is achieved

at a nonconstant smooth closed geodesic.
Proof: This is a direct consequence of Theorem 2.2.1 and Corollary 1.12.3.

71
The preceding theorem shows a nonsimply connected compact manifold always
possesses a smooth closed geodesic. To treat the simply connected case, we need
the following theorem due to Lusternik and Fet:
Theorem 2.3.2. If (M, h·, ·i) is a compact simply connected Riemannian man-
ifold, M contains a nonconstant smooth closed geodesic.
Before proving this theorem, we recall some concepts we need from homotopy
theory. A fibration is a continuous map f : E → B which has the homotopy
lifting property; this means that if the continuous map g̃ : Y → E has the
property that its projection into the base f ◦ g̃ : Y → M can be extended to a
homotopy

H : Y × [0, 1] → B with H(y, 0) = f ◦ g̃(y), for y ∈ I n ,

then this homotopy H can be lifted to

H̃ : Y × [0, 1] → E such that H̃(y, 0) = g̃(y) and f ◦ H̃ = H.

The key facts about fibrations are that the fibers Ep = f −1 (p), for p ∈ B, are
homotopy equivalent to each other, and the map f induces a long exact sequence
of homotopy groups

· · · → πk (Ep ) → πk (E) → πk (B) → πk−1 (Ep ) → · · · .

The reader can refer to [10], §§16 and 17 or [35], Theorem 4.41 for proofs of
these and related facts. In fact, the only thing needed for the exact sequence
is that f be a weak fibration, which means that it has the homotopy lifting
property for the case where Y = [0, 1]n , the n-cube, for all choices of n.
A key example concerns the path space

P = { continuous maps γ : [0, 1] → M such that γ(0) = p },

where p is some choice of base point in M . In this case, the map

π : P → M, π(γ) = γ(1)

is a fibration. Indeed, given

g̃ : Y → P and H : Y × [0, 1] → M with H(y, 0) = π ◦ g̃(y),

we can define the lift H̃ : Y × [0, 1] → P by

(  
t
g̃(y) 1−(s/2) , for 0 ≤ t ≤ 1 − (s/2),
H̃(y, s)(t) =
H(y, 2t − 2 + s), for 1 − (s/2) ≤ t ≤ 1.

One readily checks that H̃ is continuous and has the desired properties. Indeed,
when t = 1 − (s/2),
 
t
g̃(y) = g̃(y)(1) = H(y, 0) = H(y, 2t − 2 + s),
1 − (s/2)

72
so the two pieces of the function fit together continuously, while when we set
t = 1, we find that H̃(y, s)(1) = H(y, 2 − 2 + s) = H(y, s), so H̃ is indeed a lift
H.
The fiber over the base point p of this fibration is

Ωp = { continuous maps γ : [0, 1] → M such that γ(0) = p = γ(1) }

and is known as the pointed loop space. Its homotopy groups can be computed
by the long exact sequence of the fibration,

· · · → πk (Ωp ) → πk (P) → πk (M ) → πk−1 (Ωp ) → · · · .

Since P is contractible via the homotopy

H : P × [0, 1] → P, where H(γ, s)(t) = γ((1 − s)t),

we see that πk (P) = 0 for all k and this long exact sequence collapses to yield

πk (Ωp ) ∼
= πk+1 (M ).

A second example is the free loop space

C 0 (S 1 , M ) = { continuous maps γ : [0, 1] → M such that γ(0) = γ(1) },

which is the total space of a fibration

ev : C 0 (S 1 , M ) → M defined by ev(γ) = γ(0).

The fiber over p in this case is Ωp once again, so we obtain a long exact sequence

· · · → πk (Ωp ) → πk (C 0 (S 1 , M )) → πk (M ) → πk−1 (Ωp ) → · · · .

In this case, however, ev∗ : πk (C 0 (S 1 , M )) → πk (M ) possesses a right inverse

i∗ : πk (M ) → πk (C 0 (S 1 , M )) induced by the map i : M → C 0 (S 1 , M )

which takes the point p ∈ M to the constant loop located at p. Hence the exact
sequence for ev splits, and we conclude that

πk (C 0 (S 1 , M )) ∼
= πk (M ) ⊕ πk (Ωp ) ∼
= πk (M ) ⊕ πk+1 (M ).

Thus we see that the homotopy groups of the free loop space C 0 (S 1 , M ) are
quite easily determined from the homotopy groups of M .
Proof of Theorem 2.3.2: Since π1 (M ) = 0, M is orientable and hence Hn (M ; Z) 6=
0. Let q be the smallest positive integer such that Hq (M, Z) 6= 0. It follows
from the Hurewicz isomorphism theorem that

πi (M ) = 0, for 0 < i < q, and πq (M ) ∼

= Hq (M, Z) 6= 0.

73
It follows from Theorem 1.5.1 that M = L21 (S 1 , M ) is homotopy equivalent to
C 0 (S 1 , M ). Hence

πk (M) ∼
= πk (M ) ⊕ πk (Mp ) ∼
= πk (M ) ⊕ πk+1 (M ).

In particular,
πq−1 (M) ∼
= πq (M ) 6= 0.
Since M is simply connected, q ≥ 2 and πq−1 (M) ∼ = πq (M ) is abelian.
Moreover, we can identify πq−1 (M) with [S q−1 , M], the space of free homotopy
classes of maps from S q−1 to M. Choose a nonzero element α ∈ [S q−1 , M]. Let

F = {g(S q−1 ) such that g : S q−1 → M is a continuous map in [α]}.

Clearly, F is an ambient isotopy invariant family of sets. Hence Minimax(J, F)

is a critical value for J. We need only verify that Minimax(J, F) ≥  for some
 > 0. Note √ that by the Cauchy-Schwarz inequality, J −1 ([0, ]) consists of curves
of length ≤ 2.
Recall that M is isometrically imbedded in an ambient Euclidean space RN .
If p ∈ M , let Np M denote the normal space to M in RN and let νM be the
disjoint union of all the normal spaces Np M , for p ∈ M , the total space of a
smooth vector bundle over M , called the normal bundle. Let νM (δ) denote the
union of all the balls of radius δ > 0 in N − pM , for p ∈ M , and let let M (δ)
denote the open δ-neighborhood of M in RN . If δ > 0 is sufficiently small, one
can prove that the exponential map

exp : (νM )(δ) −→ M (δ),

is a diffeomorphism; this is the content of the tubular neighborhood theorem

from differential topology ([36], Chapter 4, §5). Thus if δ is sufficiently small,
M is a strong deformation retract of M (δ) and hence M is homotopy equivalent
to M (δ).
For  > 0 sufficiently
√ small, any closed curve γ on M such that J(γ) <  and
hence of length < 2 (by the Cauchy-Schwarz inequality) can be contracted
to the point γ(0) in RN without leaving M (δ). Thus if g(S q−1 ) ⊂ J −1 ([0, ]),
then g is homotopic to a smooth map

g̃ : S q−1 → M0 , where M0 = {γ ∈ M : γ is constant }.

But πq−1 (M0 ) = πq−1 (M ) = 0, and hence g is homotopic to a constant, contra-

dicting the definition of F. Hence Minimax(J, F) ≥ , and M must possess a
nonconstant smooth closed geodesic.
Remark 2.3.3. The technique we have used to prove the Lusternik-Fet Theo-
rem, based upon the minimax principle, is often called Lusternik-Schnirelmann
theory.

74
2.4 Second variation
A defect in the minimax principle is that quite different topological constraints
can in fact lead to the same minimax critical points. We need a more refined
theory that can analyze the contributions to the topology made by each individ-
ual critical point. The contributions of “nondegenerate” critical points are the
easiest to analyze. In this section, we take the first step towards understanding
nondegenerate critical points by investigating the structure of the Hessian at a
critical point.
Let M be a Banach manifold and let f : M → R be a smooth map. If
p ∈ M is a critical point for f , then the Hessian of f at p is the symmetric
bilinear map d2 f (p) : Tp M × Tp M → R defined in terms of a chart (Uα , φα ) by

d2 f (p)([α, p, v], [α, p, w]) = D2 (f ◦ φ−1

α )(φα (p))(v, w).

It is straightforward to show that this is independent of choice of chart. Indeed,

if α : (−, ) → M is a smooth curve with α(0) = p and α0 (0) = V , then a short
calculation shows that
d2

2

d f (p)(V, V ) = 2 (f ◦ α) ,
dt t=0

an expression which is clearly independent of choice of chart. Of course, once one

knows d2 f (p)(V, V ), one can determine d2 f (p)(V, W ) for arbitrary V, W ∈ Tp M
by the polarization identity:
1 2
d2 f (p)(V, W ) = [d f (p)(V + W, V + W ) − d2 f (p)(V − W, V − W )].
4

Definition. The Morse index of a critical point p for f is the maximal dimen-
sion of a linear subspace of Tp M on which d2 f (p) restricts to a negative definite
symmetric bilinear form. The nullity of the critical point p is

dim{ V ∈ Tp M such that d2 f (p)(V, W ) = 0 for all W ∈ Tp M }.

We say that the critical point is stable if its Morse index is zero. In some sense,
the index measures the extent to which a critical point fails to be stable.
We would like to calculate the Hessian at critical points for the simple me-
chanical systems we described in §2.1, where we take M to be a space of L21
maps. In the case where M = L21 (S 1 , M ), it is proven in elementary differential
geometry texts that the Hessian of the action J is given by the second variation
formula or index formula:
Proposition 2.4.1. If (M, h·, ·i) is a compact Riemannian manifold and J :
L21 (S 1 , M ) → R is the usual action, then
Z
2
d J(γ)(V, W ) = [h∇γ 0 (t) V, ∇γ 0 (t) W i − hR(V, γ 0 (t))γ 0 (t), W i]dt, (2.7)
S1

75
for V, W ∈ Tγ M.
Let us review how to establish (2.7). We consider a smooth variation of γ which
has its support within a given coordinate chart (U, t) on S 1 . Recall that such a
variation is a smooth family of maps u 7→ γu in L21 (S 1 , M ) and let

∂α
α(t, u) = γu (t), V (t) = (t) ∈ Tγ(t) M.
∂u
Differentiating twice, we obtain

d2
 Z   
2
 ∂ ∂α D ∂α 
d (J)(γ)(V, V ) = (J(γ u ))  = , dt
du2 
u=0 S 1 ∂u ∂t ∂u ∂t 
u=0
∂α D2 ∂α
Z     
D ∂α D ∂α 
= , + , 2 dt ,
S1 ∂u ∂t ∂u ∂t ∂t ∂u ∂t u=0

where D denotes the covariant derivative of the Levi-Civita connection on the

ambient Riemannian manifold M . Thus
Z    
DV DV ∂γ D DV
d2 (J)(γ)(V, V ) = , + , dt
S1 ∂t ∂t ∂t ∂u ∂t

Using the definition of the Riemann-Christoffel curvature tensor R, we obtain

Z    
2 DV DV 0 0 0 D DV
d Jγ (V, V ) = , + hγ (t), R(V, γ (t))V i + γ (t), dt.
S1 ∂t ∂t ∂t ∂u

An integration by parts and use of the fact that γ satisfies the Euler-Lagrange
equation eliminates the last term, thereby yielding (2.7).
We can write the formula for second variation as
Z
2
d J(γ)(V, W ) = hL(V ), W idt,
Σ

where Lγ is the Jacobi operator, defined by

 
D DV
L(V ) = − ◦ + R(V, γ 0 (t))γ 0 (t) . (2.8)
∂t ∂t

An element V ∈ Tγ M is called a Jacobi field along γ if Lγ (V ) = 0. Note that

if γ ∈ C ∞ (S 1 , M ), then the Jacobi operator yields a continuous linear operator

L : L2k (γ ∗ T M )) → L2k−2 (γ ∗ T M )),

for all k ≥ 1. Symmetry of d2 J implies that the Jacobi operator satisfies

Z Z
hL(V ), W idt = hV, L(W )idt, for all V, W ∈ L2k+2 (γ ∗ T M )),
S1 S1

76
and hence we say that it is formally self-adjoint. In studying the Hessian of J
it is also useful to consider, for λ ∈ C, the closely related operator

Lλ = L − λι : L2k (γ ∗ T M )) → L2k−2 (γ ∗ T M )), (2.9)

where ι : L2k (γ ∗ T M )) → L2k−2 (γ ∗ T M )) is the inclusion. We say that λ is an

eigenvalue for L if the kernel of Lλ is nonzero.
There is a similar second variation formulae for J : M → R when M =
Ω(M ; p, q). In this case, we obtain
Z 1   
DV DV
d2 J(γ)(V, V ) = , + hγ 0 (t), R(V, γ 0 (t))V i dt,
0 ∂t ∂t

except that now V is an element of

Tγ Ω(M ; p, q) = {V ∈ L21 ([0, 1], T M ) : ωπ (V ) = γ, V (0) = 0 = V (1)}.

In this case an integration by parts shows that

Z 1
d2 J(γ)(V, W ) = hL(V ), W idt, for V, W ∈ Tγ Ω(M ; p, q), (2.10)
0

where L is the differential operator defined once again by (2.8). If we set

L21,0 (γ ∗ T M ) = {X ∈ L2k (γ ∗ T M )) : X(0) = X(1) = 0},

then the Jacobi operator defines a continuous linear map

L : L21,0 (γ ∗ T M )) ∩ L2k (γ ∗ T M )) → L2k−2 (γ ∗ T M )),

for all k ≥ 1. Once again, we can define, for λ ∈ C, the closely related operator

L : L21,0 (γ ∗ T M )) ∩ L2k (γ ∗ T M )) → L2k−2 (γ ∗ T M )), (2.11)

where ι is the inclusion.

The Jacobi operators L are special cases of formally self-adjoint elliptic differ-
ential operators and there is a well-developed classical theory for the eigenvalue
problem for such operators. In order to describe this theory, we need the notion
of Fredholm operator:
Definition. A linear operator L : E → F between Banach spaces is said to be
Fredholm if
1. L has finite-dimensional kernel,

2. L has closed range, and

3. L has finite-dimensional cokernel, where the cokernel of L is the quotient
space F/(Range(L)).

77
The Fredholm index of a Fredholm operator L is defined by the formula

Fredholm index of L = dim(Kernel of L) − dim(Cokernel of L).

Note that a linear operator between finite-dimensional Banach spaces, say L :

Rn → Rp , is always Fredholm, and its Fredholm index is just the difference in
dimensions n − p.
Theorem 2.4.2 For every λ ∈ C and every integer k ≥ 1, the operators Lλ de-
fined by (2.9) and (2.11) are Fredholm maps of Fredholm index zero. Moreover,
1. for each λ ∈ C the eigenspace Wλ = Ker(Lλ ) is finite-dimensional.
2. all the elements of Wλ are C ∞ ,

3. if Wλ is empty, then Lλ possesses an inverse Gλ , which is called a Green’s

operator,
4. if Wλ is nonempty, that is, λ is an eigenvalue, then λ ∈ R, and
5. the eigenvalues can be arranged in a sequence

λ1 < λ2 < · · · < λi < · · · with λi → ∞,

and only finitely many of the eigenvalues are negative.

This is a special case of the basic theorem for second-order elliptic operators
proven in basic courses on linear PDE’s. It is worked out in various cases in
Chapter 5 of [79], Chapter 3 of [44] and in [19].

2.5 Morse nondegenerate critical points

If M is a Hilbert manifold with a Riemannian metric hh·, ·ii and f : M → R is a
C 2 map with a critical point p, it follows from the Riesz representation theorem
that there is a continuous map A : Tp M → Tp M such that

d2 f (p)(V, W ) = hhA(V ), W ii, for V, W ∈ Tp M, (2.12)

Definition. Suppose that M is a Hilbert manifold and that f : M → R is a

C 2 map. A critical point p for f is Morse nondegenerate if for some choice (and
hence every choice) of Riemannian metric hh·, ·ii the continuous map A : Tp M →
Tp M satisfying (2.12) is an isomorphism; otherwise, it is Morse degenerate.
Example 2.5.1. A Morse nondegenerate critical point has zero nullity, but the
converse is not true in general. Indeed, we can let M = L2 [0, 1] with the inner
product Z 1
hφ, ψi = φ(t)ψ(t)dt,
0

78
and define Z 1
f : L2 [0, 1] → R by f (φ) = tφ(t)2 dt.
0
Then φ = 0 is a critical point for f and

d2 f (0)(φ, ψ) = hA(φ), ψi, where A(φ(t)) = tφ(t).

Then A is injective but the range of A does not include any constant functions,
so it is not surjective. Thus the critical point 0 has zero nullity but is not Morse
nondegenerate.
In the variational problems we have been considering, however, the map A is
Fredholm and zero nullity does indeed imply nondegeneracy.
To be specific, let’s focus on the case where M is a complete Riemannian
manifold, properly and isometrically imbedded in RN and M = Ω(M, p, q). For
Riemannian metric on M, we choose the intrinsic Riemannian metric hh·, ·ii
defined by the formula
Z 1  
DV DW
hhV, W iiγ = hV (t), W (t)i + (t), (t) dt, (2.13)
0 ∂t ∂t

for V, W ∈ Tγ Ω(M, p, q), where D is the Levi-Civita connection on M . An

integration by parts shows that
Z 1
D D
hhV, W iiγ = hP (V ), W idt, where P = − ◦ + id,
0 dt dt

a formally self-adjoint second order elliptic partial differential operator. Since

the Hilbert space inner product is positive definite, it follows from Theorem 2.4.2
that
P : L21,0 (γ ∗ T M ) → L2−1 (γ ∗ T M )
has an inverse Green’s operator G.
Now consider the action function J : Ω(M, p, q) → R. It follows from the
second variation formula (2.10) that
Z 1
d2 J(γ)(V, W ) = hL(V ), W idt = hhG ◦ L(V ), W ii = hhA(V ), W ii,
0

where A is a Fredholm operator of Fredholm index zero. Thus if γ is a critical

point for J with zero nullity, then γ is a Morse nondegenerate critical point.
Definition. We say that f : M → R is a Morse function if all of its critical
points are Morse nondegenerate.
Theorem 2.5.2. Suppose that (M, h·, ·i) is a complete Riemannian manifold
and p ∈ M . For almost all choices of q ∈ M , the function

J : Ω(M ; p, q) −→ [0, ∞)

79
is a Morse function.
The proof follows from the fact that there are nonzero Jacobi fields along γ ∈
Ω(M ; p, q) vanishing at the endpoints if and only if q is a critical value for expp .
Moreover, it follows from Sard’s Theorem (which is proven in [36], Chapter 3)
that the set of regular values for expp form a set of full measure in M . Thus for
almost all choices of q, all geodesics in Ω(M ; p, q) will be Morse nondegenerate.
In the case where M is a compact Riemannian manifold and M = L21 (S 1 , M ),
we utilize the similar Riemannian metric hh·, ·ii defined by
Z   
DV DW
hhV, W iiγ = hV (t), W (t)i + (t), (t) dt,
S1 ∂t ∂t

for V, W ∈ Tγ L21 (S 1 , M )). Once again, we can integrate by parts to obtain

Z
D D
hhV, W iiγ = hP (V ), W idt, where P = − ◦ + id,
S1 dt dt
and P has a Green’s operator inverse G. The operator A = G ◦ L satisfying

d2 J(p)(V, W ) = hhA(V ), W ii, for V, W ∈ Tp L21 (S 1 , M )

is once again a Fredholm operator of Fredholm index zero.

For the space of free loops, however, J : L21 (S 1 , M ) → R is never a Morse
function. The reason is that the action J is invariant under a continuous right
action of the circle group S 1 on L21 (S 1 , M ),

φ : L21 (S 1 , M ) × S 1 −→ L21 (S 1 , M ), φ(γ, s) = γs , where γs (t) = γ(t + s),

and therefore whenever any point is critical for J, so is the entire S 1 -orbit. The
action J is also invariant under an extension of this action to an O(2)-action

φ : L21 (S 1 , M ) × O(2) −→ L21 (S 1 , M ),

the reflections in O(2) changing the orientation of the geodesic critical points.
However, after we describe Smale’s extension of Sard’s theorem to infinite-
dimensional manifolds in the next section, the theory of Fredholm maps will
allow us to perturb J to a Morse function.
We have seen that we needed to choose the L21 topology on the space of
maps in order for Condition C to hold, but we also know that for the varia-
tional problems coming from simple mechanical systems, the critical points are
automatically C ∞ . For many purposes it is convenient to have extra derivatives,
and hence to work in the space of L2k maps, for k > 1. This is admissible so
long as we don’t utilize convergence results that rely upon condition C.
For example, we can take M = L2k (S 1 , M ) with underlying Riemannian
manifold (M, h·, ·i), and consider a function f = J : M → R of the form
Z
J(γ) = L(γ)dt, where L : L2k (S 1 , M ) → L1 (S 1 , M )
Σ

80
is a sufficiently well-behaved function, called the Lagrangian of the variational
problem. We assume that we can differentiate to obtain
Z
(dJ)(γ)(V ) = hF (γ), V idA, (2.14)
S1

for some function F . The equation F (γ) = 0 is called the Euler-Lagrange equa-
tion for the variational problem, and F is called the Euler-Lagrange operator .
For example, in the case of a simple mechanical system (M, h·, ·i, φ), we take
1 0
L(γ) = hγ (t), γ 0 (t)i − φ(γ(t)),
2
and the corresponding Euler-Lagrange operator is just
F (γ) = −∇0γ γ 0 − (grad φ)(γ 0 (t)),
a nonlinear second-order differential operator.
To have an appropriate range for the nonlinear differential operator F it
is convenient to utilize pullback bundles. For k ∈ N, k > 2, we can regard
L2k−2 (S 1 , T M ) as the total space of a smooth vector bundle over L2k−2 (S 1 , M ),
and let L̃2k−2 (S 1 , T M ) denote the total space of the pullback bundle via the
inclusion ι to L2k (S 1 , M ), so

L̃2k−2 (S 1 , T M ) = {(γ, V ) ∈ L2k (S 1 , M ) × L2k−2 (S 1 , T M ) : X ∈ L2k−2 (γ ∗ T M )}.

(2.15)
However, note that the explicit construction (2.15) makes sense even if k = 2
or k = 1. Thus the Euler-Lagrange map F can be regarded as a map
F : L2k (S 1 , M ) → L̃2k−2 (S 1 , T M ),
for all k ∈ N, which is differentiable when k is large. If γ is a critical point for
J, we can differentiate to obtain a formula for the Hessian,
Z
d2 J(γ)(V, W ) = h(πV ◦ DF )(γ)(V ), W idA, (2.16)
Σ

where DF denotes the derivative with respect to γ ∈ L2k (Σ, M ) and πV is the
vertical projection into the fiber L2k−2 (γ ∗ T M ). Of course, L = πV ◦ (DF )(γ) is
just the Jacobi operator at γ. The formula shows that we can regard the Jacobi
operator L as the linearization of the Euler-Lagrange operator F at a solution
γ to the Euler-Lagrange equation.
Recall that the vector field X = −grad(J) on L21 (S 1 , M ) is defined by the
equation
hhX (γ), V ii = −dJ(γ)(V ), for V ∈ Tγ L21 (S 1 , M ). (2.17)
Thus the linearization of −X at a critical point γ ∈ L2k (S 1 , M ) (when k > 2) is
a smoothed version of the Jacobi operator. Indeed, since F (γ) = 0, it follows
from (2.14) that
−πV ◦ DX (γ) = G ◦ πV ◦ DF (γ) = G ◦ L = A,

81
where A is a restriction of the Fredholm operator satisfying (2.12).
Remark 2.5.3. The vector field X = −grad(J) is tangent to each of the
subpaces L2k (S 1 , M ) ⊆ L21 (S 1 , M ). Indeed, as mentioned before,
Z
D D
hhV, W iiγ = hPγ (V ), W idt, where Pγ = − ◦ + id.
S 1 dt dt
As γ varies, the differential operators Pγ fit together to form a vector bundle
map
P : L2k (S 1 , T M ) → L̃2k−2 (S 1 , T M ),
where L̃2k−2 (S 1 , T M ) is the total space of a vector bundle over L2k (S 1 , M ) as
described in (2.15). The vector bundle map P has a “Green’s operator” inverse
G : L̃2k−2 (S 1 , T M ) → L2k (S 1 , T M ),
a smoothing operator which increases the number of derivatives by two. If
F : L2k (S 1 , M ) → L̃2k−2 (S 1 , M )
is the Euler-Lagrange map for Jψ , then (2.17) implies that
X = −G ◦ F : L2k (S 1 , M ) → L2k−2 (S 1 , T M ).

2.6 The Sard-Smale Theorem

Sard’s theorem, so useful in understanding the transversality theory of finite-
dimensional manifolds, possesses an extension to infinite-dimensional manifolds
that is due to Smale. The standard approach to constructing Morse functions
on infinite-dimensional manifolds is based upon the Sard-Smale theorem.
Recall that a point q ∈ M2 is called a regular value for the smooth map
f : M1 → M2 if
p ∈ f −1 (q) ⇒ dfp is onto;
otherwise it is called a critical value. Any finite-dimensional manifold has a
measure which defines volume integrals on open subsets of M ; indeed, if D is
an open subset of a coordinate chart (U, (x1 , . . . xn )) on M , we can set
√
Z
Volume of D = gdx1 · · · dxn .
D

Here g is the determinant of the component matrix of the metric tensor, and it
is standard to check that the integral for volume is independent of coordinate
chart chosen. Using the Riesz representation theorem from analysis, one can
then define measurable subsets of M and speak of sets of measure zero.
Brown-Sard Theorem 2.6.1. Suppose that f : M1 → M2 is a C k map
between finite-dimensional manifolds, where
k>0 and k > dim(M1 ) − dim(M2 ).

82
Then the subset of M2 consisting of the critical values of f has measure zero.
A proof of this theorem can be found in [36], Chapter 3. The strange differen-
tiability condition cannot be weakened to k ≥ dim(M1 ) − dim(M2 ). However,
this finite-dimensional theorem actually does have a slight improvement that
was found by Bates [6]. Let C k,1 denote the class of functions which are C k and
moreover have k-th order derivatives which are Lipshitz. Bates showed that if
f : M1 → M2 is C k,1 , where

k>0 and k ≥ dim(M1 ) − dim(M2 ),

then the set of critical values of f has measure zero.

Suppose now that M1 and M2 are manifolds modeled on Banach spaces E1
and E2 respectively. A C k map f : M1 → M2 , where k ≥ 1, is said to be a
Fredholm map if its linearization

(f∗ )p : Tp M1 −→ Tf (p) M2

is a Fredholm operator, for each p ∈ M1 . The Fredholm index of a Fredholm

map f : M1 → M2 is the Fredholm index of (f∗ )p for any p ∈ M1 , this being
constant if M1 is connected. Indeed, it can be proven that the Fredholm index
is a continuous function from the space of Fredholm operators to the integers.
Note that if M1 and M2 are finite-dimensional, any C 1 map f : M1 → M2
is Fredholm with Fredholm index dim(M1 ) − dim(M2 ).
Thus the Fredholm index can be used to replace one of the hypotheses in the
Brown-Sard Theorem. We also need to replace the notion of “measure zero,”
since we have no completely satisfactory notion of measure in infinite dimensions
possessing all of the nice properties of the standard measure on submanifolds
of finite-dimensional Euclidean space. We say that a subset of M2 is residual
or generic if it is a countable intersection of open dense subsets of M2 . With
these preparations out of the way, we can now state Smale’s version of Sard’s
theorem [78]:
Sard-Smale Theorem 2.6.2. Suppose that f : M1 → M2 is a C k Fredholm
map between separable Banach manifolds, where

k>0 and k > the Fredholm index of f .

Then the set of regular values of f is residual.

The main idea of Smale’s proof is to reduce to the finite-dimensional case. We
first need to show that a Fredholm map is locally proper:
Lemma 2.6.3. If f : M1 → M2 is a Fredholm map and p ∈ M1 , then there is
an open neighborhood U of p such that the restriction of f to the closure U of
U is proper and closed.
To prove the Lemma, we use the “local representative theorem” for f near p,
a direct corollary of the inverse function theorem. According to this theorem,

83
there exist direct sum decompositions of the model spaces for M1 and M2 :

E1 = Ẽ ⊕ G1 , E2 = Ẽ ⊕ G2 ,

where G1 and G2 are both finite-dimensional, and local charts (U, φ) on M1

and (V, ψ) on M2 centered at p and f (p) respectively, such that

ψ ◦ f ◦ φ−1 : φ(U ∩ F −1 (V )) −→ E2

is of the special form

ψ ◦ f ◦ φ−1 (u, v) = (u, η(u, v)), for u ∈ Ẽ, v ∈ G1 ,

where Dη(0, 0) = 0. We denote ψ ◦ f ◦ φ−1 simply by f .

Suppose now that D1 and D2 are closed balls of radius  > 0 in Ẽ and G1
respectively. To show that f |(D1 ×D2 ) is proper, we suppose that pi = (ui , vi ) ∈
D1 × D2 and f (pi ) = qi → q ∈ V . We need to show that {pi } has a convergent
subsequence. Since D2 is compact, we can assume that vi → v ∈ D2 . On the
other hand,

f (ui , vi ) = (ui , η(ui , vi )) → q ⇒ ui → some u ∈ D1 .

Hence pi → (u, v) ∈ D1 × D2 . It follows that f |(D1 × D2 ) is proper.

To prove that f |(D1 × D2 ) is closed, we suppose that pi = (ui , vi ) ∈ D1 × D2
converges to p = (u, v). Then ui converges to some point u and η(ui , vi ) has
a subsequence which converges to some point w. Then f (ui , vi ) converges to
(u, w).
Returning to the proof of the theorem, we note that since M1 is separable, the
Lemma implies that it can be covered by a countable collection of open sets Ui
such that f |U i is proper. Thus we can reduce the proof to the case where f
is proper and closed. But this implies that the set of critical values is closed.
Thus in this case the regular values form an open set, and it suffices to prove
that the regular values are dense. In other words, it suffices to show that any
open subset of M2 contains a regular value.
It suffices to show that there is a regular value in a product neighborhood Ui
constructed as in the proof of the lemma. It follows from the finite-dimensional
version of Sard’s theorem that for each fixed choice of u0 , there is a regular
value v0 for
v 7→ ηi (u0 , v),
and (u0 , v0 ) is then a regular value, proving the theorem.
Note that by Corollary 1.2.8, if f : M1 → M2 is a C k Fredholm map between
separable Banach manifolds of Fredholm index m, where m < k, then whenever
q is a regular value of f , f −1 (q) is a submanifold of of M1 of dimension m.

84
2.7 Existence of Morse functions
We now consider an important application of the Sard-Smale theorem, the per-
turbation of a given function to a Morse function.
In the case of the action integral, the circle group S 1 acts on the free loop
space L21 (S 1 , M ) preserving the action J, and this ensures that all critical points
for J must be Morse degenerate, and in fact that all nonconstant multiples of γ 0
are Jacobi fields along γ. However, there is a simple perturbation of the action
integral whose critical points are all Morse nondegenerate, and this perturbation
is therefore a Morse function. The perturbation is obtained by making a suitable
choice of ψ ∈ C k (S 1 , RN ) for k sufficiently large, and setting
Z Z
2 1 1 0 2
Jψ : L1 (S , M ) → R by Jψ (γ) = kγ (t)k dt + γ(t) · ψ(t),
2 S1 S1

where the dot denotes the usual dot product in the ambient Euclidean space.
In this section, we will show that for most choices of ψ, Jψ is indeed a Morse
function. Recall that a subset of a complete metric space is residual if it is the
countable intersection of open dense sets.
Theorem 2.7.1. For a residual set of ψ ∈ L21 (S 1 , RN ), the function
Jψ : L21 (S 1 , M ) → R
is a Morse function; that is, all of its critical points are nondegenerate.
Note. Since critical points of Jψ are automatically elements of L23 (S 1 , M )
when ψ ∈ L21 (S 1 , RN ), it suffices to show that Jψ : L23 (S 1 , M ) → R is a Morse
function.
Proof: It is readily verified that γ is a critical point for Jψ if and only if
dJψ (X) = 0, for all X ∈ Tγ (L23 (S 1 , M )).
We can write Z
dJψ (γ)(X) = hF (γ, ψ), XidA,
Σ
where F (γ, ψ) = 0 is the Euler-Lagrange equation for Jψ , so that γ is a critical
point for Jψ if and only if F (γ, ψ) = 0.
A direct calculation shows that the Euler-Lagrange map

F : L23 (S 1 , M ) × L21 (S 1 , RN ) → L21 (S 1 , T M ) is given by

F (γ, ψ) = −∇γ 0 γ 0 + ψ T ,

where ψ T denotes the orthogonal projection of ψ into the tangent space of

M . We can regard F as a vector field on L23 (S 1 , M ), depending on a param-
eter in L21 (S 1 , RN ), which loses two derivatives, and hence takes its values in
L21 (S 1 , T M ):
−∇γ 0 γ 0 + ψ T ∈ {X ∈ L21 (S 1 , T M ) : ωπ ◦ X = γ}.

85
Lemma 2.7.2. F is transverse to the zero-section of the vector bundle

ωπ : L21 (S 1 , T M ) −→ L21 (S 1 , M ).

Proof of Lemma: Taking the partial derivative with respect to the second vari-
able, we obtain πV ◦ (D2 F )(γ, ψ)(η) = −η T , where πV denotes projection into
the vertical tangent space at (γ, 0) ∈ L21 (S 1 , T M ). The formula shows that
πV ◦ (D2 F )(ψ, γ) maps onto the fiber of ωπ over γ. Hence if γ is a critical point
for Jψ ,

(Range of πV ◦ (D2 F )(ψ, γ)) + (Tangent space to zero-section)

= Tangent space to L21 (S 1 , T M ),

which means that F is transverse to the zero-section.

Just as in the finite-dimensional case, it follows from Corollary 1.2.8 that the
inverse image of a split submersion of a submanifold is itself a submanifold, so
F −1 (zero-section) is a submanifold S of L23 (S 1 , M ) × L21 (S 1 , RN ). We can also
describe this submanifold as the solution set

S = {(γ, ψ) ∈ L23 (S 1 , M ) × L21 (S 1 , RN ) : ∇γ 0 γ 0 − ψ T = 0}.

Lemma 2.7.3. The projection on the second factor π : S → L21 (S 1 , RN ) is a

Fredholm map of Fredholm index zero.
We begin by determining the tangent space to S, obtaining

T(γ,ψ) S = {(X, η) ∈ T L23 (S 1 , M ) × L21 (S 1 , RN ) :

πV ◦ D1 F (γ, ψ)X + πV ◦ D2 F (γ, ψ)η = 0},

or equivalently,

T(γ,ψ) S = {(X, η) ∈ T L23 (S 1 , M ) × L21 (S 1 , RN ) : Lγ,ψ (X) = −η T }, (2.18)

where
Lγ,ψ = πV ◦ D1 F (γ, ψ) : Tγ L23 (S 1 , M ) → Tγ L21 (S 1 , M )
is the Jacobi operator for Jψ determined by the formula
Z
2
d Jψ (γ)(X, Y ) = hLγ,ψ (X), Y idA, for X, Y ∈ Tγ L23 (S 1 , M ).
Σ

Let B : T M × T M → N M denote the second fundamental form of M in RN

and define for n ∈ Np M ,

An : T M → T M by hAn (x), yi = B(x, y) · n.

86
Then a calculation (similar to that used to prove Proposition 2.4.1) shows that
the Jacobi operator is given by the formula

Lγ,ψ (X) = −∇γ 0 ∇γ 0 (X) − R(X, γ 0 )γ 0 − Aψ(t)⊥ (X). (2.19)

One readily verifies that Lγ,ψ is a Fredholm operator of Fredholm index zero.
We next calculate the kernel and image of the map dπ(γ,ψ) : Tγ,ψ S →
L21 (S 1 , RN ) and after a short derivation, we obtain the results:

Kernel of dπ(γ,ψ) = {(X, η) ∈ T(γ,ψ) S : η = 0, Lγ (X) = 0},

Range of dπ(γ,ψ) = {Lγ,ψ (X) + χ⊥ : X ∈ Tγ L23 (S 1 , M ), χ ∈ L21 (S 1 , RN )},

where χ⊥ denotes the orthogonal projection of ψ into the normal space to M .
Note that the kernel of dπ(γ,ψ) is isomorphic to the kernel of Lγ,ψ while the
range of dπ(γ,ψ) is a subspace of L21 (S 1 , RN ) which has the same codimension
as Range(Lγ,ψ ) ⊂ Tγ L21 (S 1 , M ).
Thus dπ(γ,ψ) is a Fredholm operator with the same Fredholm index as Lγ,ψ ,
namely zero, finishing the proof of Lemma 2.6.3.
According to the Sard-Smale Theorem 2.6.2, there is a countable intersection
of open dense subsets of ψ ∈ L22 (S 1 , RN ) consisting of regular values of π. But
if ψ is a regular value for π, then Kernel(Lγ,ψ ) = 0 at each critical point, so all
critical points are Morse nondegenerate, and Jψ is a Morse function, establishing
Theorem 2.7.1.
Definition. Suppose that M is a Banach manifold and f : M → R is a C 2
function. A compact finite-dimensional submanifold N of M is said to be a
nondegenerate critical submanifold for f if
1. every p ∈ N is a critical point for f .
2. if p ∈ N , then Tp N is the set of V ∈ Tp M such that d2 f (p)(V, W ) = 0,
for all W ∈ Tp M.
The Morse index of a connected nondegenerate critical submanifold N is the
index of any critical point p ∈ N .
Proposition 2.7.4. If M is a compact Riemannian manifold and

M0 = {γ ∈ L21 (S 1 , M ) : γ is constant },

then M0 is a nondegenerate critical submanifold of L21 (S 1 , M ) of Morse index

zero. Moreover, for some  > 0, the open neighborhood

U = {γ ∈ L21 (S 1 , M ) : |γ − M0 |C 0 < }

has M0 as a strong deformation retract and contains no critical points for the
action J other than the elements of M0 .

87
To prove this, first note that if γp denotes the constant loop at the point p ∈ M ,
then
Tγp L21 (S 1 , M ) ∼
= L21 (S 1 , Tp M )
and the tangent subspace Tγp M0 consists of the constant maps into Tp M . On
the other hand, the normal space to M0 at γp can be identified with
 Z 
Nγp M0 = V ∈ L21 (S 1 , Tp M ) : V (t)dt = 0 .
S1

According to the second variation formula (2.7),

Z
d2 J(γp )(V, W ) = hV 0 (t), W 0 (t)idt, for V, W ∈ L21 (S 1 , Tp M ),
S1

from which we conclude that d2 J(γp )(V, W ) = 0 if V is constant, that is, if V

lies in Tγp M0 . On the other hand, it is readily verified that d2 J(γp ) is positive
definite on Nγp M0 , so M0 is indeed a nondegenerate critical submanifold of
Morse index zero.
Let N M0 denote the total space of the normal bundle of M0 and for  > 0,
let
V = { V ∈ Nγp M0 for some p ∈ M such that |V |C 0 <  },
and define
φ : V → U by φ(V ) = expp (V ).
For  > 0 sufficiently small, we can define

H : U × [0, 1] → U by H(expp (V ), t) = expp (tV ).

Then H is smooth, H(γ, 1) = γ for γ ∈ U , H(γ, t) = γ for γ ∈ M0 and all

t ∈ [0, 1], so γ 7→ H(γ, 0) is the desired deformation retraction from U to M0 .
Finally, using a Taylor expansion about points in M0 , one verifies that U − M0
contains no critical points for J.
We can now alter the definition of Jψ slightly. Suppose that η : R → [0, 1] is a
smooth function such that
(
0, for t ≤ 2 /4,
η(t) =
1, for t ≥ 2 /2.

Note that since L(γ)2 ≤ 2J(γ), η(J(γ)) = 1 outside U . Given a smooth map
ψ : S 1 → RN , we consider the function Jψ defined by
Z
Jψ (γ) = J(γ) + η(J(γ)) (γ · ψ)dt. (2.20)
S1

It follows from Theorem 2.7.1 that if ψ is generic and sufficiently small, then
Jψ ≥ 0 and M0 = Jψ−1 (0) is a critical submanifold of Morse index zero. More-
over, all other critical points of Jψ are Morse nondegenerate, so the restriction
of Jψ to L21 (S 1 , M ) − M0 is a Morse function.

88
2.8 Bumpy metrics for smooth closed geodesics*
With additional work, one can show that if M is a compact manifold, then
for generic choice of Riemannian metric on M , all nonconstant smooth closed
geodesics have the property that their only Jacobi fields are those generated by
the S 1 action on L21 (S 1 , M ). This is the “Bumpy Metric Theorem” of Abraham,
and a proof due to Anosov is found in [4]; an alternate proof can be found in
[7]. We will present another proof in this section, based upon Bott’s theory of
iterated smooth closed geodesics [8]. The proof is long so some readers may
want to skip it on first reading.
We first prove a simpler version for the case where the geodesic is prime,
that is, not a nontrivial cover of a geodesic of smaller length:
Theorem 2.8.1. If M is a compact manifold, then for generic choice of Rie-
mannian metric on M , all nonconstant prime smooth closed geodesics have the
property that their only Jacobi fields are those generated by the S 1 action on
L21 (S 1 , M ).
Proof of Theorem 2.8.1: The techniques are exactly the same as those used
before. Note first that since critical points for the action function J are smooth,
we are not restricted to working in L21 (S 1 , M ), but can work for example in a
Sobolev space of more highly differentiable functions, such as L22 (S 1 , M ), which
is also a Hilbert manifold.
We need a preliminary step in the argument to deal with the fact that J is
G-equivariant, where G = S 1 . If N is a compact codimension one submanifold
of M with boundary ∂N , we let

U(N ) = { nonconstant γ ∈ L2k (S 1 , M ) : γ does not intersect ∂N

and has transversal intersection with the interior of N },

an open subset of L22 (S 1 , M ). We cover L22 (S 1 , M ) with a countable collection

Ui = U(Ni ) of open sets corresponding to a countable collection of codimension
two submanifolds Ni . If 0 is a choice of origin in S 1 , we let

Fi = {γ ∈ U(Ni ) : γ(0) ∈ Ni },

a C 1 submanifold of L22 (S 1 , M ) by the smoothness theorems in §1.6. Note that

Fi meets each nonconstant G-orbit in U(Ni ) in a finite number of points.
Let M denote the manifold of L2` Riemannian metrics on M , an open subset
of the space of L2` sections of the second symmetric power of T ∗ M , where ` is
a large integer. We claim that

S = {(γ, g) ∈ Fi × M : γ is a geodesic for the metric g}

is a smooth submanifold of Fi × M. To see this, we let L2∗ (S 1 , T M ) denote the

vector bundle over Fi whose fiber at each f ∈ L22 (S 1 , M ) is L2 (S 1 , γ ∗ T M ). It
follows from multiplication theorems from the theory of Sobolev spaces (L22 ·L2 ⊂

89
L2 ) that L2∗ (S 1 , T M ) does in fact have the structure of a smooth vector bundle
over Fi . We next define a C 1 map

F : Fi × M −→ L2∗ (S 1 , T M ) by F (γ, g) = ∇gγ 0 γ 0 ,

where ∇g is the Levi-Civita connection on T M determined by the metric g on

M . Let Z denote the image of the zero section of L2∗ (Σ, T M ) and note that
S = F −1 (Z). Our goal is to show that F is as transverse to Z as possible.
The first derivative of the action is given by the formula
Z
dJ(γ)(X) = hF (γ, g), Xidt.
S1

Differentiating once again gives us the Hessian at a critical point,

Z Z
2
d J(γ)(X, Y ) = hD1 F (γ, g)(X), Y idt = hLγ (X), Y idt, (2.21)
S1 S1

where D1 F denotes derivative with respect to the first variable

{γ ∈ L22 (S 1 , M ) : γ(0) ∈ Ni },

and

Lγ : {X ∈ L22 (S 1 , γ ∗ T M ) : X(0) ∈ Tγ(0) Ni } −→ L2 (S 1 , γ ∗ T M ) (2.22)

is the Jacobi operator. Note that at a zero (γ, g) of F the tangent space to
L2∗ (S 1 , T M ) can be divided into a direct sum

Tγ (L2∗ (S 1 , T M )) = H ⊕ V,

where H is horizontal (tangent to the zero section) and V is vertical (tangent

to the fiber). If πV denote the projection onto V along H, it follows from (2.21)
that
πV ◦ (D1 F )(γ,g) (X) = Lγ (X). (2.23)
Because it is a Jacobi field, the section T (t) = γ 0 (t) is not in the image of Lγ .
However, we claim that

{the image of πV ◦ DFγ,g } ⊕ (span of T (t)) = L2 (S 1 , γ ∗ T M ), (2.24)

or equivalently, that V is spanned by T (t), where

V = {L2 -sections X of γ ∗ T M : X is ⊥ to the image of πV ◦ DFγ,g }.

To show this, we need to calculate πV ◦ (D2 F ), where D2 F is the partial deriva-

tive with respect to the second variable g ∈ M.
Suppose that γ is a nonconstant geodesic, which will be C ` when the metric
is L2` . Choose a point t ∈ S 1 which possesses a neighborhood U such that γ
imbeds U into some open set W ⊂ M on which Fermi coordinates (x1 , . . . , xn )
are defined. Such coordinates satisfy the following conditions:

90
 1. γ is described by the equations x2 = · · · = xn = 0,
2. x1 ◦ γ = t,
gij dui duj , such that on γ(U ), gij = δij , for
P
3. the metric g takes the form
1 ≤ i, j ≤ n.
In terms of these local coordinates, the equation for geodesics becomes
d2 xk X k dxi dxj
+ Γij = 0.
dt2 i,j
dt dt

The expression on the left side of this equation is called the acceleration of the
path.
A perturbation in theP metric h ∈ Tg M with compact support in W can be
written in the form h = hij dxi dxj . Under this perturbation, the only piece
of the acceleration that changes is the Christoffel symbol
 
1 X kl ∂gil ∂gjl ∂gij
Γkij = g + − ,
2 ∂xj ∂xi ∂xl

and if Γ̇kij denotes the derivative of Γkij in the direction of the perturbation,
 
k 1 ∂hik ∂hjk ∂hij
Γ̇ij = + − .
2 ∂xj ∂xi ∂xk
We then find that
n n
X dxi dxj ∂ X ∂
πV ◦ (D2 F )(γ,g) (h) = Γ̇kij = Γ̇k11 . (2.25)
dt dt ∂xk ∂xk
i,j,k=1 k=1

We can select the perturbation so that

h11 = x2 φ, hij = 0, for other choices of indices i, j,

where φ has compact support in W and 2 ≤ r ≤ n; we then see that

1 ∂ 1
Γ̇r11 (t, 0, . . . , 0) = − (h11 )(t, 0, . . . , 0) = − φ(t, 0, . . . , 0).
2 ∂x2 2
Thus the fiber projection of the partial derivative of F with respect to the second
variable (in M) is given by the expression
n
1X r ∂ 1 ∂
πV ◦ (D2 F )f,g (h) = − Γ̇ =− φ .
2 r=2 11 ∂xr 2 ∂x2

In this manner, we can show that any vector field of the form
n
X ∂
φr
r=2
∂xr

91
where φr is a smooth function with compact support, lies in the image of πV ◦
D2 Ff,g , and hence elements of V must be tangent to γ in U .
Since a dense open subset of points of Σ can be covered by sets of the form
U , we see that elements of V must be tangent to γ over all of S 1 . On the other
hand, they must also be image of Lγ restricted to complement of the span of
T (t) by (2.23). Thus we obtain the desired result (2.24).
The remainder of the proof is similar to the proof of Theorem 2.7.1. Note
that the tangent space to S is given by the formula
Tγ,g S = {(X, h) ∈ Tγ Fi × Tg M : Lγ (X) + πV ◦ (D2 F )(h) = 0},
which makes it easy to analyze the kernel and range of the map dπ(γ,g) :
T(γ,g) S → Tg M. Indeed,
Kernel of dπ(γ,g) = {(X, h) ∈ Tγ Fi × Tg M : h = 0, Lγ (X) = 0},
while

Range of dπ(γ,g) = {h ∈ Tg M; there exists an element X ∈ Tγ L22 (S 1 , M )

such that X(0) ∈ Tγ(0) Ni and Lγ (X) = −πV ◦ (D2 F )(h)}.
Thus the kernel of dπ(γ,g) is isomorphic to the kernel of Lγ . On the other hand,
if h is an element of Tg M such that
πV ◦ (D2 F )(h) = 0, then (0, h) ∈ T(γ,g) S,
and hence h lies in the range of dπ(γ,g) . It follows that complement to the range
of dπ(γ,g) must inject to a complement to the range of Lγ , and in particular,
any such complement is finite-dimensional, so dπ(γ,g) is a Fredholm map, and
the dimension of the cokernel of dπ(γ,g) is no larger than the dimension of the
cokernel of Lγ . By the earlier transversality argument, dFγ,g maps surjectively
onto a complement to the one-dimensional space generated by the Jacobi field
T (t), so the dimension of the cokernel of dπ(γ,g) is actually one less than the
dimension of the cokernel of Lγ .
Thus the Fredholm index of dπ(γ,ψ) one more than the Fredholm index of the
map Lγ of (2.22), which is one because the restriction X(0) ∈ Tγ(0) Ni cuts down
the kernel by one. It follows that the Fredholm index of dπ itself is zero. Thus
we can use the Sard-Smale Theorem to show that a countable intersection of
open dense subsets of M consist of regular values for π. If g0 is such a “generic
metric,” all of the prime geodesics will be Morse nondegenerate, finishing the
proof of the Bumpy Metric Theorem for prime geodesics.
Implication of the above Theorem: It follows from Theorem 2.8.1, together
with condition C of Palais and Smale, that for a Riemannian metric belonging
to a residual subset, the number of S 1 -orbits of nonconstant prime geodesics of
length less than a given bound is finite.
We will next extend the argument from the previous theorem to cover the case
in which geodesics are not necessarily prime, thereby obtaining:

92
Bumpy Metric Theorem 2.8.2. If M is a compact manifold, then for generic
choice of Riemannian metric on M , all nonconstant smooth closed geodesics have
the property that their only Jacobi fields are those generated by the S 1 action
on L21 (S 1 , M ).
Of course a geodesic which is not prime covers a prime minimal geodesic and
our strategy is to study this underlying prime geodesic.
We make the following definitions following Bott [8]. Suppose that γ : R →
M is a smooth geodesic with γ(t + 2π) = γ(t). If q ∈ Z and z ∈ S 1 ⊂ C, we let
∞
Vq,z = { smooth sections X of γ ∗ T M ⊗ C : X(t + 2πq) = zX(t)},
and define an inner product
∞ ∞
h·, ·iq : Vq,z × Vq,z −→ C
by Z 2πq   
DX DȲ
hX, Ȳ iq = , + hX, Ȳ i dt,
0 ∂t ∂t
where D denotes the covariant derivative defined by the Levi-Civita connection
∞
on M . Let Vq,z denote the completion of Vq,z with respect to h·, ·iq and define
Iq (·, ·) : Vq,z × Vq,z −→ C
by Z 2πq   
DX DȲ 0 0
Iq (X, Ȳ ) = , − hR(X, γ )γ , Ȳ i dt,
0 ∂t ∂t
which is of course the restriction of d2 J(γ) to Vq,z .
Lemma 2.8.3. The inclusion
X
V1,z ⊂ Vq,1 (2.26)
z q =1

is an isomorphism.
Since the inclusion is clearly injective, it suffices to show that it is surjective. If
X ∈ Vq,1 and z is a primitive q-th root of unity, we let
q−1
1 X −j
Xz (t) = z X(t + 2πj).
q j=0

Then
q−1
1 X −j
Xz (t + 2π) = z X(t + 2π(j + 1))
q j=0
q−1
z X −(j+1)
= z X(t + 2π(j + 1)) = zXz (t),
q j=0

93
so Xz ∈ V1,z . Moreover, X
X= Xz ,
z q =1

so the inclusion (2.26) is indeed an isomorphism, proving the Lemma.

Note that if X ∈ V1,z1 and Y ∈ V1,z2 , where z1 and z2 are q-th roots of unity,
then
Z 2π   
DX DȲ
Iq (X, Ȳ ) = , − hR(X, γ 0 )γ 0 , Ȳ i dt
0 ∂t ∂t
Z 4π   
DX DȲ
+ z1 z̄2 , − hR(X, γ 0 )γ 0 , Ȳ i dt
2π ∂t ∂t
Z 2πq   
q−1 q−1 DX DȲ 0 0
+ ··· + z1 z̄2 , − hR(X, γ )γ , Ȳ i dt
2π(q−1) ∂t ∂t
 
q−1 Z 2π   
X j j DX DȲ 0 0
=  z1 z̄2  , − hR(X, γ )γ , Ȳ i dt.
j=0 0 ∂t ∂t

Thus we see that (

qI1 (X Ȳ ), if z1 = z2 ,
Iq (X, Ȳ ) =
0, if z1 6= z2 ,
P
and hence the direct sum decomposition zq =1 V1,z of Vq,1 is orthogonal with
respect to the index form Iq . Let N (z) denote the nullity of the index form I1
restricted to V1,z ,

N (z) = dimC {X ∈ V1,z : I1 (X, Ȳ ) = 0 for all Y ∈ V1,z }.

The above discussion proves the following lemma due to Bott [8], which plays
a key role in his analysis of the relationship between the index and nullity of a
prime smooth closed geodesic and the index of nullity of its multiple covers:
Lemma 2.8.4. Let γ q denote the q-fold iterate of the smooth closed geodesics
γ, so γ q (t) = γ(qt). Then
X
Nullity of γ q = N (z).
z q =1

We now return to the proof of the bumpy metric theorem itself. It suffices to
consider geodesics whose length is less than a given bound. We know that there
are only finitely many prime geodesics with length below this bound and that if
the metric is perturbed by a sufficiently small amount, no new prime geodesics
will be introduced. Our strategy is to perturb the metric in a neighborhood of
a given geodesic in such a way that the geodesic is preserved.
As in the proof of the preceding Theorem, we construct a perturbation of
the Riemannian metric on M of a specific form. Once again, we choose a point

94
t ∈ S 1 and a neighborhood U containing t such that γ imbeds U into some open
set W ⊂ M on which Fermi coordinates (x1 , . . . , xn ) are defined, coordinates
such that:
1. γ is described by the equations x2 = · · · = xn = 0,
2. x1 ◦ γ = t,
gij dui duj , such that on γ(U ), gij = δij , for
P
3. the metric g takes the form
1 ≤ i, j ≤ n.
Following Klingenberg
P ([42], Proposition 3.3.7), we construct a perturbation of
the metric ġ = ġij dxi dxj such that
n
X
ġ11 (x1 , . . . xn ) = xr xs αrs (x1 , x2 , . . . , xn ), ġij = 0 if (i, j) 6= (1, 1).
r,s=2

Here αrs are smooth functions which vanish outside a small tubular neighbor-
hood of γ. A straightforward calculation shows that the resulting changes in
the Christoffel symbols
 
1 ∂ ġik ∂ ġik ∂ ġik
Γ̇kij = + −
2 ∂uj ∂uj ∂uj
vanish unless at least two of the indices are 1, and if 2 ≤ r, s ≤ n,
X X
Γ̇r11 = − us αrs , Γ̇1r1 = us αrs .
s s

The corresponding changes in curvature components are given by the formulae

∂   ∂  
Ṙlijk = Γ̇ljk − Γ̇lik
∂ui ∂uj
X X X X
+ Γ̇lim Γmjk + Γlim Γ̇m
jk − Γ̇ljm Γm
ik − Γljm Γ̇m
ik .
m m m m

But along f0 (Σ0 ) all the Γ̇kij ’s and Γ̇kij ’s

must vanish, so the resulting change
in the curvature R1r1s along γ is given by
Ṙ1r1s (x1 , 0, . . . , 0) = αrs (x1 , 0, . . . , 0).
The Jacobi operator Lγ on normal sections can be expressed in components
as follows: If
n n
" n
#
X ∂ X d2 fr X ∂
X= fr , then Lγ (X) = − 2 + R1r1s fs .
r=2
∂xr r=2
dx1 s=2
∂xr

We can consider the family of formally self-adjoint second order differential

operators T of the form
n
" n
#
X d 2 fr X ∂
T (X) = − 2 + Trs fs .
r=2
dx 1 s=2
∂x r

95
For each such operator and each q-th root of unity z ∈ S 1 ⊂ C, we can define
the nullity
N (z) = dimC {X ∈ V1,z : L(X) = 0}.
For an open dense set of such operators T , N (z) = 0. The argument in the
preceding paragraph shows that we can perturb the Jacobi equation along any
geodesic so that N (z) = 0. Since there are a finite number of roots of unity
corresponding to covers γ q of a fixed prime geodesic which have length less than
a given bound, we can ensure that for generic metrics all such geodesics γ q will
be Morse nondegenerate. This finally proves the Bumpy Metric Theorem for
geodesics.

2.9 Adding handles

Suppose that M is a smooth manifold modeled on a Hilbert space with a com-
plete Riemannian metric, and that f : M → R is a C 2 Morse function satisfying
condition C. From the Deformation Theorem 1.11.1, we know that the topology
of
Ma = {p ∈ M : f (p) ≤ a}
does not change as a increases unless a passes a critical value for f . Our goal
now is to understand what happens to the topology of Ma when a passes a
critical value c for f such that f −1 (c) contains a finite number of critical points,
all Morse nondegenerate. In this section, we carry our that analysis for the case
of Hilbert manifolds.
Theorem 2.9.1. Suppose that M is a Hilbert manifold with a complete Rie-
mannian metric hh·, ·ii and that f : M → [0, ∞) is a smooth function satisfying
condition C. If the interval [a, b] contains a single critical value c for f , there
is exactly one critical point p for f such that f (p) = c and this critical point
is Morse nondegenerate of Morse index λ, then Mb is homotopy equivalent to
Ma with a handle of index λ attached.
To prove this, we choose a coordinate chart (U, φ) about the nondegenerate
critical point p with φ(p) = 0 ∈ E, where E is the model space for M, and let
f¯ = f ◦ φ−1 , a smooth function on the open subset φ(U ) ⊂ E which has a single
nondegenerate critical point of index λ at the origin. Expanding the smooth
function f¯ in a Taylor series yields
1
f¯(x) = c + d2 f¯(0)(x, x) + R1 (x), where kR1 (x)k ≤ 1 kxk2 , (2.27)
2
for x ∈ φ(U ), where 1 is a positive constant which can be made arbitrarily
small by making U small. We can identify Tp M with E and since p is Morse
nondegenerate, the self-adjoint bounded linear operator A on E determined by

d2 f¯(0)(x, y) = hhAx, yii, for x, y ∈ E,

96
where hh·, ·ii is the Riemannian metric, is invertible, so its spectrum is a closed
subset of the real axis, bounded away from zero. Corresponding to the restric-
tion of the vector field X = grad(f ) to U is a vector field on φ(U ) with local
representative X̄ which has the Taylor series expansion

X̄ (x) = Ax + R2 (x), where kR2 (x)k ≤ 2 kxk, (2.28)

for x ∈ φ(U ), where 2 is an arbitrarily positive constant. The vector field X̄ on

φ(U ) is closely approximated by its linearization XA which has local representa-
tive XA (x) = Ax, and one can check that this linearization is a pseudo-gradient
(as defined in § 1.11) when 2 is sufficiently small.
If the critical point p has finite index, the negative part of the spectrum is
discrete and consists of finitely many eigenvalues. In this case, we let E− be
the subspace of E generated by the negative eigenvalues of A and let E+ be
the closed orthogonal complement to E− in E. In general, the spectral theorem
allows us to divide E into a direct sum E = E+ ⊕ E− , each summand being
preserved by A, and hence we can think of A as dividing into a “block matrix”
 
A− 0
A=
0 A+

with A− and A+ self-adjoint invertible operators on the subspaces E− and

E+ respectively. The spectrum of A− lies on the negative real axis, while the
spectrum of A+ lies on the positive real axis. Thus if we suppose that x− and x+
are the orthogonal projections of x into E− and E+ respectively, so x = x− +x+ ,
the system of differential equations represented by the linearization −XA is
(
dx− /dt = −A− x− ,
dx+ /dt = −A− x+ .

This is just a linear system with constant coefficients, which has {φ̃t : t ∈ R} as
its one parameter group of diffeomorphisms, where

φ̃t (x− , x+ ) = (exp(−tA− x− ), exp(−tA+ x+ )). (2.29)

The fluid flow for −XA described by {φ̃t : t ∈ R} is expanding on the subspace
E− , contracting on E+ and closely approximates the fluid flow for X̄ .
Let V be an open subset of U such that p ∈ V ⊆ V̄ ⊆ U .
Lemma 2.9.2. There exist r1 > 0 and s1 > 0 such that
1. D− (r) × D+ (s) ⊆ V and XA is transverse to D− (r) × ∂D+ (s) when r ≤ r1
and s ≤ s1 ,
2. f¯(∂D− (r1 ) × D+ (s1 )) ≤ c − , for some  > 0, and

3. f¯−1 (c − ) is transverse to D− (r1 ) × x+ for x+ ∈ D+ (s1 ).

97
The first condition follows immediately from the explicit form (2.29) of the fluid
flow for XA .
To prove the second claim, we use the Taylor expansion (2.27) to conclude
that
1 1
f¯(x) = c + hhA− x− , x− ii + hhA+ x+ , x+ ii + R1 (x)
2 2
1 1
≤c− kx− k2 + + kA+ kkx+ k2 + 1 kxk2 .
2kA−1
− k 2

We can choose 1 smaller than (1/(4kA− k2 ). Thus if we choose r1 small and

then s1 much smaller, we can arrange that the second claim holds. The third
claim is proven in a similar fashion, and we leave it as an exercise for the reader.
In accordance with the terminology of geometric topology, we call ∂D− (r1 )
and D− (r1 ) the descending sphere and descending disk of the critical point p;
∂D− (r1 ) is a sphere of dimension λ − 1. We also have an ascending sphere
∂D+ (s1 ) and an ascending disk D+ (s1 ), both of which are infinite-dimensional.
Returning now to the proof of the theorem, we let η : M → [0, 1] be a smooth
function such that η(q) = 1 for q ∈ M − U , η(q) = 0 for q ∈ V and let

X = η grad(f ) + (1 − η) φ−1
∗ XA .

Finally, we let {φt : t ∈ R} be the one-parameter group of local diffeomorphisms

corresponding to the vector field −X . For q ∈ Mb − Mc− , let τ (q) denote the
first time t such that

φt (q) ∈ Mc− ∪ φ−1 (D− (r1 ) × ∂D+ (s1 )).

Note that τ (q) is finite by the argument for the Theorem 1.10.1 and the transver-
sality conditions of Lemma 2.8.4 show that τ (q) depends continuously on q. Let
ht : Mb → Mb by ht (q) = φtτ (x) (q). Then clearly h0 is the identity map and

h1 (Mb ) ⊂ Mc− ∪ φ−1 (D− (r1 ) × ∂D+ (s1 )).

In fact, we easily check that ht gives a deformation retraction from Mb to

Mc− ∪ φ−1 (D− (r1 ) × ∂D+ (s1 )), which is homotopy equivalent to Mc− with
a handle of index λ attached. By the Deformation Theorem 1.11.1 this has the
homotopy type of Ma with a handle of index λ attached.
Remark 2.9.3. In the proof, the Riemannian metric on M is used only to
construct the continuous isomorphism A : Tp M → Tp M and the gradient vector
field X away from the critical point p. This suggests that it might be convenient
to formulate the notion of “gradient-like” vector field, more flexible than an
ordinary gradient, a concept utilized by Milnor for finite-dimensional manifolds
in [51]:
Definition. Suppose that f : M → R is a C 2 Morse function on a Hilbert
manifold M with a complete Riemannian metric, and let K ⊆ M be the critical

98
locus of f . We say that a C 2 vector field X on M is gradient-like for f if M
possesses an open cover {U, V } such that
1. the restriction of f to M − K is a pseudogradient for f ,
2. the open set U divides into connected components Up , one for each p ∈ K,
such that Up is the domain of a coordinate system φp such that φp (p) = 0,
and
3. (φp )∗ (X |Up ) = XA , where XA has local representative

XA (x) = Ax, for x ∈ φp (U ), (2.30)

and A : Tp M → Tp M is invertible and self-adjoint with respect to some

Hilbert space inner product hh·, ·ii on Tp M.

If {ηV } ∪ {ηp : p ∈ K} is a C k partition of unity with respect to the open cover

{V } ∪ {Up : p ∈ K}, and XV is a pseudogradient for f on V , then
X
X = ηV XV + ηp [(φ−1
p )∗ XA ]
p∈K

is a C k gradient-like vector field for f on M, so long as the open neighborhoods

Up are chosen sufficiently small. Even if the Morse function f is only C 2 (so
its gradient is only C 1 ), we can construct C k gradient-like vector fields for f on
M − K so long as M admits C k partitions of unity.
Note that it was a gradient-like vector field for f that we utilized in the
proof of Theorem 2.8.1. Since pseudogradients form an open subset of the
space of vector fields, gradient-like vector fields give considerable flexibility in
constructions involving flows corresponding to Morse functions, as we will see
when constructing the Morse-Witten chain complex for a Morse function f in
the next section.
We can extend Theorem 2.9.1 to the case of several Morse nondegenerate
critical points:
Theorem 2.9.4. Suppose that M is a Hilbert manifold with a complete Rie-
mannian metric hh·, ·ii and that f : M → [0, ∞) is a smooth function satisfying
condition C. If the interval [a, b] contains a single critical value c for f and there
are finitely many critical point p1 , . . . , pk for f such that f (pi ) = c which are
Morse nondegenerate with finite Morse indices λ1 , . . . λk , then Mb is homotopy
equivalent to Ma with handles of index λ1 , . . . , λk attached.
The proof is a straightforward extension of the proof of Theorem 2.8.1.
Remark 2.9.5. Theorems 2.9.1 and 2.9.4 both hold for critical points of infi-
nite Morse index, although attaching handles of infinite index turns out to be
invisible from the homotopy theory point of view. Moreover, in the problems
we have been considering from the calculus of variations, the Morse index of

99
critical points is always finite, so the handles constructed via these theorems
are finite-dimensional.
The Bumpy Metric Theorem 2.8.2 provides motivation for extending Theo-
rems 2.9.1 and 2.9.4 to the case of nondegenerate critical submanifolds. If
N ⊂ M is a compact nondegenerate critical submanifold of finite Morse index
λ, the normal bundle νN to N is defined via the Riemannian metric hh·, ·ii on
M; its fiber at p ∈ N is

(νN )p = {V ∈ Tp M : hhV, W ii = 0 for all W ∈ Tp N }.

The normal bundle has a finite-dimensional subbundle ν− N whose fiber at p

is generated by eigenvectors corresponding to the negative eigenvalues of A :
Tp M → Tp M, the rank of the bundle ν− N being the Morse index of N . Let

D(ν− N ) = {V ∈ ν− N : kV k ≤ 1}, S(ν− N ) = {V ∈ ν− N : kV k = 1},

the unit disk and unit sphere bundles in the negative normal bundle ν− N .
Theorem 2.9.6. Suppose that M is a Hilbert manifold with a complete Rie-
mannian metric hh·, ·ii and that f : M → [0, ∞) is a smooth function satisfying
condition C. If [a, b] contains a single critical value c for f , the set of critical
of points with value c forming a nondegenerate critical submanifold N of finite
Morse index λ, then Mb is homotopy equivalent to Ma with the disk bundle
D(ν− N ) attached to Ma along S(ν− N ).
The proof is a relatively straightforward modification of the proof for Theo-
rem 2.9.1. Note that we can allow N to have more than one component.

2.10 Morse inequalities

Suppose now that M is a Hilbert manifold with a complete Riemannian metric
hh·, ·ii and that f : M → [0, ∞) is a Morse function satisfying condition C, all
critical points having finite Morse index. For a ∈ R, we let

Ma = {p ∈ M : f (p) ≤ a}.

Condition C implies that each Ma has only finitely many nondegenerate critical
points. We would like to investigate how the topology of Ωa changes as a
increases. There are two ways of tracking the changes in topology, one is via
the Morse inequalities which we discuss next, the other via the Morse-Witten
chain complex for f which will be treated in § 2.11.
For each nonnegative integer λ, we let µaλ denote the number of critical
points for f within Ma of Morse index λ and for a given choice of field F (such
as R or Z2 ), we let βλa be the dimension of H i (Ma ; F ). Then the weak Morse
inequalities state that
µaλ ≥ βλa . (2.31)

100
Instead of proving these directly, we will prove a stronger version of the Morse
inequalities in terms of the Morse polynomial for f and Poincaré polynomial for
Ma , defined respectively by
∞
X ∞
X
Ma (t) = µaλ tλ , P a (t) = βλa tλ .
λ=0 λ=1

Then (2.31) is an immediate consequence of:

Theorem 2.10.1. (Morse inequalities) M is a Hilbert manifold with a
complete Riemannian metric and that f : M → [0, ∞) is a Morse function
satisfying condition C. Then there is a polynomial
∞
X
Qa (t) = qλa tλ with qλa ≥ 0 such that Ma (t) − P a (t) = (t + 1)Qa (t).
λ=0
(2.32)

We make the simplifying assumption that there is only critical point for each
critical value, so that we can apply Theorem 2.9.1 instead of Theorem 2.9.4.
This can be arranged quite easily by simply perturbing f . Our strategy is to
apply an induction on a.
To start the induction, we note that when we set a < 0, Ma is empty and
hence βλa = µaλ = 0. We can therefore set Qa (t) = 0 in this case.
For the inductive step, observe first that if the interval [a, b] contains a single
critical value c corresponding to a Morse nondegenerate critical point of Morse
index λ, then by Theorem 2.9.1,
(
F, if k = λ,
Hk (Mb , Ma ; F ) ∼
=
0, if k 6= λ.

Thus it follows from the exact sequence

· · · → Hk (Ma ; Z) → Hk (Mb ; Z) → Hk (Mb , Ma ; Z) → · · ·

that either
P b (t) = P a (t) + tλ or P b (t) = P a (t) − tλ−1 ,
and one can check that the two cases depend on whether the descending sphere
(pushed down into Ma ) bounds or not. In the former case, the descending
disk can be completed to a cycle representing a new generator of λ-dimensional
homology, while in the latter it yields a new relation in (λ − 1)-dimensional
homology. Since Mb (t) = Ma (t) + tλ in either case, we see that

(Mb (t) − P b (t)) − (Ma (t) − P a (t))

is either 0 or tλ−1 (t + 1). Thus assuming (2.32) for a, we can arrange that

Mb (t) = P b (t) + (t + 1)Qb (t)

101
 Æ
f
q

p1
p2

Figure 2.1: If a smooth proper Morse function f : R2 → [0, ∞) has two local
minima at p1 and p2 , it must have an additional critical point (say at q) of
Morse index one. This consequence of the Morse inequalities is known as the
“mountain pass lemma.”

also holds by setting Qb (t) = Qa (t) or Qb (t) = Qa (t) + tλ−1 . This finishes the
inductive step, and the theorem follows.
If M has only finitely many critical points of each index, we can let µλ denote
the number of critical points for f of Morse index λ and let βλ be the dimension
of H i (M; F ). If we define the Morse series for f and Poincaré series for M by
∞
X ∞
X
M(t) = µλ tλ , P(t) = βλ tλ ,
λ=0 λ=0

it follows from Theorem 2.9.1 that there is a polynomial

∞
X
Q(t) = q λ tλ with qλ ≥ 0 such that M(t) − P(t) = (t + 1)Q(t). (2.33)
λ=0

This is a statement of the Morse inequalities for f : M → [0, ∞).

Corollary 2.10.2. (Lacunary Principle) If µaλ 6= 0 ⇒ µaλ−1 = 0 = µaλ+1 ,
then µaλ = βλa , for every nonnegative integer λ.
a a
Indeed, by the weak Morse inequalities, βλ−1 = 0 = βλ+1 , and hence the
a
coefficients of λ − 1 and λ + 1 in (t + 1)Q (t) must be zero. From this we
conclude that Qa (t) ≡ 0.
Similarly, we can let a → ∞ and obtain the lacunary principle for f on M: If
µλ 6= 0 ⇒ µλ−1 = 0 = µλ+1 , then µλ = βλ , for every nonnegative integer λ.

102
Example 2.10.3. Suppose that M = Sn , the n-dimensional sphere with metric
of constant curvature one, where n ≥ 3, and that p and q are points in M which
are not antipodal. Then there is exactly one geodesic from p to 1 of Morse index
k(n − 1), for each k ∈ N ∪ {0}, and hence
M(t) = 1 + tn−1 + t2(n−1) + · · · + tk(n−1) + · · · .
From the “feedback” implicit in the Morse inequalities (2.33), we see that
M(t) = P(t), so
P(t) = 1 + tn−1 + t2(n−1) + · · · + tk(n−1) + · · · ,
and we conclude that for any field F ,
(
n ∼ F, if m = k(n − 1) or k ∈ N ∪ {0},
Hm (Ω(S , p, q); F ) =
0, otherwise.

Thus if h·, ·i is any Riemannian metric on Sn , then for generic choice of p and q
in Sn , there will be infinitely many geodesics from p to q, at least one being of
index k(n − 1) for each k ∈ N ∪ {0}.
We would also like Morse inequalities for the case where M = L21 (S 1 , M ), where
M is a compact Riemannian manifold. If we give M a “generic” Riemannian
metric, Theorem 2.7.5 implies that all nonconstant critical points for the action
J : L21 (S 1 , M ) → R lie on one-dimensional nondegenerate critical submanifolds.
We can suppose that the metric is chosen so that only one O(2)-orbit of critical
points lies at each critical level.
Thus suppose that the interval [a, b] contains a unique critical value c with
J −1 (c) containing a unique O(2)-orbit of geodesics which comprise a nonde-
generate critical submanifold N . Then Theorem 2.8.6 gives an isomorphism on
cohomology
Hk (Mb , Ma ; Z) ∼
= Hk (D(ν− N ), S(ν− N ); Z).
It follows from the Thom isomorphism theorem (with twisted coefficients) that
if the Morse index of N is λ, then
Hk (D(ν− N ), S(ν− N ); Z) ∼
= Hk−λ (N ; Z ⊗ θ− ),
where θ− is the orientation bundle of N . If the bundle ν− N is orientable, twisted
homology reduces to usual homology, and hence
Hk (D(ν− N ), S(ν− N ); Z) ∼
= Hk−λ (N ; Z).
If N is a nondegenerate critical submanifold for J, we let
X
PN (t) = [dim Hλ (N ; θ− ⊗ F )]tλ .
λ

In the case where the normal bundle is orientable, this is the Poincaré polynomial
of N . If F = R and N is a nondegenerate O(2)-orbit of geodesics, then
PN (t) = 2(1 + t) or PN (t) = 0,

103
depending upon whether the normal bundle is orientable or not, while if F = Z2 ,
we always have PN (t) = 2(1 + t). The Morse polynomial of J on Ma is
X
Ma (t) = {tλN PN (t) : N is a nondegenerate critical submanifold
of Morse index λN with J(N ) ≤ a }.

As in the case of a Morse function, the Morse series is related to the Poincaré
polynomial of Ma ,
∞
X
P a (t) = dim Hλ (Ma ; F )tλ ,
λ=0

by the Morse inequalities, which state that there is a polynomial

∞
X
Qa (t) = qλa tλ with qλ ≥ 0 such that Ma (t) = P a (t) + (t + 1)Qa (t).
i=1
(2.34)
If we know the homology (or cohomology) of L21 (S 1 , M )a , the Morse inequalities
(2.34) enable us to estimate the number of smooth closed geodesics M must have
with action ≤ a when M is given a generic metric.

2.11 The Morse-Witten complex

The Morse inequalities do not usually completely determine the integer homol-
ogy H∗ (Mb ; Z). The additional information we would need for an inductive
determination of H∗ (Mb ; Z) is the boundary map in the long exact sequence

· · · → H∗ (Ma ; Z) → H∗ (Mb ; Z) → H∗ (Mb , Ma ; Z) → · · · .

However, implicit in the writings of Thom, Smale, Milnor and others, is a proce-
dure for calculating the boundary map by determining the trajectories between
critical points. This results in a chain complex (C∗ (f, X ), ∂) that depends on
the function f and on a gradient-like vector field X used to calculate trajectories
between critical points, a chain complex which calculates the homology of the
manifold Mb . This chain complex is often called the Morse-Witten chain com-
plex , because of the fact that Witten gave an important quantum-mechanical
interpretation of the boundary operator [83]. We merely sketch the ideas of the
construction here; a much more complete treatment can be found in [74] and
Chapter 6 of [40].
For the statement of the next theorem, we assume the reader is familiar with
the notion of CW complex; background on this topic can be found in Chapter 0
of [35].
Theorem 2.11.1. If f : M → [0, ∞) is a Morse function on a complete Hilbert
manifold that satisfies condition C and all of its critical points have finite index,
then for each a ∈ R, Ma has the homotopy type of a finite CW complex with
one cell of dimension λ for each critical point of index λ.

104
Sketch of proof: This is a consequence of Theorem 2.9.4 and it is an analog of
Theorem 3.5 in [50]. In fact, one could prove that M itself has the homotopy
type of a CW complex, but a rigorous proof would require a limiting procedure
as a → ∞. Such a procedure is given in Appendix A of [50].
Thus the homology of M should be just the cellular homology of the resulting
CW complex, as described in books on algebraic topology; see, for example,
[35], pages 137-141. This raises the problem of describing the cells and attaching
maps, and finding an algorithm for calculating this homology.
Such an algorithm is based upon choice of a gradient-like vector field X for
the Morse function f : M → R. Let {φt : t ∈ R} be the one-parameter group
of local diffeomorphisms corresponding to −X .
Definition. The unstable manifold Wp (f, X ) of a critical point p ∈ M consists
of the images of all trajectories t 7→ φt (q) which start at p in the remote past,
in other words, such that φt (q) → p as t → −∞. Similarly, the stable manifold
Wp∗ (f, X ) of a critical point p consists of the images of trajectories t 7→ φt (q)
such that φt (q) → p as t → ∞.
Lemma 2.11.2. The unstable and stable manifolds Wp (f, X ) and Wp∗ (f, X )
are in fact submanifolds of M.
Sketch of proof: Let us consider the case of the unstable manifold Wp (f, X ).
First one uses the explicit description (2.30) of the gradient-like vector field X
near p to show that for  > 0 sufficiently small,

Wp (f, X ) = φ−1
p ({x ∈ E− : kxk < })

is part of the unstable manifold. Then one notes that

[
Wp (f, X ) = {φt (Wp (f, X ) ) : t ≥ 0}.

The properties of smooth flows corresponding to vector fields then show that
Wp (f, X ) is a smooth manifold diffeomorphic to an open cell. The proof for
Wp∗ (f, X ) is similar, starting with

Wp∗ (f, X ) = φ−1

p ({x ∈ E+ : kxk < }) .

It follows from Condition C and the explicit form of X that any orbit q 7→ φt (q)
converges to some critical point as t → ∞ (although it may come close to several
other critical points first).
Lemma 2.11.3. We can adjust the gradient-like vector field for f so that if λp
is the index of p and λq is the index of q,
1. λp ≤ λq ⇒ Wp (X ) ∩ Wq∗ (X ) is empty, while
2. λp > λq ⇒ Wp (X ) ∩ Wq∗ (X ) is a submanifold of dimension λp − λq , when
X is sufficiently smooth.

105
Sketch of proof: If Wp (X ) ∩ Wq∗ (X ) is nonempty, then f (p) > f (q) and we can
choose a regular value c for f such that f (p) > c > f (q). Then N = f −1 (c) is
a codimension one submanifold of M, and

dim (N ∩ Wp (X )) = λp − 1, codim (N ∩ Wq∗ (X )) = λq .

Moreover, if we choose c sufficiently close to the Morse nondegenerate critical

point p, we see that S = N ∩ Wp (X ) is a (λp − 1)-dimensional sphere and lies
in an open tubular neighborhood U of N with diffeomorphism ψ : U → S × V ,
where V is an open ball in a Banach space, with ψ(S) = S × {0}. Let

N = U ∩ Wq∗ (X ), and consider g = π ◦ ψ : N → V,

where π is the projection on the second factor. We can check that g is a Fredholm
map with Fredholm index λp − 1 − λq , and

ψ −1 (S × {x}) ∩ N = {s ∈ N : g(s) = x}.

Assuming that X (and hence N ) is sufficiently smooth, we can use the Sard-
Smale Theorem 2.6.2 to choose a regular value x for g. Finally, we can choose
x as close as we want to 0, and construct an isotopy from a neighborhood of
S to itself which carries S × {0} to S × {x}. Since the conditions defining
pseudogradient are open, we can replace X by a new gradient-like vector field
so that
N ∩ Wp (X ) and N ∩ Wq∗ (X )
have transverse intersection. If λp ≤ λq , the dimension of N ∩ Wp (X ) is less
than the codimension of N ∩ Wq∗ (X ), so Wp (X ) ∩ Wq∗ (X ) is empty. If λp > λq ,
then N ∩ Wp (X ) ∩ Wq∗ (X ) is a submanifold of dimension λp − λq − 1.
In particular, if λp = λq + 1, then N ∩ Wp (X ) ∩ Wq∗ (X ) consists of a finite
number of points and Wp (X ) ∩ Wq∗ (X ) consists of a finite collection of smooth
curves from p to q.
We can now orient the unstable manifolds and define the Morse-Witten
complex of the nonnegative Morse function f . We let Cλ+1 (f, X ) be the free
Z-module generated by the critical points pλ+1,1 , pλ+1,2 . . . of f of index k and
let ∂ be the Z-module homomorphism

∂ : Cλ+1 (f, X ) −→ Cλ (f, X )

defined by X
∂(pλ+1,j ) = ajq pλ,q , (2.35)
q

where ajq ∈ Z is the oriented number of trajectories from pλ+1,j to pλ,q . The
sign of a trajectory γ : (−∞, ∞) → M from p to q is determined as follows:
First one orients each unstable manifold for f . Then one constructs a trivial
vector bundle E along γ which is transverse to T Wq∗ (X ) and of complementary
dimension such that the fiber Eγ(t) approaches a hyperplane in the negative

106
eigenspace Tp M− for d2 f (p) as t → −∞, and the positive eigenspace for d2 f (q)
as t → ∞. Using this bundle we translate the oriented negative eigenspace
Tq Wq (X ) for d2 f (q) back to p. If Tq M− denotes the resulting hyperplane in
Tp Wp (X ), then
Tp Wp (X ) = Tq M− ⊕ T,
where T is the oriented line in the direction of the trajectory leaving p. We
assign the positive sign to the trajectory γ if the orientation of Tp Wp (X ) agrees
with the direct sum orientation, the minus sign otherwise.
The next theorem says that (C∗ (f, X ), ∂) is a chain complex:
Theorem 2.11.4. Suppose that f : M → R is a nonnegative Morse function
on a complete Hilbert manifold that satisfies condition C and all of the critical
points of f have finite index. Then ∂ ◦ ∂ = 0 and the homology (with integer co-
efficients) of the Morse-Witten complex (C∗ (f, X ), ∂) is just the usual homology
H∗ (M; Z).
Sketch of proof: It suffices to prove this for the restriction of f to Ma which
has only finitely many critical points, and we henceforth use the notation M for
Ma . After generic choice of X , follows from Lemma 2.10.3 that unstable critical
points of index λ intersect only stable manifolds of critical points of index < λ.
Thus if we let M(λ) denote the union of the unstable manifolds in M of index
≤ λ, we obtain an increasing filtration

· · · ⊂ M(λ−1) ⊂ M(λ) ⊂ M(λ+1) ⊂ · · ·

of M by closed subsets. Since the unstable manifolds of index λ are in one-to-

one correspondence with generators of

H∗ (M(λ) , M(λ−1) ; Z),

we can set C∗ (f, X ) equal to this homology group. If we let ∂ 0 denote the
boundary homomorphism in the exact sequence of the triple

(M(λ) , M(λ−1) , M(λ−2) ),

one can use the usual argument from the theory of cellular homology to show
that ∂ 0 ◦ ∂ 0 = 0 and the cohomology of the Morse complex is the standard
homology of M (as in the proof of Theorem 2.35 in [35]). Thus it is relatively
easy to see that we do indeed get a chain complex, and we have reduced the
proof to the verification that ∂ 0 is the same as the homomorphism ∂ defined by
(3.44).
For this, one can follow arguments used by Milnor ([51]) to prove the h-
cobordism theorem of Smale. Note we can adjust the values of the Morse func-
tion f while simultaneously multiplying the gradient-like vector field X by a
positive function η : Ma → (0, ∞). In this way, we can replace the Morse
function f by “self-indexing” function f ∗ which satisfies the conditions f ? ≥ −1
and f ? (p) = λ, whenever p is a critical point of index λ, by following the proof

107
of Theorem 4.8 in [51], without changing critical points or the boundary maps
in the Morse-Witten complex. After this is done, M(λ) is a strong deformation
retract of  
1
Mλ+1/2 = q ∈ M : f ∗ (q) ≤ λ + .
2
To complete the sketch of the proof, we need only verify the following lemma:
Lemma 2.11.5. The Witten boundary ∂ agrees with the boundary ∂ 0 defined
by the CW decomposition.
The proof is a straightforward modification of the proof of Theorem 7.4 in [51],
to which we refer for details. Here we give only a very superficial description of
how the argument goes. The key idea is that if

N = (f ∗ )−1 (λ + 1/2),

then the CW boundary ∂ can be regarded as a composition

Hλ+1 (Mλ+3/2 , Mλ+1/2 ; Z) −→ Hλ (N ; Z) −→ Hλ (Mλ+1/2 , Mλ−1/2 ; Z)

where the first map takes the homology class corresponding to the (λ + 1)-
handle for a critical point pλ+1,j to the homology class of its boundary sphere
Sλ+1,j ⊆ N , and the second map is induced by inclusion. On the cochain level,
we have a corresponding factorization of the coboundary

H λ (Mλ+1/2 , Mλ−1/2 ; R) −→ H λ (N ; R) −→ H λ+1 (Mλ+3/2 , Mλ+1/2 ; R),

where we use real coefficients so that we can represent cycles by differential

forms. Corresponding to a critical point pλ,q we can define a “Thom form” θλ,q
(as described in [10]) which represents a cohomology class

[θλ,q ] ∈ H λ (Mλ+1/2 , Mλ−1/2 ; R)

such that Z (
1 if r = q,
θλ,q = δrq =
Wpλ,r 0 6 q.
if r =
We can think of θλ,q as Poincarë dual to the class represented by the inifinite-
dimensional stable manifold for pλ,q . One now checks that
Z
ajq = θλ,q ∈ Z
Sλ+1,j

is both the integer appearing in the Witten boundary (3.44) and the integer
appearing in the formula for the CW boundary.
Corollary 2.11.6. (Lacunary Principle) If µaλ 6= 0 ⇒ µaλ−1 = 0 = µaλ+1 ,
then the boundary ∂ in the Morse-Witten chain complex is zero.
The proof is immediate.

108
Note that applying Corollary 2.11.6 to Example 2.10.3. yields the integer ho-
mology of the loop space of Sn when n ≥ 3:
(
n ∼ Z, if m = k(n − 1) or k ∈ N ∪ {0},
Hm (Ω(S , p, q); Z) =
0, otherwise.

In spite of the fact that the Morse-Witten complex gives stronger results, the
Morse inequalities are often quite useful, since in many cases it is difficult to
calculate the boundary operator ∂.

109
Chapter 3

Harmonic and minimal

surfaces

3.1 The energy of a smooth map

Suppose now that Σ is a compact smooth Riemannian manifold of dimension m.
In terms of local coordinates (u1 , . . . , um ) on Σ we can write the Riemannian
metric and area or volume element on Σ as
m
X √
ηab dua ⊗ dub and dA = ηdu1 · · · dum ,
a,b=1

where η denotes the determinant of the matrix (ηab ). If M is a second Rieman-

nian manifold of dimension n isometrically imbedded in Euclidean space RN
and f : Σ → M ⊆ RN is a smooth map, the energy density of f at a given point
is given in terms of local coordinates by the formula
m
1 X ∂f ∂f
e(f ) = |df |2 , where |df |2 = η ab · ,
2 ∂ua ∂ub
a,b=1

ab
where (η ) denotes the matrix inverse to (ηab ) and the dot denotes the Eu-
clidean dot product in RN . The energy of a smooth map f : Σ → M ⊆ RN is
given by the Dirichlet integral
m
∂f ∂f √
Z Z X
1
E(f ) = e(f )dA = η ab a · ηdu1 · · · dum , (3.1)
Σ 2 Σ ∂u ∂ub
a,b=1

the integrand being independent of choice of local coordinates. Note that if Σ

is one-dimensional, the energy reduces to the action J that we studied in the
previous chapter.
The energy defines a smooth map
E : C 2 (Σ, M ) −→ R.

110
To find the critical points of E, we consider a variation of a given map f which
has its support within a coordinate chart (U, (u1 , . . . , um )) on Σ. Such a varia-
tion is simply a smooth family of maps t 7→ f (t) in C 2 (Σ, M ) such that f (0) = f
and f (t)(p) = f (p) for all t, when p ∈ Σ − U . In terms of the local coordinates,
we set
α(u1 , . . . , um , t) = f (t)(u1 , . . . , um ),
and a straightforward calculation shows that
  
 m 2

√
Z
d X ∂ α ∂α  1
ηη ab m
 
(E(f (t)) = 
a
· b
du · · · du
dt t=0 Σ a,b=1 ∂t∂u ∂u 

  t=0 
m   
√ ab ∂α  1
Z
∂α  X ∂ m

=− · a
ηη b
du · · · du .
Σ ∂t ∂u ∂u 
a,b=1 
t=0

We can evaluate at t = 0, setting

∂α 1
V (u1 , . . . um ) = (u , . . . um , 0),
∂t
to obtain the first variation formula,
 
m  
√ ab ∂f 
Z
1 X ∂
dE(f )(V ) = − V · √ ηη dA. (3.2)
Σ η ∂ua ∂ub
a,b=1

If f is a critical point for the energy E, then dE(f )(V ) = 0 for all such variations
V , and f must satisfy the partial differential equation

 >
 
m 
X 1 ∂ √ ∂f
 √ ηη ab b  = 0, (3.3)
η ∂ua ∂u
a,b=1

where (·)> denotes projection into the tangent space to M . Maps f ∈ C 2 (Σ, M )
which satisfy equation (3.3) are called harmonic maps. Just like the equation for
geodesics, the equation for harmonic maps is nonlinear because of the projection
into the tangent space. In fact, we can rewrite (3.3) in terms of the Levi-Civita
connection D on M as
m  
1 X D √ ab ∂f
√ ηη = 0,
η ∂ua ∂ub
a,b=1

or alternatively, in terms of local coordinates (x, . . . xn ) on M , we can write the

equation of harmonic maps as
m
√ ab ∂xi ∂xj ∂xk
  X
1 X ∂
√ a
ηη b
+ Γijk η ab a b = 0,
η ∂u ∂u ∂u ∂u
a,b=1

111
where the Γijk ’s are the Christoffel symbols.
Harmonic maps between Riemannian manifolds were introduced by Eells
and Sampson [18] in 1964, who used the heat flow approach to establish the
following theorem:
Theorem 3.1.1 (Eells and Sampson). If M is a compact connected Rie-
mannian manifold with nonpositive sectional curvatures and Σ is a compact
Riemannian manifold, then every component of C 2 (Σ, M ) contains a minimal
energy representative, which is a smooth harmonic map.
A very nice proof of this theorem using heat flow can be found at the beginning
of Chapter 5 of [45]. A fews years after Eells and Sampson, Hartman [34]
established a uniqueness result:
Theorem 3.1.2 (Hartman). If Σ and M are compact connected Riemannian
manifolds and M has negative sectional curvatures, then any component of
C 2 (Σ, M ) contains at most one harmonic map.
One should note, however, that if M has nonpositive sectional curvature, the
Hadamard-Cartan Theorem asserts that the exponential map expp : Tp M → M
is a smooth covering at any point p ∈ M , so M has Euclidean space as its
universal cover, and all of its higher homotopy groups are zero. In this case, M
is a K(π, 1) and its topology is relatively simple.
The Eells-Sampson Theorem does not hold without the curvature assump-
tion. Indeed, it fails spectacularly when the dimension of Σ is ≥ 3, and attempts
to minimize energy within a given homotopy class of maps can lead to elements
of L21 (Σ, M ) which have quite bad singularities, as discussed for example in [45].
Later we will prove the above theorems in the case where Σ has dimension
two, as part of a general theory that will give existence results even when we
make no assumptions on the curvature of the range M . In the case where Σ has
dimension two, the energy is conformally invariant, and the theory of harmonic
maps simplifies considerably due to the existence of isothermal coordinates on
Σ. A nice proof of existence of isothermal charts, based upon Hodge theory, can
be found in Chapter 5, §10 of [79]. According to this existence theorem, any two-
dimensional Riemannian manifold possesses an atlas {(Uα , (uα , vα )) : α ∈ A}
consisting of isothermal charts, charts such that the Riemannian metric on Σ
takes the form
ds2 = λ2α (duα ⊗ duα + dvα ⊗ dvα ),
where each λα is a positive smooth real-valued function on Uα . If {ψα : α ∈ A}
is a partition of unity subordinate to this atlas, then the energy of a C 2 map
f : Σ → M is given by the formula
Z    
1X ∂f ∂f ∂f ∂f
E(f ) = ψα , + , duα dvα , (3.4)
2 α Σ ∂uα ∂uα ∂vα ∂vα

where h·, ·i is the Riemannian metric on M induced by the dot product on RN ,

112
and harmonic maps are simply the maps
T
∂2f ∂2f

f : Σ −→ M such that + = 0. (3.5)
∂u2α ∂vα2

If one takes two coherently oriented isothermal charts (Uα , (uα , vα )) and
(Uβ , (uβ , vβ )), one finds by a short calculation that on the intersection Uα ∩ Uβ ,

∂uα ∂vα ∂uα ∂vα

= , =− .
∂uβ ∂vβ ∂vβ ∂uβ

These are just the Cauchy-Riemann equations which state that wα = uα + ivα
is a holomorphic function of wβ = uβ + ivβ . Thus if Σ is an oriented surface,
the atlas {(Uα , zα ) : α ∈ A} of positively oriented isothermal charts makes Σ
into a one-dimensional complex manifold, otherwise known as a Riemann sur-
face. In this way, Riemannian metrics on Σ are divided into equivalence classes,
depending upon which Riemann surface structure is defined by the isothermal
charts.
It is often convenient to choose a canonical metric on Σ within the conformal
equivalence class determined by a Riemann surface structure. The uniformiza-
tion theorem from Riemann surface theory states that any compact Riemann
surface has as its universal cover one of three simply connected Riemann sur-
faces,
S 2 = C ∪ {∞}, C or D = {z ∈ C : |z| < 1}.
The remarkable fact is that each of these model spaces possesses a Riemannian
metric of constant Gaussian curvature compatible with its conformal structure,
and the deck transformations of the universal cover of the compact Riemann
surface Σ are isometries with respect to this Riemannian metric. If Σ is closed,
we can normalize this metric by assuming that it has total area one, and it is
then unique up to diffeomorphism. It follows from the Gauss-Bonnet formula
that the sign of the curvature is positive, zero or negative when Σ has genus
zero, one, or at least two, respectively.
In terms of the complex partial differential operators
√ √
   
∂ 1 ∂ ∂ ∂ 1 ∂ ∂
= − −1 , = + −1
∂w 2 ∂u ∂v ∂ w̄ 2 ∂u ∂v

we can write (3.3) as

T
∂2f
  
D ∂f
= = 0. (3.6)
∂ w̄ ∂w ∂w∂ w̄

Note that we can regard

∂f
as a section of E = f ∗ T M ⊗ C,
∂w

113
a complex vector bundle over the Riemann surface Σ. The vector bundle E has
a Hermitian metric defined by

v, w ∈ Tp M ⊗ C 7→ hv, w̄i,

where w̄ is the conjugate of w, and the Levi-Civita connection D on f ∗ T M

extends complex linearly to a metric connection on E. The following theorem
states that this connection gives E a canonical holomorphic structure, thereby
making possible a remarkable relationship between harmonic surfaces and com-
plex analysis:
Theorem 3.1.3 (Koszul and Malgrange). If E is a complex vector bundle
with Hermitian metric over a Riemann surface Σ and D is a metric connection
on E, then there is a unique holomorphic structure on E such that if σ is a
section of E,
Dσ
σ is holomorphic ⇔ = 0,
∂ w̄
whenever w is a complex coordinate on Σ.
We will not prove this theorem here. An extension of this theorem is proven as
Theorem 5.1 in Atiyah, Hitchin and Singer [5] using the Newlander-Nirenberg
theorem on integrability of almost complex structures. Another proof is given
in Donaldson and Kronheimer [14], Theorem 2.1.53.
Thus we see that the equation for harmonic maps from oriented surfaces can
be expressed quite simply in terms of Riemann surface theory: Equation (3.6)
asserts that a map f : Σ → M is harmonic if and only if in terms of any complex
coordinate w on Σ, the section
∂f
of E = f ∗ T M ⊗ C
∂w
is holomorphic with respect to the holomorphic structure on E which is provided
by the Koszul-Malgrange Theorem.
The locally defined holomorphic section
 
∂f ∂f
may have isolated singularities, but : Σ → P(E)
∂w ∂w

extends to all of Σ, where P(E) denotes the bundle of projective spaces of fibers
of E = f ∗ T M ⊗ C. Under change of complex coordinate
∂f ∂f ∂z ∂f
= = (complex-valued function) ,
∂w ∂z ∂w ∂z
and hence the various locally defined complex derivatives of f determine a com-
plex line subbundle L of E, which is holomorphic by the Koszul-Malgrange
Theorem. The line bundle L is isomorphic to the tangent bundle to Σ if and
only if f has no branch points, in accordance with the following definition:

114
Definition. A point p ∈ Σ is a branch point for the harmonic map f : Σ → M
if (∂f /∂z)(p) = 0, where z is any complex coordinate near p.
If p is a branch point for f , and z is a local complex coordinate defined on a
small open neighborhood U of p with z(p) = 0, then we can write (∂f /∂z) = z ν g
for some positive integer ν, where g is a section of L over U such that g(p) 6= 0.
We call ν the order or multiplicity of the branch point. If we let w = z ν+1 , then
dw = (ν + 1)z ν dz and we can define a section
∂ ∂ 1 ∂
of L by = .
∂w ∂w (ν + 1)z ν ∂z
Thus we see that the restriction of L to U is obtained from the holomorphic
tangent bundle T Σ|(U − {p}) by pasting it to a trivial bundle U × C over U by
means of the transition function
1
g0p = : U − {p}) −→ C − {0}).
(ν + 1)z ν

From the trivial bundle over Σ − {p}) and this transition function one can
construct the point bundle ζpν over Σ which has first Chern number c1 (ζpν )([Σ]) =
ν. If f has a single branch point p of branching order ν then L = T Σ ⊗ ζpν .
In general, the divisor of the harmonic map f is the finite sum

(f ) = ν1 p1 + · · · + νn pm ,

where p1 , . . . , pm are the branch points of f and ν1 , . . . , νm are their branching

orders. Then the above discussion shows that

L = T Σ ⊗ ζpν11 ⊗ · · · ⊗ ζpνm
m
.

If ν denotes the total branching order of f , the total number of branch points of
f , counted with multiplicity, then the first Chern number hc1 (L), [Σ]i of the line
bundle L (also known as the degree of L in many books on Riemann surfaces)
is determined by the formula

hc1 (L), [Σ]i = 2 − 2g + ν (3.7)

where g is the genus of Σ.

Suppose, for example, that h : Σ2 → M is a smooth harmonic map without
branch points and that g : Σ1 → Σ2 is a nontrivial holomorphic branched cover.
(Thus, if Σ2 = S 2 , the Riemann sphere, we can regard g as a meromorphic
function on Σ1 .) Then the composition f = h ◦ g : Σ1 → M is also harmonic;
it is called a branched cover of the harmonic map h : Σ2 → M . We say that
the harmonic map f is prime if it is not a nontrivial branched cover of another
harmonic map. Note that if f : Σ → M is a k-fold branched cover of a harmonic
sphere h : S 2 → M which is free of branch points, then the line bundle L for f
satisfies
hc1 (L), [Σ]i = 2k.

115
3.2 Minimal surfaces

The theory of minimal surfaces in a Riemannian manifold M is concerned with

critical points of the area function
Z  
 ∂f ∂f 
A : Map(Σ, M ) −→ R defined by A(f ) =  ∧  dudv,
Σ ∂u ∂v 


which we will see is closely related to critical points for the energy function E
defined by (3.4). In contrast to E, the integrand for A is independent of the
choice of parametrization, and in particular does not depend upon the choice of
a metric or conformal structure.
Proposition 3.2.1. If Σ is given a conformal structure ω and f ∈ Map(Σ, M ),
then E(f ) ≥ A(f ), with equality holding if and only if f satisfies the conditions
     
∂f ∂f ∂f ∂f ∂f ∂f
, = , , , = 0, (3.8)
∂u ∂u ∂v ∂v ∂u ∂v
when w = u + iv is any choice of complex chart, or equivalently,
 
∂f ∂f
, = 0, (3.9)
∂w ∂w
and h·, ·i denotes the Riemannian metric on the ambient manifold M .
Proof: We utilize two well-known algebraic identities for vectors v and w in
RN :
1
|v ∧ w|2 + |v · w|2 = |vk2 |w|2 , v|2 + |w|2 ,


|v||w| ≤
2
with equality holding only if |v| = |w|. Using these two facts, we find that
  !
 ∂f ∂f   ∂f   ∂f  1  ∂f 2  ∂f 2
      
 ∂u ∧ ∂v  ≤  ∂u   ∂u  ≤ 2  ∂u  +  ∂v  , (3.10)
        

with equality holding if and only if (3.8) holds. This proves the proposition.
Definition. We say that a map f : Σ → M is weakly conformal with respect
to a Riemann surface structure ω on Σ if it satisfies (3.9) when w = u + iv is a
complex coordinate for ω.
Weak conformality of an harmonic map f : Σ → M is expressed by the vanishing
of the Hopf differential , which is the quadratic differential
 
∂f ∂f
Ω(f ) = , dw ⊗ dw. (3.11)
∂w ∂w
It follows immediately from (3.6) that
   
∂ ∂f ∂f D ∂f ∂f
, =2 , = 0,
∂ w̄ ∂w ∂w ∂ w̄ ∂w ∂w

116
so the Hopf differential of a harmonic map is indeed a holomorphic quadratic
differential, as studied in Riemann surface theory.
Theorem 3.2.2. A weakly conformal harmonic map f : Σ → M is a critical
point for the area function A.
Proof: If f is weakly conformal and harmonic, and t 7→ f (t) is a smooth family
of maps with f (0) = f , then

E(f (t)) ≥ A(f (t)) and E(f (0)) = A(f (0))

 
d  d 
⇒ E(f (t))
 = A(f (t)) ,
dt t=0 dt t=0

so f is critical for the area function.

Theorem 3.2.2 provides motivation for the following definition of parametrized
minimal surface:
Definition. A parametrized minimal surface is a weakly conformal harmonic
map f : Σ → M .
Note that if f is an immersion, we could give Σ the Riemannian metric which
makes f an isometric immersion. This gives Σ a Riemann surface structure
which makes f is weakly conformal, and with this conformal structure, f is
harmonic if and only if it is minimal.
More generally, if f : Σ → M is a weakly conformal map, then the com-
plex line bundle L we described in the preceding section is “isotropic,” that is,
hL, Li = 0. This constrains the possible singularities of f ; the differential dfp at
any point p ∈ Σ has rank two if f is an immersion in a neighborhood of p, or
rank zero if p is a branch point, but never has rank one.
We would like to consider minimal surfaces as critical points for a variational
problem which involves the Dirichlet integral and the conformal structure, rather
than the area function. In order to do this, we must allow the conformal struc-
ture on Σ to vary, as well as the map f ∈ C 2 (Σ, M ). In this and the following
section, we suppose that Σ is a compact oriented connected surface without
boundary.
The simplest case is that where Σ = S 2 , which has a unique conformal
structure by the uniformization theorem. We can regard Σ = S 2 as C ∪ {∞}
and take the atlas defined by the standard coordinate z on C and the coordinate
w = 1/z on S 2 − {0}. Then
∂f ∂f ∂w 1 ∂f
= =− 2 ,
∂z ∂w ∂z z ∂w
and
∂f ∂f
bounded near ∞ ⇒ → 0 like 1/z 2 as z → ∞. (3.12)
∂w ∂z
By the Koszul-Malgrange Theorem 3.1.3, we can regard ∂f /∂z as a holomorphic
section on S 2 − {∞} with a removeable singularity at ∞, and by the removeable

117
singularity theorem from complex analysis, ∂f /∂z extends to a holomorphic
vector field on S 2 . Thus when calculating energy on the Riemann sphere, we
are fully justified in using a single coordinate z = x + iy on C = S 2 − {∞}, and
the energy can be expressed by the improper integral
Z    
1 ∂f ∂f ∂f ∂f
E(f ) = , + , dxdy.
2 C ∂x ∂x ∂y ∂y
Conversely, if f : C → M is a harmonic map of finite energy, in other words,
such that the above integral is finite, a removeable singularity theorem of Sacks
and Uhlenbeck (Theorem 3.6 in [68]) will show that f extends to a harmonic
map from S 2 into M .
Proposition 3.2.3. A harmonic map f : S 2 → M is automatically weakly
conformal, and hence a parametrized minimal surface.
To prove this, we let z = x + iy be the standard complex coordinate on C =
S 2 − {∞}. Then it follows from (3.6) that
       
∂ ∂f ∂f D ∂f ∂f ∂f ∂f
, =2 , = 0, so ,
∂ z̄ ∂z ∂z ∂ z̄ ∂z ∂z ∂z ∂z
is a holomorphic function which extends to a C 2 function on S 2 by (3.12).
This function must be constant by the maximum modulus principle, and since
it vanishes at ∞, the constant must be zero, establishing the assertion. Note
that the argument simply shows that the Hopf differential (3.11) is zero, so one
could prove the Proposition by simply citing the well-known fact from Riemann
surface theory that S 2 has no nonzero holomorphic quadratic differentials.
We next consider the case where Σ = T 2 = S 1 × S 1 , a torus. We imagine
that we have fixed a basis (α, β) for H1 (T 2 ; Z), representing the two S 1 factors.
The covering transformations for the universal cover π : C → T 2 are invertible
holomorphic maps from C to itself, and it follows from the maximum modulus
principle they must be translations. After performing a possible rotation of
C and a uniform stretching, we can arrange that one of the translations be
horizontal with unit displacement. The other displacement can be represented
by a point ω = u + iv in the upper half-plane
H = {ω = u + iv ∈ C : v > 0}.
Thus we can arrange that the conformal structures on T 2 , for fixed a basis for
H1 (T 2 ; Z), are in one-to-one correspondence with points in the upper half-plane
H, the point ω = u + iv ∈ H corresponding to the conformal class of the torus
C/Λ, where Λ is the lattice in C generated by ω and 1, with α corresponding
to ω and β corresponding to 1. We say that H is the Teichmüller space of the
torus and write T1 = H.
A change of basis for the lattice Λ is represented by multiplication by a
matrix in SL(2, Z),
      
α a b α a b
7→ for ∈ SL(2, Z). (3.13)
β c d β c d

118
The corresponding SL(2, Z)-action on the generators ω1 and ω2 of an arbitrary
integral lattice in the complex plane is
    
ω1 a b ω1
7→
ω2 c d ω2

while the corresponding action on the coordinate ω = u + iv = ω1 /ω2 is

aω + b
ω 7→ . (3.14)
cω + d
Note that while the action of the mapping class group SL(2, Z) on H 2 (T 2 , Z) is
faithful, the action on the upper half plane H has kernel consisting of ±I. By
applying a suitable element of SL(2, Z), we can arrange that ω = u + iv lie in
the fundamental domain of T1 for this action:

D = {ω = u + iv ∈ C : |u| ≤ 1/2, v > 0, |ω| ≥ 1}.

The moduli space R1 is obtained from this fundamental domain by identify-

ing edges. It can be shown that R1 has a Riemann surface structure and is
holomorphically equivalent to the complex line C.
Thus we have two ways of looking at the Riemann surface structures on T 2 ,
the Teichmüller space T1 when a choice of basis for H1 (T 2 ; Z) is fixed, and the
moduli space R1 when we ignore choice of basis. The discrete action of SL(2, Z)
on T1 determines a branched covering T1 → R1 .
If T 2 = C/Λ, then any map f : T 2 → M can be lifted to a doubly periodic
map f˜ : C → M with periods in the lattice Λ. The standard coordinate z =
x + iy on C descends to local coordinates on T 2 , and the standard metric
dx2 +dy 2 on T 2 descends to a flat metric on T 2 within the conformal equivalence
class corresponding to ω. We can rescale the metric to (1/v)(dx2 + dy 2 ) so that
it has total area one. In terms of the coordinates (t1 , t2 ) on C defined by
(
x = t1 + ut2 ,
y = vt2 ,

the components of this flat metric and its inverse are

1 u2 + v 2
   
1 1 u ab −u
(ηab ) = , (η ) = ,
v u u2 + v 2 v −u 1

while the area element is given by dA = dt1 dt2 . Hence if f ∈ C 2 (T 2 , M ),

" 2  #
 ∂f 2
 
ab ∂f ∂f 1 2  ∂f  ∂f ∂f
X
2 2
 
|df | = η · = (u + v )  1  − 2u 1 · +   .
∂ti ∂tj v ∂t ∂t ∂t2  ∂t2 
a,b

or equivalently,
 ∂f 2 1  ∂f
   2
2 ∂f 
|df | = v  1  +  2 − u 1  ,
  
∂t v ∂t ∂t

119
From this, it is apparent that the energy
Z Z "  2 2 #
1 1 1  ∂f 

1  ∂f ∂f 
E(f, ω) = v  1  +  2 − u 1  dt1 dt2 (3.15)
2 0 0 ∂t v ∂t ∂t

is a smooth function on C 2 (T 2 , M ) × T1 .
In the following proof, it will be helpful to utilize the Euclidean coordinates
x1 = √xv and x2 = √yv , in terms of which the energy of f is expressed as
Z "  #
 ∂f 2  ∂f 2
 
1 1 2
E(f, ω) =  ∂x1  +  ∂x2  dx dx ,
   
2 ∆

where ∆ is a fundamental domain for the lattice Λ.

Proposition 3.2.4. If (f, ω) is a critical point for

E : C 2 (T 2 , M ) × T1 −→ R,

then f is a weakly conformal harmonic map, and hence a parametrized minimal

surface.
Differentiation with respect to the first variable shows that if (f, ω) is a critical
point for E, f must be harmonic. So we need only consider the derivative with
respect to the second variable. If f : T 2 → M is any map, harmonic or not, we
can construct the quadratic differential
Z   
∂f ∂f
Ω(f, ω) = , dx dx dz 2 ,
1 2
T2 ∂z ∂z
which specializes to the Hopf differential of f when f is harmonic. This differ-
ential is closely associated with a symmetric bilinear form

H(f, ω) : T0 T 2 × T0 T 2 −→ R

that we define on the tangent space to the torus at an origin 0 by

Z  
X ∂f ∂f
H(f, ω) = hab dxa dxb , where hab = , dx1 dx2 , (3.16)
T2 ∂xa ∂xb
which is easily seen to be positive definite. The trace of the quadratic form is
2E(f ), so the eigenvalues must be of the form

E(f, ω) ± a, where 0 ≤ a ≤ E(f, ω),

and a short calculation shows that a = |Ω(f, ω)|.

If we change the conformal structure from ω to ω̃, we must change the
Euclidean coordinates by a linear transformation defined by A ∈ SL(2, R),
 1  1  1  1  1
x x̃ x dx̃ dx
7→ = A 2 , and hence =A .
x2 x̃2 x dx̃2 dx2

120
It follows that the matrices H = (hij ) and H̃ = (h̃ij ) corresponding to H(f, ω)
and H(f, ω̃) transform by to the rule,

H = AT H̃A, for A ∈ SL(2, R).

If A ∈ SO(2) this reduces to the transformation for a rotation of Euclidean

coordinates for a fixed conformal structure.
Suppose now that H(f, ω) is degenerate; thus one of its eigenvalues is zero.
We can rotate coordinates so that the x1 direction corresponds to the zero
eigenvalue. Then
Z    
∂f ∂f 1 2 ∂f ∂f
, dx dx = 0 ⇒ , ≡ 0,
T2 ∂x1 ∂x1 ∂x1 ∂x1
and the map f must degenerate to a curve. If f is harmonic with respect to ω,
this curve must be a closed geodesic. The energy of the closed geodesic depends
upon its length and the choice of conformal structure ω. It is easily seen that
no such parametrized geodesic can be a critical point for E.
On the other hand, if H(f, ω) is nondegenerate, there exists an element
A ∈ SL(2, R) such that H = AT H0 A, where H0 is a scalar multiple of the
identity matrix, and hence there is a conformal structure ω0 such that

H(f, ω) = E(f, ω0 )h·, ·i.

Since H0 commutes with A, H = AT H0 A = AT AH0 . We can rotate the

Euclidean coordinates chosen for ω so that AT A and H are diagonal, and then
 λ   λ 
T e 0 e 0
A A= , H = E(f, ω0 ) ,
0 e−λ 0 e−λ

for some choice of λ. Then E(f, ω) = (cosh λ)E(f, ω0 ), and we see that the only
critical points for E are not only harmonic, but also weakly conformal.
As a byproduct of the proof, we see that there are two types of harmonic tori.
Those for which H(f, ω) is degenerate are simply parametrizations of smooth
closed geodesics, and these can never be conformal with respect to any conformal
structure. Those for which H(f, ω) is nondegenerate are either conformal or
their energy can be decreased by replacing the conformal structure ω with a
new conformal structure ω0 .
Note that the differential dz descends from C to the torus C/Λ and hence
holomorphic differentials on a torus must be of the form hdz 2 , where h is a
holomorphic function on the torus. By the maximum modulus theorem, h must
be constant. Hence the only possible Hopf differentials for harmonic tori are
cdz 2 , where c is a complex constant. If the harmonic torus f : T 2 → M is not
weakly conformal, its Hopf differential is never zero, so it cannot have branch
points. However, it may have fold points as the following example shows.
Example 3.2.5. If f is an immersion, the fibers of the line bundles L and L̄
are linearly independent at every point. However, a harmonic map f can have

121
points p at which the fibers L and L̄ coincide. At such points the rank of f can
be at most one.
Indeed, there is a degree zero harmonic map f : T 2 → S 2 when S 2 is given
the standard Riemannian metric of constant curvature one, which has “fold
points” along two circles parallel to the equator. To see how this is constructed,
we first note that the metric on S 2 ⊂ R2 with equation x2 + y 2 + z 2 = 1
is expressed in spherical coordinates z = cos φ, θ being the standard angular
coordinate in the (x, y)-plane, is

ds2 = (cos2 φ)dθ2 + dφ2 = sech2 u(dθ2 + du2 ),

where u and φ are related by the equation tanh(u/2) = tan(φ/2). In terms of

the standard coordinates (t1 , t2 ) on T 2 which correspond to the factorization
T 2 = S 1 × S 1 , the mapping f : T 2 → S 2 can expressed as

θ(t1 , t2 ) = t2 , φ(t1 , t2 ) = φ(t1 ),

where φ is a (nonconstant speed) parametrization of the geodesic θ = (constant).

Note that the circle φ = (constant) has curvature κ = 1/ cos φ and normal
curvature κn = 1. From the equation κ2g + κ2n = κ2 , where κg is the geodesic
curvature, implies that κg = ± tan φ. Moreover, the curve is traversed with
constant speed cos φ. Hence

d2 φ d2 φ 1
   
D ∂f D ∂f
0= 1 1
+ 2 2
= 2 + (tan φ)(cos2 φ) = 2 + sin(2φ).
∂t ∂t ∂t ∂t dt dt 2

The differential equation we must solve to obtain a harmonic map (the pendulum
equation except for constant factors) is equivalent to the first order system
dφ dψ 1
= ψ, = − sin(2φ).
dt dt 2
Eliminating dt yields
dψ (1/2) sin(2φ) 1
=− which integrates to ψ2 − cos(2φ) = (constant).
dφ ψ 2
Various choices of constant yield harmonic maps for appropriate conformal
structures on T 2 .
Note that the antipodal map A : S 2 → S 2 induces an orientation reversing
map A : T 2 → T 2 such that f ◦ A = A ◦ f . We can take the quotient in both
domain and range, obtaining thereby a harmonic map from a Klein bottle into
the real projective plane RP 2 , which has as its image a Möbius band.

3.3 Minimal surfaces of higher genus

Finally, we consider the case where Σ is a sphere with g handles and g ≥ 2.
Classifying the conformal structures on Σ is now somewhat more challenging.

122
We let Metk (Σg ) denote the space of L2k Riemannian metrics
2
X
η= ηab dua dub
a,b=1

on Σ, an open subset of a Hilbert space, and let Metk0 (Σg ) denote the submani-
fold of metrics of constant Gaussian curvature and total area one. It is not diffi-
cult to conclude from the Uniformization Theorem that given any η ∈ Metk (Σg ),
there is a rescaling λ2 η, where λ2 > 0, which lies in Metk0 (Σg ). Moreover, one
can show that this rescaling is unique and we thus obtain a strong deformation
retraction
π : Metk (Σg ) −→ Metk0 (Σg ). (3.17)
In addition, we consider

Diffk+1 2
+ (Σg ) = {φ ∈ Lk+1 (Σ, Σ) such that
φ is an orientation-preserving diffeomorphism },

which is a group under composition, and its subgroup

Diffk+1
0 (Σg ) = {φ ∈ Diffk+1
+ (Σg ) : φ is homotopic to the identity }.

Unfortunately, although Diffk+1 k+1

+ (Σg ) and Diff0 (Σg ) are smooth Hilbert man-
ifolds, the group multiplications are not smooth, so these are not infinite-
dimensional Lie groups. However, the maps

ψ : Diffk+1 k k
+ (Σg ) × Met (Σg ) −→ Met (Σg ), ψ(φ, η) = φ∗ η,

and its restriction

ψ : Diffk+1 k k
+ (Σg ) × Met0 (Σg ) −→ Met0 (Σg ),

are indeed smooth. Moreover, ψ(φ, π(η)) = π ◦ψ(φ, η), where π is the projection
(3.17) into constant curvature metrics. Of course, either of these maps can be
further restricted to the subgroup Diffk+1
0 (Σg ).

Lemma 3.3.1. The continuous action of Diffk+1

0 (Σg ) on Metk0 (Σg ) is free.
Moreover, the image of the map

ψ̃ : Diffk+1 k k k
+ (Σg ) × Met0 (Σg ) −→ Met0 (Σg ) × Met0 (Σg ), ψ̃(φ, η) = (η, ψ(φ, η)),
(3.18)
is closed.
Indeed, if the action of Diffk+1
0 (Σg ) on Metk0 (Σg ) were not free, there would be
two distinct isometries

id, φ : (Σg , φ∗ η) −→ (Σg , η),

123
both homotopic to the identity. Each would be harmonic, contradicting Hart-
man’s Theorem 3.1.2 which asserts that the harmonic map between two compact
surfaces of negative curvature in a given homotopy class is unique. We leave
the second statement as an easy exercise for the reader.
The orbit spaces,

Metk0 (Σg ) Metk0 (Σg )

Tg = and Rg = ,
Diffk+1
0 (Σg ) Diffk+1
+ (Σg )

are called the Teichmüller space and the Riemann moduli space of conformal
structures on Σg , respectively. The fact that the image of the map ψ̃ in (3.18)
is closed is equivalent to the orbit space Tg being Haussdorf.
Theorem 3.3.2. If g ≥ 2, the Teichmüller space Tg is a manifold of dimension
6g − 6.
This is essentially due to Earle and Eells [15], and a modern argument using
Banach manifolds is presented in Fischer and Tromba [20]. Our argument will
mostly follow Chapter 2 of [73]. Our goal is to construct local coordinates in Tg
about a base metric η ∈ Metk0 (Σg ). Let G = Diffk+1
0 (Σg ). According to a well-
known result regarding group actions (namely Theorem 2.9.10 in Varadarajan
[81]), it suffices to construct a submanifold S ⊂ Metk0 (Σg ) such that η ∈ S and

1. Tη Metk0 (Σg ) ∼
= Tη S ⊕ Tη (G · η) and
2. any G-orbit intersects S in only one point.
Such a submanifold is called a slice for the action, and makes it possible to
construct an open neighborhood U of η in S and a diffeomorphism from U × G
to an open neighborhood of the orbit through η.
To construct this slice, we first note that if
2
X ∂
X= Xa
a=1
∂ua

is a smooth vector field on Σ with local one-parameter subgroup {φt : t ∈ R},

it follows from a familiar calculation that
 2
d ∗  X
(φt η) = LX η = · · · = Xa;b dua dub ,
dt t=0 a,b=1

where the Xa;b ’s are the components of the covariant derivative of X with re-
spect to the metric η. We let h·, ·i denote the G-invariant L2 inner product on
Tη Metk (Σg ) such that
Z 2
1 X p
hη̇, η̇i = η̇ab η̇ab du1 du2 and let kη̇k = hη̇, η̇i.
Σ λ2
a,b=1

124
We suppose that we have chosen coordinates so that the components of our base
metric η are ηab = δab λ2 , for some positive function λ2 , and consider when a
one-parameter family of metrics,
ηab (t) = ηab + tη̇ab for t ∈ (, ),
is perpendicular with respect to h·, ·i at t = 0 to the orbit through η. An
integration by parts shows that
Z 2 Z 2
1 X 1 2 1 X
hLX η, η̇i = 2
η̇ ab Xa;b du du = − 2
η̇ab;b Xa du1 du2 ,
Σ λ a,b=1 Σ λ a,b=1
P
so that η̇ is perpendicular to the G-orbit when η̇ab;b = 0. This suggests a
slice for the action of G on the larger space Metk (Σg ), namely
( 2
)
X
k
S̃η = η + η̇ : η̇ ∈ Tη Met (Σg ), kη̇k <  and η̇ab;b = 0 .
b=1

To obtain a section for the action on the smaller space Metk0 (Σg ), we utilize the
following lemma:
Lemma 3.3.3. If
ηab (t) = ηab + tη̇ab for t ∈ (, )
is a one-parameter family of metrics of constant Gaussian curvature K and total
area one, η̇ab;b = 0 and
1
∆η(0) (Tr(η̇)) = −2KTr(η̇), where Tr(η̇) = (η̇11 + η̇22 )
λ2
and ∆η(0) is the Laplace operator for the base metric η(0).
To prove this, we can use geodesic normal coordinates centered for η at a given
point p ∈ Σ. In terms of such coordinates, the curvature tensor of η(t) is given
by
1
K det(ηab (t)) = R1212 (t) = (2η12,12 (t) − η11,22 (t) − η22,11 (t))
2
+ (higher order terms),
where the commas denote coordinate derivatives. Differentiating with respect
to t and setting t = 0, we obtain
1
Kλ2 (η̇11 + η̇22 ) = (2η̇12;12 − η̇11;22 − η̇22;11 ) ,
2
where we have replaced ordinary derivatives P by covariant derivatives and have
evaluated at p. Finally, we use the identity η̇ab;b = 0 to simplify the right-
hand side,  
2
1 X
Kλ2 (η̇11 + η̇22 ) = −  η̇aa;bb  ,
2
a,b=1

125
which yields the statement of the lemma:
2  
1 X 1 1
2 2
η̇aa = −2K 2 (η̇11 + η̇22 )
λ λ ;bb λ
a,b=1

Since K is negative, it follows from the Lemma and the maximum principle
applied to
Pthe operator ∆η(0) + 2K that η̇111 + η̇222 = 0. This, together with the
identity η̇ab;b = 0 implies that if w = u + iu ,

∂
(η̇11 + iη̇12 ) = 0,
∂ w̄
and hence
(η̇11 + iη̇12 )dw2
is a holomorphic quadratic differential. Thus we set

Sη = {η + η̇ : η̇ ∈ Tη Metk (Σg ), kη̇k <  and

(η̇11 + iη̇12 )dw2 is a holomorphic quadratic differential },

a convex open subset of an affine subspace of a Hilbert space. One of the

implications of the Riemann-Roch Theorem from Riemann surface theory is
that the space of holomorphic quadratic differentials on a Riemann surface of
genus g has complex dimension 3g − 3 or real dimension 6g − 6, so the affine
space in which Sη lies has dimension 6g − 6.
A straightforward (if slightly tedious) calculation shows that the identity
map
id : (Σg , η) −→ (Σg , η + η̇)
is harmonic with Hopf differential (1/2)(η̇11 +iη̇12 )dw2 . If the G-orbit intersected
Sη in more than one point, we would have homotopic harmonic maps from
(Σg , η) to the same target with two different Hopf differentials, contradicting
Theorem 3.1.2 on uniqueness for harmonic maps. The section Sη is only tangent
to the submanifold Metk0 (Σg ), but since the projection π to the space Metk0 (Σg )
commutes with the G-action, S = π(Sη ) is a slice for the G-action on Metk0 (Σg )
of dimension 6g − 6, completing the proof of the theorem.
Remark 3.3.4. In contrast to the Teichmüller space Tg , the Riemann moduli
space Rg is not even a manifold in general, but an orbifold. It is the quotient
of Teichmüller space by the properly discontinuous action of the mapping class
group Γg = Diff+ (Σg )/Diff0 (Σg ).
Choose a base Riemann surface (Σ, η0 ) of genus g and let O(κ2 ) denote the
space of holomorphic quadratic differentials on (Σ, η0 ), a real vector space of
dimension 6g − 6. We can define a map

Ψ : Tg → O(κ2 ) by Ψ(Σ, η) = Ω(f ),

126
where Ω(f ) is the Hopf differential of the unique harmonic map f from (Σ, η0 )
to (Σ, η).
Theorem 3.3.5. (Teichmüller) If g ≥ 2, the map Ψ : Tg → O(κ2 ) is a
diffeomorphism, and hence Tg is diffeomorphic to R6g−6 .
Teichmüller’s original proof that Tg is homeomorphic to R6g−6 was based upon
the theory of quasiconformal maps, and was later simplified by Bers, as pre-
sented in [37]. Later a proof was found using harmonic maps, and we will
return to discuss it later.
Let η0 ∈ Metk0 (Σg ) be a base Riemannian metric. If η is any element of Tg , it
follows from Theorems 3.1.1 and 3.1.2 that there is a unique harmonic map

s(η) : (Σ, η) → (Σ, η0 ),

which depends smoothly on η in terms of the above coordinates. Using a variant

of the Bochner Lemma to be treated in the next section, Schoen and Yau were
able to prove that s(η) is a diffeomorphism ([73], Chapter 1, §8). Thus we can
define a map

Metk0 (Σg ) → Metk0 (Σg ) by σ(η) = [s(η)−1 ]∗ (η).

Since φ ◦ s(φ∗ η) = s(η), the metric s(η) defines a section σ : Tg → Metk0 (Σg ).
We can now consider the energy as a map of two variables E : C 2 (Σ, M ) ×
Metk0 (Σg ) → R defined by
 
∂f ∂f √
Z X
1
E(f, (ηab )) = η ab , ηdu1 du2 , (3.19)
2 Σ ∂ua ∂ub
a,b

where (η ab ) is the matrix inverse of (ηab ) and η = det(ηab ), the integrand being
independent of choice of local coordinates. If φ ∈ Diffk+1
0 (Σg ), then

E(f ◦ φ, φ∗ η) = E(f, η),

and hence E descends to a map on the space of orbits

C 2 (Σ, M ) × Metk0 (Σg )

.
Diffk+1
0 (Σg )

However, using the map σ described above, one can exhibit this quotient as
simply a product C 2 (Σ, M ) × Tg .
Theorem 3.3.6. If (f, ω) is a critical point for the map

E : C 2 (Σ, M ) × Tg −→ R,

then f is harmonic and weakly conformal with respect to ω, and hence a

parametrized minimal surface.

127
Differentiation with respect to f shows that f is harmonic. We need to calculate
the derivative with respect to the metric, so we take a perturbation of a given
metric with support in a given isothermal coordinate system (u1 , u2 ) on Σ.
Suppose that the perturbation is given by the formula
ηab (t) = ηab + tη̇ab = ηab + tψ(u1 , u2 )ρab (u1 , u2 ),
where ψ(x1 , x2 ) has compact support, the variation in the metric is trace-free
(η̇11 + η̇22 = 0) and the initial metric is given by the formulae
√
ηab = δab λ2 , η = λ2 ,
for some conformal factor λ2 . Then a direct calculation shows that

d d
= λ2 (η̇11 + η̇22 ) = 0,

ηab (t) = η̇ab , η(t)
dt dt t=0

From this, we can easily calculate

√ 11 √ 12   
d ηη ηη  −2 η̇22 −η̇12
√ 21 √ 22   =λ .
dt ηη ηη t=0
− η̇21 η̇11
Thus we find that
   
d √ ab 
Z X
d ∂f ∂f
du1 du2

E(f, ηab (t))
 = ηη  ,
a ∂ub
dt t=0 Σ a,b dt t=0 ∂u
Z       
η̇11 ∂f ∂f ∂f ∂f 2η̇12 ∂f ∂f
=− 2
, − , + 2 , du1 du2
Σ λ ∂u1 ∂u1 ∂u2 ∂u2 λ ∂u1 ∂u2
Z   
∂f ∂f 1 1 2
= −4 Re (η̇11 + iη̇12 ) , 2
du du , (3.20)
Σ ∂w ∂w λ
where w = u1 + iu2 . Since this is true for all trace-free variations in the metric,
we conclude that
     
∂f ∂f ∂f ∂f ∂f ∂f
, − , =0= , ,
∂u1 ∂u1 ∂u2 ∂u2 ∂u1 ∂u2
and hence the Hopf differential Ω(f ) must vanish. This finishes the proof of the
theorem.
Of course, it is also true that E(f ◦ φ, φ∗ η) = E(f, η) for φ in the larger group
of orientation-preserving diffeomorphisms Diffk+1+ (Σg ), but this group does not
act freely on C 2 (Σ, M ) × Metk0 (Σg ) and the quotient space

C 2 (Σ, M ) × Metk0 (Σg )

M(Σ, M ) =
Diffk+1
+ (Σg )

is not a manifold, but only an orbifold in general. We have two ways of looking
at parametrized minimal surfaces of genus g, as critical points for the energy
E : M(Σ, M ) −→ R

128
or as Γg -orbits of critical points for
E : C 2 (Σ, M ) × Tg −→ R,

where Γg = Diffk+1 k+1

+ (Σg )/Diff0 (Σg ) is the mapping class group.

3.4 The Bochner Lemma

The key step behind the existence theory for harmonic is the Bochner Lemma
which gives an estimate for the Laplacian of the energy density in terms of cur-
vature. We describe the Bochner Lemma in this section, following the treatment
in [45]. Let Σ and M be compact Riemannian manifolds with metrics (ηab ) and
(gij ) respectively, with M as usual being isometrically imbedded in Euclidean
space RN . We suppose that
f : Σ −→ M ⊆ RN
is a smooth map and for given choice of coordinates (u1 , . . . , um ) on Σ, we
consider the vector-valued maps
∂f ∂f i
fa = ∈ RN which have components fai = ,
∂ua ∂ua
with respect to coordinates (x1 , . . . , xn ) on M . Recall that the energy density
is given by
1 X ab 1X
e(f ) = η fa · fb = gij η ab fai fbj .
2 2
Finally, we let
RΣ = (R̃abcd ) and RM = (Rijkl )
be the curvature tensors of Σ and M , and let RicΣ = (R̃ab ) be the Ricci cur-
vature of Σ. Recall that if we divide the Laplacian of the vector-valued map
f : Σ → RN into tangential and normal components,
∆η f = (∆η f )> + (∆η f )⊥ ,
the condition that f be harmonic is just (∆η f )> = 0. On the other hand,
the normal component of the Laplacian is expressed in terms of the second
fundamental form
α(f )(p) : Tp M × Tp M −→ Np M
of M within RN by
X
(∆η f )⊥ (p) = η ab α(f )(p)(fa , fb ).

Theorem 3.4.1 (Bochner Lemma). If f : Σ → M is a smooth map, then

∆η (e(f )) = |∇df |2 + hd[(∆η f )> ], df i

X X
+ η ac η bd R̃ab fc · fd − η ac η bd Rijkl fai fbj fck fdl . (3.21)

129
where ∇ denotes covariant derivative with respect to the connection in T ∗ Σ ⊗
f ∗ T M . Hence if f is harmonic,
X X
∆η (e(f )) = |∇df |2 + η ac η bd R̃ab fc · fd − η ac η bd Rijkl fai fbj fck fdl . (3.22)

To prove this, we use normal coordinates centered a point p ∈ Σ and similar

coordinates at f (p) in M . Then
X X
∆η (e(f )) = |fa;b |2 + hfa , fa;bb i
X X X
= |fa;b |2 + hfa , fb;ba i − hfa , R̃bcab fc i
X X X
= |fa;b |2 + hfa , (∆η f )a i + R̃ab hfa , fb i. (3.23)

On the other hand,

hfa , (∆η (f ))a i = hfa , [(∆η f )> ]a i + hfa , [(∆η f )⊥ ]a i

X
= hfa , [(∆η f )> ]a i + hfa , (α(f )(fb , fb )a i.

Now we use the fact that

hfa , α(f )(fbi , fbi )i = 0

to conclude that
X
hfa , (∆η (f ))a i = hfa , [(∆η f )> ]a i − h∆η (f ), (α(f )(fb , fb )i
X
= hfa , [(∆η f )> ]a i − h(α(f )(fa , fa ), (α(f )(fb , fb )i. (3.24)

Finally, we note that

X X
|fa;b |2 = |∇df |2 + hα(f )(fa , fb ), α(f )(fa , fb )i. (3.25)

Finally, we substitute (3.24) and (3.25) into (3.24) and obtain

X
∆η (e(f )) = |∇df |2 + hd[(∆η f )> ], df i + R̃ab fa · fb
hX X i
− hα(f )(fa , fa ), (α(f )(fb , fb )i − hα(f )(fa , fb ), α(f )(fa , fb )i .

This last equation implies (3.21) by the Gauss equation.

If Σ is a compact surface with Gaussian curvature K̃, then (3.22) simplifies to
X
∆η (e(f )) = |∇df |2 + K̃|df |2 − Rijkl fai fbj fak fbl . (3.26)

In terms of isothermal coordinates (u, v) = (u1 , u2 ) on Σ, we can write

 
X
a b 2 2 2 1  ∂f ∂f 
ηab du du = λ (du + dv ), a(f ) = 2  ∧ .
λ ∂u ∂v 

130
Then (3.26) becomes
∆η (e(f )) ≥ |∇df |2 + 2K̃e(f ) − 2K(σ)a(f )2 , (3.27)
where K(σ) is the sectional curvature of the two-plane generated by ∂f /∂u and
∂f /∂v. It follows from inequality (3.10) that if K0 ≥ 0, then
K(σ) ≤ K0 ⇒ ∆η (e(f )) ≥ |∇df |2 + 2K̃e(f ) − 2K0 e(f )2 , (3.28)
equality holding when f is conformal.
Corollary 3.4.2. If M has nonpositive sectional curvatures, there are no har-
monic maps from the two-sphere S 2 into M and any harmonic map from the
torus T 2 into M must be totally geodesic. If M has negative sectional curva-
tures, there are no harmonic maps from S 2 or T 2 into M .
Proof: If Σ = S 2 , we can choose the Riemannian metric on Σ so that K̃ ≡ 1.
The continuous function e(f ) must assume a maximum at some point. At this
point the Bochner inequality shows that ∆(e(f )) > 0, a contradiction.
If Σ = T 2 , we can choose the Riemannian metric on Σ so that K̃ ≡ 0. In
this case, the inequality must be an equality. Hence ∇df = 0 which implies that
∂f ∂f
and
∂u1 ∂u2
are parallel sections of the pullback of the bundle f ∗ T M . This implies of course
that f is totally geodesic.
The case where M has negative sectional curvatures is proven in a similar
fashion.

3.5 The α-energy

In order to apply critical point theory to a function on an infinite-dimensional
manifold, we need the function to be continuous and satisfy condition C of
Palais and Smale. For the usual energy, however, the latter condition would
require that we complete the space C 2 (Σ, M ) with respect to the L21 -topology,
a topology which is too weak for the usual techniques of global analysis, since
the Sobolev inequality barely fails to show that L21 functions are continuous.
Thus Sacks and Uhlenbeck [68], [69] were led to consider perturbations of
the energy. Among the simplest such perturbations are functions
Z
F : C ∞ (Σ, M ) → R defined by F (f ) = φ(|df |2 )dA,
Σ

where φ : R → R is a smooth function. By calculating the differential of F and

integrating by parts (assuming that f is sufficiently differentiable), we obtain
the Euler-Lagrange equation for critical points
   
D 0 2 ∂f D 0 2 ∂f
φ (|df | ) + φ (|df | ) = 0, (3.29)
∂u ∂u ∂v ∂v

131
whenever (u, v) is a set of isothermal coordinates on Σ.
Example 3.5.1. We could take φ(t) = tα , where α > 1, which has continuous
derivatives up to order two when regarded as a function on L2α
1 (Σ, M ), a Banach
space which does lie within Sobolev range. Thus for a given Riemannian metric
η on Σ, we could define
Z
1
Êα,η : L2α
1 (Σ, M ) → R by Êα,η (f ) = |df |2α dA, (3.30)
2 Σ

However, the critical points of Êα,η are not necessarily smooth. Indeed, in the
special case where M = R, the Euler-Lagrange equation
   
∂ ∂f ∂ ∂f
|df |2α−2 + |df |2α−2 =0
∂u ∂u ∂v ∂v

can be rewritten in terms of polar coordinates (r, θ) as

   
1 ∂ 2α−2 ∂f 1 ∂ 2α−2 ∂f
r|df | + 2 |df | = 0,
r ∂r ∂r r ∂θ ∂θ

which is satisfied by f (r, θ) = r(2α−2)/(2α−1) . Although this function is not C 1 ,

it can be checked that it does lie in L2α 2α
1 . Thus we see that critical points in L1
∞
for this choice of φ are not necessarily C .
However, the slight modification in which φ(t) = (1 + t)α provides a function
which has C ∞ critical points, as we will see later, and that is the choice that
we will make.
Definition. Suppose that M is a smooth manifold with Riemannian metric
g. Given a Riemannian metric η on Σ, the α-energy is the function Eα,η :
L2α
1 (Σ, M ) → R defined by
Z
1
Eα,η (f ) = (1 + |df |2 )α dA, (3.31)
2 Σ
for α > 1, where dA is calculated with respect to η and |df | is calculated with
respect to η and g. For ω ∈ T , the (α, ω)-energy Eα,ω is just the function Eα,η ,
where η is chosen to be the unique constant curvature metric of total area one
on Σ within the conformal class ω.
By the normalization we have chosen for the metric, the α-energy Eα,ω ap-
proaches Eω + (1/2) as α → 1.
Note that the (α, ω)-energy on M is a composition of the map ωi , induced via
Theorem 1.4.7 (the so-called ω-lemma) by an isometric imbedding i : M → RN ,
with the (α, ω)-energy on RN . The latter map, Eα,ω : L2α N
1 (Σ, R ) → R, is
2
clearly continuous, and it is also C with derivatives
Z
dEα,ω (f )(V ) = α (1 + |df |2 )α−1 df · dV dA (3.32)
Σ

132
and
Z
d2 Eα,ω (f )(V1 , V2 ) = α (1 + |df |2 )α−1 dV1 · dV2 dA
Σ
Z
+ 2α(α − 1) (1 + |df |2 )α−2 (df · dV1 )(df · dV2 )dA, (3.33)
Σ

by the same argument we used in Example 1.2.3. Since the composition of C 2

maps is C 2 we see that the map Eα,ω on L2α 2
1 (Σ, M ) is also C .
We will see shortly that Eα,ω satisfies condition C and its critical points are
smooth. If we let ω ∈ T vary and set Eα (f, ω) = Eα,ω (f ), we get a C 2 function

Eα : L2α
1 (Σ, M ) × Tg → R.

Definition. A critical point f ∈ L2α 1 (Σ, M ) for Eα,ω is called an (α, ω)-
harmonic map, or sometimes an α-harmonic map. If (f, ω) ∈ L2α 1 (Σ, M ) × T is
a critical point for Eα , f is called a (parametrized) α-minimal surface.
In complete analogy with ω-harmonic maps, we can calculate the first variation
formula for (α, ω)-harmonic maps. Indeed, suppose that Σ is a compact Rie-
mann surface. By the same argument that led to (3.29), we obtain the formula
Z        
D ∂f D ∂f
dEα,ω (f )(V ) = − γ2 ,V + γ2 ,V dudv,
Σ ∂u ∂u ∂v ∂v

where γ 2 = (1+|df |2 )α−1 . In the integrand, (u, v) can be any choice of conformal
coordinates. This leads to the Euler-Lagrange equation
   
D ∂f D ∂f
γ2 + γ2 = 0. (3.34)
∂u ∂u ∂v ∂v

This equation can also be written in terms of the standard Laplace operator
acting on f ,
   
D ∂f D
∂f
+
∂u ∂u ∂v
∂v
 
∂ 2 ∂f ∂ 2 ∂f
= −(α − 1) (log(1 + |df | )) + (log(1 + |df | )) . (3.35)
∂u ∂u ∂v ∂v

Our next goal is to show that the function Eα,ω : L2α 1 (Σ, M ) → R satisfies
condition C. Recall that condition C asserts that if {fi } is a sequence of points
in L2α
1 (Σ, M ) such that

Eα,ω (fi ) ≤ E0 where E0 is some constant, and kdEα,ω (fi )k → 0,

then {fi } possesses a subsequence which converges in L2α

1 (Σ, M ) to a critical
point for Eα,ω .

133
Theorem 3.5.2. If α > 1, the function Eα,ω : L2α
1 (Σ, M ) → R satisfies
condition C.
The proof is very similar to the one given before for the action integral in the
theory of smooth closed geodesics. If V is tangent to L2α
1 (Σ, M ),
Z    
2 α−1 DV ∂f DV ∂f
dEα,ω (f )(V ) = 2α (1 + |df | ) , + , dudv,
Σ ∂u ∂u ∂v ∂v

or equivalently,
Z      
D 2 ∂f D 2 ∂f
dEα,ω (f )(V ) = −2α γ + γ ,V dudv,
Σ ∂u ∂u ∂v ∂v

for V ∈ Tf (L2α 2 2 α−1

1 (Σ, M )), where γ = (1 + |df | ) and D denotes covariant
derivative. More generally, we can consider an element V ∈ Tf (L2α N
1 (Σ, R )) and
2α
project it into the tangent space to L1 (Σ, M ). Starting with the orthogonal
projection P : i∗ T RN → T M into the tangent space, we use the ω-Lemma to
construct a smooth map
∗
ωP : L2α N 2α
1 (Σ, i T R ) −→ L1 (Σ, T M ).

∗
Lemma 3.5.3. If V ∈ Tf (L2α N
1 (Σ, i T R ),

Z 
∂V ∂f ∂V ∂f
dEα,ω (f )(ωP (V )) = α (1 + |df |2 )α−1 · + ·
Σ ∂u ∂u ∂v ∂v
    
∂f ∂f ∂f ∂f
−V · α , +α , dudv,
∂u ∂u ∂v ∂v

where α : T M × T M → N M is the second fundamental form of M in RN .

In the proof, we can suppose that f and V are C ∞ , since the C ∞ maps are dense
⊥ ⊥
in L2α
1 . We write V = ωP (V ) + V , where ωP (V ) and V are the components
of V which are tangent and normal to M . Then
        
∂ 2 ∂f ∂ 2 ∂f ∂ 2 ∂f
γ ·V = γ · ωP (V ) + γ ·V⊥
∂u ∂u ∂u ∂u ∂u ∂u
    
D 2 ∂f ∂f ∂f
= γ 2
· ωP (V ) + γ α , · V ⊥,
∂x ∂x ∂x ∂x

and similarly
       
∂ ∂f D ∂f ∂f ∂f
γ2 ·V = γ2 · ωP (V ) + γ 2 α , · V ⊥,
∂v ∂v ∂v ∂v ∂v ∂v

where the dot denotes the dot product in the ambient Euclidean space RN . It

134
follows that

dEα,ω (f )(ωP (V ))
Z    
∂ 2 ∂f
∂ 2 ∂f
= −2α γ + γ ·V
Σ ∂u ∂u ∂v ∂v
     
2 ∂f ∂f ∂f ∂f ⊥
+γ α , +α , ·V dudu.
∂u ∂u ∂v ∂v
An integration by parts now yields the statement of the Lemma.
Lemma 3.5.4. There is a constant C depending only on M such that
∗
kωP (V )kL2α
1
≤ C(1 + Eα,ω (f ))kV kL2α
1
for V ∈ Tf (L2α N
1 (Σ, i T R ).

This is a straightforward consequence of applying the Leibniz rule to the equa-

tion
(ωP (V ))(p) = Pf (p) V (p),
where Pf (p) is the projection from RN into the tangent space Tf (p) M , to obtain
   
∂ ∂P ◦ f ∂V
(ωP (V ))(p) = (V (p)) + Pf (p) (p) ,
∂u ∂u ∂u
   
∂ P ◦ ∂f ∂V
(ωP (V ))(p) = (V (p)) + Pf (p) (p) .
∂v ∂v ∂v

Thus that after possibly replacing the usual L2α

1 norm with an equivalent norm,
we can write
kωP (V )kL2α
1
= kωP (V )kL2α + kdωP (V )kL2α ,
but
   
∂P ◦ f ∂V
kdωP (V )kL2α =
(V (p))du + Pf (p) (p) du
∂u ∂u 2α
    L
∂P ◦ f ∂V
+ (V (p))dv + Pf (p) (p) dv
∂v ∂v 2α
L
≤ C1 sup{k(V (p))k : p ∈ Σ} kdf kL2α + C2 kDV kL2α ,

where C1 and C2 are positive constants. Thus we see that

h i
kωP (V )kL2α
1
≤ C3 kV k 2α
L1 E α,ω (f ) + C4 kDV kL 2α + C 5 kV kL 2α ,

where C3 , C4 and C5 are yet other positive constants and we have used the
fact that the C 0 -norm is less than some constant times the L2α
1 -norm. The last
estimate yields the statement of the lemma.
Let us turn now to the proof of Theorem 3.5.2 itself. Let {fi } be a sequence from
L2α 2α N
1 (Σ, M ) ⊂ L1 (Σ, R ) such that Eα,ω (fi ) is bounded and kdEα,ω (fi )k →

135
0. By the Sobolev Lemma 1.4.4 each fi ∈ C 0 (Σ, M ). Moreover, since M is
compact, {fi } is uniformly bounded, and it follows from the Hölder estimate in
Lemma 1.4.4 that {fi } is equicontinuous. Therefore, the Arzela-Ascoli Theorem
from real analysis implies that a subsequence of {fi } converges uniformly to a
continuous map f∞ : Σ → M .
Since Eα,ω (fi ) is bounded and fi is bounded in C 0 , it follows that fi is
bounded in L2α N
1 (Σ, R ). It then follows from Lemma 3.5.3 that ωP (fi − fj ) is
2α N
bounded in L1 (Σ, R ). Since kdEα,ω (fi )k → 0,

|dEα,ω (fi )(ωP (fi − fj )) − dEα,ω (fj )(ωP (fi − fj ))| → 0 as i, j → ∞.

Using Lemma 3.5.2, we can rewrite this as

 Z     
2α (1 + |dfi |2 )α−1 ∂fi · ∂fi − ∂fj + ∂fi · ∂fi − ∂fj


Σ ∂u ∂u ∂u ∂v ∂v ∂v
     
∂fi ∂fi ∂fi ∂fi
− α , +α , · (fi − fj ) dudv
∂u ∂u ∂v ∂v
Z     
∂fj ∂fi ∂fj ∂fj ∂fi ∂fj
−2α (1 + |dfj |2 )α−1 · − + · −
Σ ∂u ∂u ∂u ∂v ∂v ∂v
      
∂fj ∂fj ∂fj ∂fj 
− α , +α , · (fi − fj ) dudv  → 0
∂u ∂u ∂v ∂v
as i and j approach infinity. Note that since energy is bounded,
Z   
 α ∂fi , ∂fi · (fi − fj )dudv  ≤ (constant) sup |fi − fj | → 0
 

Σ ∂u ∂u 
Z   
 ∂fi ∂fi 
and  α , · (fi − fj )dudv  ≤ (constant) sup |fi − fj | → 0
Σ ∂v ∂v

as i and j approach infinity, and hence

Z
 (1 + |dfi |2 )α−1 hdfi , dfi − dfj idA


Σ
Z 
2 α−1

− (1 + |dfj | ) hdfj , dfi − dfj idA → 0
Σ

as i and j approach infinity.

To proceed further, we need:
Lemma 3.5.5. If α > 1, there is a constant c ≥ (1/16) such that

(|v|2α−2 v − |w|2α−2 w) · (v − w) ≥ c|v − w|2α , for all v, w ∈ R2N +1 .

136
Proof: It suffices to establish this inequality when both v and w are nonzero.
If we set f (v) = |v|2α , then f is C 2 on R2N +1 − {0}, and a direct calculation
shows that
Df (v)(w) = 2α|v|2α−2 v · w,
D2 f (v)(w, w) = 4α(α − 1)|v|2α−4 (v · w)2 + 2α|v|2α−2 w · w ≥ 2α|v|2α−2 w · w.
Hence
Z 1
Df (v)(v − w) − Df (w)(v − w) = D2 f (w + t(v − w))(v − w, v − w)dt
0
Z 1
≥ 2α |w + t(v − w)|2α−2 |v − w|2 dt ≥ c|v − w|2α ,
0

for some constant c > 0. To see that c can be chosen to be larger than 1/16,
note that by the triangle inequality, either
1 1 1
|w| ≥ |v − w| ⇒ |w + t(v − w)| ≥ |w| ≥ |v − w| for t ∈ [0, 1/4],
2 2 4
or
1 1 1
|v| ≥ |v − w| ⇒ |v + (1 − t)(w − v)| ≥ |v| ≥ |v − w| for t ∈ [3/4, 1].
2 2 4
This proves the lemma.
To apply Lemma 3.5.5, we set
   
∂fi ∂fi ∂fj ∂fj
v = 1, , , w = 1, , .
∂u ∂v ∂u ∂v
Then Lemma 3.5.5 implies that
Z
|dfi − dfj |2α dA → 0 as i, j → ∞.
Σ

Since fi → f∞ in C 0 , it follows that {fi } is a Cauchy sequence in L2α N

1 (Σ, R ).
2α N
By completeness of the Banach space L1 (Σ, R ), there exists an element
f˜∞ ∈ L2α N ˜ 2α N ˜
1 (Σ, R ) such that fi → f∞ in L1 (Σ, R ). Clearly, f∞ = f∞ ∈
2α
L1 (Σ, M ), and Theorem 3.5.2 is proven.
Of course, there are versions of Theorem 3.5.2 which hold for perturbations of
the α-energy. For example, if α > 1 and ψ ∈ L2k (Σ, RN ) for k ≥ 2, the perturbed
function
Z
2α
Eα,ψ,ω : L1 (Σ, M ) → R, defined by Eα,ψ,ω (f ) = Eα,ω (f ) + f · ψdA
Σ

satisfies condition C. In fact, even more generally, we could consider the function
β
Eα,ψ,ω : L2α
1 (Σ, M ) → R defined by
Z Z
β 1 2 2 α
Eα,ψ,ω (f ) = (β + |df | ) dA + f · ψdA. (3.36)
2 Σ Σ

137
 β
Theorem 3.5.5. If α > 1 and β ∈ [0, 1], then the function Eα,ψ,ω defined by
(3.36) satisfies condition C.
The proof is a straightforward modification of that given for Theorem 3.5.2.
Note that the argument works even if β = 0, in which case the critical points
are not necessarily smooth.
Thus we can use Corollary 1.11.3 to show that any component of L2α 1 (Σ, M )
β
contains an element f which minimizes Eα,ω or more generally Eα,ψ,ω . More-
over, we can use Theorem 1.11.2 to prove existence of minimax critical points
for these functions corresponding to various algebraic topology constraints. We
note that
β 2α
E(f ) ≤ Eα0 (f ) ≤ Eαβ (f ) − , (3.37)
2
the second of these inequalities following from the fact that

ψ(t) = (β 2 + t2 )α − (β 2α + t2α ), then ψ 0 (t) > 0, for t > 0,

an immediate consequence of differentiating ψ with respect to t.

Of course, our goal is to use these existence results to obtain existence of
ω-harmonic maps in the limit as the various perturbations are turned off. Thus,
for example, we could let β → 0 first, obtaining a function that has simpler
behavior under rescaling, and then let α → 1. Condition C is only lost when
taking the second limit.

3.6 Regularity of (α, ω)-harmonic maps

At some point, it becomes useful to know that the (α, ω)-harmonic maps con-
structed by means of Condition C are actually smooth maps. Indeed, if the
ambient manifold M is C ∞ , so is every (α, ω)-harmonic map into M . This is
the content of the following theorem:
Theorem 3.6.1. If 1 < α < 3/2, any critical point f ∈ L2α 1 (Σ, M ) for Eα,ω is
smooth. Moreover, if ψ ∈ L2k (Σ, RN ), for some k ≥ 3, and β ∈ (0, 1], then any
β
critical point f ∈ L2α 2
1 (Σ, M ) for Eα,ψ,ω is Lk+2 .

The proof given by Sacks and Uhlenbeck [68] relies on results from Morrey
[54], a classic book which contains proofs of many such regularity theorems. A
complete proof of this result goes beyond the scope of these notes. However,
we hope the following outline will be helpful in giving an idea as to what is
involved. Some readers may wish to skip the following sketch, which makes use
of the Hölder spaces described in §5.1 of [19].
Following [68], we divide the proof into three steps: First we show that the
critical point f is in L22 , then in a Hölder space C 1,β , and finally, we use the
Schauder theory [25] together with elliptic bootstrapping to prove that f is
β
C ∞ . We will deal only with Eα,ω , the modification necessary for Eα,ψ,ω being
relatively straightforward.

138
Step 1. We show that f lies in L22 (Σ, M ). To do this, we can use regularity
results for variational problems presented in Evans [19], §8.3.
Regularity is a local condition, and we need only show that a critical point f
is regular near a given point p ∈ Σ. Let U and V be small open neighborhoods of
p with V̄ ⊂ U . Our strategy is to work in local coordinates (x1 , . . . , xn ) defined
on a neighborhood of F (U ) in M and a local conformal parameter (u1 , u2 ) on
U ⊂ Σ. The coordinate system on M allows us to regard f |UP as an Rn -valued
map.PWe abbreviate the composition x ◦ f to x , and let η = ηab dua dub and
i i

g= gij dxi dxj denote the Riemannian metrics on Σ and M respectively. Let
2 n
∂xi X X
pia = so |df |2 = gij (x1 , . . . xn )η ab pia pjb .
∂ua
a,b=1 i,j=1

Then
Z
1
Eα,ω (f |U ) = L(p, x, u)du1 du2 ,
2 U
p
where L(p, x, u) = (1 + |df |2 )α det(ηab ).

The fact that f |V is (α, ω)-harmonic is expressed by the Euler-Lagrange condi-

tion  
Z X ∂L ∂ X ∂L

i ∂ua
(ζ 2 v i ) + i
ζ 2 v i  du1 du2 = 0, (3.38)
U a,i
∂p a i
∂x

for every smooth test function v = (v 1 , . . . , v n ), ζ : Σ → [0, 1] being a suitable

smooth cutoff which is one on V and zero outside U .
Note that the first derivatives of the Lagrangian L are given by
∂L ∂
= α(1 + |df |2 )α−1 i (|df |2 ),
∂pia ∂pa

where
2 n
∂ 2
XX
1 n ab j
p
(|df | ) = gij (x , . . . x )η p b det(ηab ).
∂pia j=1
b=1

Similarly, the second derivatives of L are given by

∂2L ∂2
= α(1 + |df |2 )α−1 (|df |2 )
∂pia ∂pjb ∂pia ∂pjb
∂ ∂
+ α(α − 1)(1 + |df |2 )α−2 (|df |2 ) j (|df |2 ),
∂pia ∂pb

where
∂2 p ∂2
(|df |2 ) = gij (x1 , . . . xn )η ab det(ηab ) (|df |2 ).
∂pia ∂pjb ∂pia ∂xk

139
Note that by scaling the coordinates appropriately, we can make |∂gij /∂xk | and
|∂ 2 gij /∂xk ∂xl | less than  for any given  > 0. As a consequence we find that
the Euler-Lagrange operator for the (α, ω)-energy is uniformly elliptic:
2 n 2 n
X X ∂2L i j
X X
gij (x1 , . . . xn )η ab det(ηab )ξai ξbj ,
2
p
j
ξ a ξ b ≥ αγ
i
∂pa ∂pb
a,b=1 i,j=1 a,b=1 i,j=1

where γ 2 = (1 + |df |2 )α−1 . It is exactly this estimate on uniform ellipticity that

allows us to apply the difference quotient approach described in Evans [19] to
show that u|V is in L22 . Since Σ can be covered by finitely many neighborhoods
V , it follows that f ∈ L22 (Σ, M ), finishing Step 1.
β
We remark that it is uniform ellipticity that fails for the function Eα,ψ,ω when
β = 0 and this explains why critical points of this function are not necessariy
smooth.
Step 2. For the second step, we recall that equation (3.35) for α-harmonic
maps can be written in the form

(∆f )> = −(α − 1)d(log(1 + |df |2 )) · df,

   
Df ∂f Df ∂f
where (∆f )> = + ,
∂x1 ∂x1 ∂x2 ∂x2
D denoting the Levi-Civita connection on M . Since we know by Step 1 that f is
L22 , we are justified in differentiating on the right-hand side, thereby obtaining
the following result after a short calculation:

∂2f ∂2f B(d2 f, df )

2 + 2 = α(f )(df, df ) − (α − 1) df,
∂x1 ∂x2 1 + |df |2

where α(f ) and B are bilinear maps, α(f ) being the familiar second fundamental
form of M in RN . We can put the last term on the left-hand side obtaining
L(f ) = α(f )(df, df ), where

∂2u ∂2u B(d2 u, df )

L(u) = 2 + 2 + (α − 1) df,
∂x1 ∂x2 1 + |df |2
L being a second-order differential operator with coefficients in L∞ . Moreover,
since α − 1 < 1/2, the operator L is uniformly elliptic. The linear operator L
defines a bounded linear map from L42 to L4 , which can be regarded as a small
perturbation of the scalar Laplace operator when α is close to one.
If we restrict L to a small disk D about a given point p and u ∈ Ker(L),
each component of u will assume its maximum value on the boundary ∂D of
D. Thus if we impose Dirichlet boundary conditions that u and f agree on ∂D,
the map L : L42 → L4 will be injective. On the other hand, ∆ : L42 → L4 has
a continuous inverse G0 : L4 → L42 , and we can think of G0 as an approximate
inverse to L when α − 1 is small. Given h ∈ L4 , then for appropriate choices
of norms on L42 and L4 , the map u 7→ G0 (Lu − h) + u is a contraction, and it

140
must therefore possess a unique fixed point. This implies that L is surjective,
and the open mapping theorem implies that the inverse G to L is continuous.
Since the critical point f is continuous and df ∈ L2 , α(f )(df, df ) is in L4 ,
and there is a unique u ∈ L42 (D, M ) satisfying the Dirichlet boundary conditions
such that L(u) = α(f )(df, df ). This u must of course be f and hence f ∈ L42 .
We thus conclude that the restriction of f to a small neighborhood of any point
lies in L42 . Since L42 is contained in the Hölder space C 1,β for suitable β, we
conclude that f ⊂ C 1,β (Σ, M ).
Step 3. The final step is via the technique of ”elliptic bootstrapping” to the
differential equation
L(f ) = α(f )(df, df ), (3.39)
using the Schauder theory. The necessary Schauder estimates are described in
[25], Chapter 8 for scalar operators, the general case being a standard extension
within PDE theory. The result is that:

f ∈ C 1,β ⇒ right side of (3.39) ∈ C 0,β ⇒ f ∈ C 2,β

⇒ right side of (3.39) ∈ C 1,β ⇒ f ∈ C 3,β ⇒ · · · .

This shows that f is C ∞ and completes our sketch of the argument for the
theorem.

3.7 Morse theory for the perturbed energy

Once we have condition C, we can apply Liusternik-Schnirelmann theory to the
functions Z Z
1
Eα,ψ,ω (f ) = (1 + |df |2 )α dA + f · ψdA,
2 Σ Σ
but we would like to establish Morse inequalities and define a Morse-Witten
complex when the critical points are Morse nondegenerate. This requires an
extension of Morse theory to certain functions on Banach manifolds, and such a
theory was in fact developed by Uhlenbeck [80]. In this section, we present the
results of that theory, specialized to the perturbations of the energy we have
been studying.
β
Note that if f is a critical point for Eα,ψ,ω , then a calculation similar to that
which yields (3.32) shows that
Z Z
dEα,ψ,ω (f )(V ) = α (1 + |df |2 )α−1 hdf, DV idA + ψ · V,
Σ Σ

for all V ∈ Tf L2α

1 (Σ, M ), where DV denotes the covariant differential of V with
respect to the Levi-Civita connection on M .
We would also like a formula for the second derivative such as that given
in Proposition 2.4.1. In order to state this formula, we define a linear operator

141
K : Tf L2α 2α
1 (Σ, M ) → Tf L1 (Σ, M ) in terms of the Riemann curvature R of M
by the formula
     
∂f ∂f ∂f ∂f
hK(V ), W idA = R V, + R V, , W dudv,
∂u ∂u ∂v ∂v

where (u, v) are isothermal coordinates on M .

Proposition 3.7.1. If f is a critical point for Eα,ψ,ω , then
Z
d2 Eα,ψ,ω (f )(V, W ) = α (1 + |df |2 )α−1 [hDV, DW i − hK(V ), W i]dA
Σ
Z
+ 2α(α − 1) (1 + |df |2 )α−2 hdf, DV ihdf, DW idA
Σ
Z
+ ψ · α(V, W )dA, (3.40)
Σ

for all V, W ∈ Tf L2α

1 (Σ, M ), where α is the second fundamental form of M in
RN .
Although (3.40) may look complicated at first, note that it specializes to (3.40)
in the case where the Riemannian manifold M is just Euclidean space and
ψ = 0. Moreover, we can set α = 1, and obtain the formula for the Hessian of
the ordinary energy Eω at a critical point:
Corollary 3.7.2. If f is ω-harmonic, then
Z
2
d Eω (f )(V, W ) = [hDV, DW i − hK(V ), W i]dA, (3.41)
Σ

for all V, W ∈ Tf L2α

1 (Σ, M ).

The proof of Proposition 3.7.1 is quite similar to the proof of Proposition 2.4.1.
For simplicity, we assume that ψ = 0. We consider a variation of f which has
its support within a given coordinate chart (U, (x, y)) on Σ. Recall that such a
variation is a smooth family of maps t 7→ f (t) in L2k (Σ, M ) with f (0) = f , and
let
∂α
α(x, y, t) = f (t)(x, y), V (x, y) = (x, y, 0) ∈ Tf (x,y) M.
∂t
Setting γ(t)2 = (1 + |df (t)|2 )α−1 , we obtain

d2

2

d Eα,ω (f )(V, V ) = 2 (Eω (f (t))
dt
Z   t=0    
∂ 2 ∂α D ∂α 2 ∂α D ∂α 
=α γ , +γ , dxdy 
Σ ∂t ∂x ∂t ∂x ∂y ∂t ∂y t=0

142
where D as usual denotes the covariant derivative of the Levi-Civita connection
on the ambient Riemannian manifold M . We can rewrite this as
Z    
2 2 D ∂α D ∂α D ∂α D ∂α
d Eα,ω (f )(V, V ) = α γ , + ,
Σ ∂t ∂x ∂t ∂x ∂t ∂y ∂t ∂y
∂α D2 ∂α ∂α D2 ∂α
    

+ , 2 + , 2 dxdy 
∂x ∂t ∂x ∂y ∂t ∂y
Z 2
    t=0
dγ ∂α D ∂α ∂α D ∂α 
+2α(α − 1) , + , dxdy  . (3.42)
Σ dt ∂x ∂t ∂x ∂y ∂t ∂y t=0

When we evaluate at t = 0, we find that

dγ 2 hdf, DV i
= ,
dt 1 + |df 2
where V is the variation field, so we can rewrite (3.42) as
Z    
DV DV DV DV
d2 Eα,ω (f )V, V ) = α γ2 , + ,
Σ ∂x ∂x ∂y ∂y
   
∂f D DV ∂f D DV
+ , + , dxdy
∂x ∂t ∂x ∂y ∂t ∂y
Z
+ 2α(α − 1) (1 + |df |2 )α−2 hdf, DV i2 dA.
Σ

Finally, applying the definition of the Riemann-Christoffel curvature tensor R,

we obtain
Z    
DV DV DV DV
d2 Eα,ω (f )(V, V ) = γ2 , + ,
Σ ∂x ∂x ∂y ∂y
       
∂f ∂f ∂f ∂f
+ , R V, V + , R V, V
∂x ∂x ∂y ∂y
   
∂f D DV ∂f D DV
+ , + , dxdy
∂x ∂x ∂t ∂y ∂y ∂t
Z
+ 2α(α − 1) (1 + |df |2 )α−2 hdf, DV i2 dA.
Σ

An integration by parts and use of the Euler-Lagrange equation eliminates the

third term, thereby yielding (3.40) in the case where ψ = 0.
If f is a critical point for the function Eα,ψ,ω , we can define a second-order
linear partial-differential operator
Z
L : L21 (f ∗ T M ) → L2−1 (f ∗ T M ) by d2 Eα,ψ,ω (f )(V, W ) = hL(V ), W idA,
Σ

and we call L the Jacobi operator. The Jacobi operator L restricts to a con-
tinuous linear map L : L2k (f ∗ T M ) → L2k−2 (f ∗ T M ) for all k ≥ 2, and if

143
ι : L2k−2 (f ∗ T M ) → L2k (f ∗ T M ) is the inclusion and λ ∈ C, we have an as-
sociated operator Lλ = L − λι. The fundamental Theorem 2.4.2 on formally
self-adjoint elliptic operators now applies to the Jacobi operator L, and hence
L is a Fredholm operator, which satisfies the additional conditions:

1. for each λ ∈ C the eigenspace Wλ = Ker(Lλ ) is finite-dimensional.

2. all the elements of Wλ are C ∞ ,

3. if Wλ is empty, then Lλ possesses a Green’s operator inverse,

4. if Wλ is nonempty, that is, λ is an eigenvalue, then λ ∈ R, and
5. the eigenvalues can be arranged in a sequence

λ1 < λ2 < · · · < λi < · · · with λi → ∞,

and only finitely many of the eigenvalues are negative.

On the other hand, if f is a critical point for Eα,ψ,ω , we can also define an
inner product

hh·, ·ii : Tf L2α 2α

1 (Σ, M ) × Tf L1 (Σ, M ) −→ R

by the somewhat complicated formula,

Z
hhV, W ii = α (1 + |df |2 )α−1 [hDV, DW i + hV, W i]dA
Σ
Z
+ 2α(α − 1) (1 + |df |2 )α−2 hdf, DV ihdf, DW idA,
Σ

which is useful because it so close to (3.41). Since f is a smooth function, this

is actually equivalent to the simpler L21 inner product hh·, ·ii0 defined by
Z
hhV, W ii0 = [hDV, DW i + hV, W i]dA.
Σ

We can then define a second-order linear partial-differential operator

Z
J : L21 (f ∗ T M ) → L2−1 (f ∗ T M ) by hhV, W ii = hJ(V ), W idA.
Σ

Once again J restricts to a Fredholm operator J : L2k (f ∗ T M ) → L2k−2 (f ∗ T M )

for all k ≥ 2. However, since hh·, ·ii is positive-definite, the eigenvalues of J
are all positive and J itself has a Green’s operator inverse. It is here that the
complicated formula for the L21 -inner product on L21 (f ∗ T M ) pays off, because
L − J is of zero order, and hence L − J : L2k (f ∗ T M ) → L2k−2 (f ∗ T M ) is a
compact operator, for all k ≥ 1.

144
 In parallel with the theory of smooth closed geodesics, we can define a con-
tinuous linear map

A : L21 (f ∗ T M ) → L21 (f ∗ T M ) by hhAV, W ii = d2 Eα,ψ,ω (f )(V, W ). (3.43)

Then the operator

A = J −1 ◦ L : L21 (f ∗ T M ) → L21 (f ∗ T M )

is Fredholm, and restricts to a Fredholm operator on each L2k (f ∗ T M ), for k ≥ 2.

By the Lp theory for elliptic operators, it also restricts to a Fredholm operator
on each Lpk (f ∗ T M ), for p ≥ 2 and k ≥ 2.
But if we let p = 2α and choose q so that (1/p) + (1/q) = 1, then we can also
define the operator A directly on Lp1 (f ∗ T M ) by (E:explicitformulaA). Indeed,
any V ∈ Lp1 (f ∗ T M ) defines a continuous linear functional
0
W 7→ d2 Eα,ω (f )(V, W ), for W ∈ Lq1 (f ∗ T M ),

and by the duality between Lp1 and Lq1 this linear functional defines an element
AV ∈ Lp1 (f ∗ T M ). This defines a continuous linear map

A : Tf L2α 2α
1 (Σ, M ) −→ Tf L1 (Σ, M ).

such that
A − id : Tf L2α 2α
1 (Σ, M ) −→ Tf L1 (Σ, M )

is a compact operator, and thus A is a Fredholm operator as before.

Definition. We say that the critical point f for Eα,ψ,ω is Morse nondegenerate
if A is an isomorphism. The Morse index of f is the sum of the dimensions of
the eigenspaces of L for negative eigenvalues. Finally, Eα,ψ,ω : L2α
1 (Σ, M ) → R
is a Morse function if all of its critical points are Morse nondegenerate.
As in the theory of smooth closed geodesics, we can choose ψ so that Eα,ψ,ω is
a Morse function:
Theorem 3.7.3. For a residual set of ψ ∈ L2k (Σ, RN ), the function Eα,ψ,ω is a
Morse function.
The argument is virtually identical to the argument we presented for Theo-
rem 2.7.1.
As in the theory of smooth closed geodesics, we can set

M0 = {f ∈ L2α
1 (Σ, M ) : f is constant },

and then M0 is a nondegenerate critical submanifold for Eα,ω of Morse index

0
zero. Just as as at the end of §2.7, we can perturb Eα,ω to a function Eα,ω :
2α
L1 (Σ, M ) → R such that
0
1. Eα,ω = Eα,ω on a neighborhood U of M0 ,

145
 0
2. Eα,ω = Eα,ψ,ω outside a larger neighborhood V of M0 , for some ψ ∈
L2k (Σ, RN ), and
0
3. all critical points of Eα,ω either lie within M0 or are Morse nondegenerate
and do not lie in V .
The Handle Addition Theorem 2.9.1 can now be modified so that it applies
0
to the Morse function Eα,ω , and hence the Morse inequalities of §2.10 can be
0
extended to Eα,ω . Let M = L2α 1 (Σ, M ) with its standard Finsler metric, and
let
0
Ma = {f ∈ L2α 1 (Σ, M ) : Eα,ω (f ) ≤ a}.

0
Theorem 3.7.4. If the interval [a, b] contains a single critical value c for Eα,ω ,
0 0
there is exactly one critical point p for Eα,ω such that Eα,ω (p) = c and this
critical point is Morse nondegenerate of Morse index λ, then Mb is homotopy
equivalent to Ma with a handle of index λ attached.
Indeed, the proof presented in §2.9 was designed precisely so that it would apply
to this case. The main difference is that now the model space E is a Banach
space instead of a Hilbert space. and one must make some small modifications
to the argument to account for this.
Moreover, the notion of gradient-like vector field presented in §2.9 was designed
0
so that it would apply directly to the function Eα,ω : L2α
1 (Σ, M ) → R. There-
fore, we can define stable and unstable manifolds of nondegenerate critical points
just as we did before.
0
Suppose that Eα,ω is a perturbation of Eα,ω which agrees with Eα,ω in a
neighborhood of M0 and satisfies the condition that all its critical points are
either in M0 or are Morse nondegenerate. If X is a gradient-like vector field
0 0
for Eα,ω , we let Ck (Eα,ω , X ) denote the free Z-module generated by the critical
0
points fk,1 , fk,2 . . . for Eα,ω of index k and let ∂ be the Z-module homomorphism
0 0
∂ : Ck (Eα,ω , X ) −→ Ck−1 (Eα,ω ,X)

defined by X
∂(fk,j ) = ajq fk−1,q , (3.44)
q

where ajq ∈ Z is the oriented number of trajectories from fk,j to fk−1,q , the
orientation being determined as in §2.11.
Theorem 3.7.5. The Z-module homomorphisms thus defined satisfy the iden-
tity ∂ ◦ ∂ = 0 and the resulting Morse-Witten complex
0 0 0
· · · → Ck+1 (Eα,ω , X ) → Ck (Eα,ω , X ) → Ck−1 (Eα,ω ,X) → ···

calculates the homology of the pair (C 0 (Σ, M ), M0 ).

The proof is a straightforward modification of the argument presented in §2.11.

146
3.8 Local control of energy density
Our next goal is to investigate the limit of α-energy critical points as α → 1. The
key technique here is the Bochner Lemma, which gives a uniform C 1 estimate
on harmonic and α-harmonic maps whose total energy is small (less than some
positive constant 0 ). This in turn is crucial in understanding a key feature of
harmonic and α-harmonic maps, the phenomenon of bubbling. Throughout this
section, we use the notation
 2  2 !
1 2 1  ∂f   ∂f 
e(f ) = |df | = +   dudv,
2 2  ∂u  ∂v

for isothermal parameters (u, v) on Σ.

Lemma 3.8.1. Suppose that f : D1 → M is a harmonic map, where D1 is
the unit disk in the complex plane, with the standard Euclidean metric ds2 and
M is a compact Riemannian manifold with sectional curvatures satisfying the
inequality K(σ) ≤ 1. Then for any r0 > 0,
Z
e(f )dA < π(r02 − 8r04 ) ⇒ max σ 2 sup e(f ) < 4r02 . (3.45)
D1 σ∈(0,1] D1−σ

Proof: Choose σ0 ∈ (0, 1] so that

σ02 sup e(f ) ≥ σ 2 sup e(f ), for all σ ∈ (0, 1],

D1−σ0 D1−σ

and choose p0 ∈ D1−σ0 so that

e0 = e(f )(p0 ) = sup{e(f )(p) : p ∈ D1−σ0 }.

If σ02 e0 < 4r02 , the lemma is proven. So it suffices to assume that σ02 e0 ≥ 4r02 ,
and derive a contradiction. But then
 σ 2
0
sup e(f ) ≤ σ02 sup e(f ) = σ02 e0 ⇒ sup e(f ) ≤ 4e0 .
2 D1−σ0 /2 D1−σ0 D1−σ0 /2

Moreover, σ02 e0 ≥ 4r02 implies that

r σ
√0 ≤ 0 ,
e0 2
so we can define a new harmonic map g : Dr0 → M by
 
q
g(q) = f p0 + √ such that e(g)(0) = 1 and e(g) ≤ 4.
e0

It now follows from (3.28), which follows from the Bochner Lemma, that

∆(e(g)) ≥ −2(e(g))2 ≥ −32.

147
 Thus if we define a smooth function h : Dr0 → R in terms of the standard
coordinates (u, v) by
h = e(g) − [1 − 16(u2 + v 2 )],
then we find that
∆h = ∆(e(g)) + 32 ≥ 0,
and hence by the maximum principle,
Z Z
0 = h(0) ≤ hdA ≤ [e(g) − (1 − 16(u2 + v 2 ))]dA
Dr0 Dr0
Z Z 2π Z r0
≤ e(g)dA − (1 − 16r2 )rdrdθ,
Dr0 0 0

and an integration yields the inequality

Z
e(f )dA > π(r02 − 8r04 ).
D1

Thus we have proven the contrapositive of the assertion in the lemma.

Corollary 3.8.2. Suppose that f : D1 → M is a harmonic map. Then
Z Z
π 2 8
e(f )dA < ⇒ max σ sup e(f ) < e(f )dA.
D1 16 σ∈(0,1] D1−σ π D1

To prove this we note that if r0 ≤ (1/4),

Z
π 2
e(f )dA = (r ) ≤ π(r02 − 8r04 )
D1 2 0
Z
8
⇒ max σ 2 sup e(f ) < 4r02 = e(f )dA.
σ∈(0,1] D1−σ π D1

Corollary 3.8.3. Suppose that Dr is the disk of radius r in the complex plane
and f : Dr → M is a harmonic map. Then
Z Z
π 8
e(f )dA < ⇒ max σ 2 sup e(f ) < e(f )dA.
Dr 16 σ∈(0,r] Dr−σ π Dr

To prove this, simply note that under rescaling from D1 to Dr , the energy
integral remains unchanged, while
1 1
|df | 7→ |df |, e(f ) 7→ e(f )
r r2
and
 σ 2
max σ 2 sup e(f ) 7→ max sup e(f ) = max τ 2 sup e(f ).
σ∈(0,1] D1−σ σ∈(0,1] r Dr−rσ σ∈(0,r] Dr−τ

148
Of course, we would like versions of the previous lemmas for the α-energy Eα,ω
when α is sufficiently close to one.
Lemma 3.8.4. Suppose that M is a compact Riemannian manifold whose
sectional curvatures satisfy the bound K(σ) ≤ 1. There exists α0 ∈ (1, ∞) and
for each α ∈ [1, α0 ] a homogeneous second-order elliptic operator Lα such that
such that the coefficients of Lα tend uniformly to those of ∆ as α → 1, and

Lα (e(f )) ≥ [2K̃e(f ) − 2e(f )2 ],

where K̃ is the Gaussian curvature of the Riemannian metric on Σ.

Proof: It follows from (3.21) and the discussion leading to (3.28) that

∆η (e(f )) = |∇df |2 + hd[(∆η f )> ], df i + 2̃Ke(f ) − 2(e(f ))2 .

Recall that equation (3.35) for α-harmonic maps can be written in the form

(∆η f )> = −(α − 1)d(log γ 2 ) · df, where γ 2 = 1 + |df |2 ,

Since
d(|df |2 ) 2h∇df, df i
d(log γ 2 ) = = ,
1 + |df |2 1 + |df |2
∇d(|df |2 )(1 + |df |2 ) + (d(|df |2 ))2 ∇d(|df |2 ) 4h∇df, df i2
∇d(log γ 2 ) = 2 2
= + ,
(1 + |df | ) (1 + |df | ) (1 + |df |2 )2
2

we conclude that there is a homogeneous second-order differential operator Q

(with uniformly bounded coefficients which depend on df ) such that

h∇df, df i2
|h∇(∆f ), df i| ≤ (α − 1)Q(e(f )) + (α − 1)(constant) ,
(1 + |df |2 )2

the constant being independent of f and α. The second term can be absorbed
in |∇df |2 when α is sufficiently close to one, while the first term can be absorbed
in the elliptic operator Lα .
One can now use Lemma 3.8.4 in place of the standard Bochner Lemma to prove
a version of Lemma 3.8.3 for α-harmonic maps.
Lemma 3.8.5. Suppose that M is a compact Riemannian manifold whose
sectional curvatures satisfy the bound K(σ) ≤ 1 and that Dr is the disk of
radius r in the complex plane, with the standard Euclidean metric ds2 . There
exists an α0 > 1 and an 0 > 0 with the following property: If f : Dr → M is
an α-harmonic map,
Z Z
e(f )dA < 0 ⇒ max σ 2 sup e(f ) < 3 e(f )dA. (3.46)
Dr σ∈(0,r] Dr−σ Dr

149
The proof is a straightforward modification of the proof of Lemmas 3.8.1 to
3.8.3, in which we use Lemma 3.8.4 for the elliptic operator Lα instead of the
ordinary Bochner Lemma for the Laplace operator ∆η .
We can next relax the condition that the Riemannian metric on Σ be flat and
rescale the metric on M so that it satisfies the condition K(Σ) ≤ K0 for K0 > 0:
Lemma 3.8.6. Suppose that M is a compact Riemannian manifold whose
sectional curvatures satisfy the bound K(σ) ≤ K0 where K0 > 0 and that Dr
is the disk of radius r in the complex plane, a Riemannian metric ds2 which is
related to the flat metric ds20 by inequalities of the form
1 2
ds ≤ ds2 ≤ L2 ds20 ,
L2 0
L being a constant. There exists an α0 > 1 and constants 0 > 0 and CL
depending on L such that: If f : Dr → M is an α-harmonic map with α ∈
(1, α0 ],
Z Z
0 2
e(f )dA < √ ⇒ max σ sup e(f ) < CL e(f )dA. (3.47)
Dr K0 σ∈(0,r] Dr−σ Dr

Proof: The estimates (3.46) holds for the energy density e0 (f ) with respect to
the flat metric ds20 . Now simply use the inequalities
1
e0 (f ) ≤ e(f ) ≤ L2 e0 (f )
L2
√
to obtain (3.47) for K0 = 1. Now note that multiplying the metric by K0 also
multiplies the energy density by the same factor.
Inequality (3.47) is typically used as follows: Let Σ be a compact Riemann
surface and M a compact Riemannian manifold, r0 > 0 a fixed radius. Then
there are constants 0 > 0, C > 0 and α ∈ [1, α0 ] which depends upon Σ, M
and r0 such that if Dr is a disk of radius r ≤ r0 in Σ and f : Dr → M is an
α-harmonic map,
Z Z
C
e(f )dA < 0 ⇒ sup e(f ) < 2 e(f )dA. (3.48)
Dr Dr/2 r Dr

In other words, if the energy on a ball of small radius is sufficiently small, it

gives a bound on the energy density itself on a ball of half the radius.
Theorem 3.8.7. Suppose that M is a compact Riemannian manifold and Σ
is a closed connected Riemann surface. If {ωm } is a sequence of conformal
structures ∈ T such that ωm → ω∞ ∈ T and {fm : Σ → M } is a sequence of
(αm , ωm )-harmonic maps such that αm → 1 and E(fm ) ≤ E0 for some constant
E0 > 0, then there is a finite collection of points

{p1 , . . . , pl } ⊆ Σ,

150
and a subsequence of {fm } (which we still denote by {fm }), such that {fm }
converges uniformly in C k on compact subsets of Σ − {p1 , . . . , pl }, for any non-
negative integer k, to an ω∞ -harmonic map

f∞ : Σ − {p1 , . . . , pl } −→ M.

Proof: We begin by choosing 0 > 0 so that when r ≤ r0 , (3.48) implies that

Z
C0
e(f )dA < 0 ⇒ sup e(f ) < 2 . (3.49)
Dr Dr/2 r

Let rm = (1/2)m , for each positive integer m and let Om be an open cover of
Σ by disks Drm (pi ), so that each each point of Σ is covered by at most h disks
and the disks Drm /2 (pi ) still cover M . (We can choose h to be independent of
m.) Then
XZ
e(fm ) ≤ hE0 .
i Drm (pi )

Hence there are at most hE0 /0 of the disks in the cover Om on which
Z
e(fm ) ≥ 0 .
Drm (xi )

For each j we let {p1m , . . . , plm } be the center points of these disks. After
possibly passing to a subsequence, we can arrange that

p1m → p1 , . . . , plm → pl .

Then the {fm }’s are uniformly bounded and equicontinuous on compact subsets
of Σ − {p1 , . . . , pl }. By Arzela’s Theorem the {fm }’s converge to a continuous
function f∞ on Σ − {p1 , . . . , pl }, and the convergence is uniform on compact
subsets.
To finish the proof, we use the process of “elliptic bootstrapping” to show
that fm → f∞ in C k on Σ − {p1 , . . . , pl }, uniformly on compact subsets, for any
k ≥ 0. Indeed, if M is isometrically imbedded in E N , the equation for which f
must satisfy to be an α-harmonic map is

∂2f ∂2f B(d2 f, df )

2
+ 2 = −α(f )(df, df ) − (α − 1) df, (3.50)
∂x ∂y 1 + |df |2

where α(f ) and B are bilinear maps, α(f ) being the familiar second fundamental
form. It follows from (3.49) that {dfm } is bounded in Lp for all p on any disk
within Σ − {p1 , . . . pl }. It then follows from multiplication theorems for Sobolev
spaces that the right-hand side of (3.50) is bounded in Lp for all p on any such
disk, and hence elliptic estimates give that f is bounded in Lp2 for all p. But
now one of the Sobolev imbedding theorems implies that a subsequence of {fm }

151
converges in C 1,β for some β > 0 on any compact subset of Σ − {p1 , . . . pl }. Now
we can apply Schauder estimates to see that a subsequence of {fm } converges
in C 2,β , then in C 3,β , and so forth. Finally we obtain a subsequence of {fm }
which converges uniformly in every C k on compact subsets of Σ − {p1 , . . . , pl },
and the theorem is proven.
Theorem 3.8.8. Suppose that M is a compact Riemannian manifold with
nonpositive sectional curvatures. If ω is a conformal structure on the compact
Riemann surface Σ, there is an ω-energy minimizing harmonic map in any free
homotopy class [Σ, M ].
In this case, Lemma 3.8.6 implies that for some α0 ∈ (1, ∞) and some r0 ≥ 0,
whenever Dr is a disk of radius r ≤ r0 in Σ and f : Dr → M is an α-harmonic
map with α ∈ (1, α0 ],
Z
sup e(f ) < 4CL e(f )dA. (3.51)
Dr/2 Dr

Thus we do not need the complicated covering argument utilized in the proof
of Theorem 3.8.7.
Instead, we simply take a sequence αm → 1 and for each αm choose an
αm -harmonic map fm : Σ → M which lies in a given component of [Σ, M ] and
minimizes the αm -energy on that component. Then (3.51) implies that e(fm )
is bounded on any ball Dr/2 , which implies that {fm } converges uniformly on
compact subsets. Elliptic bootstrapping then implies that the sequence {fm }
converges uniformly in all C k , proving the theorem.
Theorem 3.8.8 is just the Eells-Sampson Theorem 3.1.1 in the special case where
Σ has dimension two. Similarly, we can prove Hartman’s Theorem 3.1.2 in the
special case where Σ is a surface:
Theorem 3.8.9. Suppose that M is a compact Riemannian manifold with
negative sectional curvatures. If (Σ, ω) is any Riemann surface, there is at most
one ω-harmonic map in any free homotopy class [Σ, M ].
Indeed, the corresponding statement for (α, ω)-harmonic maps follows from the
fact that since
     
∂f ∂f ∂f ∂f
R V, + R V, , V ≤ 0,
∂u ∂u ∂v ∂v

any (α, ω)-harmonic map is stable by (3.40). If there were two (α, ω)-harmonic
maps in the same component of C 0 (Σ, M ), there would have to be an unstable
critical point by the Morse inequalities, which cannot happen.
To prove the corresponding statement for the limiting case of ω-harmonic
maps, we can use the implicit function theorem. For this, we define a smooth
map
Eω? : L2k (Σ, M ) × [1, α0 ) −→ R by Eω? (f, α) = Eα,ω (f ),

152
and observe that at α = 1 we obtain the usual energy up to a constant,
Eω? (f, 1) = Eω (f ) + (1/2). We can differentiate to obtain the Euler-Lagrange
map
Fω∗ : L2k (Σ, M ) × [1, α0 ) −→ L2k−2 (Σ, T M ).
If f is an (α, ω)-harmonic map, we can compose the differential DFω∗ of Fω∗ at
(f, α) with the projection onto the vertical obtaining the corresponding Jacobi
operator,

L = πV ◦ DFω∗ (f, α) : Tf L2k (Σ, M ) ⊕ R −→ Tf L2k−2 (Σ, M ).

Note that L is a Fredholm operator with Fredholm index one, and that f is a
Morse nondegenerate critical point for Eα,ω if and only if L is surjective.
Suppose now that f : Σ → M is an ω-conformal harmonic map, and note
that f must be Morse nondegenerate by Corollary 3.7.2. It follows from the
implicit function theorem that if U is a sufficiently small C 2 neighborhood of
(f, 1) in L2k (Σ, M ) × [1, α0 ), then

(Fω∗ )−1 (zero section of T (Map(Σ, M )))

is a smooth submanifold of U of dimension one. This submanifold consists of

critical points of Eα,ω for α varying in some interval [1, α0 ) for which the corre-
sponding Jacobi operators L are surjective. All the points of this submanifold
are Morse nondegenerate critical points for Eα,ω .
Thus two distinct ω-harmonic maps f, g : Σ → M in the same component
of C 0 (Σ, M ) would give rise to two distinct one-parameter families fα , gα of
(α, ω)-harmonic maps in the same component of C 0 (Σ, M ), contradicting the
uniqueness of (α, ω)-harmonic maps. This finishes the proof of Theorem 3.8.9.

3.9 Bubbling
In §3.7 we have shown existence of many α-energy critical points, corresponding
to numerous topological constraints, while in §3.8 we have shown that given a
sequence {fm : Σ → M } of αm -harmonic maps with αm > 1 such that αm → 1
and E(fm ) ≤ E0 for some constant E0 > 0, there is a collection of points
{p1 , . . . , pl } and a subsequence of {fm } (which we still denote by {fm }) which
converges uniformly on compact subsets of Σ − {p1 , . . . , pl }, together with the
first k derivatives to a map

f∞ : Σ − {p1 , . . . , pl } −→ M, (3.52)

which is harmonic on Σ − {p1 , . . . , pl }. The following theorem implies that f∞

extends to a harmonic map defined on the entire Riemann surface Σ:
Removeable Singularity Theorem 3.9.1. Let D be the unit disk in the
complex plane C. If
f : D − {0} −→ M

153
is a harmonic map of finite energy, then f extends to a smooth harmonic map
on the entire disk D.
We apologize to the reader for not giving a proof of this theorem in these notes.
It is proven as Theorem 3.6 in [68].
Returning now to our sequence {fm } of αm -harmonic maps with αm → 1
and E(fm ) ≤ E0 , we note that as measures, the sequence of Radon measures
{e(fm )dA} converges weakly to
X
e(f∞ )dA + mi δpi dA,
i=1l

where mi is a nonnegative constant and δpi is the Dirac delta-function located at

pi . If mi is zero, we can throw the point pi away, since in this case Lemma 3.8.6
implies that after possibly passing to a subsequence {fm } will converge on a
neighborhood or pi . If mi > 0 is nonzero, we call pi a bubble point of the
sequence and mi measures the amount of energy lost at the bubble point.
We next investigate what happens at the bubble points p1 , . . . , pl . For each
m, choose qim ∈ Drm (pim ) so that

e(fm )(qim ) = sup{e(fm )(q) : q ∈ Drm (pim )}.

Note that qim → xi as j approaches ∞, and let

p
bim = e(fm )(qim ).

If no subsequence of the fim ’s converge near pi then bim → ∞. Assume that

the latter alternative holds, and define
 
p
f˜m : Drm bim (0) → M by f˜m (q) = fm qim + .
bim

One readily verifies that e(fim ) ≤ 1 and e(fim )(0) = 1. Therefore a subsequence
of the fm ’s converges to a nonconstant harmonic map

f˜∞ : C − {q1 , . . . qm } −→ M,

the convergence being uniform in C k for all k on compact subsets of C −

{q1 , . . . qm }. At the finitely many points {q1 , . . . qm }, further bubbling can occur.
Recall that as a Riemann surface the complex plane is just the Riemann
sphere S 2 minus a point. Therefore the removable singularity theorem implies
that f˜∞ extends to a smooth harmonic map fˆ∞ : S 2 → M .
It is clearly of interest to understand what α-energy critical points look like
near bubble points. This problem has been studied by Parker and Wolfson [64],
[63] and Chen and Tian [12], among others.
Remark 3.9.2. Each harmonic two-sphere which bubbles off carries with it a
certain minimal amount of energy. Indeed, a harmonic two-sphere h : S 2 → M

154
is automatically conformal and hence a minimal surface, and from the Gauss
equation, its Gaussian curvature K : S 2 → R is bounded above by K0 , the
maximum value of all sectional curvatures on M . Hence by the Gauss-Bonnet
theorem,
Z
4π
4π = KdA ≤ K0 (Area of h(S 2 )) ⇒ E(h) ≥ .
S2 K 0

In particular, if the energy of a sequence of αm -harmonic maps is bounded by

E0 , the number of harmonic two-spheres that can bubble off in the limit is
≤ (4πE0 )/K0 .
We can now prove a theorem due to Sacks and Uhlenbeck [68], which can be
thought of as the analog within minimal surface theory of the Fet-Lusternik
Theorem 2.3.2 on existence of smooth closed geodesics:
Theorem 3.9.3. If M is a compact simply connected Riemannian manifold,
then M contains at least one nonconstant minimal two-sphere.
Proof: In many ways, the argument is very similar to that given in §2.3. There
is a least integer q ≥ 2 such that Hq (M̃ ; Z) 6= 0, and it follows from the Hurewicz
theorem that

πi (M ) = 0, for 0 < i < q, and πq (M ) ∼

= Hq (M, Z) 6= 0.

Let M be the space of smooth maps from S 2 to M and let π : M → M by

π(f ) = f (0), the evaluation of f at the basepoint 0 ∈ S 2 = C∪{∞}. It is a well-
known fact from homotopy theory that π is a fibration with fiber Mp = π −1 (p),
the set of basepoint preserving maps from S 2 to M . Moreover,

πk (Mp ) ∼
= πk+2 (M )

and the fibration π induces a long exact sequence

· · · → πk (Mp ) → πk (M) → πk (M ) → πk−1 (Mp ) → · · · .

We note, moreover, that π∗ : πk (M) → πk (M ) possesses a right inverse

i∗ : πk (M ) → πk (M) induced by the inclusion i:M →M

which takes a point to the constant map at that point. Hence the long exact
sequence splits and we conclude that

πk (M) ∼
= πk (M ) ⊕ πk (Mp ) ∼
= πk (M ) ⊕ πk+2 (M ).

Thus the homotopy groups of M are completely determined by the homotopy

groups of M . In particular,

πq−2 (M) ∼
= πq (M ) 6= 0.

155
 Since M is simply connected, q ≥ 2 and πq−1 (M) = ∼ πq (M ) is abelian.
Moreover, we can identify πq−2 (M) with [S q−2 , M], the space of free homotopy
classes of maps from S q−2 into M. Choose a nonzero element α ∈ [S q−2 , M].
Let

F = {g(S q−2 ) such that g : S q−2 → M is a continuous map in [α]}.

Clearly, F is an ambient isotopy invariant family of sets. Hence Minimax(Eα , F)

is a critical value for Eα .
Recall that M is isometrically imbedded in an ambient Euclidean space
RN and let M (δ) denote the open δ-neighborhood of M in RN for δ > 0.
By the tubular neighborhood theorem, if δ is sufficiently small, M is a strong
deformation retract of M (δ). Moreover, for  > 0 sufficiently small, any map
f : S 2 → M of energy <  can be contracted to its center of mass in RN without
leaving M (δ).
Suppose now that g(S q−1 ) ⊂ J −1 ([0, ]). Then g is homotopic to a smooth
map
g̃ : S q−1 → M0 , where M0 = {γ ∈ M : γ is constant }.
Hence Minimax(Eα , F) ≥ . We now take a decreasing sequence of real num-
bers {αm } with αm → 1. The Minimax Theorem 1.11.2 gives a corresponding
sequence {fm } of critical points for Eαm which have energy ≥  > 0. Either
a subsequence converges to a nonconstant harmonic map or a bubble forms in
the limit. In the latter case, the bubble provides a nonconstant harmonic map.
In either case we get a nonconstant harmonic map f∞ : S 2 → M which is au-
tomatically conformal, since there is only one conformal structure on S 2 . The
image is a nonconstant minimal two-sphere, which proves the theorem.

3.10 Existence of minimizing spheres

Let M be a compact manifold and suppose that f : D1 → M is a map with
f (∂D1 ) = p which represents an element [f ] ∈ π2 (M, p). Suppose, moreover,
that γ : [0, 1] → M is a smooth map with γ(0) = p = γ(1) which represents an
element [γ] ∈ π1 (M, p). We can then define γ ? f as follows: Let (r, θ) be polar
coordinates on D1 and set
(
f (2r, θ), for 0 ≤ r ≤ 1/2,
g(r, θ) = (γ ? f )(r, θ) =
γ(2r − 1), for 1/2 ≤ r ≤ 1.

Then [γ], [f ] 7→ [g] = [γ ? f ] gives an action of π1 (M, p) on π2 (M, p), which

makes π2 (M, p) into a Z[π1 (M, p)]-module, where Z[π1 (M, p)] is the group ring
of π1 (M, p). This action is discussed in more detail in Chapter 4 of [35].
Theorem 3.10.1 (Sacks and Uhlenbeck). Suppose that M is a compact
Riemannian manifold. Then a generating set for π2 (M, p) as a Z[π1 (M, p)]-
module can be represented by minimal two-spheres, possibly with branch points.

156
Note that the minimal two-spheres need not pass through the point p. If M
is simply connected, it follows from the Hurewicz Theorem that π2 (M, p) ∼
=
H2 (M ; Z), and hence we obtain:
Corollary 3.10.2. If M is a compact simply connected Riemannian man-
ifold, then a set of generators for H2 (M ; Z) can be represented by minimal
two-spheres.
Before proving Theorem 3.10.1, we make a few remarks regarding the symmetry
of the functions E, Eα : C 2 (S 2 , M ) → R. We regard S 2 as the one-point
compactification C ∪ {∞} of the complex plane C with the standard coordinate

z = x + iy = reiθ = eu+iθ ,

and note that E is invariant under all the linear fractional transformations: If
φ : S 2 → S 2 is the diffeomorphism defined by
 
az + b a b
φ(z) = , for ∈ SL(2, C),
cz + d c d

then E(f ◦ φ) = E(f ). Note that

   
a b −1 0
= ⇒ f ◦ φ = f,
c d 0 −1

so the group of linear fractional transformations is actually

SL(2, C)
G = P SL(2, C) = .
±I
In contrast, when α > 1, the maps Eα0 , Eα : C 2 (S 2 , M ) → R defined by
Z Z
1 1
Eα0 (f ) = |df |2α dA and Eα (f ) = (1 + |df |2 )α dA.
2 Σ 2 Σ

are invariant only under the smaller group of isometries SO(3) ⊆ P SL(2, C).
Indeed, critical points of Eα0 must be parametrized so that the “center of mass”
is zero:
Lemma 3.10.3. Let X : S 2 → R3 denote the standard inclusion, and let
0 denote the origin in R3 . If f ∈ C 2 (S 2 , M ) is a critical point for Eα0 with
α ∈ (1, ∞), then Z
X|df |2α dA = 0.
S2

Moreover, there is a smooth function ψα : [0, ∞) → R such that ψα (0) = 0,

ψα (t), ψα0 (t) > 0 for t > 0, and if f is a critical point for Eα , then
Z
Xψα (|df |2 )dA = 0. (3.53)
S2

157
In the argument for this lemma, we utilize the standard metric on S 2 of constant
curvature one, expressed in terms of our standard coordinates by
4 1
ds2 = (dr2 + r2 dθ2 ) = (du2 + dθ2 ),
(1 + r2 )2 cosh2 u
where r = e−u . (This metric is related by scaling to the constant curvature
metric of total area one.) For t ∈ R, we define a linear fractional transformation

φt : S 2 → S 2 by u ◦ φt = u + t, θ ◦ φt = θ,

so that t 7→ φt is a one-parameter subgroup. The energy density is then given

by the formula
 2  2 !2
1  ∂f   ∂f 
e(f ◦ φt ) = +   cosh2 (u + t),
2  ∂u  ∂θ

and it is straightforward to calculate its derivative at t = 0:

 
d  1 d 2

e(f ◦ φt )
 = |d(f ◦ φt )| 
dt t=0 2 dt t=0
 2  2 !
 ∂f   ∂f 
=   +   sinh u cosh u = |df |2 tanh u. (3.54)
∂u ∂θ

Using this fact, we can perform a straightforward calculation to obtain

 Z
d 0
|df |2α (tanh u)dA.

E (f ◦ φt )
 = (α − 1)
dt α t=0 S 2

In terms of the standard Euclidean coordinates (x, y, z) on R3 , S 2 is represented

by the equation x2 + y 2 + z 2 = 1, and a straightforward computation using
stereographic projection from the north pole to the (x, y)-plane shows that

r2 − 1 sinh u
z= = = tanh u,
r2 + 1 cosh u
so if f is a critical point for Eα0 ,
 Z
d 0
|df |2α zdA.

0= Eα (f ◦ φt ) = (α − 1)
dt t=0 S 2

But we can take the z-axis to be any line passing through the origin, and hence
Z
X|df |2α dA = 0,
S2

for the metric of constant curvature one, and hence for the constant curvature
metric of total area one.

158
 More generally, we can use (3.54) to calculate the derivative of Eα to obtain
 Z 
d  1 d 2 α 2

Eα (f ◦ φt )
 = (1 + |d(f ◦ φt )| ) sech (u + t)dudθ
dt t=0 2 dt §2 t=0
Z Z
2 α−1 2 2 α
=α (1 + |df | ) tanh u|df | dA − (1 + |df | ) tanh udA.
S2 S2

Thus, if f is a critical point for Eα ,

Z
0= [α(1 + |df |2 )α−1 |df |2 − (1 + |df |2 )α ]zdA,
S2
Z
= [α(1 + |df |2 )α−1 |df |2 − (1 + |df |2 )α − 1]zdA,
S2

where we have used the fact that the average value of z on S 2 is zero. We obtain
(3.53) by setting
t
α(1 + t)α−1 t − (1 + t)α − 1
Z
τ
ψα (t) = = α dτ. (3.55)
α−1 0 (1 + τ )2−α

One easily verifies that this function has the desired properties, and has a smooth
limit as α → 1, namely Z t
τ
ψ1 (t) = dτ. (3.56)
0 (1 + τ)

Remark 3.10.4. It is sometimes convenient to demand that harmonic maps

f : S 2 → M satisfy the side condition
Z
Xψ1 (|df |2 )dA = 0,
S2

thereby reducing the symmetry group of the problem from P SL(2, C) to a max-
imal compact subgroup SO(3).
Lemma 3.10.5. Suppose that M is a compact Riemannian manifold and K0 >
0 is an upper bound for the sectional curvatures of M and α0 ∈ (1, ∞) is
sufficiently close to one. Then there is an 0 > 0 such that every nonconstant
α-harmonic map f : S 2 → M satisfies
1
Eα (f ) − ≥ E(f ) ≥ 0 .
2

We give S 2 the constant curvature metric of total area one, which has Gaussian
curvature K̃ = 2π. Then it follows from Lemma 3.8.4 that there is an elliptic
operator Lα such that

Lα (e(f )) ≥ 2(2π − e(f ))e(f ).

159
Moreover, it follows from Lemma‘3.8.5 that there is an 0 > 0 such that

α ∈ (1, α0 ] and E(f ) < 0 ⇒ e(f ) < 2π.

This yields a contradiction to the maximum principle for Lα unless e(f ) is

identically zero.
Lemma 3.10.6. Suppose that {fm } is a sequence of nonconstant critical points
for Eαm with αm → 1 such that fm converges to a harmonic map f∞ : S 2 → M
on S 2 − {0}, uniformly on compact subset in C k for all k. Then f∞ cannot be
a map to a point.
Indeed, it follows from the previous lemma that Eα (fm ) ≥ 0 and hence if f∞
were constant a nontrivial bubble would form near the point 0 ∈ S 2 . Thus en-
ergy density would have to concentrate at the point 0 and this would contradict
(3.53).
Proof of Theorem 3.10.1: The proof is an induction which makes use of a suc-
cession of variational problems. Let

F1 = {f ∈ C 2 (S 2 , M ) : the free homotopy class of f is nontrivial },

and let
µ1 = inf{E(f ) : f ∈ F1 }.
Suppose that αm → 1 and let fm be an element of F1 which achieves a minimum
for Eαm on F1 . Either a subsequence of {fm } converges without bubbling to a
minimal two-sphere f∞ : S 2 → M which lies within F1 , in which case we have
a nonconstant minimal two-sphere f ∈ F1 such that E(f1 ) = µ1 , or at least one
nonconstant minimal minimal two-sphere h : S 2 → M bubbles off at a point
p1 ∈ S 2 . We need to show that the second possibility leads to a contradiction.
Recall that a subsequence of {fm } will converge to a harmonic map f∞ :
S 2 → M , uniformly on compact subsets of S 2 − {p1 , . . . , pl }, where p1 , . . . , pl
are the bubble points. By Lemma 3.10.6, either f∞ is nonconstant or there is at
least one other bubble point. Let us suppose that f∞ is nonconstant. (The other
case can be treated in a similar fashion.) We will perform surgeries on small
circles about p1 to divide each αm -harmonic map into a base map fˆm : S 2 → M
and a bubble ĥm : S 2 → M .
In more detail, let Drm (p1 ) be a disk of radius rm chosen so that rm → 0 as
m → ∞,
Z
e(fm )dA ≥ 0 and fm (Drm (p1 ) − D(1/3)rm (p1 )) ⊆ Nδ (q)
Drm (p1 )

where q = f∞ (p1 ) and Nδ (q) is the domain of a geodesic coordinate system of

radius δ in M . Then define a map fˆm : S 2 → M by
(
ˆ fm (p), for p ∈ S 2 − Drm (p1 ),
fm (p) =
expq (η(p)(expq )−1 fm (p)), for p ∈ Drm (p1 ),

160
where η : S 2 → [0, 1] is a smooth function such that
(
1 on S 2 − D(2/3)rm (p1 ),
η≡
0 on D(1/3)rm (p1 ),

Let Rm : S 2 → S 2 be the reflection through the circle ∂D(1/2)rm (p1 ), and define
ĥm : S 2 → M by
(
expq (η ◦ Rm (p)(expq )−1 Rm ◦ fm (p)), for p ∈ S 2 − Drm (p1 ),
ĥm (p) =
fm (p), for p ∈ Drm (p1 ),

Thus fˆm agrees with fm outside Drm (p1 ) while Rm ◦ ĥm agrees with fm inside
Drm (p1 ).
We can clearly arrange that
0
Eαm (fˆm ) + Eαm (ĥm ) < Eαm (fm ) +
2
if m is sufficiently large, where 0 is the constant appearing in Lemma 3.10.5. At
least one of the maps fˆm or ĥm must be homotopically nontrivial with energy
less than the infimum over F1 , and this provides the desired contradiction.
Thus we obtain a homotopically nontrivial minimal two-sphere f1 : S 2 → M
and we let Γ1 denote the Z[π1 (M, p)]-submodule of π2 (M, p) generated by [f1 ].
If Γ1 6= π2 (M, p), we let

F2 = {f ∈ C 2 (S 2 , M ) : the free homotopy class of f is not in Γ1 },

set µ2 = inf{E(f ) : f ∈ F2 }, and proceed exactly as before. One thereby

obtains a minimal two-sphere f2 : S 2 → M which does not lie in Γ1 . we then
let Γ2 be the Z[π1 (M, p)]-submodule of π2 (M, p) generated by [f1 ] and [f2 ],
and so forth. For the inductive step, we suppose we have already constructed
a Z[π1 (M, p)]-submodule Γk−1 of π2 (M, p) generated by minimal two-spheres
[f1 ], . . . [fk−1 ]. If Γk−1 6= π2 (M, p), we let

Fk = {f ∈ C 2 (S 2 , M ) : the free homotopy class of f is not in Γk−1 },

let µk = inf{E(f ) : f ∈ Fk }, and verify that there is a nonconstant minimal

two-sphere fk ∈ Fk such that E(fk ) = µk . Theorem 3.10.1 follows by induction.
The above theorem can be applied to compact three-dimensional Riemannian
manifolds. In this case, surfaces which minimize area are free of branch points
by a theorem of Osserman and Gulliver (see [58] and [31]). In fact the minimal
spheres which generate π2 (M, p) are either imbedded or double coverings of
projective planes:
Theorem 3.10.7 (Meeks and Yau). Suppose that M is an oriented com-
pact three-dimensional Riemannian manifold. Then there is a finite collection
{f1 , . . . , fl } of generators for π2 (M, p) as a Z[π1 (M, p)]-module, each of which

161
is either an embedded minimal two-sphere or a doubly covered imbedded pro-
jective plane.
We refer the reader to [48] for the proof, which is based upon the tower con-
struction of Papakyriakopoulos.
Thus one can divide a compact orientable three-dimensional manifold M along
the homotopically nontrivial minimal two-spheres which separate M , obtaining
a connected sum decomposition
M = M0 ]M1 ] · · · ]Mk . (3.57)
Suppose that an element π2 (Mi ) is nonzero for some i. Then there exists an
imbedded two sphere N ⊂ Mi and two points close to N but on opposite sides
can be connected by an arc C within Mi . A small neighborhood of N ∪ C will
have boundary diffeomorphic to a separating two-sphere. It then follows from
the positive resolution to the Poincaré conjecture that Mi is diffeomorphic to
S 1 × S 2 . Thus each summand in (3.57) either has vanishing π2 or is S 1 × S 2 .
A compact orientable three-dimensional manifold is said to be prime if it
cannot be expressed as a nontrivial connected sum. It also follows from the
Poincaré conjecture that each factor in the connected sum decomposition (3.57)
is prime, and the decomposition must be the prime decomposition for a compact
oriented three-manifold M discovered by Kneser and Milnor. The remarkable
fact is that the decomposition occurs along imbedded two-spheres which are
minimal with respect to any preassigned Riemannian metric.
Example 3.10.8. Here is an explicit example illustrating how much gets lost
of the critical point theory for α-harmonic maps as α → 1. Starting with a lens
space L(3, 1) of constant curvature one, we consider the connected sum M =
L(3, 1)]L(3, 1)]L(3, 1) along isolated embedded minimal two-spheres N1 and
N2 of very small radius of curvature which minimize within their free homotopy
classes. From van Kampen’s theorem it follows that the fundamental group
is generated by elements a, b and c, with the relations a3 = b3 = c3 = 1.
Explicit construction of the universal cover shows that π2 (M ; p) is generated as
a π1 (M )-module by the two imbedded minimal spheres
f1 : S 2 → N1 , f2 : S 2 → N2 .
We can construct an additional embedded two-sphere N = N1 ]N2 by connecting
N1 with N2 by a very thin tube and a corresponding imbedding f : S 2 → N
which represents the homotopy class [f1 ] + [f2 ], which is not freely homotopic
to any multiple of [f1 ] or [f2 ]. If α0 is sufficiently close to one, then for each
α ∈ (1, α0 ] there is a minimizing α-minimal two-sphere fα in the component of
Map(S 2 , M ) representing this free homotopy class. As α → 1 a subsequence of
these α-minimal two-spheres should approach a configuration consisting of N1 ,
N2 and a minimal geodesic connecting N1 and N2 of some length bounded away
from zero.
Although π2 (M, p) is generated by two imbedded spheres as a Z[π1 (M, p)]-
module, it is interesting to note that as an abelian group, π2 (M, p) has infinitely

162
many generators, as one sees by noting that it is isomorphic to the second
homotopy group of its universal cover.

3.11 Existence of minimal tori

Suppose we want to investigate the topology of a compact oriented three-
dimensional manifold. We start by using π2 (M ) to express M as a direct sum
decomposition (3.57) in which each of the summands is prime. One can show
that each prime summand is one of three types: it is diffeomorphic to S 1 × S 2 ,
it is finitely covered by S 3 , or all of its homotopy groups are trivial except for
]pi1 = π, so it is a K(π, 1).
The next step in exploring the topology is to try to divide up the K(π, 1)
summands along tori. An imbedded torus or Klein bottle in M is incompress-
ible if the inclusion induces an injection on π1 . To construct incompressible
tori, we look for subgroups of π1 (M ) which are isomorphic to Z ⊕ Z and are
maximal among subgroups with this property. Indeed, according to Thurston’s
geometrization program (as described, for example, in [54]), any prime compact
oriented three-manifold can be decomposed along incompressible imbedded tori
and Klein bottles into manifolds which have locally homogeneous Riemannian
structures. It is natural to ask whether the incompressilble tori and Klein bot-
tles used in the torus decomposition can be taken to be minimal with respect
to any Riemannian metric on M .
This question provides some motivation for the next existence theorem for
minimal surfaces, due to Schoen and Yau [72] or Sacks and Uhlenbeck [69] with
different proofs. But the theory also applies to minimal tori in Riemannian
manifolds of arbitrary dimension.
For k ∈ {0} ∪ N, we let

Map(k) (T 2 , M ) = {f ∈ C 2 (T 2 , M ) : the image of

f] : π1 (T 2 ) −→ π1 (M ) is an abelian group with k generators }.

For example, if f] (π1 (T 2 )) ∼

= Z ⊕ Z2 , then f ∈ Map(2) (T 2 , M ). Note that the
mapping class group Γ = SL(2, Z) preserves Map(k) (T 2 , M ), so Eα induces a
map

Map(k) (T 2 , M ) × T
Eα : M(k) (T 2 , M ) −→ R, where M(k) (T 2 , M ) = .
Γ

Moreover, if f ∈ Map(2) (T 2 , M ),

f ◦φ=f for some φ∈Γ ⇒ φ = identity,

so the mapping class group SL(2, Z) acts freely on Map(2) (T 2 , M ) × T . Thus

(k)
if Mα (T 2 , M ) denotes the completion with respect to the L2α 1 norm, then

163
 (k)
Mα (T 2 , M ) will be a smooth Banach manifold. Recall that the α-energy
descends to a C 2 map on the quotient

Eα : M(k) 2
α (T , M ) −→ R.

(1)
We expect that sequences tending to a minimum for α-energy in Mα (T 2 , M )
would degenerate to a closed geodesic.
(2)
Lemma 3.11.1. The map Eα : Mα (T 2 , M ) → R satisfies Condition C.
What Condition C asserts is that if [fi , ωi ] is a sequence of points in M(2) (Σ, M )
on which Eα is bounded and for which kdEα ([fi , ωi ])k → 0, and if for each i,
(fi , ωi ) ∈ Map(2) (T 2 , M ) × T is a representative for [fi , ωi ], then there are
elements φi ∈ Γ such that a subsequence of (fi ◦ φi , φ∗i ωi ) converges to a critical
point for Eα on Map(2) (T 2 , M ) × T .
To prove this, we recall that for the torus, the Teichmüller space T is the
upper half plane, and after a change of basis we can arrange that an element
ω ∈ T lies in the fundamental domain

D = {u + iv ∈ C : v > 0, −(1/2) ≤ u ≤ (1/2), u2 + v 2 ≥ 1} (3.58)

for the action of the mapping class group Γ = SL(2, Z). The moduli space
R is obtained from D by identifying points on the boundary. The complex
torus corresponding to ω ∈ T can be regarded as the quotient of C by the
abelian subgroup generated by d and ωd, where d is any positive real number, or
alternatively, this torus is obtained from a fundamental parallelogram spanned
by d and ωd by identifying opposite sides. The fundamental parallelogram of
area one can be regarded as the image of the unit square {(t1 , t2 ) ∈ R2 : 0 ≤
ti ≤ 1} under the linear transformation
 1      1
t x 1 1 u t
7→ =√ ,
t2 y v 0 v t2

where z = x + iy is the usual complex coordinate on C. A straightforward

calculation gives a formula for the usual energy
Z  2  2 !
1  ∂f 
  +  ∂f  dxdy
 
E(f, ω) =
2 P
 ∂x   ∂y 
2 !
 ∂f 2 1  ∂f
Z   
1 ∂f
v  1  +  2 − u 1  dt1 dt2 ,

=
2 P ∂t v ∂t ∂t

P denoting the image of the unit square. The only way that ω can approach the
boundary of Teichmüller space while remaining in the fundamental domain D
is for v → ∞. The rank two condition implies that the maps t1 7→ f (t1 , b) must
be homotopically nontrivial for each choice of t2 = b, and hence the length in

164
M of t1 7→ f (t1 , b) is bounded below by a positive constant c. This implies that

1 1  ∂f 2
Z Z  
1
E(f, ω) ≥ v  1  dt1 dt2
2 0 0 ∂t
v 1 c2 v
Z
≥ (length of t1 7→ f (t1 , b))2 db ≥ (3.59)
2 0 2

by the Cauchy-Schwarz inequality, and hence Eα (f, ω) (which is ≥ E(f, ω))

must approach infinity.
Suppose now that [fi , ωi ] is a sequence of points in M(2) (T 2 , M ) on which
Eα is bounded and for which kdEα ([fi , ωi ])k → 0, and for each i, (fi , ωi ) ∈
Map(T 2 , M ) × T is a representative for [fi , ωi ]. Then the projection [ωi ] ∈ R of
ωi ∈ T is bounded, and must therefore have a subsequence which converges to
an element [ω∞ ] ∈ R. Hence there are elements φi ∈ Γ such that a subsequence
of φ∗i ωi converges to an element ω∞ ∈ T . Then Eα,ω∞ (fi ◦ φi ) is bounded
and kdEα,ω∞ (fi ◦ φi )k → 0, so by Condition C for Eα,ω∞ , a subsequence of
{(fi ◦ φi , φ∗i ωi )} converges to a critical point for Eα on Map(T 2 , M ) × T . This
establishes Condition C for the function Eα .
Just as for a generic perturbation of Eα,ω , one can establish Morse inequalities
for a generic perturbation Eα0 of Eα . Thus we have a Morse-Witten complex for
this function, and we can use it to investigate critical points of E in the limit
as α → 1.
The following theorem was proven by Schoen and Yau [72] and Sacks and
Uhlenbeck [69]:
Theorem 3.11.2. Every component of M(2) (Σ, M ) contains an element [f, ω]
which minimizes the function

E : M(2) (T 2 , M ) −→ R

If (f, ω) is a representative, then f : T 2 → M is conformal and harmonic with

respect to ω, and hence a minimal surface.
Proof: Let C be a component of M(2) (T 2 , M ). Choose a decreasing sequence
αm → 1 and for each αm , a corresponding critical point

[fm , ωm ] ∈ M(2) (T 2 , M ) which minimizes Eαm on C.

We can assume that

Eαm ([fm , ωm ]) → inf{E(f, ω) : [f, ω] ∈ C}. (3.60)

Since [ωm ] must lie in a bounded region of the Riemann moduli space R1 ,
we can arrange that [ωm ] converges after passing to a subsequence, and hence
after choosing suitable representatives, arrange that ωm converges to an element
ω ∈ T1 . By Theorem 3.8.7, we can pass to a further subsequence so that either
{fm } converges uniformly on compact subsets of T 2 to an ω-harmonic map

165
f : T 2 → M or nonconstant minimal two-spheres bubble off as α → 1. But in
the latter case, we can perform a surgery just like we described in the proof of
Theorem 3.10.1 and obtain a new parametrized torus fˆm : T 2 → M which lies
in the same component C and has smaller energy
0
Eαm (fˆm , ωm ) < Eαm (fm , ωm ) − ,
2
contradicting (3.60). So no bubbling can occur, and we obtain a critical point
(f, ω) for E which minimizes E within the component C.
Remark 3.11.3. The local regularity result of Osserman and Gulliver implies
that the minimal tori found by Theorem 3.11.2 are immersed. Using a variant of
the tower construction, it was proven by Freedman, Hass and Scott [24] that the
tori and Klein bottles needed for the torus decomposition of a prime compact
oriented three-manifold M can be taken to be minimal.

3.12 Higher genus minimal surfaces*

The existence theory for incompressible minimal tori presented in the previous
section can be extended to surfaces of higher genus. Suppose that Σ is a compact
oriented compact surface of genus g ≥ 2. We set

Map0 (Σ, M ) = {f ∈ C 2 (Σ, M ) : f] : π1 (Σ) → π1 (M ) is injective },

define M0 (Σ, M ) to be the quotient space

Map0 (Σ, M ) × T
M0 (Σ, M ) = ,
Γ
where Γ is the mapping class group, and let M0α (Σ, M ) be the completion of
M0 (Σ, M ) with respect to the L2α
1 norm.

Lemma 3.12.1. If Σ is a compact oriented surface of genus g ≥ 2, then the

map Eα : M0α (Σ, M ) → R satisfies Condition C.
The proof makes use to two ingredients, a collar theorem of Halpern and Keen,
and the structure of the Bers compactification of the moduli space R of confor-
mal structures on Σ.
We suppose that Σ is given the hyperbolic metric of Gaussian curvature
−1 corresponding to the conformal structure ω ∈ T . The collar theorem is a
fundamental result of Riemann surface theory, and states that if γ is a closed
geodesic of this metric of length ≤ k1 , where k1 is a positive constant, then
there is a collar region C ⊆ Σ of fixed area k2 > 0 about γ. Indeed, we can
arrange that γ lifts to the map
 
2 iπ
γ̃ : [0, l] → H defined by γ̃(t) = exp t + ,
2

166
and the collar region is of the form C = π(C̃), where

k2
C̃ = {reiθ ∈ H2 : 1 ≤ r < el , π − θ0 < θ < θ0 }, where cot θ0 = . (3.61)
2l

Lemma 3.12.2. Suppose that Σ has a closed geodesic γ whose length with
respect to the hyperbolic metric corresponding to ω is l ≤ k1 , and that C is the
collar region about γ described above. If f : Σ → M is any smooth map, then
the ω-energy of f |C is at least k4 /l, for some positive constant k4 .
We consider the lift f˜ : C̃ → M , which has energy density

1 1 |(f˜ ◦ γ̃θ )0 (t)|2

e(f˜) = |df |2 ≥ , where γ̃θ (t) = exp (t + iθ) .
2 2 |γ̃θ0 (t)|2

Straightforward calculation shows that

 2
1  ∂ f˜
|γ̃θ0 (t)|2 = and |(f˜ ◦ γ̃θ ) (t)| = r   ,
0 2 2
 
sin2 θ  ∂r 
 2
r2 sin2 θ  ∂ f˜
so e(f ) ≥   .
 
2  ∂r 

Thus we find that

Z π−θ0 Z el
drdθ
E(f˜|C̃) = e(f˜)
θ0 1 r sin2 θ
Z π−θ0 Z el  2 Z  2
r  ∂ f˜ 1 π−θ0 l  ∂ f˜
Z
≥   drdθ =   dtdθ,
θ0 1 2  ∂r  2 θ0 0  ∂t 

where we have set r = et in the last integral. But by the Cauchy-Schwarz

inequality,
Z l  ˜ Z l  ˜2
∂ f  ∂f 
L(f˜ ◦ γ̃θ )2 =   dt ≤ l   dt,
 
0  ∂t  0  ∂t 

and since f˜ ◦ γ̃θ is a closed curve in M which is not homotopic to a constant,

L(γθ ) ≥ k3 , for some positive constant, k3 . Hence

k 2 (π − 2θ0 )
E(f, ω) ≥ E(f˜|C̃) ≥ 3 ,
2l
which yields the assertion of the lemma, since θ0 → 0 as l → 0.
To finish the proof of Lemma 3.12.1, we suppose that {[fm , ωm ]} is a sequence
of points in M0α (Σ, M ) such that Eα ([fm , ωm ]) is bounded. We claim that the
sequence {[ωm ]} must stay in a compact region of the moduli space R. To

167
see this, we make use of the Bers compactification of the moduli space R, as
described in Appendix B of [37]. After passing to a subsequence, we can assume
that {[ωm ]} converges to a point in the compactification. But points in the
compactification are Riemann surfaces with nodes and and as one approaches a
Riemann surface with nodes from inside the moduli space, some homotopically
nontrivial loop must have its length go to zero, and then the energy will go to
infinity by Lemma 3.12.2. Thus {[ωm ]} must remain bounded, and after passing
to a subsequence we can assume that [ωm ] → [ω∞ ] ∈ R. Now we use the fact
that Eα,ω∞ satisfies Condition C to conclude that Eα : M0α (Σ, M ) → R satisfies
Condition C.
Thus we have a well-behaved Morse theory for suitable perturbations of the
α-energy on M0α (Σ, M ), just as in the case of the torus.
Theorem 3.12.3. If Σ is a compact oriented Riemann surface of genus g ≥ 2,
then every component of the space M0 (Σ, M ) of incompressible maps from Σ
to M contains an element [f, ω] which minimizes the function

E : M0 (Σ, M ) −→ R

If (f, ω) is a representative, then f : Σ → M is conformal and harmonic with

respect to ω, and hence a minimal surface.
To prove this, we use essentially the same argument as we used in the proof of
Therem 3.11.2.

3.13 Complex form of second variation

In order to apply the apply the Morse theory of α-harmonic maps to derive
geometric consequences, it is useful to be able to estimate the Morse index of
a harmonic map. Recall that by Corollary 3.7.2, the second variation of energy
at a harmonic map f : Σ → M is given by the formula
Z
2
d Eω (f )(V, W ) = hLf (V ), W idA,
Σ

where Lf is the Jacobi operator , defined by

 
1 D D D D
Lf (V ) = − 2 ◦ + ◦ + K(V ) . (3.62)
λ ∂x ∂x ∂y ∂y

Recall that we say that an element V ∈ Tf M is a Jacobi field along f if

Lf (V ) = 0.
Theorem 3.13.1. The Jacobi operator can be written in the following complex
form:    
4 D D ∂f ∂f
Lf (V ) = − 2 ◦ + R ·, ,
λ ∂z ∂ z̄ ∂z ∂ z̄

168
where the Riemann-Christoffel curvature tensor R has been extended to be
complex linear.
Straightforward calculations show that on the one hand,

D D √
 
D D D D D D D D
4 ◦ = ◦ + ◦ + −1 ◦ − ◦
∂z ∂ z̄ ∂x ∂x ∂y ∂y ∂x ∂y ∂y ∂ x̄
D D √
 
D D ∂f ∂f
= ◦ + ◦ + −1R , ,
∂x ∂x ∂y ∂y ∂x ∂y
while on the other,
     
∂f ∂f ∂f ∂f ∂f ∂f
4R ·, = R ·, + R ·,
∂z ∂ z̄ ∂x ∂x ∂y ∂y
√
     
∂f ∂f ∂f ∂f
+ −1 R ·, − R ·,
∂x ∂y ∂y ∂x
∂f ∂f √
     
∂f ∂f ∂f ∂f
= R ·, + R ·, − −1R , ,
∂x ∂x ∂y ∂y ∂x ∂y

the last step following from the Bianchi symmetry. Substitution into (3.62) now
yields the Theorem.
Corollary 3.13.2. The second variation of ω-energy is given by the formula:
Z " #
 DV 2
   
2 ∂f ∂f
d Eω (f )(V, V̄ ) = 4  ∂ z̄  − R V ∧ ∂z , V̄ ∧ ∂ z̄ dxdy,
 
Σ

for sections V of the complex vector bundle E, where the bar indicates complex
conjugate and we use the notation hR(x ∧ y), v ∧ wi = hR(x, y)w, vi.
Proof: Simply substitute into (3.62) and integrate by parts.
Corollary 3.13.3. A section V of the subbundle L of f ∗ T M ⊗ C, characterized
by the fact that ∂f /∂z is a locally defined section of L, is a Jacobi field if and
only if it is holomorphic.
Proof: If σ is a holomorphic section of L, then σ = g(∂f /∂z) where g is a
holomorphic function. It follows that
 
D D ∂f
◦ (σ) = 0 and R σ, = 0.
∂z ∂ z̄ ∂z
We leave the converse to the reader.
Remark 3.13.4. Note first that since Lf is a real operator, real and imaginary
parts of Jacobi fields are also Jacobi fields.
Remark 3.13.5. The dimension of the space of holomorphic sections O(L) of
the line bundle L can often be estimated by the Riemann-Roch theorem from the

169
theory of Riemann surfaces. (One can apply the usual Riemann-Roch theorem
to L since L has a canonical holomorphic structure by the Koszul-Malgrange
¯ ¯
Theorem 3.1.3. Alternatively, one can apply the Atiyah-Singer Index Theorem
directly to the operator Z 7→ (DZ/∂ z̄)dz and use the fact that this operator has
¯
the same symbol as the ∂-operator, avoiding the proof of the Koszul-Malgrange
Theorem.) First consider the case in which Σ has genus zero. In this case, one
recalls that there is exactly one holomorphic line bundle Hk over Σ = S 2 (up
to isomorphism) whose Chern class satisfies hc1 Hk ), [Σ]i = k, and the space
of holomorphic sections of this bundle has complex dimension k + 1 if k ≥ 0.
The reason for the notation Hk is that the bundle of first Chern class k is the
k-th tensor power of a bundle H called the hyperplane bundle over S 2 . Thus
L = Hk , where k = c1 (L)[S 2 ].
According to (3.7), the line bundle L defined by the harmonic map f has
first Chern class given by the formula
X
hc1 (L), [Σ]i = 2 + {νp : p is a branch point of f }.

In particular, hc1 (L), [Σ]i is always positive and L always has a space of holo-
morphic section which has complex dimension at least three. A holomorphic
section of L can be identified with a meromorphic section σ of the holomorphic
tangent bundle T Σ with the property that its divisor (σ) satisfies
X
(σ) + {νp p : p is a branch point of f } ≥ 0.

Counting up the possibilities, we see that there is a holomorphic section of L

which has every possible principal part at every branch point of f .
We can do a similar analysis for line bundles over a torus T 2 , except that this
time the line bundles of a given Chern class form a torus of dimension two, and
if c1 (L)[T 2 ] ≥ 1, the dimension of holomorphic sections of L will miss by one
the dimension that would be possible if every principal part were realized. Thus
in the case of the torus, there is a section σ of L such that the corresponding
section of the tangent bundle to T 2 has arbitrary principal part at all but one
branch point.

3.14 Isotropic curvature

Just as the theory of geodesics uncovers relations between curvature and topol-
ogy of Riemannian manifolds (through classical theorems with names such as
Synge’s theorem, Myers’ theorem and the theorem of Hadamard and Cartan
among others), one might hope to apply the theory of minimal surfaces to cur-
vature and topology. Pursuing this idea, however, leads to a different notion of
curvature. Just as studying the stability of geodesics leads to sectional curva-
ture, we find stability theory for minimal surfaces leads to the notion of isotropic
curvature.

170
 Recall that in normal coordinates (x1 , . . . , xn ) centered at p on a Riemannian
manifold M , the Riemannian metric can be expressed by a Taylor series
1X
gij = δij − Rikjl (p)xk xl + higher order terms,
3
k,l

where the Rikjl ’s are components of the Riemann-Christoffel curvature tensor.

The curvature operator is defined in terms of these components to be the linear
map
R : Λ2 Tp M −→ Λ2 Tp M
such that
  ! X  
∂  ∂  ∂  ∂ 
R ∧ = R ijkl (p) ∧ .
∂xi p ∂xj p ∂xk p ∂xl p
k,l

If z and w are linearly independent elements of Tp M ⊗C, the sectional curvature

of the complex two-plane σ spanned by z and w is
hR(z ∧ w), z̄ ∧ w̄i
,
hz ∧ w, z̄ ∧ w̄i
where the bar denotes complex conjugation. The complex two-plane σ is said
to be isotropic if hz, zi = hw, wi = hz, wi = 0.
Definition. The Riemannian manifold M is said to have positive isotropic
curvature if K(σ) > 0, whenever σ is an isotropic complex two-plane.
To see why isotropic curvature is related to stability properties of minimal sur-
faces, we first recall the second variation formula for a harmonic map f : S 2 →
M (Corollary 3.13.3):
Z " #
 DV 2
   
2 ∂f ∂f
d Eω (f )(V, V̄ ) = 4  ∂ z̄  − R V ∧ ∂z , V̄ ∧ ∂ z̄ dxdy. (3.63)
 
Σ

We need one further ingredient:

Grothendieck Theorem 3.14.1. Any holomorphic line bundle over the Rie-
mann sphere S 2 = CP 1 divides into a holomorphic direct sum of holomorphic
line bundles.
This theorem, proven in [30], allows us to write E = f ∗ T M ⊗ C as a direct sum
of line bundles,

E = L1 ⊕ L2 ⊕ · · · ⊕ Ln , where c1 (L1 )[S 2 ] ≥ · · · ≥ c1 (Ln )[S 2 ].

Since the Riemannian metric is invariant under the Levi-Civita connection, it

extends to a holomorphic complex bilinear form

h·, ·i : E × E −→ C,

171
and in particular, E is isomorphic to its dual. Thus the line bundles in the
above sequence can be arranged so that c1 (Li ) = −c1 (Ln−i+1 ) for each i.
If V is a holomorphic section of one of the line bundles Li inthe direct sum
decomposition of E, then
 
2 ∂f
hV, V i : S → C and V, : S2 → C
∂z

are holomorphic, and hence

 
∂f
hV, V i = (constant), V, = (constant).
∂z

In particular, if V is a holomorphic section of Li where Li has positive first Chern

class, or more generally if V is any holomorphic section such that hV, V i = 0,
then    
∂f ∂f ∂f
hV, V i = V, = , = 0,
∂z ∂z ∂z
and
∂f
V and span an isotropic two-plane
∂z
at every point where neither vanishes. If M has positive isotropic curvatures, it
therefore follows from the index formula (3.63) that d2 Eω (f )(V, V̄ ) < 0.
Let m be the number of line bundle summands Li of E with positive first
Chern number,

c1 (L1 )[S 2 ] ≥ · · · ≥ c1 (Lm )[S 2 ] ≥ 1, c1 (Lm+1 )[S 2 ] ≤ 0

and let E0 be the direct sum of all of the line bundle summands with zero Chern
class; the dimension of the space O(E0 ) of holomorphic sections of E0 is n − 2m.
If V1 , . . . , Vm are nonzero holomorophic sections of L1 , . . . , Lm respectively, then
hVi , E0 i = 0 for 1 ≤ i ≤ m. If {W1 , . . . , Wl } is a basis for O(E0 ), then hWi , Wj i
is constant, and therefore there is a maximal isotropic holomorphic subbundle
I0 ⊂ E0 ) which has rank ≥ (1/2)(rank of E0 )-1).
If V is the space of holomorphic sections of L1 ⊕ · · · Lm ⊕ I0 , then dim V ≥
(1/2)(dim M − 1) and

V ∈V ⇒ d2 Eω (f )(V, V̄ ) < 0.

Thus we have proven a lemma which will be needed for the proof of the next
theorem:
Lemma 3.14.2. Suppose that M is a Riemannian manifold which has positive
isotropic curvature. Then the Morse index of any harmonic two-sphere f : S 2 →
M is at least (1/2)(dim M − 1).
The following theorem [49] generalizes the earlier classical sphere theorem due
to Berger, Klingenberg and Toponogov:

172
Sphere Theorem 3.14.3. Suppose that M is a compact smooth simply
connected Riemannian manifold of dimension at least four which has positive
isotropic curvature. Then M is homeomorphic to a sphere.
The idea behind the proof is to show that M is a homotopy sphere and apply
the solutions to the generalized Poincaré conjecture in dimensions greater than
four to conclude that M is homeomorphic to a sphere.
By definition, a compact connected manifold M of dimension n is a homo-
topy sphere if πq (M ) = 0, for all integers q such that 0 < q < n. Thus if M is
not a homotopy sphere there must be an integer q with 0 < q < n such that
πq (M ) 6= 0, and we choose the smallest such integer. By the Hurewicz isomor-
phism theorem, q is also the smallest positive integer such that Hq (M ; Z) 6= 0.
By Poincaré duality, we can assume that q ≤ n/2.
Just as in §5.4, we let M be the space of smooth maps from S 2 to M
and let π : M → M be the evaluations map defined by π(f ) = f (0) where
0 ∈ S 2 = C ∪ {∞} is the basepoint. Since the fibration π possesses a section,
the long exact homotopy sequence of π splits, and if Mp denotes the subset of
M consisting of maps such that f (0) = p, we conclude that

πk (M) ∼
= πk (M ) ⊕ πk (Mp ) ∼
= πk (M ) ⊕ πk+2 (M ),

which implies that

πq−2 (M) ∼
= πq (M ) 6= 0.
We now apply Morse theory to a perturbation Eα,ψ of the energy E, where ψ
is chosen so that all nonconstant critical points of Eα,ψ are Morse nondegenerate,
and let α → 1 and ψ → 0, in accordance with Theorem 2 from §3.5. We thereby
obtain a sequence {fm } of critical points for Eαm ,ψm , each having Morse index
no more than q − 2. By the bubbling argument presented in §3.9, we find that a
subsequence (still denoted by {fm }) converges uniformly in every C k on every
compact subset of S 2 − {p1 , · · · , pl }, where p1 , . . . , pl are a finite number of
bubble points, to a harmonic map on Σ − {p1 , · · · , pl }, which can be extended
to a smooth harmonic map f∞ : Σ → M by the Sacks-Uhlenbeck removeable
singularity theorem.
Suppose first that f∞ is nonconstant. In that case, we claim that the Morse
index of f∞ is no larger than q − 2. For this, we need the following lemma:
Lemma 3.14.4. Suppose that fm : Σ → M is a sequence of αm -harmonic maps
in M which converge in C k on compact subsets of Σ − {x1 , · · · , xl } to a smooth
harmonic map f∞ : Σ → M . Then

Morse index of f∞ ≤ lim inf Morse index of fm .

To prove the Lemma, we first recall the formula the second variation formula

173
(3.40) for α-harmonic two-spheres:
Z
2
d Eα (f )(V, W ) = α (1 + |df |2 )α−1 [h∇V, ∇W i − hK(V ), W i]dA
Σ
Z
+ 2α(α − 1) (1 + |df |2 )α−2 hdf, ∇V ihdf, ∇W idA.
Σ

Note that the second term is dominated by the first and goes to zero when α
is close to one, and the first term approaches d2 E(f )(V, W ) when the support
of V and W does not contain any bubble points. We claim that if d2 E(f∞ ) is
negative definite on a fixed linear space V of dimension q − 2, so is d2 Eα (fm )
when m is sufficiently large.
To see this, we make use of a cutoff function near the bubble points. First,
define a map φ : R → R by

0,
 if r ≤ 2 ,
φ(r) = (log( ) − log r)/(log ), if 2 ≤ r ≤ ,
2

1, if  ≤ r,


so that 
0,
 if r ≤ 2 ,
dφ
(r) = (−1)/(r log ), if 2 ≤ r ≤ ,
dr 
0, if  ≤ r,


and 2
Z 2π Z  Z 
dφ 2π 2π
(r) rdrdθ = 2
dr = − .
0 0 dr 2 r(log ) log 
Thus if we define ψi : S 2 → R so that it is one outside an -neightborhood of
the bubble point xi and in terms of polar coordinates (ri , θi ) about the bubble
point xi satisfies the condition ψi = φ ◦ ri , then
−C
Z
|dψi |2 dA ≤ , where C is a positive constant,
S 2 log 
and a similar estimate holds for ψ = ψ1 · · · ψl , a cutoff function which vanishes
at every bubble point.
If  > 0 is chosen sufficiently small, then

V ∈V ⇒ d2 E(f )(ψV, ψV ) < 0 ⇒ d2 Eα (f )(ψV, ψV ) < 0.

This shows that d2 Eα (f ) is negative definite on a space of dimension q − 2 and

proves the Lemma.
It follows from Lemma 3.14.4 that if f∞ is nonconstant, it must have index
≤ q − 2 ≤ (1/2)(dim M − 2) which contradicts Lemma 3.14.2.
But if f∞ is constant a nonconstant sphere must bubble off; that is there
must be a family of conformal reparametrizations gm : S 2 → S 2 , such that

174
fm ◦ gm converges to a nonconstant harmonic sphere fˆ∞ in C k on compact
subsets of S 2 − {p1 , · · · , pl }, where p1 , . . . , pl are a finite number of bubble
points. We can choose pm ∈ S 2 such that

kdfm (pm )k = sup{kdfp k : p ∈ S 2 },

and after rotations we can arrange that all the pm ’s are equal and in fact that
z(pm ) = 0, where z is the standard coordinate on S 2 − {∞} = C. Let rm =
kdfm (pm )k, and let gm : S 2 → S 2 be the conformal map expressed in terms of
the standard coordinate as hm (z) = rm z. Finally, let hm = fm ◦ gm , a sequence
of maps which converge to fˆ∞ on compact subsets of S 2 − {p1 , · · · , pl }.
We must now replace d2 Eα (fm ) by d2 Eα (fm ◦ gm ) in the above argument.
Once again, one obtains a nonconstant two-sphere fˆ∞ in the limit. Moreover,
a calculation similar to that for f∞ leads to the conclusion that fˆ∞ has index
≤ q − 2 ≤ (1/2)(dim M − 2), which once again contradicts Lemma 3.14.2.
Thus M must be a homotopy sphere, and follows from positive resolutions
of the generalized Poincaré conjecture ([51] and [23]) in dimensions ≥ 4 that M
is homeomorphic to a sphere.
Remark 3.14.5. It can be shown that if the real sectional curvatures K(σ) of
a Riemannian manifold satisfy the inequalities
1
< K(σ) ≤ 1, (3.64)
4
then the manifold has positive isotropic curvatures, but not conversely. The
sphere theorem proven by Berger, Klingenberg and Toponogov made the hy-
pothesis (3.64) on real sectional curvatures and is therefore weaker than the
sphere theorem we have proven. The complex projective space with the stan-
dard Fubini-Study metric is simply connected, has real sectional curvatures
lying in the range [1/4, 1] and has nonnegative isotropic curvature, but is not
homeomorphic to a sphere.
Remark 3.14.6. Conformally flat four-manifolds of positive scalar curvature
automatically have positive isotropic curvature. It is possible to construct a
conformally flat metric of positive scalar curvature on the connected sum of
a finite number of S 3 × S 1 ’s. Therefore the fundamental group of a compact
manifold of positive isotropic curvature can be a free group of arbitrary rank.
However, a recent article of Fraser [22] shows that the fundamental group of
such a manifold cannot contain a free abelian group of rank two.

175
Bibliography

[1] R. Abraham, Lectures of Smale on differential topology, Columbia Univer-

sity, 1963.

[2] R. Abraham, J. Marsden and T. Ratiu, Tensor analysis, Second Edition,

Addison-Wesley, 1988.
[3] M. Anderson, Geometrization of 3-manifolds via the Ricci flow , Notices
Amer. Math. Soc. 51 (2004),184-193.

[4] V. Anosov, On generic properties of closed geodesics, Math. USSR Izvestiya

21 (1983), 1-29.
[5] M. Atiyah, N. Hitchin and I. Singer, Self-duality in four-dimensional Rie-
mannian geometry, Proc. Roy. Soc. London Ser. A 362 (1978), 425-261.
[6] S. M. Bates, Toward a precise smoothness hypothesis in Sard’s theorem,
Proc. Amer. Math. Soc. 117 (1993), 279-283.
[7] L. Biliotti, M. A. Javaloyes and P. Piccione, On the semi-Riemannian
bumpy metric theorem, to appear.
[8] R. Bott, On the iteration of closed geodesics and the Sturm intersection
theory, Comm. Pure and Applied Math. 9 (1956), 171-206.
[9] R. Bott, Lectures on Morse theory, old and new , Bull. Amer. Math. Soc. 7
(1982), 331-358.
[10] R. Bott and L. Tu, Differential forms in algebraic topology, Springer Verlag,
New York, 1982.

[11] H. Brezis, The interplay between analysis and topology in some nonlinear
PDE problems, Bull. Amer. Math. Soc. 40 (2003), 179-201.
[12] J. Chen and G. Tian, Compactification of moduli space of harmonic map-
ping, Comm. Math. Helv. 74 (1999), 201-237.

[13] P. Deligne, P. Griffiths, J. Morgan and D. Sullivan, Real homotopy theory

of Kähler manifolds, Inventiones mathematicae 29 (1975), 245-274.

176
[14] S. Donaldson and P. Kronheimer, The geometry of four-manifolds, Claren-
don Press, Oxford, 1990.
[15] C. Earle and J. Eells, A fiber bundle description of Teichmüller theory, J.
Differential Geometry 3 (1969), 19-43

[16] J. Eells, A setting for global analysis, Bull. Amer. Math. Soc. 72 (1966),
751-807.
[17] J. Eells and S. Salamon, Twistorial construction of harmonic maps, Ann.
Scuola Norm. Sup. Pisa Cl. Sci. (4) 12 (1986), 589-640.
[18] J. Eells and J. Sampson, Harmonic maps of Riemannian manifolds, Amer.
J. Math. 86 (1964), 109-160.
[19] L. C. Evans, Partial differential equations, Amer. Math. Soc., Providence,
RI, 1998.
[20] A. Fischer and A. Tromba, On a purely “Riemannian” proof of the struc-
ture and dimension of the unramified moduli space of a compact Riemann
surface, Math. Ann. 267 (1984), 311-345.
[21] A. Fischer and A. Tromba, A new proof that Teichmüller space is a cell ,
Trans. Amer. Math. Soc. 303 (1987), 257-262.
[22] A. Fraser, Fundamental groups of manifolds with positive isotropic curva-
ture, Ann. of Math. 158 (2003), 345-354.
[23] M. H. Freedman and F. Quinn, Topology of 4-manifolds, Princeton Univ.
Press, Princeton NJ, 1990.
[24] M. Freedman, J. Hass and P. Scott, Least area incompressible surfaces in
3-manifolds, Inventiones math. 71 (1983), 609-642.
[25] D. Gilbarg and N. Trudinger, Elliptic partial differential equations of second
order , second edition, Springer, New York, 1983.
[26] P. Griffiths and J. Morgan, Rational homotopy theory and differential
forms, Birkhäuser, Boston, 1981.

[27] D. Gromoll and W. Meyer, Periodic geodesics on compact Riemannian

manifolds, J. Differential Geometry 3 (1969), 493-510.
[28] D. Gromoll and W. Meyer, On differentiable functions with isolated critical
points, Topology 8 (1969), 361-369.

[29] M. Gromov, Homotopical effects of dilatation, J. Differential Geometry 13

(1978), 303-310.
[30] A. Grothendieck, Sur la classification des fibres holomorphes sur la sphere
de Riemann, Amer. J. Math. 79 (1957), 121-138.

177
[31] R. Gulliver, R. Osserman and H. Royden, A theory of branched immersions
of surfaces, Amer. J. Math 95 (1973), 750-812.
[32] R. Hamilton, The inverse function theorem of Nash and Moser , Bull. Amer.
Math. Soc. 7 (1982), 65-222.

[33] F. Hang and F. Lin, Topology of Sobolev mappings II , Acta Math. 191
(2003), 55-107.
[34] P. Hartman, On homotopic harmonic maps, Canadian Math. J. 19 (1967),
673-687.
[35] A. Hatcher, Algebraic topology, Cambridge Univ. Press, Cambridge, UK,
2002.
[36] M. Hirsch, Differential topology, Sixth Edition, Springer, New York, 1997.
[37] Y. Imayoshi and M. Taniguchi, An introduction to Teichmüller spaces,
Springer, New York, 1992.

[38] W. Jäger and H. Kaul, Uniqueness and stability of harmonic maps and
their Jacobi fields, Manuscripta Math. 28 (1979), 269-291.
[39] J. Jost, Two-dimensional geometric variational problems, John Wiley and
Sons, New York, 1991.

[40] J. Jost, Riemannian geometry and geometric analysis, Fifth edition,

Springer, New York, 2008.
[41] J. Jost and M. Struwe, Morse-Conley theory for minimal surfaces of varying
topological type, Invent. math. 102 (1990), 465-499.
[42] W. Klingenberg, Lectures on closed geodesics, Springer, New York, 1978.

[43] S. Lang, Differential and Riemannian manifolds, Springer, New York, 1995.
[44] H. B. Lawson and M. L. Michelsohn, Spin geometry, Princeton Univ. Press,
Princeton, NJ, 1989.
[45] F. Lin and C. Wang, The analysis of harmonic maps and their heat flows,
World Scientific, Singapore, 2008.
[46] D. McDuff and D. Salamon, J-holomorphic curves and symplectic topology,
AMS Colloquium Publications 52, Amer. Math. Soc., Providence Rhode
Island, 2004.

[47] Y. Matsumoto, An introduction to Morse theory, Translations of Math-

ematical Monographs 208, Amer. Math. Soc., Providence Rhode Island,
2002.

178
[48] W. Meeks and S. T. Yau, The topology of three-dimensional manifolds
and embedding problems in minimal surface theory, Annals of Math. 112
(1980), 441-484.
[49] M. Micallef and J. D. Moore, Minimal two-spheres and the topology of
manifolds with positive curvature on totally isotropic two-planes, Annals of
Math. 127 (1988), 199-227.
[50] J. Milnor, Morse theory, Annals of Math. Studies 51, Princeton Univ.
Press, Princeton, NJ, 1963.
[51] J. Milnor, Lectures on the h-cobordism theorem, Princeton Univ. Press,
Princeton, NJ, 1965.
[52] J. D. Moore, Bumpy Riemannian metrics and closed parametrized min-
imal surfaces in Riemannian manifolds, revised version, arXiv preprint
1012.3906.
[53] J. D. Moore, Nondegeneracy of coverings of minimal tori in Riemannian
manifolds, Pacific J. Math. 230 (2007), 147-166.
[54] J. Morgan, Recent progress on the Poincaré conjecture and the classification
of 3-manifolds, Bull. Amer. Math. Soc. 42 (2004), 57-78.
[55] M. Morse, Relations between the critical points of a real function of n in-
dependent variables, Trans. Amer. Math Soc. 27 (1925), 345-396.
[56] M. Morse, The calculus of variations in the large, American Math. Soc.
Colloquium Publications 18, Ann Arbor, Mich., 1934.
[57] J. Nash, The imbedding problem for Riemannian manifolds, Annals of
Math. 63 (1956), 20-63.

[58] R. Osserman, A proof of the regularity everywhere of the classical solution

to Plateau’s problem, Annals of Math. 91 (1970), 550-569.
[59] R. Palais, Lusternik-Schnirelmann theory on Banach manifolds, Topology
5 (1966), 115-132.

[60] R. Palais, Foundations of global nonlinear analysis, Benjamin, New York,

1968.
[61] R. Palais, Critical point theory and the minimax principle, Proc. Symposia
in Pure Math. 15 (1970), 185-212.

[62] R. Palais and S. Smale, A generalized Morse theory, Bull. Amer. Math.
Soc. 70 (1964), 165-171.
[63] T. Parker, Bubble tree convergence for harmonic maps curvature flow , J.
Differential Geometry 44 (1996), 595-633.

179
[64] T. Parker and J. Wolfson, Pseudo-holomorphic maps and bubble trees, J.
Geometric Analysis 3 (1993), 63-98.
[65] M. Reed and B. Simon, Methods of mathematical physics I: Functional
analysis, Academic Press, New York, 1980.
[66] H. L. Royden, Real analysis, Third edition, Prentice-Hall, New York, 1988.
[67] W. Rudin, Real and complex analysis, Third Edition, McGraw-Hill, New
York, 1986.
[68] J. Sacks and K. Uhlenbeck, The existence of minimal immersions of 2-
spheres, Annals of Math. 113 (1981), 1-24.
[69] J. Sacks and K. Uhlenbeck, Minimal immersions of closed Riemann sur-
faces, Trans. Amer. Math. Soc. 271 (1982), 639-652.
[70] J. H. Sampson, Some properties and applications of harmonic mappings,
Annales Scientifiques de l’École normale Supérieure 11 (1978), 211-228.
[71] R. Schoen and S. T. Yau, On univalent harmonic maps between surfaces,
Inventiones math. 44 (1978), 265-278.
[72] R. Schoen and S. T. Yau, Existence of incompressible minimal surfaces
and the topology of three dimensional manifolds with non-negative scalar
curvature, Annals of Math. 110 (1979), 127-142.
[73] R. Schoen and S. T. Yau, Lectures on harmonic maps, International Press,
Boston, 1997.
[74] M. Schwarz, Morse homology, Birkhäuser Verlag, Basel, 1993.
[75] J. P. Serre, Homologie singulière des espaces fibrées, Annals of Math. 54
(1951), 425-505.
[76] S. Smale, Morse theory and a nonlinear generalization of the Dirichlet prob-
lem, Annals of Math. 80 (1964), 382-396.
[77] S. Smale, On the Morse index theorem, J. Math. Mech. 14 (1965), 1049-
1055; Corrigendum 16 (1967), 1069-1070.
[78] S. Smale, An infinite-dimensional version of Sard’s theorem, Amer. J.
Math. 87 (1966), 861-866.
[79] M. Taylor, Partial differential equations: basic theory, Springer, New York,
1996.
[80] K. Uhlenbeck, Morse theory on Banach manifolds, J. Functional Analysis
10 (1972), 430-445.
[81] V. S. Varadarajan, Lie groups, Lie algebras and their representations,
Prentice-Hall, Englewood Cliffs, NJ, 1974.

180
[82] M. Vigué-Poirrier and D. Sullivan, The homology theory of the closed
geodesic problem, J. Differential Geometry 11 (1976), 633-644.
[83] E. Witten, Supersymmetry and Morse theory, J. Differential Geometry 17
(1982), 661-692.

181