Академический Документы
Профессиональный Документы
Культура Документы
For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance
Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy
is not required from the publisher.
Printed in Singapore
In memory of Professors
S. S. Chern, T. Nagano, T. Otsuki and K. Yano
who had the most important influence on my research
v
b2530 International Strategic Relations and China’s National Security: World at the Crossroads
Foreword
For almost half a century now I am being a privileged mind witness of the
ongoing exceptional geometrical creativity of Distinguished MSU Professor
Bang-Yen Chen. Similarly as has been attempted in the Forewords to Pro-
fessor Chen’s previous books “Pseudo-Riemannian Geometry, δ-Invariants
and Applications” (2011) and “Total Mean Curvature and Submanifolds of
Finite Type” (2015), also in the Foreword to his present book on the geom-
etry of warped product manifolds and submanifolds I will try to describe a
personal perception of its contents from a somewhat general cultural point
of view.
Quoting from Chern’s Introduction to “Handbook of Differential Geome-
try. Volume 1” (eds. Franki Dillen et al.) that “While algebra and analysis
provide the foundations of mathematics, geometry is at the core.”, it may
be well in this context to look at the following citations (i) from Newton’s
“Philosophiae Naturalis Principia Mathematica” and (ii) from Neumann’s
“The Mathematician” and (iii) from Freudenthal’s “Initiation to Geome-
try”: (i) “Geometry (...) is, in fact, nothing other than that very part of the
totality of mechanics which forms the basis of and precisely determines the
art of measurement.” and (ii) “The most vitally characteristic fact about
mathematics is, in my opinion, its quite peculiar relationship to the natural
sciences, or, more generally, to any science which interprets experience on
a higher than purely descriptive level. Most people, mathematicians and
others, will agree that mathematics is not an empirical science, or at least
that it is practised in a manner which differs in several decisive respects
from the techniques of the empirical sciences. And, yet, its development is
very closely linked with the natural sciences. Some of the best inspirations
of modern mathematics (I believe, the best ones) clearly originated in the
natural sciences. (...). There is a peculiar duplicity in the nature of math-
vii
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page viii
ematics. One has to realize this duplicity, to accept it, and to assimilate it
into one’s thinking on the subject. This double face is the face of mathemat-
ics, and I do not believe that any simplified, unitarian view of the thing is
possible without sacrificing the essence.” and (iii) “Symmetry in its broad
sense probably was the first mathematical idea that caught the attention
of human beings, in particular before quantifications and countings of any
kind, human beings appear to have been interested in similarities of forms,
of shapes, in its wide sense.” ; (and it seems to me to be rather worthwhile
to be aware of Chern’s above quoted assertion whenever coming in contact
with mathematics).
In its main notions and in its main statements about these notions,
classical geometry – roughly speaking: Euclidean and conformal and pro-
jective geometry – essentially deals with abstractions and generalisations of
some primitive objects and their properties in accordance with the way that
these objects and their properties are perceived by human vision in “our”
surrounding world, statically as well as dynamically and directly as well as
through natural processes like various sorts of projections and of sections.
Caused by the uncomfort brought about by the finding out of the existence
of pairs of mutually incommensurable line segments, likely in the school of
Pythagoras, such as the sides and diagonals in the regular 5-gons with as
ratio the irrational golden section, (which is clearly present for instance in
the vertical : horizontal scale of “the screens” of human’s instantaneous vi-
sual fields and in “the basic construction” of the 10-step units of the human
DNA-molecules), serious attention has been given to the establishment of
the known mathematics in a logically sound deductive way in axiomatical
systems. Next, a more subtle than before appreciation of the axiomatical
method in mathematics and indeed of the very nature of mathematics as
a whole resulted from the solution of the parallel postulate problem of pla-
nar Euclidean geometry by Lobachevsky and Bolyai and Gauss with their
development of the classical non-Euclidean 2D geometries. And also fur-
ther along in this direction of looking for security about “mathematical
truths”, in particular the contributions of Hilbert and Gödel brought us to
the present day understanding of the logical basis of mathematics, which
may be summarised in the concluding sentence “It is better to be aware of
our limitations than to live in a fool’s paradise.” of Stewart’s chapter on
“The Shape of Logic” in his “Taming the Infinite”. At this stage thinking
back at Neumann’s double face of mathematics, simultaneously one may
recall Lemaı̂tre’s closing words of his Antwerp lecture on “Cosmic Radia-
tion and Cosmology” when referring to “the primaeval atom”: “Birth of
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page ix
Foreword ix
space and origin of multiplicity: all the familiar notions on which we reflect
and base our knowledge lose their meaning and disappear; space vanishes
to a point and multiplicity reduces to unity. (...). Let us modestly admit
our essential limitations!”.
At this stage it seems well to repeat the above quotation of Chern and
to add to it its following line in the “Handbook of Differential Geometry”:
“While algebra and analysis provide the foundations of mathematics, ge-
ometry is at the core. This was already recognized by Euclid whose book
contains a geometrical treatment of the number system.”, and in further
trying to prepare the setting in which later to put the spotlights on warped
products – at least in their most special manifestations – next follows Ein-
stein’s “erkenntnistheoretisches Credo” : “Ich sehe auf der einen Seite die
Gesamtheit der Sinnen-Erlebnisse, auf der andern Seite die Gesamtheit
der Begriffe und Sätze, die in Büchern niedergelegt sind. Die Beziehungen
zwischen den Begriffen und Sätzen unter einander sind logischer Art, und
das Geschäft des logischen Denkens ist strikte beschränkt auf die Herstel-
lung der Verbindung zwischen Begriffen und Sätzen unter einander nach
festgesetzten Regeln, mit denen sich die Logik beschäftigt. Die Begriffe
und Sätze erhalten “Sinn” bezw. “Inhalt” nur durch ihre Beziehung zu
Sinnen-Erlebnissen. Die Verbindung der letzteren mit den erstenen ist
rein intuitiv, nicht selbst von logischer Natur. Der Grad der Sicherheit,
mit der diese Beziehung bezw. intuitive Verknüpfung vorgenommen wer-
den kann, und nichts anderes, underscheidet die leere Phantasterei von
der wissenschaftlichen “Wahrheit”. Das Begriffssystem ist eine Schöpfung
des Menschen samt den syntaktischen Regeln, welche die Struktur der Be-
griffssysteme ausmachen. Die Begriffssysteme sind zwar an sich logisch
gänzlich willkürlich, aber gebunden durch das Ziel, eine möglichst sichere
(intuitive) und vollständige Zuordnung zu der Gesamtheit der Sinnen-
Erlebnisse zuzulassen; zweitens entstreben sie möglichste Sparsamkeit in-
bezug auf ihre logisch unabhängingen Elemente (Grundbegriffe und Axiome)
d. h. nicht definierte Begriffe und nicht erschlossene Sätze. Ein Satz ist
richtig, wem er innerhalb eines logischen Systems nach den acceptierten lo-
gischen Regeln abgeleitet ist. Ein System hat Wahrheitsgehalt, entsprechend
der Sicherheit und Vollständigkeit seiner Zuordnungs-Möglichkeit zu der
Erlebnis-Gesamtheit. Ein richtiger Satz erborgt seine “Wahrheit” von dem
Wahrheit-Gehalt des Systems, dem er angehört”, taken from his “Autobi-
ographisches”.
And, last in this preparation, here comes one more citation, “One ex-
pects a mathematical theorem or a mathematical theory not only to describe
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page x
and to classify in a simple and elegant way numerous and a priori disparate
special cases. One also expects “elegance” in its “architectural”, structural
makeup. Ease in stating the problem, great difficulty in getting hold of it
and in all attempts at approaching it, then again some very surprising twist
by which the approach, or some part of the approach, becomes easy, etc.
(...). I think that it is a relatively good approximation to truth – which is
much too complicated to allow anything but approximations – that math-
ematical ideas originate in empirics, although the genealogy is sometimes
long and obscure. But, once they are so conceived, the subject begins to live
a peculiar life of its own and is better compared to a creative one, governed
by almost entirely aesthetic motivations, than to anything else and in par-
ticular, to an empirical science. There is, however, a further point which,
I believe, needs stressing. As a mathematical discipline travels far from its
empirical source, or still more, if it is a second and third generation only
indirectly inspired by ideas coming from “reality”, it is beset with very grave
dangers. It becomes more and more purely aestheticizing, more and more
purely “l’art pour l’art”. This need not be bad, if the field is surrounded by
correlated subjects, which still have closer empirical connections, or if the
discipline is under the influence of men with an exceptionally well-developed
taste. But there is a grave danger that the subject will develop along the
lines of least resistance, that the stream, so far from its source, will separate
into a multitude of insignificant branches, and that the discipline becomes
a disorganized mass of details and complexities. In other words, at a great
distance from its empirical source, or after much “abstract” inbreeding, a
mathematical subject is in danger of degeneration.”, from Neumann’s “The
Mathematician”.
The purpose of this Foreword, then, consists in an attempt to make clear
the importance of warped products in the above described general setting
and to show in particular, against the background of the above mentioned
warning for the danger of mathematical decadence when traveling far from
empirical sources and worse, that actually exactly to the contrary the ge-
ometry of warped products clearly is situated at some of the main empirical
sources of geometry as such, namely at the sources of the very inspirations
of “our space” and of “our spacetime”. The following presentation to this
end will be done in a qualitative way, sometimes including too rough formu-
lations and oversimplifications; for underlying technical considerations, see
the textbook “Differentialgeometrie. Kurven-Flächen-Mannigfaltigkeiten”
of Kühnel (of which there exists an English translation too), Freudenthal
and Steiner’s Chapter 13 on “Group Theory and Geometry” in “Funda-
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page xi
Foreword xi
sor g = ghk dxh dxk constitutes the holonomy of (tangent 1D) directions on
M after parallel transport fully around infinitesimal co-ordinate parallelo-
grams and similarly the (0, 6) curvature tensor R·R (whereby here the first
R stands for the curvature operator which acts as a derivation on the (0, 4)
curvature tensor R) constitutes the holonomy of the Riemann sectional cur-
vatures after such same transformations. And while the knowledge of the
Riemann curvature tensor R is equivalent to the knowledge of the Riemann
sectional curvatures K(p, π) – for all (non-degenerate) tangent 2-planes π
at the points p of M – similarly the knowledge of the tensor R · R is equiv-
alent to the knowledge of the Deszcz double sectional curvatures L(p, π, π̄)
– for all pairs of (curvature dependent) tangent 2-planes π and π̄ at the
points p of M – . Thus, the locally Euclidean or locally flat spaces (R = 0,
or, equivalently, K(p, π) = 0 for all points p in M and for all proper tan-
gent 2-planes π to M at p) are the spaces (M n , g) which, in the sense of
Weyl, satisfy the symmetry property that all their (tangent) directions at
all their points remain invariant under the parallel transport fully around
all infinitesimal co-ordinate parallelograms cornered at these points. And,
as is well known, by applying projective transformations to locally Euclidean
spaces one obtains real space forms and the class of real space forms is closed
under projective transformations. Here, by real space forms are meant the
spaces (M n , g) of constant sectional curvatures K(p, π) = c, (c = 0, or,
c > 0, or, c < 0), or, still, in abbreviation, the CC-spaces, denoted by
M n (c), and these spaces are characterised by the fact that R = 2c g ∧ g,
i.e. that their (0, 4) Riemann curvature tensor R is constantly proportional
to the Kulkarni-Nomizu square of their metric tensor g. According to the
Lemma of Schur, for dimensions n ≥ 3, the real space forms M n (c) are
characterised as the spaces (M n , g) having the property that at all their
points p their sectional curvatures K(p, π) for all 2-plane sections π at p are
the same, or, for short, by the isotropy of their sectional curvature function
K(p, π) – i.e. the numerical value of this function being independent of
the tangent 2D directions π implies its constancy, or, still, that the com-
mon value K(p, π) for all planes π at any point p for n ≥ 3 automatically
moreover is independent of the points p as well – . And thus further, the
semi symmetric or Szabó symmetric spaces (R · R = 0, or, equivalently,
L(p, π, π̄) = 0 for all points p in M and for all proper pairs of tangent
2-planes π and π̄ to M at p) are the spaces (M n , g) which, in the sense of
Weyl, satisfy the symmetry property that all their Riemann sectional curva-
tures at all their points remain invariant under the parallel transport fully
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page xiii
Foreword xiii
Foreword xv
Foreword xvii
Foreword xix
Foreword xxi
Leopold Verstraelen
De Haan (Belgium); 29 - 10 - 2016.
b2530 International Strategic Relations and China’s National Security: World at the Crossroads
Preface
Warped products are the most natural and the most fruitful generalization
of Cartesian products. More precisely, a warped product is a manifold
equipped with a warped product metric of the form:
X X
g= gij (y)dy i ⊗ dy j + f (y) gst (x)dxs ⊗ dxt ,
i,j s,t
where the warped geometry decomposes into a product of the “y” geometry
and the “x” geometry, except that the second part is warped, i.e., it is
rescaled by a scalar function of the other coordinates “y”. If one substitutes
the variable y for the time variable t and x for a 3-dimensional spatial space,
then the first part becomes the effect of time in Einstein’s curved space.
How it curves space will define one or the other solution to a spacetime
model. For that reason different models of spacetime in general relativity
are often expressed in terms of warped geometry. Consequently, the notion
of warped products plays very important roles not only in geometry but
also in mathematical physics, especially in general relativity.
The term of “warped product” was introduced by R. L. Bishop and B.
O’Neill in [Bishop and O’Neill (1964)], who used it to construct a large
class of complete manifolds of negative curvature. However, the concept
of warped products appeared in the mathematical and physical literature
before [Bishop and O’Neill (1964)]; for instance, warped products were
called semi-reducible spaces in [Kruchkovich (1957)]. Nevertheless, inspired
by Bishop and O’Neill’s article, many important works on warped products
from intrinsic point of view were done during the last fifty years.
According to the famous Nash embedding theorem published in 1956,
every Riemannian manifold can be isometrically embedded in some Eu-
clidean spaces. Nash’s theorem shows that every warped product N1 ×f N2
can be embedded as a Riemannian submanifold in some Euclidean spaces
xxiii
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page xxiv
with sufficiently high codimension. Due to this fact, the author asked the
following basic question (see, e.g., [Chen (2002a)]).
Question: What can we conclude from an isometric immersion of an ar-
bitrary warped product into a Euclidean space or into a space form with
arbitrary codimension with arbitrary codimension?
The study of warped products from this extrinsic point of view was
initiated around the beginning of this century by the author in a series of
his articles. Since then the study of warped product submanifolds from
extrinsic point of view has become a very active research subject in differ-
ential geometry and many nice results on this subject have been obtained
by many geometers.
The main purpose of this book is thus to provide an extensive and
comprehensive survey on the study of warped product manifolds and sub-
manifolds from intrinsic and extrinsic points of view done during the last
few decades. It is the author’s hope that the reader will find this book
both a good introduction to the theories of warped product manifolds and
of warped product submanifolds as well as a useful reference for recent and
further research of both areas.
In concluding the preface, the author would like to thank World Scien-
tific Publishing for the invitation to undertake this project. He also would
like to express his appreciation to Professors D. E. Blair, I. Dimitric, O.
J. Garay, I. Mihai, M. Petrović-Torgašev, B. Sahin, B. Suceava, J. Van
der Veken, and S. W. Wei for reading parts of the manuscript and offering
many valuable suggestions. In particular, the author thanks Professor L.
Verstraelen for writing an excellent foreword for this book.
November 1, 2016
Bang-Yen Chen
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page xxv
Contents
Foreword vii
Preface xxiii
2. Submanifolds 23
2.1 Embedding theorems . . . . . . . . . . . . . . . . . . . . . 24
2.2 Formulas of Gauss and Weingarten . . . . . . . . . . . . . 26
2.3 Equations of Gauss, Codazzi and Ricci . . . . . . . . . . . 30
2.4 Existence and uniqueness theorems of submanifolds . . . . 34
2.5 Reduction theorems . . . . . . . . . . . . . . . . . . . . . 35
2.6 Totally geodesic submanifolds . . . . . . . . . . . . . . . . 37
2.7 Totally umbilical submanifolds . . . . . . . . . . . . . . . 38
2.8 Pseudo-umbilical submanifolds . . . . . . . . . . . . . . . 41
2.9 Cartan’s structure equations . . . . . . . . . . . . . . . . . 46
xxv
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page xxvi
Contents xxvii
Contents xxix
Bibliography 451
General Index 473
Author Index 481
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 1
Chapter 1
1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 2
and product f1 f2 . The usual algebraic rules hold for these two operations,
which make F (M ) a commutative ring. We denote by X(M ) the set of all
smooth vector fields on M .
For V, W ∈ X(M ), the bracket [V, W ] is defined by
[V, W ]x (f ) = Vx (W f ) − Wx (V f )
at each x ∈ M and f ∈ F (M ). The bracket operation [ , ] on X(M ) is a
R-bilinear and skew-symmetric, which also satisfies the Jacobi identity:
[X, [Y, Z] ] + [Y, [Z, X] ] + [Z, [X, Y ] ] = 0.
These makes X(M ) an infinite-dimensional Lie algebra.
for k = 1, . . . , n.
Proof. Since both ∇ and the bracket operation on vector fields are local
operations, it suffices to work on any neighborhood of x. Moreover, be-
cause the identities are tensor equations, u, v, w, z can be extended to local
vector fields U, V, W, Z on some neighborhood of x in any convenient way.
In particular, we may choose the extensions in such way that all of their
brackets are zero.
Since R(U, V )W = [∇U , ∇V ]W − ∇[U,V ] W and the bracket operation is
skew-symmetric, (1.7) follows immediately from the definition of the cur-
vature tensor.
For (1.8) we only need to show that hR(u, v)w, zi = 0 by polarization.
By Theorem 1.1(5), we have
hR(U, V )W, W i = h∇U ∇V W, W i − h∇V ∇U W, W i
= h∇U W, ∇V W i − V h∇U W, W i + h∇V W, ∇U W i − U h∇V W, W i
1 1
= U V hW, W i − V U hW, W i = 0.
2 2
This proves (1.8), since [U, V ] = 0. For (1.9) we consider S, the sum of
cyclic permutations of U, V, W , to find
R(U, V )W + R(V, W )U + R(W, U )V
= SR(U, V )W
= S∇U ∇V W − S∇V ∇U W
= S∇U ∇V W − S∇U ∇W V
= S∇U [Y, W ] = 0.
If we put
S(u, v, w, z) = hR(u, v)w, zi + hR(v, w)u, zi + hR(w, u)v), zi ,
then a direct computation shows that
0 = S(u, v, w, z) − S(v, w, z, u) − S(w, z, u, v) + S(z, u, v, w)
= hR(u, v)w, zi − hR(v, u)w, zi − hR(w, z)u, vi + hR(z, w)u, vi .
Thus, by applying (1.7), we obtain (1.10).
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 12
(1) K is constant;
(2) a ≤ K or K ≤ b;
(3) a ≤ K ≤ b on indefinite planes;
(4) a ≤ K ≤ b on definite planes,
Remark 1.1. The Ricci curvature provides one way of measuring the de-
gree to which the geometry determined by the given Riemannian metric
on M differ from that of the Euclidean n-space. The scalar curvature rep-
resents the amount by which the volume of a geodesic ball in a curved
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 15
The point x0 is called the center of Ssm (x0 , c) and Hsm (x0 , c). If x0 is
the origin o, we simply denote Ssk (o, c) and Hsk (o, c) by Ssk (c) and Hsk (c),
respectively.
The pseudo-Riemannian manifolds Eks , Ssk (c), Hsk (c) are the standard
models of the indefinite real space forms. In particular, Ek1 , S1k (c), H1k (c)
are the standard models of Lorentzian space forms. Topologically, a de
Sitter spacetime S1k is R × S k−1 . Thus when k ≥ 3 a de Sitter spacetime is
simply-connected.
The S14 and H14 are known as the de Sitter spacetime and anti-de Sitter
spacetime, respectively; named after Willem de Sitter (1872-1934), a Dutch
mathematician, physicist and astronomer.
When s = 0, the manifolds Ek , S k (c) and H k (c) are of constant curva-
ture, called real space forms. The Euclidean k-space Ek , the k-sphere S k (c)
and the hyperbolic k-space H k (c) are simply-connected complete Rieman-
nian manifolds of constant curvature 0, c > 0 and c < 0, respectively.
A complete simply-connected pseudo-Riemannian k-manifold, k ≥ 3,
of constant curvature c and index s is isometric to Eks , or Ssk (c) or Hsk (c)
according to c = 0, or c > 0 or c < 0, respectively.
We denote a k-dimensional indefinite space form of curvature c and
index s simply by Rsk (c). We simply denote the indefinite space form R0k (c)
with index s = 0 by Rk (c).
Xn X
∂f ∂f
df = duj and ∇f = g ij ∂j . (1.23)
j=1
∂uj i,j
∂ui
Axiom 3. The Lie derivative obeys the Leibniz rule with respect to con-
traction:
LX (T (Y1 , . . . , Yn )) = (LX T )(Y1 , . . . , Yn ) + T (LX Y1 , . . . , Yn )
+ T (Y1 , . . . , LX Yn ).
Axiom 4. The Lie derivative commutes with exterior derivative d on
functions: [LX , d] = 0.
Under the flow of a Killing vector field X, the metric tensor does not
change. Thus a Killing vector field is an infinitesimal isometry.
Remark 1.2. It follows immediately from Lemma 1.4 and (1.30) that every
concircular vector field is a conformal vector field.
Example 1.1. Let I be an open interval of the real line R and let ϕ(s), s ∈
I, be a function on I which is nowhere zero. Consider a warped product
manifold of the form:
I ×ϕ(s) F, (1.31)
where F is a Riemannian manifold. The metric tensor g of I ×ϕ(s) F is
given by g = ds2 + ϕ2 (s)gF , where gF is the metric tensor of the second
factor F . Consider the vector field given by
∂
v = ϕ(s) . (1.32)
∂s
It is easy to verify that the vector field v satisfies (1.30) with µ = ϕ′ (s).
Thus v is a concircular vector field.
Chapter 2
Submanifolds
23
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 24
Submanifolds 25
Submanifolds 27
Submanifolds 29
Hence
˜ X ξ, η i + h ξ, ∇
h DX ξ, η i + h ξ, DX η i = h ∇ ˜ X η i = Xhξ, ηi.
The length of the mean curvature vector is called the mean curvature.
Submanifolds 31
Submanifolds 33
Proposition 2.5 has some nice applications, e.g., it implies the following.
Submanifolds 35
Theorem 2.5. (Uniqueness) Let φ, φ′ : Mtn → Rsm (c) be two isometric im-
mersions of a pseudo-Riemannian n-manifold Mtn into an indefinite space
for Rsm (c) of constant curvature c with normal bundles ν and ν ′ equipped
with their canonical bundle metrics, connections and second fundamental
forms, respectively. Suppose there is an isometry φ : Mtn → Mtn such that
φ can be covered by a bundle map φ̄ : ν → ν ′ which preserves the bundle
metrics, the connections and the second fundamental forms. Then there is
an isometry Φ of Rsm (c) such that Φ ◦ φ = φ′ .
Let Rni,j denote the affine n-space with the metric whose canonical form is
Oj
−Ii ,
In−i−j
where Ik is the k × k identity matrix and Oj is the j × j zero matrix.
The metric is non-degenerate if and only if j = 0. The j in Rni,j measures
the degenerate part. The metric of Rni,1 = R0 × Ein−1 vanishes on the first
factor R0 and it is the standard pseudo-Euclidean metric with index i on
the second factor Ein−1 . Denote the natural embedding ι : Rni,1 → En+1 i+1 of
Rni,1 into En+1
i+1 given by
ι((x1 , x2 , . . . , xn )) = (x1 , x2 , . . . , xn , x1 ) ∈ En+1
i+1
Submanifolds 37
Proof. Under the hypothesis, assume that Mtn is totally umbilical in Nsm .
If γ : I → Mtn is a null geodesic of Mtn , then γ ′ (t) is a null vector for each
t ∈ I. Then it follows from (2.19) that σ(γ ′ (t), γ ′ (t)) = 0. Thus ∇ ˜ γ ′ γ ′ = 0,
m
which shows that γ is also a geodesic of Ns .
Conversely, if null geodesics of Mtn are geodesics of Nsm , then σ(v, v) = 0
for null vectors v of Mtn . At a point x ∈ Mtn , let us choose an orthonormal
basis {e1 , . . . , en } of Tx Mtn such that hei , ei i = −1 for i = 1, . . . , t and
hej , ej i = 1 for j = t+1, . . . , n. Then ei ±ej are null vectors for i ∈ {1, . . . , t}
and j ∈ {t + 1, . . . , n}. Thus σ(ei ± ej , ei ± ej ) = 0, which implies that
σ(ei , ej ) = 0, σ(ei , ei ) + σ(ej , ej ) = 0. (2.35)
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 39
Submanifolds 39
√
If t ≥ 2, then ei1 + ei2 + 2en is a null vector for 1 ≤ i1 6= i2 ≤ s. Thus we
find σ(ei1 , ei2 ) = 0 by applying (2.35).
Similarly, if n − t ≥ 2, we have
σ(ej1 , ej2 ) = 0, t + 1 ≤ j1 6= j2 ≤ n.
Consequently, we obtain (2.19). Thus Mtn is totally umbilical in Nsm .
Submanifolds 41
Proof. Under the hypothesis, the positive vector field is a normal vector
˜ Xx = X
field of M which is normal to Ssm (r−2 ) or to Hsm (−r−2 ). Since ∇
for X ∈ T M , the Weingarten formula yields
Ax = −I, DH = D′ H ′ , (2.40)
where A is the Weingarten map in Em+1
s or in Em+1
s+1 . Hence it follows from
the formula of Gauss and (2.40) that
˜ X Y = ∇′ Y − ǫ x = ∇X Y + σ ′ (X, Y ) − ǫ x,
∇ (2.41)
X
r2 r2
which gives (2.38). By taking the trace of (2.38) we get (2.39).
(1) φ has parallel mean curvature vector if and only if x = ι ◦ φ has parallel
mean curvature vector.
(2) φ is a parallel immersion if and only if x = ι◦φ is a parallel immersion.
(3) φ is totally umbilical if and only if x = ι ◦ φ is totally umbilical.
Case (2) occurs only when s ≥ 1 and case (3) occurs only when s ≥ 1 and
m ≥ dim M + 2.
Proof. Assume that M is a pseudo-umbilical submanifold with parallel
mean curvature vector in Em
s . Then
XhH, Hi = 2 hH, DX Hi = 0
for any X ∈ T M . Thus hH, Hi is constant.
Case (a): hH, Hi 6= 0. We put
ǫ
hH, Hi =
, (2.43)
r2
where ǫ = 1 or −1 depending on H is spacelike or timelike. On the other
hand, it follows from (2.42) that AH = λI for some function λ ∈ F (M ).
Thus we find from (2.43) that ǫ = λr2 . Let us put
φ̂ = φ + ǫr2 H. (2.44)
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 43
Submanifolds 43
φ = (f, z, f ), (2.45)
where f is function on N and z : N → Em−2
is an isometric immersion.
s−1
Now, by applying the Laplace operator ∆ to (2.45) we find from Beltrami’s
formula that nH = (−∆f, nHz , −∆f ), where Hz is the mean curvature
vector of z. Since H = (1, 0, . . . , 0, 1), we find Hz = 0 and ∆f = −nr.
Thus z is a minimal immersion and ∆f is a nonzero constant.
The converse can be verified easily.
Submanifolds 45
from (2.47) and Lemma 2.3 that the mean curvature vector of M in Ssm (1)
is exactly η. Therefore M is minimal in H1 .
The converse is easy to verify.
Chapter 3
One of the most fruitful generalizations of the notion of Cartesian (or direct)
products is the notion of warped products defined in [Bishop and O’Neill
(1964)]. The concept of warped products appeared in the mathematical
and physical literature before [Bishop and O’Neill (1964)]. For instance,
warped products were called semi-reducible spaces in [Kruchkovich (1957)].
Many exact solutions of the Einstein field equations and modified field
equations are warped products. For instance, the Schwarzschild solution
and Robertson-Walker models are warped products. While the Robertson-
Walker models describes a simply-connected homogeneous isotropic ex-
panding or contracting universe, the Schwarzschild solution is the best rel-
ativistic model that describes the outer space around a massive star or a
black hole. The Schwarzschild model laid the groundwork for the descrip-
tion of the final stages of gravitational collapse and the objects known today
as black holes.
Twisted products and convolution manifolds are two natural extensions
of warped product manifolds. In the last three sections of this chapter we
discuss both twisted products and convolution manifolds.
47
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 48
Based on Lemma 3.1, we may simply write λ for λ ◦ π and grad λ for
grad(λ ◦ π). The Levi-Civita connection ∇ of M = B ×f F is related with
the Levi-Civita connections of B and F as follows.
Proposition 3.1. For X, Y ∈ L(B) and V, W ∈ L(F ), we have on B ×f F
that
(1) ∇X Y ∈ L(B) is the lift of ∇X Y on B;
(2) ∇X V = ∇V X = (X ln f )V ;
hV, W i
(3) nor(∇V W ) = σ(V, W ) = − ∇f ;
f
(4) tan(∇V W ) ∈ L(F ) is the lift of ∇′V W on F , where ∇′ is the Levi-Civita
connection of F .
Proof. Property (1) can be proved as follows. From Koszul’s formula we
find 2 h∇X Y, V i = hV, [X, Y ]i − V hX, Y i due to [X, V ] = [Y, V ] = 0. Since
X, Y are lifts from B, hX, Y i is constant on fibers. Because V is vertical,
V hX, Y i = 0. But [X, Y ] is tangent to leaves, hV, [X, Y ]i = 0. Hence,
h∇X Y, V i = 0 for all V ∈ L(F ). This shows that ∇X Y is horizonal. Since
each π|B×q is an isometry, we obtain properties (1).
From [X, V ] = 0, we find ∇X V = ∇V X. Since these vector fields are
vertical, we obtain h∇X V, Y i = − hV, ∇X , Y i = 0. Thus, by the Koszul
formula, we get
2 h∇X V, W i = XhV, W i . (3.5)
On the other hand, by the definition of warped product metric, we find
hV, W i(p,q) = f 2 (p) hVq , Wq i .
So, after writing f for f ◦ π, we have hV, W i = f 2 (hV, W i ◦ η). Hence
XhV, W i = X(f 2 (hV, W i ◦ η) = 2f Xf (hV, W i ◦ η) = 2(X ln f ) hV, W i .
Combining this with (3.5) give property (2). By property (2) we find
h∇V W, Xi = − hW, ∇V Xi = −(X ln f ) hV, W i .
Thus, after applying Lemma 3.1, we find Xf = h∇f, Xi on M as on B.
Hence, for any X, we obtain
h∇V W, Xi f = − hV, W i h∇f, Xi
which implies property (3).
Since V and W are tangent to all fibers, tan(∇V W ) is the fiber covariant
derivative applied to the restrictions of V and W to that fiber. Therefore,
we have property (4).
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 50
Lemma 3.2. [Kim and Kim (2003)] Let f be a smooth function on a Rie-
mannian manifold B. Then for any vector X, the divergence of the Hessian
tensor H f satisfies
Proof. Taking the trace of both sides of the formula given in Proposition
3.4(2) gives
k
2τ = mλ − ∆f, (3.11)
f
where τ is the scalar curvature of B defined by (1.16).
The second Bianchi identity implies
dτ = div(Ric). (3.12)
From (3.11) and (3.12), we obtain
k
div(Ric(X)) = {(∆f )df − f d(∆f )}(X). (3.13)
2f 2
On the other hand, by definition we have
f
H 1 1
div (X) = − 2 H f (∇f, X) + divH f (X)
f f f
for any vector field X and an orthonormal frame E1 , . . . , Em of B. Since
1
H f (X, ∇f ) = (∇X df )(∇f ) = d(|∇f |2 )(X),
2
the last equation becomes
f
H 1 1
div (X) = − 2 d(|∇f |2 )(X) + divH f (X)
f 2f f
for a vector field X on B. Thus, it follows from (3.7) and condition (2) of
Proposition 3.4 that
f
H 1
div = {(k − 1)d(|∇f |2 ) − 2f d(∆f ) + 2λf df }. (3.14)
f 2f 2
But condition (2) gives div(Ric) = div fk H f . Therefore (3.13) and (3.14)
imply
d(λf 2 − f ∆f + (k − 1)|∇f |2 ) = 0.
Thus condition (3) in Proposition 3.4 holds for some constant µ. Hence the
first part of the proposition is proved. For a compact Einstein k-manifold
(F, gF ) with FRic = µgF , we can construct a compact Einstein warped
product B ×f F by the sufficiencies of Proposition 3.4
The following are examples of Einstein warped product manifolds.
Notice that En∗ , M1 and M−1 are non-complete Riemannian manifolds, but
they are open dense submanifolds of En , S n (1) and H n (−1), respectively.
For complete Einstein warped products B ×f F with dim B = 1, we
have the following.
Question 3.1. Does there exist a compact Einstein warped product with
non-constant warping function?
The necessary and sufficient condition for the warped product of two
conformally flat manifolds to be conformally flat was obtained in [Ogawa
(1978)]. We also have the following result from [Brozos-Vázquez et al.
(2005)] for conformally flat warped product manifolds.
g = f 2 (f −2 gB + gF ).
N1 × N2 × · · · × Nℓ with f2 = · · · = fℓ = 1
then we have
ℓ
X
i i
∇fX Y − ∇0X Y = X , Y Ůi − h Ůi , X i Y i − h Ůi , Y i X i , (3.34)
i=2
ℓ
X
Rf (X, Y ) − R0 (X, Y ) = (∇fX 1 Ůi − h Ůi , X i Ůi ) ∧ Y i (3.35)
i=2
k
X ℓ
X
+ X i ∧ (∇fY 1 Ůi − h Ůi , Y i Ůi ) − h Ůi , Ůj i X i ∧ Y j ,
i=2 i,j=2
Remark 3.1. The Euclidean 3-space E3 has two famous warped product
representations; namely, the cylindrical and the spherical coordinates. S.
Nölker generalized these these two examples as far as possible: starting
from algebraic initial data, he constructed for every standard space Rn (k) of
constant curvature k an isometry ψ of a warped product N1 ×ρ1 × · · ·×ρk Nk
onto an open dense subset of Rn (k). Such a warped product representation
ψ has many “rotational” symmetries for i = 2, . . . , k.
The product foliation induced by Ni is an orbit foliation with respect
to the action of a suitable subgroup of the isometry group of Rn (k). Every
leaf of this foliation is a complete totally umbilical submanifold of Rn (k).
The foliation induced by N1 is even totally geodesic (but its leaves are in
general not complete).
For the details on possible warped product representations of standard
spaces of constant curvature, see [Nölker (1996)].
Proof. Follows from Lemma 3.4 and the definition of totally umbilical
immersions.
The following is the indefinite version of Moore’s lemma [Moore (1971)]
stated in [Magid (1984)].
where pi ∈ Ni , 1 ≤ i ≤ ℓ.
(a) The partial mean curvature vector H1 is equal to the mean curvature
vector of φ1 : N1 → M1 ; thus, φ is N1 -minimal if and only if φ1 :
N1 → M1 is a minimal immersion.
(b) φ is N2 -minimal if and only if φ2 : N2 → M2 is a minimal immersion
and (∇ ln ρ)|N1 = ∇ ln f holds.
(c) φ = (φ1 , φ2 ) is a minimal immersion if and only if φ2 : N2 → M2 is a
minimal immersion and the mean curvature vector of φ1 : N1 → M1 is
given by n−1
1 n2 D ln ρ.
On the other hand, the equation of Gauss implies that the curvature
tensor R of N1 ×f N2 satisfies
hR(X, Y )Z, W i = hσ(X, W ), σ(Y, Z)i − hσ(X, Z), σ(Y, W )i
(3.61)
+ c {hX, W i hY, Zi − hX, Zi hY, W i}
for vectors X, Y, Z, W ∈ T (N1 ×f N2 ). Using (3.57), (3.60), and (3.61), we
obtain
∆f
h H1 , σ(Z, Z) i = −c (3.62)
n1 f
for any unit Z ∈ L(N2 ). Thus, by applying polarization, we find
h H1 , σ(Z, W ) i = 0 (3.63)
for orthonormal vectors Z, W ∈ L(N2 ). Equations (3.62) and (3.63) imply
that the shape operator at H1 satisfies
∆f
AH1 Z = −c Z (3.64)
n1 f
for Z ∈ L(N2 ). Thus we have statement (1).
It follows from (3.57) and (3.58) that the covariant derivative of the
second fundamental form satisfies
¯ X σ)(Y, Z) = DX σ(Y, Z) − σ(∇X Y, Z) − σ(Y, ∇X Z)
(∇ (3.65)
= −σ(∇X Y, Z) = 0,
due to the fact that N1 is totally geodesic in N1 ×f N2 .
On the other hand, by applying (3.57) and (3.58), we also find
¯ Z σ)(X, Y ) = DX σ(X, Y ).
(∇ (3.66)
Therefore, after applying (3.65), (3.66), and the equation of Codazzi, we
obtain statement (2).
By applying equations (3.39) and (3.62) we obtain
∆f
h H1 , H 2 i = −c (3.67)
n1 f
which gives statement (3). It follows from equations (3.48) and (3.49) that
the partial mean curvature vectors H1 and H2 are perpendicular to each
other if and only if we have either (i) H1 = 0 or (ii) (∇ ln ρ)|N1 = ∇ ln f .
According to Theorem 3.12(a), the first case occurs when and only when
φ1 is a minimal immersion. By combining these results with statement
(3), we obtain statement (4). Obviously, statement (5) is a special case of
statement (3). If φ1 is a non-minimal immersion and if the two partial mean
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 71
Remark 3.3. Many results given in this section were extended to doubly
warped product immersions φ : f1 N1 ×f2 N2 → ρ1 M1 ×ρ2 M2 in [Faghfouri
and Majidi (2015)].
Theorem 3.15. [Kazan and Sahin (2013)] Let B×b F be the twisted product
of pseudo-Riemannian manifolds with dim B > 1 and dim F > 1. Then
B ×b F can be expressed as a warped product of (B, gB ) and (F, gF ) if and
only if B is conformally flat along F , where gF is a metric tensor conformal
to gF .
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 74
The notion of twisted products was extended to the notion of doubly twisted
products in [Ponge and Reckziegel (1993)].
The following two results were proved in [Ponge and Reckziegel (1993)].
Proposition 3.7. [Ponge and Reckziegel (1993)] Let (M1 , g1 ) and (M2 , g2 )
be two pseudo-Riemannian manifolds and g the metric of doubly twisted
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 77
product M1 ×(f1 ,f2 ) M2 . Put Ui = −∇(ln fi2 ), where the gradient of ln fi2
is calculated with respect to g. Then the Levi-Civita connection ∇ and
curvature tensor R of the doubly twisted product M1 ×(f1 ,f2 ) M2 is related
to the Levi-Civita connection ∇ ˜ and curvature tensor R of the direct product
of (M1 , g1 ) and (M2 , g2 ) by
X
∇X Y = ∇ ˜ XY + {g(Pi X, Pi Y )Ui − g(X, Ui )Pi Y − g(Y, Ui )Pi X}
i
X
R(X, Y ) = R̃(X, Y ) + {(∇X Ui − g(X, Ui )Ui ) ∧ Pi Y
X
− (∇Y Ui − g(Y, Ui )Ui ) ∧ Pi X} + g(Ui , Uj )Pi X ∧ Pj Y,
i,j
where u ∧ v is the linear map w 7→ g(v, w)u − g(u, w)v for all u, v ∈ Tx M
and Pi : T M → ζi is the vector bundle projection related to the splitting
T M = ζ1 ⊕ ζ2 with ζ1 = ker(π2∗ ) and ζ2 = ker(π1∗ ).
Definition 3.18. [Chen (2003e)] Let (N1 , g1 ) and (N2 , g2 ) be two pseudo-
Riemannian manifolds and let f ∈ F (N1 ) and h ∈ F (N2 ). The symmetric
tensor h g1 ∗ f g2 on N1 × N1 defined by
We put Em m n n
∗ = E − {0} and Cs∗ = Cs − {0} for 0 ≤ s < n.
The next result from [Chen (2003e)] shows that the notion of convolution
manifolds arises very naturally from tensor product immersions.
Chapter 4
Astronomers were unsure of the size of our galaxy at the beginning of the
20th century. Generally, they believed it was not much greater than a few
tens of thousands of light years across. However, astronomers had noted
fuzzy patches of light in the night sky, which are called nebulae. Some
astronomers thought these could be distant galaxies.
E. Hubble observed in the 1920s that some of these nebulae were indeed
distant galaxies comparable in size to our own Milky Way. Hubble also
made the remarkable discovery that our Universe is expanding. Hubble’s
observations that the light from nebulae showed a red shift increasing with
distance ruled out the possibility that Einstein’s static model1 as well as de
Sitter’s static model without matter represented the real universe. A new
estimate of the mass of our galaxy made in 1927 caused de Sitter to re-
examine his assumption. At a meeting in London of the Royal Astronomical
Society in early 1930, de Sitter admitted that neither his nor Einstein’s
solution to the field equations could represent the observed universe.
In fact, a few astronomers had been looking for other solutions to Ein-
stein’s field equations. Back in 1922 A. Friedmann had published a set of
possible mathematical solutions that gave a non-static universe. In 1927, G.
Lemaı̂tre arrived independently at similar results as Friedmann. During the
1930s, H. Robertson and A. G. Walker explored the problem further. They
rigorously proved in 1935 that the Robertson-Walker metric is the only one
on a spacetime that is spatially homogeneous (all places look the same) and
isotropic (all spatial direction the same). This is the birth of Robertson-
Walker spacetimes. In the last section we discuss Schwarzschild’s solution
to Einstein’s field equations briefly.
1 A static model is a cosmological model in which the universe is both spatially infinite
81
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 82
(1) ∇∂t ∂t = 0;
(2) ∇∂t V = ∇V ∂t = (ln f )′ V ;
(3) h ∇V W, ∂t i = − hV, W i (ln f )′ ;
(4) (∇V W )− is the lift of ∇′V W on Rm−1 (k),
f ′′
(1) R(∂t , V )∂t = V;
f
f ′′
(2) R(V, ∂t )W = − hV, W i ∂t ;
f
(3) R(V, W )∂t = 0;
k + f ′2
(4) R(U, V )W = {hV, W i U − hU, W i V }
f2
Proof. (1) follows immediately from Lemma 4.1. From (4.3) we get
[X, Y ] = [X, Y ]− ∈ D⊥ . Thus, we obtain (2) by Frobenius’ theorem.
For (3) let us observe that Lemma 4.1(1) implies that the rank-one
distribution spanned by ∂t on N is a totally geodesic distribution of N .
Moreover, it follows from (1) that the integral manifolds of D⊥ are totally
umbilical hypersurfaces of N with constant mean curvature. Thus, Hiepko’s
theorem implies that N is locally the warped product of an open subinterval
I ′ ⊂ I and an integral manifold P n−1 of D⊥ with respect to the warping
function fˆ = f |I ′ .
Since P n−1 is perpendicular to I ′ ⊂ I, P n−1 lies in some rest space
S(t0 ), t0 ∈ I ′ . Without loss of generality, we may assume that P n−1 is a
submanifold of Rm−1 (k). This proves statement (3).
Statement (4) follows from statement (3) and Lemma 4.1.
S(t0 )
(3) the normal curvature tensor RD of N in Rm (k, f ) and RD of N
D DS(t0 )
in S(t0 ) satisfy R (X, Y )ξ = R (X, Y )ξ for X, Y ∈ X(N ) and ξ
orthogonal to ∂t .
Proof. Statements (1) and (2) follow from Lemma 4.1; and statement
(3) is an easy consequence of statement (2).
The next two corollaries follow immediately from Lemma 4.5(1).
of Lm
1 (k, f ) is a parallel submanifold if and only if it is one of the following:
Remark 4.2. Most results given in sections 4.2-4.4 are based on [Chen
and Van der Veken (2007); Chen and Wei (2008b)].
Then
(j+1)-th
z}|{ (4.23)
−1
φs = (1, bf (s) , 0, . . . , 0), φuj = (0, . . . , 0, 1 , 0, . . . , 0),
for j = 2, . . . , n. The metric tensor induced from (4.21) via φ is
Xn
g = (b2 − 1)ds2 + f 2 (s) du2j , (4.24)
j=2
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 92
and
r-th
1 z}|{
ξ1 = √ 2 b, f −1 , 0, . . . , 0 , ξr = f −1 0, . . . , 0, 1 , 0, . . . , 0 ,
b −1
r = n + 2, . . . , m,
are orthonormal normal vector fields to N in Lm 1 (0, f ). A straightforward
computation shows that the second fundamental form of φ satisfies
√
b b2 − 1f ′
σ(∂s , ∂s ) = ξ1 , σ(∂s , ∂ui ) = 0, σ(∂ui , ∂uj ) = 0,
f
(4.25)
bf f ′
σ(∂u2 , ∂u2 ) = · · · = σ(∂un , ∂un ) = √ ξ1 ,
b2 − 1
for 2 ≤ i 6= j ≤ n. Hence φ is a totally umbilical immersion such that
bf ′
σ(X, Y ) = √ hX, Y i ξ1 . (4.26)
b2 − 1f
Since f ′′ is nowhere zero, (4.26) implies that φ is totally umbilical with
non-constant mean curvature. Therefore, we have DH 6= 0.
Proposition 4.6. Assume that Lm 1 (k, f ) contains no open subsets of con-
stant curvature and N is a totally umbilical pseudo-Riemannian submani-
fold of Lm
1 (k, f ) with dim N ≥ 3. Then N is an Einstein manifold if and
only if N is a transverse submanifold.
Proof. If Lm 1 (k, f ) contains no open subsets of constant curvature and
N is a totally umbilical in Lm 1 (k, f ) with dim N ≥ 3, then it follows from
(4.18) and Lemma 4.3(4) that the sectional curvature K N (u ∧ v) of N with
respect to orthonormal vectors u, v is given by
k + f ′2 k
K N (u ∧ v) = hH, Hi + + (ǫ ϕ2
u u + ǫ ϕ
v u
2
) − (ln f ) ′′
.
f2 f2
If we choose an orthonormal basis e1 , . . . , en of Tx N such that e2 , . . . , en
are perpendicular to ∂t , then the Ricci curvature of N satisfies
k + f ′2 2 k ′′
Ric(e1 ) = (n − 1) hH, Hi + + ǫ1 ϕe1 − (ln f ) ,
f2 f2
(4.27)
k + f ′2 2 k ′′
Ric(ej ) = (n − 1) hH, Hi + + ǫ1 ϕe1 − (ln f ) ,
f2 f2
for j = 2, . . . , n. Since n ≥ 3, it follows from (4.27) and the Einstein
condition that ϕe1 = 0. Hence N is a transverse submanifold.
Conversely, if N is a totally umbilical transverse submanifold of
Lm1 (k, f ), then N lies in a rest space S(t0 ) for some t0 ∈ I. Since S(t0 ) is
a real space form, Gauss’ equation implies that N is of constant curvature.
Hence, N is an Einstein manifold.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 93
The last condition implies that the spacetime is a vacuum, i.e., it is Ricci
flat. Based on those conditions, Schwarzschild found a coordinate chart
(t, r, θ, φ), known as Schwarzschild’s chart for his spacetime.
First, he derived from the static and spherically symmetry conditions
that the metric is a warped product metric of the form:
g = −F (r)dt2 + G(r)dr2 + r2 (dθ2 + sin2 θdφ2 ),
(4.42)
t ∈ R, r ∈ I, 0 < θ < π, −π < φ < π,
where I is an open interval. Hence, on each rest space given by t constant,
the surface r constant has the line element
dσ 2 = r2 (dθ2 + sin2 θdφ2 ), (4.43)
which is a round sphere with Gauss curvature r−2 and surface area 4πr2 .
Obviously, sufficiently far away from the source of gravitation, the influ-
ence of the star becomes arbitrary small. Therefore one may require that
the metric approaches the physical Minkowski metric (in special relativ-
ity) as r → ∞, i.e., the metric is Minkowski at infinity, which in spherical
coordinates is given by
−c2 dt2 + dr2 + r2 (dθ2 + sin2 θdφ2 ),
where c denotes the speed of light and t is the time coordinate measured
by a stationary clock located infinitely far from the massive body. Hence
F (r) → −c2 and G(r) → +1 as r → ∞. (4.44)
By applying the Ricci flat condition on the spacetime and using (4.44),
one may determine functions F and G as
rs
F = −c2 λ, G = λ−1 , λ(r) = 1 − ,
r
where rs is a constant, known as the Schwarzschild radius. It is known that
rs is related to the mass M of the star by rs = 2GM/c2 , where G is the
gravitational constant. Consequently, the Schwarzschild metric in term of
Schwarzschild’s chart is the following warped product metric
rs 2 rs −1 2
g = −c2 1 − dt + 1 − dr + r2 (dθ2 + sin2 θdφ2 ). (4.45)
r r
Conversely, it was shown in [Jebsen (1921); Birkhoff (1923)] that any
local spherically symmetric solution of vacuum Einstein field equations is a
part of Schwarzschild’s metric. It follows from (4.45) that Schwarzschild’s
spacetime has singularities at r = 0 and r = rs .
Since the Schwarzschild metric is only expected to be valid for radii
larger than the radius R of the gravitating body, there is no problem as
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 98
long as R > rs , which is always the case for ordinary stars. For example,
the radius of the Sun is approximately 700,000 km, while the Schwarzschild
radius rs of the Sun is only 3 km.
On the other hand, any physical object whose radius R is less than or
equal to the Schwarzschild radius will undergo gravitational collapse. For
this reason any object whose radius is smaller than its Schwarzschild radius
is often called a black hole and the Schwarzschild radius is often called the
radius of a black hole.
The singularity at r = rs divides the Schwarzschild spacetime into two
disconnected portions. The exterior one with r > rs is the one that is
related to the gravitational field of the star (cf. [Hawking and Ellis (1973)]).
On the other hand, the interior one with 0 ≤ r < rs which contains the
singularity at r = 0, is completely separated from the outer one by the
singularity at r = rs .
The Schwarzschild coordinates give no physical connection between the
exterior and interior portions. Therefore they can be viewed as separate
solutions of Einstein’s field equations. The hypersurface r = rs gives rise
to the event horizon of the black hole. It represents the point past which
light can no longer escape the gravitational field.
The case r = 0 is different. If one asks that the solution be valid for all
r, then it runs into a gravitational singularity at the origin. At this point
the spacetime is no longer well-defined.
The existence of the singularity can be verified by noting that the in-
variant given by the square norm ||R||2 of the Riemann curvature tensor
R is infinite. (||R||2 is also known as the Kretschmann invariant in general
relativity.) A more recent better understanding of general relativity led to
the realization that such singularities were a generic feature of the theory
and not just an exotic special case (see, e.g. [Henry (2000); Cherubini et
al. (2002)]).
Remark 4.5. The Reissner-Nordström spacetime, discovered soon after
Schwarzschild’s article, is a static solution of the Einstein field equations
corresponding to the gravitational field of a charged, non-rotating, spheri-
cally symmetric body with the warped product metric
−1
2 rs e2 2 rs e2
g = −c 1 − + 2 dt + 1 − + 2 dr2 + r2 (dθ2 + sin2 θdφ2 ),
r r r r
where e is the electric charge. If e = 0, it reduces to the Schwarzschild
solution. Therefore Reissner-Nordström’s solution can be regarded as a
generalization of Schwarzschild’s solution of Einstein’s field equations.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 99
Chapter 5
99
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 100
5.3 Submersions
Escobales did not know whether case (2) in Theorem 5.3 was empty or
not. By applying homotopy theory it was shown in [Ucci (1983)] that case
(2) cannot occur.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 107
Remark 5.2. It was proved in [Bejancu and Duggal (1993)] that every
connected Sasakian real hypersurface of E2n+2
2s is either an open part of
n+1 2n+1
S2s or an open part of H2s−1 (−1).
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 111
Case (i): R2n+1 (-3). Consider the canonical fibration π : R2n+1 (−3) → Cn
defined by
1
π(x1 , . . . , xn , y1 , . . . , yn , z) = (y1 , . . . , yn , x1 , . . . , xn ). (5.17)
2
Let i : N → Cn be a Lagrangian isometric immersion of a Riemannian
n-manifold N into Cn . Then there is a covering map τ : N̂ → N and a
horizontal immersion î : N̂ → R2n+1 (−3) such that i ◦ τ = π ◦ î. Therefore
each Lagrangian immersion can be lifted (locally) (or globally if N is simply-
connected) to a Legendre immersion of the same manifold.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 114
Chapter 6
115
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 116
Consider R̂ = R − Ro . Then
g(R̂(v, Jv)Jv, v) = 0 (6.7)
for v ∈ T M with g(v, v) 6= 0. Since {v ∈ Tp M : g(v, v) 6= 0} is dense in
Tp M , g(R̂(v, Jv)Jv, v) = 0 holds in general. Thus, by applying polariza-
tion, we find R̂ = 0.
The converse is trivial.
The following result shows that there exist no non-trivial concircular vector
fields on every pseudo-Kähler manifold of complex dimension > 1.
with α0 + · · · + αn = α.
The Segre embedding Shp : CP h (4)×CP p (4) → CP h+p+hp (4) is defined
by (cf. [Segre (1891)])
Shp (z0 , . . . , zh , w0 , . . . , wp )
= (z0 w0 , . . . , z0 wp , . . . , zh w0 , . . . , zh wp ) (6.35)
= zj wt 0≤j≤h,0≤t≤p ,
Proof. ˜ = 0 that
It follows from (2.5), (2.13), (6.38), (6.39) and ∇J
˜ X (JY ) − J ∇
0 =∇ ˜ XY
= ∇X (P Y ) + σ(X, P Y ) − AF Y X + DX (F Y ) (6.45)
− P (∇X Y ) − F (∇X Y ) − tσ(X, Y ) − f σ(X, Y )
for X, Y ∈ T N .
The tangential components of (6.45) yield
(∇X P )Y = AF Y X + tσ(X, Y ).
(a) ∇F = 0;
(b) F ∇X Y = DX F Y for X, Y ∈ T N ;
(c) σ(X, P Y ) = f σ(X, Y ) for X, Y ∈ T N .
for X, Y, Z, W tangent to N .
Due to Lemma 6.2(2), the equation (6.56) of Ricci is nothing but the
equation (6.54) of Gauss. Consequenlty, after applying a result obtained
in [Eschenburg and Tribuzy (1993)], we obtain the following existence and
uniqueness theorems for Lagrangian submanifolds in (indefinite) complex
space forms.
Proof. Lemma 6.3 implies that every complex extensor of the unit hy-
persphere centered at the origin in En is a Lagrangian submanifold in Cn .
Now we prove that every complex extensor of the unit hypersphere of
En is an H-umbilical submanifold of Cn as follows.
For a unit speed curve F : I → Cn , we may put F ′ (s) = e iζ(s) for some
real-valued function f on I. Thus F takes the following form:
Z s
F (s) = e iζ(t) dt (6.63)
a
for some real number a. Let {x2 , . . . , xn } be a local coordinate system on
S n−1 . Then {s, x2 , . . . , xn } is a local coordinate chart on I × S n−1 .
Because ι is the unit hypersphere, (6.61) and (6.63) imply
φs = e iζ(s) ⊗ ι, Y φ = F ⊗ Y, (6.64)
and
φss = iζ ′ (s)e iζ(s) ⊗ ι, Y φs = e iζ(s) ⊗ Y,
(6.65)
Y Zφ = F ⊗ ∇Y Z − hY, Zi (F ⊗ ι),
where Y, Z are vectors fields tangent to the second component of I × S n−1 .
Since ι is the unit hypersphere in En , (6.64) implies that e1 = ∂/∂s is
a unit vector field tangent to the first component of I × S n−1 ; moreover,
for each Y tangent to the second component of I × S n−1 , φs and Y φ are
orthogonal. Thus, by applying (6.64) and (6.65) we may conclude that
the second fundamental form of the complex extensor satisfies (6.59) with
respect to a suitable orthonormal local frame field. Moreover, we have
iζ
′ e , iF
λ = ζ (s), µ = , (6.66)
hhF, F ii
where hh , ii denotes the canonical scalar product of the complex plane.
Therefore φ = F ⊗ ι is an H-umbilical submanifold unless φ is totally
geodesic, which occurs only when F (s) = (s + a)c for some a ∈ R and some
unit complex number c ∈ C.
Remark 6.6. It follows from (6.64) that the complex extensor F ⊗ ι of
ι : S n−1 → En via the curve F given by (6.63) is the warped product
I ×f S n−1 with warping function f = |F |.
Remark 6.7. It was proved in [Castro and Urbano (2004)] that Lagrangian
pseudo-spheres are the only branched Lagrangian immersions of an ordinary
sphere in C2 with constant length mean curvature vector. For an extension
of these results in 2-dimensional complex space forms, see [Li et al. (2008)].
from (6.73) and (6.74) that we may choose the initial conditions:
Ls (0, . . . , 0) = (1, 0, . . . , 0),
1
Lu2 (0, . . . , 0) = 0, ,...,0 ,
2bi
(6.90)
......
1
Lun (0, . . . , 0) = 0, . . . , 0, .
2bi
By using (6.87) and (6.88) we obtain
n n−1
e2ibs + 1 Y Y
L= cos uj , sin u2 , sin u3 cos u2 , · · · , sin un cos uj .
2ib j=2 j=2
∂ f′ ∂
∇∂ = ,
∂s ∂u f ∂uk
k
∂ ∂
∇∂/∂u2 = −f f ′ ,
∂u2 ∂s
∂ ∂ (6.106)
∇ ∂ = − tan ui , 2 ≤ i < j,
∂ui ∂u ∂uj
j
Y j−1
∂ ∂
∇ ∂ = −f f ′ cos2 uℓ
∂uj ∂uj ∂s
ℓ=2
j−1 j−1
!
X sin 2uk Y ∂
+ cos2 ul , j > 2.
2 ∂uk
k=2 l=k+1
n−1 n
! (6.116)
Y Y
· · · + cn−1 sin un cos uj + cn cos uj
j=2 j=2
n
for some constant vectors c1 , . . . , cn ∈ C . Since N is Lagrangian, in view
of (6.105), we may choose the initial conditions (6.90). Then, by using
(6.116)we obtain
Z s R
s
L = α+ ei λ(t)dt ds
Yn n−1
Y (6.117)
× cos uj , sin u2 , sin u3 cos u2 , · · · , sin un
cos uj .
j=2 j=2
Since U is dense in N , (6.117) and continuity imply that, up to rigid motions
of Cn , N is the complex extensor of the unit hypersphere in En .
Remark 6.8. Flat H-umbilical submanifolds in the complex Euclidean
spaces have been completely determined in [Chen (1999c)]. Furthermore,
H-umbilical submanifolds in complex projective spaces and also in complex
hyperbolic spaces have been classified in [Chen (1999a)].
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 141
Chapter 7
Slant Submanifolds
Let E2m be the Euclidean 2m-space with the Euclidean metric. An almost
complex structure J on E2m is called compatible if (E2m , J) is complex
analytically isometric to the complex number space Cm .
We denote by J0 and J1− (when m is even) the compatible almost com-
plex structures on E2m defined respectively by
J0 (a1 , . . . , am , b1 , . . . , bm ) (7.2)
= (−b1 , . . . , −bm , a1 , . . . , am )
−
J1 (a1 , . . . , am , b1 , . . . , bm ) (7.3)
= (−a2 , a1 , . . . , −am , am−1 , b2 , −b1 , . . . , bm , −bm−1 ).
141
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 142
Example 7.5. Let k be any positive number and (g(s), h(s)) a unit speed
plane curve. Then
x(u, s) = (−ks sin u, g(s), ks cos u, h(s))
√
defines a non-minimal, proper slant flat surface with slant angle k/ 1 + k 2 .
Example 7.6. For any nonzero real numbers p and q, we consider the
following immersion form R × (0, ∞) into C2 defined by
x(u, v) = (pv sin u, pv cos u, v sin qu, v cos qu).
Then the immersion x gives rise to a complete flat slant surface in C2 .
April 18, 2017 12:18 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 143
Example 7.7. For any real number k ∈ (0, 1), the map
Remark 7.2. From Remark 6.5 we see that Corollary 7.4 is false if the
totally umbilical slant surface is non-proper.
Theorem 7.6. [Sahin (2009b)] Every totally umbilical proper slant sub-
manifold in a Kähler manifold is totally geodesic.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 150
−AJH X + DX JH = −P AH X − F AH X + tDX H + f DX H.
Hence we get
hDX JH, F Y i = − hF AH X, F Y i + hf DX H, F Y i . (7.20)
Since f DX H ∈ ν and hF Y, JHi = 0, we derive from (7.20) that
˜ X F Y, JH i = sin2 θ hAH X, Y i = sin2 θ hX, Y i , hH, Hi .
h∇ (7.21)
Combining (7.19) and (7.21) gives cos2 θ hX, Y i hH, Hi = 0. Consequently,
H = 0 and N is totally geodesic.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 151
Similarly, we have Jˆ2 ((0, X)) = −(0, X). Thus, Jˆ2 = −I. By using (7.28),
ˆ JY
it is easy to verify that h JX, ˆ i = hX, Y i . Thus, (J,
ˆ h , i) is a Hermitian
structure on T N ⊕ T N .
Now, we define A, σ and D by
AY ∗ X = csc θ{(∇X P )Y − α(X, Y )}, (7.29)
∗
σ(X, Y ) = −(csc θ)α (X, Y ), (7.30)
∗ ∗ 2 ∗ ∗
DX Y = (∇X Y ) + csc θ{P α (X, Y ) + α (X, P Y )}, (7.31)
for vector fields X, Y tangent to N .
It is easy to verify that each AY ∗ is an endomorphism on T N , σ is a
(T N )∗ -valued symmetric bilinear form on T N , and D is a metric connection
of the vector bundle (T N )∗ over N .
Let ∇ ˆ denote the canonical connection on T N ⊕ T N induced from the
Levi-Civita connection on T N . Then, from (7.28)-(7.31), we have
(∇ ˆ = (∇
ˆ X J)Y ˆ ∗ = 0,
ˆ X J)Y (7.32)
for vector fields X, Y ∈ T N .
Let RD denote the curvature tensor associated with the connection D
on (T X)∗ , i.e.,
RD (X, Y )Z ∗ = DX DY Z ∗ − DY DX Z ∗ − D[X,Y ] Z ∗ , (7.33)
for X, Y, Z ∈ T N . Then, by (7.22), (7.26), (7.31), (7.33) and a simple
computation, we may obtain
RD (X, Y )Z ∗ = (R(X, Y )Z)∗
+ {cP [hY, P Zi X − hX, P Zi Y − 2 hX, P Y i Z]
+ c[ Y, P 2 Z X − X, P 2 Z Y − 2 hX, P Y i P Z] (7.34)
2
+ csc θ[(∇X P )α(Y, Z) − (∇Y P )α(X, Z)
− α(X, (∇Y P )Z) + α(Y, (∇X P )Z) ] }∗ .
Also, (7.29) yields
sin2 θ h[AZ ∗ , AW ∗ ]X, Y i = h(∇Y P )Z, (∇X P )W i
− h(∇X P )Z, (∇Y P )W i + h(∇X P )Z, α(Y, W )i
+ h(∇Y P )W, α(X, Z)i − h(∇Y P )Z, α(X, W )i (7.35)
− h(∇X P )W, α(Y, Z)i + hα(X, W ), α(Y, Z)i
− hα(X, Z), α(Y, W )i .
From (7.23) we have
hα(Y, Z), P W i + hP α(Y, Z), W i = 0. (7.36)
April 18, 2017 12:18 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 153
(1) θ = π2 ,
(2) there exists a point p of M such that P1 = P2 ,
6 0,
(3) c =
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 154
Taking the inner product of (7.43) with a vector W and using (7.42),
we deduce that
− hα(Y, Z), α(X, P W )i + hα(X, Z), α(Y, P W )i
+ hα(X, W ), P α(Y, Z)i − hα(Y, W ), P α(X, Z)i
(7.44)
+ c sin4 θ hX, P Zi hY, W i − hY, P Zi hX, W i
+ 2 hX, P Y i hZ, W i = 0.
If α vanishes identically at a point, then a contradiction follows from
(7.44) since c 6= 0. Now, we take a fixed point p ∈ N and look at the
function f defined on the set of all unit tangent vectors U Np at p by
f (v) = hα(v, v), vi .
Since U NP is compact there exists a vector u such that f attains an absolute
maximum at the vector u. Let w be a unit vector orthogonal to u. Then
the function
f (t) = f (g(t)), g(t) = (cos t)u + (sin t)w,
satisfies f (0) = 0 and f ′′ (0) ≤ 0. The first condition implies that
′
where S2k is the permutation group of order 2k and sign(τ ) is the signature
of raw permutation τ .
Proof. Put
P = πV ◦ (J |V ) : V → V, (7.60)
⊥ ⊥
P = J |V − P : V → V , (7.61)
2
Q = P : V → V. (7.62)
Then we have
J |V = P + P ⊥ . (7.63)
April 26, 2017 14:23 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 160
via the natural identification of G(2k, 2m) with D1 (2k, 2m), then we derive
Ω0 (ea , eb ) = −δa∗ b cos θ for a < b, (7.72)
∗ ∗
where (2i) = 2i − 1 and (2i − 1) = 2i for i = 1, . . . , k.
By (7.71), Lemma 7.4, and (7.72), we find
(2k)! Ωk0 (V ) = (2k)! Ω0 (e1 ∧ . . . ∧ e2k )
X
= sign(τ ) Ω0 (eτ (1) , eτ (2) ) · · · Ω0 (eτ (2k−1) , eτ (2k) )
τ ∈S2k
2k
X 12···(2k)
= δa1 ···a2k Ω0 (ea1 , ea2 ) · · · Ω0 (ea2k−1 , ea2k )
a1 ,...,a2k =1
2k
X 12······(2k) (7.73)
= δa1 a∗ ···ak a∗ Ω0 (ea1 , ea∗1 ) · · · Ω0 (eak , ea∗k )
1 k
a1 ,... ak =1
X X 12······(2k)
= 2k ··· δa1 a∗ ···ak a∗ Ω0 (ea1 , ea∗1 ) · · · Ω0 (eak , e∗ak )
1 k
a1 <a∗
1 ak <a∗
k
X X 12·······(2k)
= 2k (− cos θ)k ··· δa1 a∗ ···ak a∗
1 k
a1 <a∗
1 ak <a∗
k
for any fixed 2k-vector ζ ∈ ∧2k (E2m ), where dV is the volume element of
N with respect to the metric induced from the immersion f .
Let ζ = ζ̂0 , where ζ̂0 is defined by (7.66). Then Lemma 7.6 and (7.75)
imply that
ck vol(N ) cosk θ = 0.
But this contradicts to the assumption cos θ 6= 0. Therefore we must have
θ = π2 and φ is a totally real immersion.
Theorem 7.9 implies the following non-existence result from [Chen and
Tazawa (1991)].
We know from section 7.1 that there exist ample examples of proper slant
minimal surfaces in C2 . It is also known that there exist many examples
of proper slant surfaces in complex projective plane CP 2 and in complex
hyperbolic plane CH 2 (cf. [Chen (1998b, 1999b)]).
Now, we present the following non-minimality theorem for proper slant
surfaces in CP 2 and in CH 2 .
Theorem 7.10. [Chen and Tazawa (2000)] Every proper slant surfaces in
a complex space form M̃ 2 (4c) with c 6= 0 is non-minimal.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 163
Thus we find
2j 2i
(cot θ)(ω2i−1 − ω2j−1 )(X) = Ae(2i−1)∗ e2j−1 − Ae(2j−1)∗ e2i−1 , X .
Hence, by using ωir (X) = hAer ei , Xi, we get (7.99).
Similarly, after taking the derivatives of the following equations:
hJe2i−1 , e2j i = (cos θ)δij , hJe2i , e2j i = 0,
Je(2i−1)∗ , e(2j−1)∗ = 0, Je(2i)∗ , e(2j)∗ = 0,
Je2i−1 , e(2j−1)∗ = (sin θ)δij ,
Je2i−1 , e(2j)∗ = 0,
Je2i , e(2j)∗ = (sin θ)δij , Je(2i)∗ , e(2j)∗ = 0,
we obtain (7.100)-(7.107), respectively.
Lemma 7.10. Let N be a proper slant submanifold of Cn with dim N =
n = 2k. Then with respect to a slant frame we have
(2j)∗ (2j−1)∗ (2i)∗ (2i−1)∗
ω2i + ω2i−1 = ω2j + ω2j−1 , (7.109)
∗ ∗
(2j) (2j−1)
2j 2j−1
ω(2i)∗ − ω(2i−1)∗ = ω2i − ω2i−1 , (7.110)
2i−1 (2i−1)∗ 2j−1 (2j−1)∗
ω2j − ω(2j)∗ = ω2i − ω(2i)∗ (7.111)
for any i, j = 1, . . . , k.
Proof. Formula (7.109) follows from (7.99) and (7.101). (7.110) follows
from (7.104) and (7.106). And (7.111) follows from (7.101) and (7.103).
Now we return to the proof of Theorem 7.12. From the definition we
have
X2k
∗
Θ= ωℓℓ . (7.112)
ℓ=1
Thus from the structure equations we have
Xk Xk
2j (2i)∗ 2j−1 (2i)∗
−dΘ = ω2i ∧ ω2j + ω2i ∧ ω2j−1
i,j=1 i,j=1
k
X k
X
(2j)∗ (2i)∗ (2j−1)∗ (2i)∗
+ ω2i ∧ ω(2j)∗ + ω2i ∧ ω(2j−1)∗
i,j=1 i,j=1
(7.113)
k
X k
X
2j (2i−1)∗ 2j−1 (2i−1)∗
+ ω2i−1 ∧ ω2j + ω2i−1 ∧ ω2j−1
i,j=1 i,j=1
k
X k
X
(2j)∗ (2i−1)∗ (2j−1)∗ (2i−1)∗
+ ω2i−1 ∧ ω(2j)∗ + ω2i−1 ∧ ω(2j−1)∗ .
i,j=1 i,j=1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 168
k
X k
X
2j (2i−1)∗ (2j)∗ (2i−1)∗
+ ω2i−1 ∧ ω2j + ω2i−1 ∧ ω(2j)∗
i,j=1 i,j=1
k
X k
X
2j−1 (2i)∗ 2i−1 (2j−1)∗
= ω2i ∧ ω2j−1 − ω2j ∧ ω2i
i,j=1 i,j=1
k
X k
X
(2j−1)∗ (2i−1)∗ (2j)∗ (2i−1)∗
− ω2i ∧ ω(2j)∗ + ω2i−1 ∧ ω(2j)∗
i,j=1 i,j=1
k
X
2i−1 (2i−1)∗ (2j)∗ (2j−1)∗
= (ω2j − ω(2j)∗ ) ∧ (ω2i−1 − ω2i ).
i,j=1
2i−1 (2i−1)∗
Since ω2j − ω(2j)∗ is symmetric in i, j according to Lemma 7.10 and
(2j)∗ (2j−1)∗
ω2i−1 − ω2i is skew-symmetric in i, j by (7.100) in Lemma 7.9, we find
k
X k
X
2j−1 (2i)∗ (2j−1)∗ (2i)∗
ω2i ∧ ω2j−1 + ω2i ∧ ω(2j−1)∗
i,j=1 i,j=1
(7.115)
k
X k
X
2j (2i−1)∗ (2j)∗ (2i−1)∗
+ ω2i−1 ∧ ω2j + ω2i−1 ∧ ω(2j)∗ = 0.
i,j=1 i,j=1
Then with respect to the local orthonormal frame e1 , e2 , e1∗ , e2∗ on the
open subset of S 2 on which θ 6≡ π2 and ϕ 6≡ 0 (mod π), we have
∗ ∗ sin θ sin 2ϕdθ − 2 cos θdϕ
Θ = ω11 + ω22 = .
2 − 2 sin2 θ cos2 ϕ
Now it is direct to verify that dΘ 6= 0.
Remark 7.6. Pointwise slant submanifolds have been defined and studied
in [Etayo (1998)] under the name of quasi-slant submanifolds.
Example 7.13. Let E4n = (R4n , g0 ) denote the Euclidean 4n-space en-
dowed with the standard Euclidean metric g0 and let {J0 , J1 } be a pair of
almost complex structures on E4n satisfying J0 J1 = −J1 J0 .
Assume that J0 , J1 are orthogonal almost complex structures, i.e., they
are compatible with the Euclidean metric g0 ; thus
g0 (Ji X, Ji Y ) = g0 (X, Y ), i = 0, 1,
for X, Y ∈ T (E ). Let us denote (R4n , J0 , g0 ) by C2n
4n
0 .
For any real-valued function f : E4n → R, we define an almost complex
structure Jf on E4n by
Jf = (cos f )J0 + (sin f )J1 . (7.124)
Then C2n 4n
= (R , Jf , g0 ) is an almost Hermitian manifold.
f
For any given pointwise slant submanifold (resp. slant submanifold)
M in (E4n , J0 , g0 ), M is also a pointwise slant submanifold (resp. slant
submanifold) of C2n f = (R4n , Jf , g0 ). In particular, if M is a complex
2n
submanifold of C0 , then M is a pointwise slant minimal submanifold in
C2n
f whose slant function θ is the restriction of f on M , i.e., θ = f |M .
Hence φ : N → (M̃ , J, e2f g̃) is a pointwise slant immersion with the same
slant function as φ : N → (M̃ , J, g̃).
The following conformal property of slant immersions is an immediate
consequence of Proposition 7.5.
Corollary 7.9. For each integer n ≥ 1, there exist infinitely many 2n-
dimensional totally umbilical proper slant submanifolds in locally conformal
Kähler 2n-manifolds.
Proof. Follows from Theorem 7.16 and the fact that the shape operator
A vanishes identically for totally geodesic submanifolds.
Remark 7.8. Corollary 7.10 have been proved in [Etayo (1998)] under an
additional assumption that the totally geodesic pointwise slant submanifold
is complete.
Remark 7.9. Proposition 7.6(2) shows that each totally umbilical surface
of the Euclidean complex plane is a non-slant, pointwise slant surface.
Remark 7.10. It follows from Corollary 7.9 that Proposition 7.6(2) is false
if the Kähler surface were replaced by a locally conformal Kähler surface.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 177
Chapter 8
179
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 180
8.2 Integrability
Remark 8.1. Theorem 8.2 is false if the ambient space is a complex space
form with arbitrary codimension.
(2) If N is not purely real, then H ∈ F H⊥ and both F H⊥ and ν are parallel
in the normal bundle.
P H⊥ = H⊥ , T ⊥ N = F H⊥ ⊕ ν, t(T ⊥ N ) = t(F H⊥ ) = H⊥ .
Theorem 8.3. [Chen and Verheyen (1983)] Every totally umbilical sub-
manifold N of a positively (or negatively) curved Kähler manifold M̃ is
either a totally geodesic complex submanifold or a purely real submanifold.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 189
Thus we have
2h
X p
X
||σ||2 = 4hp + |σ(Xi , Xj )|2 + |σ(Zα , Zβ )|2 , (8.43)
i,j=1 α,β=1
Next, let us assume that m = h + p + hp. Then for any unit vector
X, Y ∈ H and Z ∈ H⊥ the equation of Gauss and the curvature tensor of
CP m (4c) imply that
hσ(X, Y ), σ(JX, Z)i = hσ(X, Z), σ(JX, Y )i . (8.48)
On the other hand, since σ(H, H⊥ ) ⊥ F H⊥ , Lemmas 8.2 and 8.5 give
hσ(X, Z), σ(JX, Y )i = hσ(X, Z), Jσ(X, Y )i
= − hJσ(X, Z), σ(X, Y )i
(8.49)
= − hJσ(X, Z), Jf σ(X, Y )i
= − hσ(JX, Z), σ(X, Y )i .
By combining (8.48) and (8.49) we find
hσ(X, Y ), σ(JX, Z)i = hσ(X, Z), σ(JX, Y )i = 0. (8.50)
Hence, after applying linearity we derive from (8.50) that
hσ(X, Z), σ(Y, W )i = hσ(Y, Z), σ(X, W )i = 0 (8.51)
for X, Y, W ∈ H and Z ∈ H⊥ . Since m = h + p + hp and (8.48) spans ν, we
obtain from (8.51) that σ(H, H) ⊂ F H⊥ . After combining this with (8.36)
we find σ(H, H) = {0}. Thus N T is immersed as a totally geodesic complex
submanifold in CP m (4). Consequently, M T is an open submanifold of a
complex projective h-space CP h (4).
By applying Theorem 8.7 we may obtain the following result from [Chen
(1981b,e)].
Chapter 9
195
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 196
According to Lemma 9.1(a) the first two terms belong to D. Also according
to Lemma 9.1(b) the last two terms vanish. Hence the complex subbundle
D defined by (9.1) is an involutive distribution.
Remark 9.1. It was shown in [Blair and Chen (1979)] that Theorem 9.2
is false for general CR-submanifolds in Hermitian manifolds. In fact, Blair
and Chen constructed examples of CR-submanifolds of a Hermitian mani-
fold on which their totally real distributions are not integrable.
Remark 9.2. Corollary 8.1, Theorem 8.5 and Theorem 8.7 imply the fol-
lowing.
(a) There do not exist proper CR-products in any complex hyperbolic
space.
(b) Every CR-product N T × N ⊥ in Cm is locally the direct product of a
complex submanifold and a totally real submanifold of some complex
Euclidean spaces.
(c) For every CR-product N T × N ⊥ in CP m , we have m ≥ h + p + hp,
where h = dimC N T and p = dim N ⊥ .
Remark 9.3. The cohomology class given in (9.13) is also known in the
literature as the Chen class (see, e.g. [Dragomir and Ornea (1998)]).
From this we conclude that the canonical class c(N ) and the fundamental
class [Ω] of N are related by
[Ω]h = (−1)h (h!)c(N ). (9.19)
Now, if H is integrable and H⊥ is minimal in N , then Theorem 9.6 and
(9.19) imply H 2k (N ; R) 6= 0 for k = 1, . . . , h.
The next two results were proved in [Blair and Chen (1979)].
Chapter 10
207
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 208
≥ 0,
with equality holding if and only if (10.2) holds.
An immediate consequence of Lemma 10.1 is the following.
a1 + a2 = a3 = · · · = an .
The following result provides the first solution to the following basic ques-
tion proposed in [Chen (2002a)].
where
n+1 n+1
ā1 = σ11 , ā2 = σ22 + · · · + σnn+1
1 n1
,
ā3 = σnn+1
1 +1n1 +1
n+1
+ · · · + σnn .
Applying Corollary 10.1 to (10.11) yields
X X
n+1 n+1 n+1 n+1
σjj σkk + σss σtt
1≤j<k≤n1 n1 +1≤s<t≤n
X m n (10.13)
δ n+1 2 1 X X r 2
≥ + (σαβ ) + (σαβ ) ,
2 2 r=n+2
1≤α<β≤n α,β=1
1 X X r 2 1 X 2
m m X
r δ
− σjj − σtt −
2 r=n+2 2 r=n+2 2
1≤j≤n1 n1 +1≤t≤n
2
n(n − 1) δ n 2
≤ τ− c + n1 n2 c − = H + n1 n2 c,
2 2 4
which proves inequality (10.4). From (10.14) and (10.16) we see that the
equality sign of (10.4) holds if and only if we have
r
σjt = 0, n + 1 ≤ r ≤ m, 1 ≤ j ≤ n1 ; n1 + 1 ≤ t ≤ n
r
σ11 + · · · + σnr 1 n1 (10.17)
= σnr 1 +1n1 +1 + ···+ r
σnn = 0, n + 2 ≤ r ≤ m.
Condition (10.17) shows that the second fundamental form σ of N1 ×f N2
in Rm (c) satisfies σ(D1 , D2 ) = {0}. Thus the immersion φ is mixed totally
geodesic. Hence, by using a result of [Nölker (1996)], we know that locally
there exists a warped product representation M1 ×ρ M2 of Rm (c) such that
φ : N1 ×f N2 → M1 ×ρ M2 = Rm (c) is a warped product immersion of
φ1 : N1 → M1 and φ2 : N2 → M2 ; so that we have
φ(x1 , x2 ) = (φ1 (x1 ), φ(x2 )) for x1 ∈ N1 , x2 ∈ N2 .
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 212
As before, let S n+1 (c) (c > 0) denote the hypersphere of radius c−1 in the
Euclidean space Rn+2 centered at the origin, i.e.,
n+1 n+2 2 2 1
S (c) = x ∈ R : x1 + · · · + xn+2 = ,
c
P
where Rn+2 equips with the standard Euclidean metric ds2 = n+2 2
i=1 dxi .
n+1
Then S (c) has constant positive sectional curvature c.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 216
Suppose that Rn+2 is the Lorentzian space equipped with the Lorentzian
metric ds2 = dx21 + · · · + dx2n+1 − dx2n+2 . Let us consider the hypersurface
H n+1 (c) (c < 0) defined by
1
H n+1 (c) = x ∈ Rn+2 : x21 + · · · + x2n+1 − x2n+2 = , xn+2 > 0 .
c
Then H n+1 (c) with the induced Riemannian metric has constant negative
curvature c.
In this section, let Rn+1 (c) denote a complete simply-connected Rie-
mannian manifold with constant sectional curvature c. Thus Rn+1 (c) will
be the Euclidean space En+1 for c = 0; the hypersphere S n+1 (c) for c > 0;
and the hyperbolic space H n+1 (c) for c < 0.
Besides Theorems 10.1 and 10.2, there is another general optimal inequality
for a warped product N1 ×f N2 in a real space form Rm (c).
∆f = cf. (10.43)
Now, we can easily verify from (10.41), (10.42) and (10.43) that the rotation
hypersurface I ×f M n−1 (ǫ) in Rn+1 (c) satisfies the equality case of (10.25)
identically.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 222
where ||σ||2 is the squared norm of the second fundamental form σ and
Pn 2
||P ||2 = i,j=1 hei , P ej i , where P is defined by defined by (6.38).
If we put
n2 2
η = 2τ − H + n(n − 1) + 3||P ||2 , (10.48)
2
then we obtain from (10.47) and (10.48) that
n2 H 2 = 2η + 2||σ||2 . (10.49)
If we choose a local field of orthonormal frame en+1 , . . . , e2m of the normal
bundle so that en+1 is in the direction of the mean curvature vector, then
(10.49) becomes
!2
Xn n Xn
X X2m Xn
o
n+1 n+1 2 n+1 2 r 2
σii =2 η+ σii + σij + σij ,
i=1 i=1 i6=j r=n+2 i,j=1
X 2m
X n
X
n+1
2 r
2
≥η+2 σij + σij
(10.52)
1≤i<j≤n r=n+2 i,j=1
X X
n+1 n+1 n+1 n+1
−2 σαα σββ − 2 σss σtt
2≤α<β≤n1 n1 +1≤s<t≤n
which is nothing but
X X η
n+1 n+1 n+1 n+1
σαα σββ + σss σtt ≥
2
1≤α<β≤n1 n1 +1≤s<t≤n
X 2m n (10.53)
n+1
2 1 X X r 2
+ σjk + σjk .
2 r=n+2
1≤j<k≤n j,k=1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 224
X 2m
X X (10.55)
2 r r r 2
+ 3 hP es , et i − σαα σββ − (σαβ )
s<t r=n+1 α<β
2m
X X
r r r 2
− σss σtt − (σst ) .
r=n+1 s<t
1 X X r 2 1 X X r 2
2m 2m
− σαα − σ
2 r=n+2 α 2 r=n+2 t tt
X 2
X 2 η
+ 3 hP eα , eβ i + 3 hP es , et i − ,
s<t
2
α<β
where the equality case of the inequality holds if and only if condition
(10.54) is satisfied. From (10.56) we find
n2 ∆f n(n − 1) η
≤τ + − n1 n2 −
f 2 2
X 2
X 2
(10.57)
+ 3 hP eα , eβ i + 3 hP es , et i
α<β s<t
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 225
for r = n + 2, . . . , 2m.
Finally, by applying (10.48) and (10.57), we derive that
n2 ∆f n2 2 X 2 n2 2
= H − n1 n2 − 3 hP eα , et i ≤ H − n1 n2 (10.59)
f 4 α,t
4
(a) n1 = n2 = 1,
(b) f is an eigenfunction of the Laplacian of N1 with eigenvalue 4, and
(c) φ is totally geodesic and holomorphic immersion.
The following examples show that Theorem 10.6, Corollary 10.8 and
Corollary 10.9 are sharp.
Example 10.6. Consider the same warped product I ×(cos 2s)/2 S 1 (1) as in
Example 10.5. Let φ : CP 1 (4) → CP m (4) be a standard totally geodesic
embedding of CP 1 (4) into CP m (4). Then the restriction of φ to I ×(cos 2s)/2
S 1 (1) gives rise to a minimal isometric immersion of I ×(cos 2s)/2 S 1 (1) into
CP m (4) which satisfies the equality case of (10.61) on I ×(cos 2s)/2 S 1 (1)
identically.
Example 10.7. Let g1 be the standard metric on the unit (n − 1)-sphere
S n−1 (1) and let N1 ×f N2 be the warped product with N1 = (− π2 , π2 ), N2 =
S n−1 (1) and f = cos s. Then the warping function of this warped product
satisfies ∆f = n1 f identically.
Moreover, it is direct to verify that this warped product is isometric to
a dense open subset of S n (1). Let
projection totally geodesic
φ : S n (1) −−−−−−→ RP n (1) −−−−−−−−−−→ CP n (4)
2:1 totally real
Chapter 11
229
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 230
as a CR-warped product.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 235
We put
Ch(α) = (z1 , . . . , zh ) ∈ Ch : |z1 |2 + · · · + |zα |2 6= 0 . (11.26)
If z : N T = Ch∗ ֒→ Ch and w : S p (1) ֒→ Ep+1 are the inclusion maps, then
(11.25) is the map
(α)
Shp : Ch(α) ×f S p (1) → Ch+αp (11.27)
defined by
(α)
Shp (z, w) = w1 z1 , . . . , wp+1 z1 , . . . , (11.28)
w1 zα , . . . , wp+1 zα , zα+1 , . . . , zh
Pp+1
for z = (z1 , . . . , zh ) ∈ Ch , w = (w1 , . . . , wp+1 ) ∈ S p (1) with j=1 wj2 = 1.
The warping function of the CR-warped product Chα ×f S p (1) is
p
f = |z1 |2 + · · · + |zα |2 .
(α)
Definition 11.3. The embedding Shp defined by (11.28) is called a partial
Segre embedding. Also a CR-warped product defined via (11.25) is called a
partial Segre CR-warped product.
h
Remark 11.2. The map ψhp = Shp defined by
⊥
σ (Z, W ), X = h∇W Z, Xi . (11.34)
Combining (11.33) and (11.34) yields
σ ⊥ (Z, W ) = − hZ, W i ∇ ln f. (11.35)
Now, assume that the equality case of (11.30) holds identically. Then
we obtain from (11.31) that
σ(H, H) = 0, σ(H⊥ , H⊥ ) = 0, σ(H, H⊥ ) ⊂ JH⊥ . (11.36)
Since N T is a totally geodesic submanifold in M , the first condition in
(11.36) implies that N T is totally geodesic in M̃ .
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 237
defined by
φ(p1 , p2 , p3 ) = (z(p1 ), z(p2 )) ⊗ j(p3 ),
(11.56)
∀p1 ∈ N T , p2 ∈ N1⊥ , p3 ∈ N2⊥ .
On (Cm ⊕ Cℓ ) ⊗ Eq we define a complex structure J by
√
J((B, E) ⊗ F ) = (iB, iE) ⊗ F, i = −1, (11.57)
for any B ∈ Cm , E ∈ Cℓ and F ∈ Eq . Then (Cm ⊕ Cℓ ) ⊗ Eq becomes a
complex Euclidean (m + ℓ)q-space C(m+ℓ)q .
We put N ⊥ = N1⊥ × N2⊥ . We denote by |z| the distance function from
the origin of Cm to the position of N T in Cm via z; and by |w| the distance
ℓ ⊥ ℓ
function from the origin
p of C to the position of N1 in C via w. We define
a function λ by λ = |z|2 + |w|2 . Then λ > 0 is a differentiable function
on N T × N ⊥ , which depends on both N T and N ⊥ = N1⊥ × N2⊥ .
Let M denote the twisted product N T ×λ N ⊥ with twisting function λ.
Clearly, M is not a warped product.
φ∗ (X) = Xφ = X ⊗ j, (11.60)
φ∗ (Z1 ) = Z1 φ = Z1 ⊗ j, (11.61)
φ∗ (Z2 ) = (z, w) ⊗ Z2 . (11.62)
Theorem 11.9. [Al-Solamy and Khan (2008b); Khan et al. (2008)] There
do not exist non-trivial warped product submanifolds N 0 ×f N T in a Kähler
manifold M̃ such that N T is a complex submanifold and N 0 is a proper
purely real submanifold of M̃ .
April 18, 2017 12:18 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 243
where
gMθ = 3(dv 2 + dw2 ) and gM ⊥ = du2 (11.77)
are the metric tensors of Mθ and M⊥ , respectively. Consequently,
M = Mθ ×f M⊥
is a warped
√ product hemi-slant submanifold of C3 with warping function
f = v 2 + w2 and whose bi-slant angle satisfies θ 6= 0, π2 . Hence M is
warped product hemi-slant.
Theorem 11.12. [Sahin (2006a)] There do not exist warped product sub-
manifolds of the form: N T ×f N θ in a Kähler manifold M̃ such that N T
is a complex submanifold and N θ is a proper slant submanifold of M̃ .
Pointwise bi-slant immersions are defined as follows (cf. [Chen and Uddin
(2016)]).
Definition 11.7. A submanifold N of an almost Hermitian manifold M̃
is called pointwise bi-slant if there exists a pair of orthogonal distributions
H1 and H2 of M such that the following three conditions hold.
(a) T N = H1 ⊕ H2 ;
(b) JH1 ⊥ H2 and JH2 ⊥ H1 ;
(c) Each distribution Hi is pointwise slant with slant function θi (i = 1, 2).
A pointwise bi-slant submanifold is a bi-slant submanifold if both slant
functions θ1 and θ2 are constant.
Analogous to CR-warped products, we define warped product pointwise
bi-slant submanifolds as follows.
Definition 11.8. A warped product N1 ×f N2 in an almost Hermitian
manifold M̃ is called a warped product pointwise bi-slant submanifold if
both factors N1 and N2 are pointwise slant submanifolds of M̃ .
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 253
Example 11.11 and Example 11.12 illustrate that there exist pointwise
bi-slant submanifolds and bi-slant submanifolds in Kähler manifolds.
On the other hand, for warped product bi-slant submanifolds we have
the following classification result.
Remark 11.9. Examples 11.7 and 11.9 and Examples 11.11 and 11.12
illustrate that both case (1) and case (2) of Theorem 11.14 do occur.
for X, Y ∈ Γ(T M̃ ).
The 1-form ω is called the Lee form and its metrically equivalent vector
field B̃ = ω ♯ is called Lee vector field, where ♯ means the rising of the indices
with respect to g, namely
g(X, B̃) = ω(X), (11.85)
for all X ∈ M̃ .
It is known that (M̃ , J, g) is globally conformal Kähler (respectively,
Kähler) if the Lee-form ω is exact (respectively ω = 0). It is also known
that the Levi-Civita connections ∇U of the local metrics gU glue up to a
globally defined torsion free linear connection ∇ˆ on M̃ given by
n o
∇ˆ XY = ∇˜ X Y − 1 ω(X)Y + ω(Y )X − g(X, Y )B̃ (11.86)
2
for X, Y ∈ X(M̃ ), where ∇ ˜ is the Levi-Civita connection of g. Moreover,
we have ∇g ˆ = ω ⊗ g and ∇J ˆ = 0. This connection ∇ ˆ is called the Weyl
connection of the locally conformal Kähler manifold M̃ .
Considering the anti-Lee form θ = ω ◦ J and the anti-Lee vector field
A = −JB, one can obtain a third equivalent definition in terms of the Levi-
Civita connection ∇ ˜ of the metric g (see [Dragomir and Ornea (1998)]).
Namely, (M̃ , J, g) is a locally conformal Kähler manifold if and only if the
following equation is satisfied:
˜ X J)Y = 1 θ(Y )X − ω(Y )JX − g(X, Y )A − Ω(X, Y )B̃
(∇ (11.87)
2
for any X, Y ∈ X(M̃ ).
A submanifold M of a locally conformal Kähler manifold (M̃ , J, g) is
called a CR-submanifold if there exists a differentiable distribution
H : x → Hx ⊂ Tx M
on M satisfying the following conditions:
Chapter 12
257
April 18, 2017 12:18 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 258
For statement (b), first we observe that the second equation in (11.36)
and the umbilicity of N ⊥ in N T ×f N ⊥ imply that N ⊥ is totally umbilical
in M̃ (4c). Hence, it follows from Gauss’ equation that N ⊥ is of constant
curvature, say ǫ ≥ c. It follows from (11.35) that ǫ = c occurs only when
the warping function is constant. Therefore we have statement (b).
⊥
Let RN denote the Riemann curvature tensor of N ⊥ . Then Proposition
3.2(5) gives
⊥
R(Z, W )V = RN (Z, W )V − |∇ ln f |2 {hW, V i Z − hZ, V i W }, (12.3)
for vectors Z, W, V ∈ T N ⊥ . Therefore, statement (c) follows from (11.36),
(12.3), Gauss’ equation, and statement (b).
The next result from [Chen (2001b)] shows that in order to study CR-
warped products satisfying the basic equality in a complex space form, it
is sufficient to consider anti-holomorphic submanifolds.
Ry
for some function η = η(x1 , x2 , y2 , . . . , xh , yh ), where µ = 2 1 ψdy1 .
From (12.13) we find
ηx1 µy 1
φx1 = , φy1 = , (12.14)
2(η + µ) 2(η + µ)
ηx x (η + µ) − ηx21
φx1 x1 = 1 1 . (12.15)
2(η + µ)2
By applying (12.7), (12.14) and (12.15) we derive
2(η + µ)ηx1 x1 = ηx21 + µ2y1 . (12.16)
Similarly, from (12.6) with j = k = 1 and (12.13), we also have
2(η + µ)µy1 y1 = ηx21 + µ2y1 . (12.17)
By combining (12.16) and (12.17) we find ηx1 x1 = µy1 y1 . Since η and µ are
independent of y1 and x1 respectively, we find
ηx1 x1 = µy1 y1 = 2F (x2 , y2 , . . . , xh , yh ), (12.18)
for some positive function F = F (x2 , y2 , . . . , xh , yh ). Hence, after solving
(12.18), we obtain
η = F (x2 , . . . , yh )x21 + G(x2 , . . . , yh )x1 + H(x2 , . . . , yh ),
(12.19)
µ = F (x2 , . . . , yh )y12 + K(x2 , . . . , yh )y1 + L(x2 , . . . , yh ),
for some functions G, H, K, L of 2n − 2 variables.
Substituting (12.19) into (12.16) gives
4F (H + L) = G2 + K 2 .
Hence, by (12.19), we get
1
η+µ= (2F x1 + G)2 + (2F y1 + K)2 .
4F
Combining this with (12.13) yields
1 n o
φ = ln (ax1 + β)2 + (ay1 + δ)2 , (12.20)
2
where
√ G K
a(x2 , . . . , yh ) = F > 0, β(x2 , . . . , yh ) = √ , δ(x2 , . . . , yh ) = √ .
2 F 2 F
From (12.20) we find
a(ax1 + β)
φx1 = ,
(ax1 + β)2 + (ay1 + δ)2
a(ay1 + δ)
φy1 = ,
(ax1 + β)2 + (ay1 + δ)2
(12.21)
(ax1 + β)(axj x1 + βxj ) + (ay1 + δ)(axj y1 + δxj )
φxj = ,
(ax1 + β)2 + (ay1 + δ)2
(ax1 + β)(ayj x1 + βyj ) + (ay1 + δ)(ayj y1 + δyj )
φyj = ,
(ax1 + β)2 + (ay1 + δ)2
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 261
for j = 2, . . . , h.
Hence, by applying (12.7) for φx1 xj , we obtain
h
∂
t−1
Y X
∂ ∂
∇ ∂ =− cos2 us f fxk + f fyk
∂ut ∂ut s=1
∂xk ∂yk
k=1
t−1 t−1
! (12.38)
X sin 2uq Y ∂
+ cos2 us , t = 1, . . . , p.
q=1
2 s=q+1 ∂uq
(12.45) we find
{(a1 x1 + · · · + ah xh )2 + (a1 y1 + · · · + ah yh )2 }Âjut
= aj a1 (x1 − iy1 ) + · · · + ah (xh − iyh )
( h h
) (12.50)
X X
k k
× Âut xk + B̂ut yk + Cut ,
k=1 k=1
for j = 1, . . . , h; t = 1, . . . , p. Therefore, after comparing the coefficients of
x21 , xk yℓ and x1 in (12.50) respectively, we derive
a21 Âjut = a1 aj Â1ut , aj ak B̂uℓ t = iaj aℓ Âkut , Cu1 = · · · = Cup = 0 (12.51)
for j, k, ℓ = 1, . . . , h.
The last equation in (12.51) shows that C is a constant vector in Cm .
Hence we may choose C = 0 by applying a suitable translation on Cm if
necessary. Since a1 6= 0, the first two equations in (12.51) with j = k = 1
imply that
aj
Âjut = Â1ut , B̂uj t = i Âjut , j = 1, . . . , h, t = 1, . . . , p. (12.52)
a1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 266
Case (a): p = 1. In this case, after substituting (12.54) into (12.48) with
t = 1, we find
h
X
Au1 u1 ak (xk + iyk )
k=1
h n
X (12.55)
=− ak (a1 x1 + · · · + ah xh )(ak A + γk )
k=1
o
+ ak (a1 y1 + · · · + ah yh )(iak A + δk )
which implies that
Au1 u1 + A = 0, (12.56)
h
X h
X
a k γk = ak δk = 0, (12.57)
k=1 k=1
Ph
where b = j=1 a2j .
Now, we obtain from (12.9) that
a1
h
X
x = x1 + iy1 + (cos u1 − 1) ak (xk + iyk ), . . . ,
b
k=1
X h
ah
xh + iyh + (cos u1 − 1) ak (xk + iyk ), (12.63)
b
k=1
h
X
sin u1 ak (xk + iyk ) .
k=1
From (12.63) we find
1
hxx1 , xx1 i = 1 + 1− sin2 u1 .
a21 (12.64)
b
From (12.33) with (12.64) we get b = 1. So, we find (12.29) with p = 1.
Case (b): p > 1. In this case, we get |∇f |2 = 1 from Proposition 12.1.
Hence we obtain a21 + a22 + · · · + a2h = 1 from (12.42).
By substituting (12.4) into (12.47), we find
Aus ut = −(tan us )Aut , 1 ≤ s < t ≤ p. (12.65)
Solving (12.65) for t = p and s = 1 gives
A(u1 , . . . , up ) = D1 (u2 , . . . , up ) cos u1 + E1 (u1 , . . . , up−1 ) (12.66)
for some functions D and E. Hence, we derive from (12.4) that
h
X
x = D1 (u2 , . . . , up ) cos u1 + E1 (u1 , . . . , up−1 ) ak (xk + iyk )
k=1
h
X
+ (γk xk + δk yk )
k=1
By substituting this into (12.47) with s = 2, t = p, we find
D1 (u2 , . . . , up ) = D2 (u3 , . . . , up ) cos u2 + E2 (u2 , . . . , up−1 ).
Continuing such procedure (p − 1)-times we get
n
x = Dp−1 (up ) cos u1 · · · cos up−1 + E1 (u1 , . . . , up−1 )
+ E2 (u2 , . . . , up−1 ) cos u1 + · · ·
oX
h
+ Ep−1 (up−1 ) cos u1 · · · cos up−2 ak (xk + iyk ) (12.67)
k=1
h
X
+ (γk xk + δk yk ).
k=1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 269
Put Cm+1
1∗ = Cm+1
1 − {0}. Consider the C∗ -action on Cm+1
1∗ by
λ · (z0 , . . . , zm ) = (λz0 , . . . , λzm ).
The set of equivalent classes obtained from this action is denoted by CH m .
Just like CP m , there is an alternate way to view CH m as follows: Let
H12m+1 (−1) = {z = (z1 , z2 , . . . , zm+1 ) ∈ Cm+1
1 : hz, zi = −1},
where h , i is the inner product on Cm+1 1 induced from the pseudo-
Euclidean metric g0 . H12m+1 (−1) is known as the anti-de Sitter spacetime.
We put
Tz′ = z ∈ Cm+1
1 : Re hu, zi = Re hu, izi = 0 .
Then we have an U (1)-action on H12m+1 defined by z 7→ λz. At each
point z ∈ H12m+1 (−1), the vector iz is tangent to the flow of the action.
The orbit lies in the negative definite plane spanned by z and iz. The
quotient space H12m+1 (−1)/ ∼ under the U (1)-action is exactly the complex
hyperbolic space CH m with constant holomorphic sectional curvature −4.
The complex structure J on CH m is induced from the canonical complex
structure J on Cm+1
1 via the Riemannian submersion:
ϕ : H12m+1 (−1) → CH m (−4),
which has totally geodesic fibers. This submersion is called the hyperbolic
Hopf fibration.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 272
for any U ∈ D̆ and Z ∗ ∈ D̆⊥ . Hence we find (12.85) from (12.79), (12.80)
and (12.93). Moreover, from (12.79), (12.80), (12.94), and f˘ = f ◦ π, we
obtain (12.86). Consequently, the basic equality ||σ||2 = 2p|∇(ln f )|2 must
holds for N in CP m (4).
Proof. Follows from Theorem 12.1 and Propositions 12.2 and 12.4.
where ||σ|| is the norm of the second fundamental form, dVT is the volume
element of N T and vol(N T ) is the volume of N T .
The equality sign of (12.118) holds identically if and only if we have:
(a) The warping function f is constant.
(b) (N T , gN T ) is holomorphically isometric to CP h (4) and it is isometri-
cally immersed in CP m (4) as a totally geodesic complex submanifold.
(c) (N ⊥ , f 2 gN ⊥ ) is isometric to an open portion of the real projective p-
space RP p (1) of constant sectional curvature one and it is isometrically
immersed in CP m (4) as a totally geodesic totally real submanifold.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 281
for each q ∈ N ⊥ .
Obviously, inequality (12.124) implies (12.118), with the equality sign
holding if and only if we have
(a) f is constant and
(b) ||σ||2 = 4hp holds identically.
When f is constant, the warped product is the Riemannian product of
(N T , gN T ) and (N ⊥ , f 2 gN ⊥ ). Therefore, N T ×f N ⊥ can be regarded as a
CR-product in CP m (4). Consequently, we may applying Theorem 6.6 to
conclude that
(1) N T is the complex projective h-space CP h (4) immersed as a totally
geodesic complex submanifold.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 282
Chapter 13
For CR-warped product in a complex space form, we also have the following
optimal inequality from [Chen (2003a)].
283
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 284
From (13.17) and (13.18) we know that the tangent space of Ch∗ × S p at
a point (z, w) is spanned by vectors given by (13.17) and (13.18). Since S p
is the unit hypersphere centered at the origin, it follows from (13.17) and
(α)
(13.18) that the induced metric on Ch∗ × S p via Shp is the warped product
qP
α
metric g = g0 + f 2 g1 with warping function f = j=1 zj z̄j , where g0 and
g1 are the metrics of Ch∗ and S p , respectively.
It follows from (13.17) that Ch∗ is immersed as a complex submanifold
of Cαp+h . From (13.17) and (13.18) we also know that S p is immersed as
a totally real submanifold of Cαp+h . Hence we have statement (1).
Applying (11.27) and (13.17)-(13.22) gives
(α)
XY Shp = ∇ ˜ X (1) Y (1) ⊗ j, ∇˜ X (2) Y (2) , (13.23)
(α)
ZW §hp = z (1) ⊗ ∇ ˜ Z W, 0, . . . , 0 , (13.24)
(α)
XZShp = X (1) ⊗ Z, 0, . . . , 0 , (13.25)
˜ is the
for vector fields X, Y tangent to Ch∗ and Z, W tangent to S p , where ∇
Levi-Civita connection for Euclidean space as well as for complex Euclidean
space. From (13.17)-(13.18) and (13.23)-(13.25), we find
(1)
σ(X, Y ) = σ(Z, W ) = 0, σ(X, Z) = Xz(1) ⊥ ⊗ Z, 0, . . . , 0 (13.26)
for vector fields X, Y tangent to Ch∗ and Z, W tangent to S p . Hence the
squared norm of the second fundamental form is given by
Xα
2p(2α − 1)
||σ||2 = , f 2
= zj z̄j . (13.27)
f2 j=1
(2) If the CR-warped product satisfies the equality case of (13.29), then
(2.a) N T is an open portion of Ch∗ ;
(2.b) N ⊥ is an open portion of S p ;
(2.c) There exist a natural number α ≤ h and a complex coordinate system
. . , zh } on Ch∗ such that the warping function f is given by
{z1 , .q
Pα
f= j=1 zj z̄j ;
(2.d) Up to rigid motions of Cm , the immersion φ of the CR-warped
product in Cm is given by
(α)
φ(z, w) = Shp (z, w), 0, . . . , 0 ,
∂2 φ
where φzj z̄k = ∂zj ∂ z̄k , . . . ,
etc., and
∂φ 1 ∂ ∂ ∂ 1 ∂ ∂
= −i , = +i . (13.43)
∂zj 2 ∂xj ∂yj ∂ z̄j 2 ∂xj ∂yj
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 289
It follows from (13.56) that (γj k̄ ) is Hermitian, i.e., γ̄j k̄ = γkj̄ . Hence
the Spectral Theorem in Matrix Theory shows that there exists a unitary
matrix which diagonalizes (γj k̄ ). So there is a suitable complex Euclidean
coordinate system {z1 , . . . , zh } on Ch such that (13.57) reduces to
h
X
2
f = bj zj z̄j + H(z1 , . . . , zh ) + K(z1 , . . . , zh ). (13.59)
j=1
Since f is a real-valued function, we may put
H = X + iY, K = U − iY, (13.60)
for some real-valued functions X, Y, U . From (13.58) and (13.60), we obtain
the following Cauchy-Riemann equations:
∂X ∂Y ∂Y ∂X
=− , = ,
∂xj ∂yj ∂xj ∂yj
(13.61)
∂U ∂Y ∂Y ∂U
= , =− .
∂xj ∂yj ∂xj ∂yj
From (13.61) we find that H + K = X + U is constant, say δ. Hence (13.59)
reduces to
Xh
f2 = bj zj z̄j + δ.
j=1
We may put δ = 0 by applying a suitable translation on Cm . Thus
Xh
f2 = a2j zj z̄j , (13.62)
j=1
for some real numbers a1 , . . . , ah ≥ 0, since we have f > 0.
Now, by combining (13.46) and (13.62) we find
∂ 2 Aj
= −a2j Aj , j = 1, . . . , h, (13.63)
∂u21
∂2B
= 0. (13.64)
∂u21
Since f > 0, there exists at least one aj greater than zero. Without loss of
generality, we may assume
a1 , . . . , aα > 0, aα+1 = · · · = ah = 0. (13.65)
for some 1 ≤ α ≤ h. From (13.63), (13.64) and (13.65), we obtain
Aj = Dj (u2 , . . . , up ) cos(aj u1 ) + Ej (u2 , . . . , up ) sin(aj u1 ), (13.66)
Ak = Dk (u2 , . . . , up )u1 + Ek (u2 , . . . , up ), (13.67)
B = F (u2 , . . . , up )u1 + G(u2 , . . . , up ) (13.68)
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 292
where ar+1 = · · · = aα = 1.
If r > 1, then (13.88) implies
cos u1 Dj cos(aj u1 ) + Ej sin(aj u1 )
(13.89)
+ sin u1 Dj sin(aj u1 ) − Ej cos(aj u1 ) = 0
for j = 1, . . . , r. Since a1 , . . . , ar 6= 1, (13.89) implies
D1 = · · · = Dr = E1 = · · · = Er = 0
which is a contradiction. Hence a1 = · · · = aα = 1. Thus we find from
(13.88) that ∂ 2 Dj /∂u22 = −Dj holds for j = 1, . . . , α. After solving these
equations we get
Dj = Fj (u3 , . . . , up ) cos u2 + Gj (u3 , . . . , up ) sin u2 .
Consequently, (13.86) becomes
Xα
φ= Fj (u3 , . . . , up ) cos u1 cos u2
j=1
+ Gj (u3 , . . . , up ) cos u1 sin u2 + Ej sin u1 zj (13.90)
h
X
+ Ek zk + G.
k=α+1
By substituting (13.90) into (13.41) with s = 2 and t > 2, we conclude that
G1 , . . . , Gα are constant vectors.
Continuing these procedures sufficiently many times, we find
α
( p
X j
Y
φ= c1 cos ut + c2j sin u1 + c3j sin u2 cos u1
j=1 t=1
p−1
) (13.91)
Y h
X
j
+ · · · + cp+1 sin up cos ut zj + Ek zk + G,
t=1 k=α+1
(1 + αp + α)-th
z}|{
φzα+1 (1, 0, . . . , 0) = (0, . . . , 0, 1 , 0, . . . , 0),
······
(αp + h)-th
z}|{
φzh (1, 0, . . . , 0) = (0, . . . , 0, 1 , 0, . . . , 0),
(p + 3)-th (1 + αp − p + α)-th
z}|{ z}|{
φu1 (1, 0, . . . , 0) = (0, 1, . . . , 0, 1 , 0, . . . , 0, 1 , 0, . . . , 0),
······
(p + 1)-th α(p + 1)-th
z}|{ z}|{
φup (1, 0, . . . , 0) = (0, . . . , 0, 1 , 0, . . . , 0, 1 , 0, . . . , 0).
Now, by applying (13.91) and (13.92) we derive that
φ = w0 z1 , . . . , wp z1 , . . . , w0 zα , . . . , wp zα ,
(13.93)
zα+1 , . . . , zh , 0, . . . , 0 ,
where
p
Y
w0 = cos ut , w1 = sin u1 , w2 = sin u2 cos u1 , . . . ,
t=1
p−1
Y
wp+1 = sin up cos ut .
t=1
Since φ is an immersion, (13.93) implies that N T is contained in Ch∗ . This
completes the proof of the theorem.
The next theorem from [Chen (2003a)] classifies all CR-warped product
submanifolds in the complex projective m-space CP m (4) satisfying the
equality case of inequality (13.1) with c = 1.
in Cm+1
∗ . Therefore Theorem 13.3 implies that, up to rigid motions, the
immersion of M̆ is the φ̆ defined defined in (2.c) of this theorem for some
natural number α satisfying α ≤ h. Consequently, up to rigid motions, φ
is given by the composition π ◦ φ̆.
Conversely, it is straightforward to verify that the immersion φ̆ is a
CR-warped product Ch+1 ∗ ×f S p in Cm+1 which is invariant under the C∗ -
action. Therefore the projection π ◦ φ̆ of φ̆ under π : Cm+1∗ → CP m (4)
m
defines a submanifold M in CP (4).
It is direct to verify that M is a CR-warped product CP h (4) ×f˜ S p in
CP m (4) for some suitable warping function f˜. Moreover, it follows from
(13.98) that the CR-warped product M satisfies the condition (13.104).
Consequently, after applying (13.14), we conclude that M = π(Ch+1 ∗ ×f S p )
satisfies the equality case of (13.94) identically. This completes the proof
of the theorem.
The next theorem from [Chen (2003a)] classifies all CR-warped product
submanifolds in the complex hyperbolic m-space CH m (−4) which satisfy
the equality case of (13.1) with c = −1.
and u ∈ R.
where e2n−2 is a unit vector tangent to N1 and e2n−1 a unit vector tangent
to N2 . It follows from (13.118) that
Je2n−1 = − sin αξ − cos αe2n−3 (13.119)
for some unit vector e2n−3 ∈ T N1 with he2n−3 , e2n−2 i = 0.
Since hJe2n−3 , ξi = − he2n−3 , Jξi = 0, (13.118) gives
Je2n−3 = − sin αe2n−2 + cos αe2n−1 . (13.120)
Using (13.118) and (13.119) , we find
Je2n−2 = − cos αξ − sin αe2n−3 . (13.121)
Clearly, e2n−3 , e2n−2 , e2n−1 , ξ span a complex 2-plane Hx at each point
x ∈ M and Dx1 is the orthogonal complementary subspace of Hx .
Since N2 is totally geodesic in M , (13.120) and the formulas of Gauss
and Weingarten imply
Jσ(V, e2n−1 ) = J ∇ ˜ V e2n−1 = ∇˜ V Je2n−1
= −(cos α){(V α)ξ + ∇V e2n−3 + σ(V, e2n−3 )} (13.122)
+ sin α{(V α)e2n−3 + AV },
for V ∈ T M , where A = Aξ is the shape operator. Using (13.122), we get
(V α) = − hAV, e2n−3 i , V ∈ T M. (13.123)
Also, by taking the inner product of (13.122) with e2n−2 , we get
h∇V e2n−3 , e2n−2 i = tan α hAV, e2n−2 i − hAV, e2n−1 i , V ∈ T M. (13.124)
Moreover, by taking the inner product of (13.122) with X ∈ D1 , we find
h∇V e2n−3 , Xi = tan α hAV, Xi , X ∈ D1 , V ∈ T M. (13.125)
In particular, if V = e2n−1 , (13.124) and (13.125) reduce to
σ(e2n−1 , e2n−1 ) = (tan α)σ(e2n−1 , e2n−2 ), σ(e2n−1 , X) = 0,
for X ∈ D1 .
We summary the above results as the following.
Lemma 13.2. Let N1 × N2 be a real hypersurface of a non-flat complex
space form M̃ n (4c). If dim N2 = 1 and n ≥ 3, then we have
V α = − hAV, e2n−3 i , (i)
h∇V e2n−3 , e2n−2 i = tan α hAV, e2n−2 i − hAV, e2n−1 i , (ii)
h∇V e2n−3 , Xi = tan α hAV, Xi , (iii)
σ(e2n−1 , e2n−1 ) = (tan α)σ(e2n−1 , e2n−2 ), (iv)
σ(e2n−1 , X) = 0 (v)
1
for X ∈ D and V ∈ T M.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 304
It follows from (13.204) that σ(e1 , Je2 ), σ2 n+1 , σ2 n+3 6= 0. Hence we derive
from the equation of Gauss that
0 = R̃(e2 , en+2 ; en+1 , en+3 ).
On the other hand, from (6.5), (13.200), (13.201), and (13.202) we get
R̃(e2 , en+2 ; en+1 , en+3 ) = c cos α 6= 0,
which is a contradiction. Hence we must have sin δ cos δ 6= 0, too.
Finally, from (13.197), (13.199), and Lemma 13.4(d), we get
σjn+2 σ(e1 , Je2 ) = 2cδ2j sin γ sin α ξ,
(13.205)
σjn+3 σ(e1 , Je2 ) = cδ1j sin γ sin α ξ
for j = 1, 2, 3. From (13.205) we find σ3n+2 = σ3n+3 = 0. Hence, by Gauss’
equation, we get
0 = R̃(e3 , en+2 ; en+3 , en+2 ).
On the other hand, from (6.5), (13.198), and (13.200), we get
R̃(e3 , en+2 ; en+3 , en+2 ) = 3c cos2 γ cos δ sin δ 6= 0,
which is a contradiction. Therefore case (2-b) is impossible as well. Hence
the real hypersurface must be irreducible. This completes the proof of the
theorem.
Remark 13.1. For any nonzero function λ(t) there exists a (unit speed)
Legendre curve z = z(t) in S 3 (c) satisfying equation (13.206). Such a
Legendre curve is unique if one imposes the initial conditions: z(0) = z0 ∈
S 3 (c) and z ′ (0) = u for some unit vector tangent to S 3 (c).
Similar to Lemma 13.5, we also have the following result from [Chen
(1997a)].
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 316
for j, k = 1, . . . , n.
Clearly, every constant function φ is a solution of the PDE system
(13.218). However, if φ is constant, the warping function f is constant.
In this case the hypersurface is a CR-product. Thus, by applying Theorem
4.6 of [Chen (1981b)], we know that the real hypersurface is locally the
product of Cn and a curve in C1 .
It follows from Proposition 12.3 that the solution of system (13.218) is
given by
1
φ = ln f, f = {hα, zi2 + hiα, zi2 } (13.219)
2
for some vector α ∈ Cn+1 . So, after choosing a suitable Euclidean complex
coordinates on Cn+1 , we may obtain
α = (b1 + ib2 , 0, . . . , 0).
With respect to this vector α, we have
f = {(b1 x1 + b2 y1 )2 + (−b2 x1 + b1 y1 )2 }1/2 ,
which can also be expressed as
q q
f = a x21 + y12 , a = b21 + b22 . (13.220)
From Gauss’ formula, (13.209), (13.211)-(13.214), (13.216) and (13.220),
we know that the immersion x of N ×f I in Cm satisfies
xxj xk = xxj yk = xyj yk = 0, j, k = 1, . . . , n, (13.221)
x1 − iy1 y1 + ix1
xx1 t = 2 xt , xy1 t = 2 xt , (13.222)
x1 + y12 x1 + y12
xxj t = xyj t = 0, j = 2, . . . , n, (13.223)
2
xtt = −a (x1 xx1 + y1 xy1 ) + iλxt . (13.224)
It is straightforward to verify from (13.221)-(13.224) that the immersion
x satisfies
xxj tt = xttxj , xyj tt = xttyj , j = 1, . . . , n
hold if and only if ∂λ/∂xj = ∂λ/∂yj = 0 and xyj = ixxj for j = 1, . . . , n.
Hence λ = λ(t) is a function of t. Therefore (13.224) reduces to
xtt = −a2 z1 xx1 + iλ(t)xt . (13.225)
Solving (13.221) gives
n
X n
X
k
x= Â (t)xk + B̂ k (t)yk + C(t) (13.226)
k=1 k=1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 319
where c1 and c2 are constant vectors in Cn+1 and A1 (t), A2 (t) are two
independent solutions of (13.234). Thus the image of γ(t) must lie in the
complex plane, say C2 , spanned by c1 and c2 .
Clearly, C2 is the complex plane defined by
{(w1 , w2 , 0, . . . , 0) : w1 , w2 ∈ C}.
Hence, if we denote the curve γ by γ(t) = (Γ1 (t), Γ2 (t)), then the hyper-
surface is given by
The following theorem from [Chen (2002c)] classifies all warped products
real hypersurfaces of the form N ×f I in complex projective spaces.
Theorem 13.8. Let a be a positive number and γ(t) = (Γ1 (t), Γ2 (t)) be a
unit speed Legendre curve γ : I → S 3 (a2 ) ⊂ C2 on an open interval I. If
x : S∗2n+1 × I → Cn+2 is the map defined by
n
X
x(z0 , . . . , zn , t) = aΓ1 (t)z0 , aΓ2 (t)z0 , z1 , . . . , zn , zk z̄k = 1, (13.241)
k=0
then we have:
(1) x induces an isometric immersion ψ : S∗2n+1 ×a|z0 | I → S 2n+3 .
(2) The image ψ(S∗2n+1 ×a|z0 | I) in S 2n+3 is invariant under the action of
U (1).
(3) the projection ψπ : π(S∗2n+1 ×a|z0 | I) → CP n+1 of ψ via π is a warped
product hypersurface CP0n ×a|z0 | I in CP n+1 .
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 322
Proof. Statement (1) is easy to verify, since γ(t) = (Γ1 (t)Γ2 (t)) is a unit
speed Legendre curve in S 3 (a2 ). Statement (ii) follows from (13.241) and
the definition of the U (1)-action.
Since γ = γ(t) is a Legendre curve in S 3 (a2 ), for each z = (z0 , . . . , zn )
P
with zk z̄k = 1 (13.241) implies that the curve Γz defined by
Γz (t) = aΓ1 (t)z0 , aΓ2 (t)z0 , z1 , . . . , zn
is horizontal in S 2n+3 . So, π : Γz → π(Γz ) is isometric. It is clear that the
restriction of π on S∗2n+1 is also a Riemannian submersion. Therefore the
projection of
ψπ : ψ(S∗2n+1 ×a|z0 | I) → CP n+1 (4)
of ψ : S∗2n+1 ×a|z0 | I → S 2n+3 is a warped product real hypersurface
CP n+1 (4) ×a|z0 | I into CP n+1 (4).
Conversely, assume that M = N ×f I is a warped product hypersurface
of CP n+1 (4), where N is a complex hypersurface of CP n+1 (4) and I is
an open interval. Then, according to Theorem 13.6, the warping function
f is non-constant.
Let H1 and H2 denote the distributions on M spanned by vectors tan-
gent to the N and I, respectively. Trivially, H1 and H2 are integrable
distributions. From Lemma 11.1(1) we know that the second fundamen-
tal form σ of M in CP n+1 (4) satisfies σ(H1 , H1 ) = 0. Since N is totally
geodesic in N ×f I, N is thus totally geodesic in CP n+1 (4). Hence, N is
holomorphically isometric to an open part of a CP n+1 (4).
Let ∇ˆ and ∇ denote the Riemannian connections of M̂ and M respec-
tively. And let σ̂ denote the second fundamental form of M̂ in S 2n+3 . Then
we have
ˆ X ∗ Y ∗ = (∇X Y )∗ − hP X, Y i V,
∇ (13.242)
ˆ V X∗ = ∇
∇ ˆ X ∗ V = (P X)∗ , (13.243)
ˆ V V = 0,
∇ (13.244)
∗ ∗ ∗
σ̂(X , Y ) = (σ(X, Y )) , (13.245)
∗ ∗
σ̂(X , V ) = (F X) , (13.246)
σ̂(V, V ) = 0, (13.247)
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 323
which implies that each integral curve of Ĥ2 is a circle in M̂ , i.e., a order
two Frenet curve with constant curvature in M̂ . Thus, a result of [Hiepko
(1979)] implies that locally M̂ is a warped product S 2n+1 ×fˆI with warping
function fˆ.
Let M̆ denote the punctured cone over M̂ with 0 as its vertex defined
by
M̆ = tw ∈ Cn+2 : w ∈ M̂ = S 2n+1 ×fˆ I ⊂ S 2n+3 ⊂ Cn+2 , t > 0 .
Since the tangent vector field ∂/∂t on M̆ is parallel to the position vector
field of M̆ in Cn+2 and V is tangent to the first component of S 2n+1 ×fˆ I,
we see that locally M̆ is the warped product Cn+1 ×tfˆ I, where Cn+1 is a
complex hyperplane of Cn+2 . Since the warping function is non-constant,
Theorem 4.1 implies that, up to rigid motions, M̆ is given by
x(z1 , . . . , zn , t) = aΓ1 (t)z1 , aΓ2 (t)z1 , z2 , . . . , zn , z1 6= 0,
for some positive number a and a unit speed Legendre curve γ(t) =
(Γ1 (t), Γ2 (t)) in S 3 (a2 ). Consequently, up to rigid motions, the warped
product hypersurface in CP n+1 (4) is the projection of ψ given by (13.241)
via the Hopf fibration.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 324
Chapter 14
325
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 326
14.1 δ-invariants
( )2
X X
+ ai1 − ai3 + ···
1≤ai1 ≤n1 n1 +n2 +1≤ai3 ≤n1 +n2 +n3
( )2
X X
+ aik−1 − aik
n1 +···+nk−2 +1≤ai1 ≤n1 +···+nk−1 n1 +···+nk−1 +1≤aik ≤n
≥ 0,
with equality holding if and only if (14.3) holds.
An immediate consequence of Lemma 14.1 is the following.
n+1 P
where ai = σii ,i = 1, . . . , n, and γ = n + k − nj .
We put
∆1 = {1, . . . , n1 },
...
∆k = {n1 + · · · + nk−1 + 1, . . . , n1 + · · · + nk }.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 330
for α2 , β2 ∈ ∆2 , . . . , αk , βk ∈ ∆k , where
ā1 = a1 , ā2 = a2 + · · · + an1 ,
ā3 = an1 +1 + · · · + an1 +n2 ,
...
āk+1 = an1 +···+nk−1 +1 + · · · + an1 ···+nk ,
āk+2 = an1 ···+nk +1 ,
...
āγ+1 = an .
Applying Corollary 14.1 to (14.14) yields
X X X
aα1 aβ1 + aα2 aβ2 + · · · + aαk aβk
α1 <β1 α2 <β2 αk <βk
X m n (14.15)
η n+1 2 1 X X r 2
≥ + (σij ) + (σ ) ,
2 i<j 2 r=n+2 i,j=1 ij
1 X X X r 2
m m k
1 X X
r
+ (σαβ )2 + σαj αj (14.17)
2 r=n+1 2 r=n+2 j=1
/ 2
(α,β)∈∆ αj ∈∆j
k
η X nj (nj − 1)
≥ + c,
2 j=1 2
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 331
where
∆ = ∆1 ∪ · · · ∪ ∆k , ∆2 = (∆1 × ∆1 ) ∪ · · · ∪ (∆k × ∆k ).
From (14.1), (14.11), and (14.17), we obtain (14.7). If the equality in
(14.5) holds at a point p, then the inequalities in (14.15) and (14.17) are
actually equalities at x. In this case, after applying Corollary 14.1, (14.14),
(14.15), (14.16) and (14.17), we obtain (14.8) and (14.9).
The converse can be verified by a direct computation.
Since δ(n − 1)(p) = maxu∈Tp1 (M) Ric(u), Theorem 14.1 implies
Corollary 14.2. For any n-dimensional submanifold M of Rm (c) with
n ≥ 3, the Ricci curvature Ric of M satisfies
n2 2
Ric(u) ≤ H + (n − 1)c
4
for any unit vector u ∈ T 1 M , where T 1 M denotes the unit tangent bundle
of M .
It follows immediately from the equation of Gauss that a necessary
condition for a Riemannian manifold to admit a minimal immersion in a
Euclidean space is “Ric ≤ 0”. For many years before the invention of δ-
invariants, this was the only known Riemannian obstruction for a general
Riemannian manifold to admit a minimal immersion into a Euclidean space
with arbitrary codimension.
An immediate application of Theorem 14.1 is the following theorem
which provides many new obstructions to minimal immersions.
Theorem 14.2. [Chen (1998a, 2000a)] Let N be a Riemannian n-
manifold. If there exist a point p and a k-tuple (n1 , . . . , nk ) ∈ S(n) with
δ(n1 , . . . , nk )(p) > 0, then N never admits a minimal immersion into any
Riemannian manifold with non-positive sectional curvature. In particular,
N never admits a minimal immersion into any Euclidean space for any
codimension.
Remark 14.2. There exist many Riemannian manifolds with Ricci tensor
Ric ≤ 0, but with some positive δ-invariants. For instance, the following
example was constructed in [Suceavă (2001)].
Let ε > 0 and put U = {(x, y, u, v) ∈ R4 : y > 2ε and v > 0}. Denote by
N 4 the Riemannian 4-manifold (U, g) with the warped product metric
1 2ε tan−1 y
g = 2 (dx2 + dy 2 ) + (du2 + dv 2 ).
y v2
Then, for sufficiently small ε > 0, N 4 satisfies Ric < 0 and δ(2, 2) > 0.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 332
Proof. This theorem follows from the fact that the proof of Theorem
14.1 is based only on the equation of Gauss and the Gauss equation for
submanifolds of a real space form Rm (c) is the same as the Gauss equation
of a totally real submanifold in a complex space form M̃ m (4c).
Let N be a Lagrangian submanifold of a complex space form M̃ n (4c).
Let p ∈ N and V be a d-dimensional subspace of Tp N . Denote by πV :
Tp N → V the orthogonal projection. For each vector v ∈ V , we define a
symmetric endomorphism AVJv on V by AVJv = πV ◦ AJv , where AJv is the
shape operator at Jv.
In order to prove the next theorem, we need the following two lemmas
from [Chen (2000d)].
We define a function fx by
fx : V1 → R : u 7→ fx (u) = hσ(u, u), Jui , (14.24)
where V1 is the unit hypersphere of V consisting of all unit vectors in V .
Since V1 is a compact set, there exists a vector v in V1 such that fx attains
an absolute maximum at v. We have fx (v) > 0.
By taking the directional derivative of fx with respect to w, we also
have hσ(v, v), Jwi = 0 for all w ∈ V orthogonal to v. So, by (2.14), we
know that v is an eigenvector of the symmetric operator AVJv . If we put
ε1 = v and choose ε2 , . . . , εn so that {ε1 , . . . , εn } is an orthonormal basis of
V and each εi is an eigenvector of AVJε1 with eigenvalue λi , then we obtain
(14.23).
Since fx attains an absolute maximum at ε1 , the function fi , i ∈
{2, . . . , d}, defined by fi (t) = fx (cos t ε1 + sin t εi ) has a relative maximum
at t = 0. So, after computing fi′′ (0), we get λ1 ≥ 2λi .
Since fx (v) > 0, we get λ1 > 0. Hence we find λ1 > λi , for i ≥ 2. In
particular, this implies that the eigenspace of AJ ē1 with eigenvalue λ1 is
1-dimensional. This proves the lemma.
Now, assume N is a Lagrangian submanifold of a complex space form
M̃ n (4c) satisfying the equality case of (14.20) identically for some k-tuple
(n1 , . . . , nk ) ∈ S(n). Thus, according to Theorem 14.3, there exists an
orthonormal basis {e1 , . . . , en } of Tx N at each x ∈ N such that, for each
normal vector ξ at x, the shape operator Aξ with respect to {e1 , . . . , en }
takes the form:
Aξ1 . . . 0
.. . . ..
.
. .
Aξ = 0 . . . Aξ
0 , (14.25)
k
0 µξ I
Proof. If hσ(u, v), Jwi = 0 for all tangent vectors u, v and w at x, then
we get σ = 0 by polarization. In this case there is nothing to prove. So
we may assume that this is not the case. By applying Lemma 14.2 to
V = Tx M n , we know that there exists an orthonormal basis {ε1 , . . . , εn } of
V such that
AJε1 εi = λi εi , i = 1, . . . , n, (14.31)
where λ1 , . . . , λn satisfy λ1 ≥ 2λj , j = 2, . . . , n. Moreover, we have
(a) λ1 > 0 unless AJv = 0 for all v ∈ V and
(b) if λ1 > 0, then the multiplicity of λ1 is one.
Assume that N satisfies equality (14.20) at a point x. Then Theorem
14.3 implies that there exists an orthonormal basis {e1 , . . . , en } of Tx N such
that, for each normal vector ξ at x, the shape operator Aξ takes the form
of (14.25) so that condition (14.26) holds.
Define the subspaces L1 , . . . , Lk of Tx M n by (14.27) and Lk+1 by
Lk+1 = Span {en1 +···+nk +1 , . . . , en }.
Obviously, we have Tp N = L1 ⊕ · · · ⊕ Lk+1 .
Now, we claim that ε1 must lies in one of L1 , . . . , Lk . This can be seen
as follows. First we put
ε1 = v1 + · · · + vk + vk+1 , (14.32)
where v1 ∈ L1 , . . . , vk+1 ∈ Lk+1 . Then, we have
λ1 ε1 = AJε1 ε1 = AJε1 v1 + · · · + AJε1 vk + AJε1 vk+1 . (14.33)
From (14.25) we have AJε1 vi ∈ Li . Thus (14.32) and (14.33) imply
AJε1 v1 = λ1 v1 , . . . , AJε1 vk+1 = λ1 vk+1 . (14.34)
Since the eigenspace of AJε1 with eigenvalue λ1 is 1-dimensional, (14.34)
implies that exactly one of v1 , . . . , vk+1 is nonzero.
April 18, 2017 12:18 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 336
Theorem 14.4 suggests that inequality (14.20) given in Theorem 14.3 is not
optimal, i.e., that the coefficient of H 2 can be replaced by a smaller value.
In this respect, the following two theorems were proved.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 337
6 XX r r
− σ σ
p − 1 r s<t ss tt
6(p + 2) X r 2 3 XX r r 2
= (σst ) + (σss − σtt )
p − 1 r<s<t p−1 s<t r6=s,t
1 X r r 2
+ (σrr − 3σss )
p−1
s6=r
≥ 0.
Hence inequality (14.39) follows from (14.45). Moreover, it is easy to verify
that the equality sign of (14.39) holds if and only if (14.40) holds. This
proves the theorem.
CR-warped products in Ch+p satisfying the equality case of inequality
(14.39) have been completely classified as follows.
Theorem 14.8. [Chen (2012)] If ψ : N T ×f N ⊥ → Ch+p is a CR-warped
product in Ch+p with h = dimC N T ≥ 1 and p = dim N ⊥ ≥ 2, then
2 2(p + 2) p∆f
H ≥ δ(H) − . (14.46)
(2h + p)2 (p − 1) f
The equality sign of (14.46) holds identically if and only if, up to dila-
tions and rigid motions of Ch+p , one of the following three cases occurs:
(2h + p)2 2 1
= H − 2h2 |HH |2 − ||σH⊥ ||2 ,
2 2
where ||σH⊥ ||2 is defined by
2h+p
X
||σ⊥ ||2 = |σ(er , es )|2 . (14.52)
r,s=2h+1
By combining (14.50) and (14.51) we find
(2h + p)2 2 p
δ(H) = H + (4h + p − 1)c − 2h2 |HH |2
2 2
X2h 2h+p
X (14.53)
1
− |σ(ei , er )|2 − ||σH⊥ ||2 .
i=1
2
r=2h+1
It follows from statement (2) of Lemma 6.1 that the coefficients of the
second fundamental form σ satisfy
r s t
σst = σrt = σrs . (14.54)
Hence we find from (14.47), (14.52) and (14.54) that
2h+p 2h+p
!2
X X
2 2 2 r
(p + 2)||σH⊥ || − 3p |HH⊥ | = (p − 1) σss
r=2h+1 s=2h+1
X
r 2
+ 3(p + 1)(σss )
2h+1≤r6=s≤2h+p
X
r 2
+ 6(p + 2)(σst )
2h+1≤r<s<t≤2h+p
2h+p
X X
r r
+ 2(p + 2)σss σtt (14.55)
r=2h+1 2h+1≤s<t≤2h+p
2h+p
X X
r 2 r 2
= (p − 1)(σrr ) + 3(p + 1)(σss )
r=2h+1 2h+1≤r6=s≤2h+p
X
r 2
+ 6(p + 2)(σst )
2h+1≤r<s<t≤2h+p
2h+p
X X
r r
− 6σss σtt
r=2h+1 2h+1≤s<t≤2h+p
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 343
X X
r 2 r r 2
= 6(p + 2)(σst ) + (σrr − 3σss )
2h+1≤r<s<t≤2h+p 2h+1≤s6=r≤2h+p
X X
r r 2
+ 3(σss − σtt )
r6=s,t 2h+1≤s<t≤2h+p
≥ 0.
Thus we get
3p2
||σH⊥ ||2 ≥ |H ⊥ |2 , (14.56)
p+2 H
with equality holding if and only if
r r
σrr = 3σss , for 2h + 1 ≤ r 6= s ≤ 2h + p,
r
(14.57)
σst = 0, for distinct r, s, t ∈ {2h + 1, . . . , 2h + p}.
Now, by combining (14.53) and (14.56), we obtain
(2h + p)2 2 p
H + (4h + p − 1)c − δ(H)
2 2
X2h 2h+p
X 3p2
≥ 2h2 |HH |2 + |σ(ei , er )|2 + |HH⊥ |2
i=1
2(p + 2)
r=2h+1
2h 2h+p
X X
= 2h2 |HH |2 + |σ(ei , er )|2
i=1 r=2h+1
(
3
+ (2h + p)2 H 2 − 4h2 |HH |2
2(p + 2)
) (14.58)
X2h 2h+p
X
−2 |σ(ei , er )|2
i=1 r=2h+1
2
3(2h + p) 2 2h2 (p
− 1)
= H + |HH |2
2(p + 2) p+2
2h 2h+p
p−1X X
+ |σ(ei , er )|2
p + 2 i=1
r=2h+1
3(2h + p)2 2
≥ H .
2(p + 2)
It is obvious that the equality of the last inequality in (14.58) holds if and
only if N is H-minimal and mixed totally geodesic. Consequently, we have
inequality (14.48) from (14.58). It is direct to verify that the equality sign
of (14.48) holds identically if and only if conditions (a), (b) and (c) of
Theorem 14.9 are satisfied.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 344
for some suitable function λ with respect to some suitable orthonormal local
frame field {e2h+1 , . . . , e2h+p } of T N ⊥ .
If λ = 0, then N ⊥ is an open portion of a totally geodesic totally real
p-plane in Cp . Hence, in this case N is a totally geodesic anti-holomorphic
submanifold.
If λ 6= 0, it follows from (14.60) that, up to dilations and rigid motions,
N ⊥ is an open part of the Whitney p-sphere in Cp (cf. [Borrelli et al.
(1995); Chen (2011b)]). Consequently, up to dilations and rigid motions
of Ch+p , the anti-holomorphic submanifold is locally given by the product
immersion:
φ : Ch × S p (1) → Ch+p ; (z, x) 7→ (z, w(x)), (14.61)
for z ∈ Ch and x ∈ S p (1), where w : S p (1) → Cp is the Whitney p-sphere.
The converse is easy to verify.
Theorem 14.11. [Al-Solamy et al. (2014)] Let N be an anti-holomorphic
submanifold in a complex space form M̃ 1+p (4c) with h = dimC H = 1 and
p = dim H⊥ ≥ 2. Then we have
(p − 1)(p + 2)2 2 p
δ(H) ≤ H + (p + 3)c. (14.62)
2(p + 2) 2
The equality case of (14.62) holds identically if and only if c = 0 and
either
for i = 1, . . . , 2h. Then it follows from (14.37) and Gauss’ equation that
2h
X 2h
X
δ(H) = σi,i σ2h+1,2h+1 − (σi,2h+1 )2 + 2hc. (14.68)
i=1 i=1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 347
Proof. Let π : S 2m+1 (1) → CP m (4) be the Hopf fibration and put ξ = iz
ˆ and ∇
as before. Denote by ∇ ˇ the Levi-Civita connections of S 2m+1 (1) and
CP (4), respectively. For vector fields X, Y tangent to CP m (4), we have
m
ˆ X̄ Ȳ = ∇
∇ ˇ X Y − hJX, Y i ξ, (14.91)
ˆ X̄ ξ = ∇
∇ ˆ ξ X̄ = JX. (14.92)
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 352
The next result from [Alegre et al. (2014)] provides the necessary and
sufficient condition for π −1 (N ) to satisfy the equality case of the inequality
(14.89), where N is a CR-submanifold of CP m (4).
Theorem 14.20. There exist many non-totally geodesic totally real sub-
manifolds of CP m (4) which satisfy the equality case of inequality (14.89)
identically.
Theorem 14.21. [Chen et al. (1994)] Let S 2n+1 (1) be the unit hypersphere
of Cn+1 and consider the orthogonal decomposition
Cn+1 = C3 ⊕ J(En−2 ) ⊕ En−2 .
Let f : M 2 → S 5 (1) ⊂ C3 be a minimal, isometric, C-totally real immer-
sion and consider the hypersphere S n−3 (1) in En−2 . Then we have:
(1) F : (0, π2 ) ×cos t M 2 ×sin t S n−3 (1) → S 2n+1 (1) defined by
F (t, p, q) 7→ (cos t)ψ(p) + (sin t)q
is a minimal δ(2)-ideal C-totally real immersion.
(2) If ψ has no totally geodesic points, then the dimension of Nx is exactly
n − 2 for each point x.
(3) If we extend F to a map
F̃ : − π2 , π2 × M 2 × S n−3 (1) → S 2n+1 (1)
defined by
F̃ (t, p, q) = (cos t)ψ(p) + (sin t)q.
Then F̃ fails to be immersive at t = 0, but the image of F̃ is an im-
mersed minimal C-totally real submanifold. If ψ is not totally geodesic,
then this image can not be extended further.
Proof. The first part of the theorem follows from the discussion above.
F is δ(2)-ideal because the dimension of N is at least n − 2. In case f has
no totally geodesic points, we have
N = T 0, π2 ⊕ T S n−3 (1) and N ⊥ = T M 2 .
Obviously the conditions (1) and (2) are satisfied in this case.
In order to prove the second part, we introduce a map
n−2
γ : − π2 , π2 × M 2 × S n−3 (1) → S+ × M2
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 358
Chapter 15
359
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 360
The best known example of nearly Kähler manifolds, but not Kählerian
manifold, is S 6 with the nearly Kählerian structure induced from the vector
cross product on the space of purely imaginary Cayley numbers O (see
§15.2).
It was proved that any 6-dimensional strict nearly Kähler manifold is
an Einstein manifold and it has vanishing first Chern class [Nagy (2002)]
(in particular, this implies spin). It was shown in [Nagy (2002)] that any
complete, strict nearly Kähler manifold is locally a Riemannian product of
homogeneous nearly Kähler spaces, twistor spaces over Kähler manifolds
and 6-dimensional nearly Kähler manifolds. Moreover, P.-A. Nagy proved
in [Nagy (2002)] the following properties for a strict and complete nearly
Kähler manifold M :
It is also known that the (2, 1)-tensor G satisfies the following properties
[Sekigawa (1983)]:
G(X, JY ) = −JG(X, Y ), (15.11)
˜
(∇G)(X, Y, Z) = g(Y, JZ)X + g(X, Z)JY − g(X, Y )JZ, (15.12)
g(G(X, Y ), Z) + g(G(X, Z), Y ) = 0, (15.13)
g(G(X, Y ), G(Z, W )) = g(X, Z)g(Y, W ) − g(X, W )g(Z, Y ) (15.14)
+ g(JX, Z)g(Y, JW ) − g(JX, W )g(Y, JZ).
Theorem 15.1. [Gray (1969)] With respect to the usual nearly Kähler
structure, S 6 has no 4-dimensional complex submanifolds.
Proof. From (15.23) and (15.24) we see that the normal connections and
the shape operators coincide. Thus, by applying the standard uniqueness
result for minimal immersions in real space forms we get the theorem.
for arbitrary tangent vector fields X,Y and Z and for orthonormal vector
fields Y1 , Y2 and Y3 .
Next, we take X and Y orthonormal. Then we have
h∇Z (X ∧ Y ), Xi = − hX ∧ Y, ∇Z Xi
= h(∇Z X) ∧ Y, Xi
= h(∇Z X) ∧ Y + X ∧ (∇Z Y ), Xi
h∇Z (X ∧ Y ), Y i = − hX ∧ Y, ∇Z Y i
= hX ∧ ∇Z Y, Y i
= h(∇Z X) ∧ Y + X ∧ (∇Z Y ), Y i
h∇Z (X ∧ Y ), (X ∧ Y )i = 0
h(∇Z X) ∧ Y + X ∧ (∇Z Y ), X ∧ Y i = 0.
So, we have
∇Z (X ∧ Y ) = (∇Z X) ∧ Y + X ∧ (∇Z Y ). (15.26)
A straightforward computation shows that (15.26) remains valid if X and
Y are arbitrary tangent vector fields.
Now, we take N M = T M as bundle over M . We define a connection
⊥
∇ on N M by
∇⊥
X J0 Y = J0 ∇X Y + J0 X ∧ Y,
for X and Y tangent to M , where J0 is the identification between T M and
N M . We define the second fundamental form σ by
σ : T M × T M → N M : (X, Y ) 7→ σ(X, Y ) = J0 α(X, Y ),
and the Weingarten operator A by AJ0 X Y = α(X, Y ) for X, Y ∈ T M .
A straightforward computation shows that the Gauss, Codazzi and Ricci
equations are satisfied and hence, by the existence and uniqueness theorem
for immersions into a real space form, there exists an isometric immersion
x : M 3 → S 6 with second fundamental form σ, normal connection ∇⊥ and
Weingarten operator A.
Let us now prove that x is a Lagrangian immersion. In order to do so,
we consider S 6 as a hypersphere in E7 . Let p ∈ M . We define a vector
cross product on E7 by
p × X = J0 X,
p × J0 X = −X,
X × Y = J0 X ∧ Y, (15.27)
J0 X × Y = (X ∧ Y ) − hX, Y i p,
J0 X × J0 Y = −X ∧ Y.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 369
Remark 15.1. Theorem 15.5 was first given in a preprint of [Ejiri (1981)].
For minimal totally real surfaces in the nearly Kähler S 6 we have the fol-
lowing.
The following result gives the integrability condition for the holomorphic
distribution of a CR-submanifold in nearly Kähler manifold.
The next result gives integrability condition for the totally distribution
of a CR-submanifold in nearly Kähler manifold.
The following result from [Khan et al. (2007); Sahin and Günes (2008);
Al-Luhaibi et al. (2009)] is analogous to Theorem 11.1 for warped product
CR-submanifolds of a Kähler manifold.
which implies
˜ JX J)JX + J ∇
hAJZ X, JXi = − h JZ, (∇ ˜ JX JX i .
The next result from [Sahin and Günes (2008); Al-Luhaibi et al. (2009)]
is analogous to Theorem 11.3.
Proof. This can be done in a similar way as Theorem 11.3 for CR-warped
products in Kähler manifolds.
Remark 15.2. It was proved in [Khan et al. (2009)] that there does not
exist a proper doubly warped product CR-submanifold f1 M T ×f2 M ⊥ in a
nearly Kähler manifold such that M T is a complex submanifold and M ⊥
is a totally real submanifold.
Theorem 15.13. There does not exist any proper CR-product in the nearly
Kähler S 6 .
Then we have
hσ(E, JE), σ(Z, Z)i = 0. (15.67)
From (15.60) and (15.67) we get
hσ(E, Z), σ(JE, Z)i = 0. (15.68)
Now, we claim that σ(E, Z) 6= 0. Suppose that σ(E, Z) = 0. Then it
follows from (15.59) that
hσ(E, E), σ(Z, Z)i = −1. (15.69)
From (15.59) and (15.66) we obtain
hσ(E, Z), σ(E, Z)i = max {hσ(X, Z), σ(X, Z)i}.
|X|=1,X∈Hx
Thus we find hσ(JE, Z), σ(JE, Z)i = 0, and hence from (15.57) and (15.59)
it follows that
−1 = hσ(JE, JE), σ(Z, Z)i = − hσ(E, E), σ(Z, Z)i ,
which contradicts (15.69). Hence we must have σ(E, Z) 6= 0.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 379
defined by
ϕ(t, p) = (sin t)N + cos(t)ψ(p).
The next proposition from [Chen et al. (1995a)] shows that the warped
product immersion ψ satisfies some nice properties.
Chapter 16
383
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 384
Proof. The proof of (16.3) is the same as that of (16.3). (16.4) follows
from (1.10), (16.1), (16.3) and P 2 = I. (16.5) follows from (16.4).
From (16.1) we find
g(Pv, w) + g(v, Pw) = 0, v, w ∈ Tp M̃ , p ∈ M̃ . (16.6)
Thus g(v, Pv) = 0. When {v, Pv} determines a non-degenerate plane, the
sectional curvature H P (v) = K(v ∧ Pv) is called the para-holomorphic
sectional curvature of the P-plane section spanned by v.
with
1 1
hv, vi 4 hu, ui 4
a= 1 1 , b = 1 1 .
hu, ui 4 hu, vi 2 hv, vi 4 hu, vi 2
Thus
P r (B) ∼
= {(u, v) ∈ Rr+1 × Rr+1 : hu, ui = hv, vi , hu, vi = 1}. (16.8)
Since Z(qz) = qZ(z) for all Z ∈ U (B, r + 1), z ∈ B0r+1 ,
q ∈ B+ , the action
of U (B; r + 1) pass to the quotient P r (B).
For a nonzero real number c, consider on B0r+1 the tensor field
( r r
2 X X
g̃c = duj ⊗ dvj + dvj ⊗ duj
c hu, vi
j=0 j=0
r
)
1 X
− ui vj (dvi ⊗ duj + duj ⊗ dvi ) .
hu, vi
i,j=0
The P r (B), together with the metric (16.9) and the almost product struc-
ture P given by (16.10), is a para-Kähler manifold of constant para-
holomorphic sectional curvature c 6= 0.
Moreover, if r > 1, then P r (B) is complete, connected, simply connected
para-Kähler space form. To obtain a complete simply-connected model of
para-Kähler space form for r = 1, it is enough to extend the above structure
on P 1 (B) = S 1 ×R to its universal covering, which is R2 . We simply denote
the triple (P r (B), P, gc ) by Pcr (B).
where a1 , . . . , an are real numbers, not all zero. Such null-isotropic invariant
submanifolds satisfy Sσ = 0.
Corollary 16.2. If the direct product Rr2r × Pp2p of two para-Kähler man-
ifolds can be isometrically immersed in a non-flat para-Kähler space form
2(r+p+rp)
Mr+p+rp (4c) as an invariant submanifold, then both Rr2r and Pp2p are of
constant para-holomorphic sectional curvature 4c.
Moreover, both Rr2r and Pp2p are immersed as totally geodesic invariant
2(r+p+rp)
submanifolds of Mr+p+rp (4c).
Proof. Under the hypothesis of the corollary, let us consider the PR-
2(r+p+rp)
submanifold Rr2r ×N ⊥ of Mr+p+rp (4c) such that N ⊥ is a Lagrangian sub-
manifold of Pp2p . Then, by Theorem 16.5, Rr2r is totally geodesic invariant
2(r+p+rp)
submanifold of Mr+p+rp (4c). Thus Rr2r is of constant para-holomorphic
sectional curvature 4c.
2(r+p+rp)
Similarly, Pp2p is a totally geodesic submanifold of Mr+p+rp (4c).
2p
Hence Pp is also of constant para-holomorphic sectional curvature 4c.
First we give the following result from [Chen and Munteanu (2012)].
Then we have
g(σ(D, D), F D⊥ ) = 0, (16.48)
g(σ(P X, Z), F W ) = −(Xµ)g(Z, W ), (16.49)
⊥
for vectors X, Y ∈ D and Z, W ∈ D . Thus Theorem 16.3(5) and (16.48)
imply that D is integrable and its leaves are totally geodesic in M .
On the other hand, from Lemma 16.4 and (16.49) we obtain
g(∇Z X, W ) = −g(∇Z W, X)
= −g(P APW Z, X)
(16.50)
= g(σ(P X, Z), PW )
= −(Xµ)g(Z, W )
⊥
for X in D and Z, W in D .
Since the distribution D⊥ of M is always integrable, (16.50) together
with W µ = 0, W ∈ D⊥ , imply that leaves of D⊥ are extrinsic spheres in M ,
i.e., they are totally umbilical submanifolds with parallel mean curvature
vector. Hence, after applying a result of [Hiepko (1979)], we conclude that
M is locally the warped product N T ×f N ⊥ of a invariant submanifold
and an anti-invariant submanifold N ⊥ of M , where NT is a leaf of D and
N ⊥ is a leaf of D⊥ and f is a certain warping function. Therefore M is a
P R-warped product.
The next result was proved in [Chen and Munteanu (2012)].
We obtain
P
h p
P
g̃ σ(D, D⊥ ), σ(D, D⊥ ) = − b 2
(σia b
) − (σi∗a )2
i=1 a,b=1
q
(16.57)
P p P
h P
α 2 α∗ 2 α 2 α∗ 2
+ (σia ) − (σia ) − (σi∗a ) + (σi∗ ) .
i=1 a=1 α=1
This yields
b
σia = P Xi (ln f )δab ,
b
(16.59)
σi∗a = Xi (ln f )δab .
By combining (16.57), (16.58) and (16.59) we get
h
X 2 2
g(σ(D, D⊥ ), σ(D, D⊥ )) = p Xi (ln f ) − P Xi (ln f )
i=1
q
p X
(16.60)
h X
X α 2 α∗ 2
+2 (σia ) − (σia ) .
i=1 a=1 α=1
As
g(σ(X, Z), f ξ) = −g(∇⊥
X F Z, ξ)
σ(D, D⊥ ) ⊂ PD⊥ .
α α∗
Hence σia and σia vanish. Thus
g̃ σ(D, D⊥ ), σ(D, D⊥ ) = p g (∇ ln f, ∇ ln f ) . (16.61)
Finally, we study g(σ(D⊥ , D⊥ ), σ(D⊥ , D⊥ )). We write
c α α∗
σ(Za , Zb ) = σab F Zc + σab ξα + σab ξα∗
and hence
p X
X q p
X
α 2
g(σ(D⊥ , D⊥ , σ(D⊥ , D⊥ )) = α∗ 2
(σab ) − (σab ) − c 2
(σab ) .
a,b=1 α=1 a,b,c=1
As g(σ(Z, W ), f ξ) = −g(∇⊥ ⊥
Z F W, ξ) and using ∇D ⊥ PD
⊥
⊂ PD⊥ , we
⊥ ⊥ ⊥ α α∗
derive σ(D , D ) ⊂ PD . Thus we have σab = σab = 0. Consequently, we
obtain
X p
g(σ(D⊥ , D⊥ ), σ(σ(D⊥ , σ(D⊥ )) = − c 2
(σab ) . (16.62)
a,b,c=1
(1) Φ : D1 ×f S p → P h+p ;
h
X h
X
Φ(z, w) = z1 + v̄1 (w0 − 1) vj zj , . . . , zh + v̄h (w0 − 1) vj zj ,
j=1 j=1
h
X h
X
w1 jvj zj , . . . , wp jvj zj , h ≥ 2,
j=1 j=1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 408
p
with warping function f = hv̄, zi2 − hjv̄, zi2 , where v = (v1 , . . . , vh ) ∈
S 2h−1 ⊂ Dh , w = (w0 , w1 , . . . , w
p
p ) ∈ S , z = (z1 , . . . , zh ) ∈ D1 and
D1 = z ∈ Dh : hv̄, zi2 > hjv̄, zi2 .
(2) Φ : D1 ×f H p → P h+p ;
X h h
X
Φ(z, w) = z1 + v̄1 (w0 − 1) vj zj , . . . , zh + v̄h (w0 − 1) vj zj ,
j=1 j=1
h
X h
X
w1 jvj zj , . . . , wp jvj zj , h ≥ 1,
j=1 j=1
p
with the warping function given by f = hv̄, zi2 − hjv̄, zi2 , where v =
(v1 , . . . , vh ) ∈ H 2h−1 ⊂ Dh , w = (w0 , w1 , . . . , wp ) ∈ H p and z =
(z1 , . . . , zh ) ∈ D1 .
(3) Φ(z, u) : D1 ×f Ep −→ P h+p ;
v̄1 X 2 X v̄h X 2 X
p h p h
Φ(z, u) = z1 + ua vj zj , . . . , zh + ua vj zj ,
2 a=1 j=1
2 a=1 j=1
Xh X h
u1 jvj zj , . . . , up jvj zj , h ≥ 2,
j=1 j=1
p
with warping function f = g(v̄, z)2 −g(jv̄, z)2 ,
where v = (v1 , . . . , vh )
is a light-like vector in Dh , z = (z1 , . . . , zh ) ∈ D1 and u = (u1 , . . . , up )
∈ Ep . Moreover, in this case, each leaf Ep is quasi-minimal in P h+p .
(4) Φ(z, u) : D2 ×f Ep −→ P h+p ;
p p
!
v1 X 2 vh X 2 v0 v0
Φ(z, u) = z1 + u , . . . , zh + u , u1 , . . . , up ,
2 a=1 a 2 a=1 a 2 2
h ≥ 1,
p √ √
with the warping function f = −hv, zi, where v0 = b1 + ǫj b1 with
b1 > 0, ǫ = ±1, z = (z1 , . . . , zh ) ∈ D2 and u = (u1 , . . . , up ) ∈ Ep ,
D2 = {z ∈ Dh : hv, zi < 0}, and
v = (v1 , . . . , vh ) = (b1 + ǫjb1 , . . . , bh + ǫjbh ).
2(h+p)
In each of the four cases the warped product is minimal in Eh+p .
Remark 16.5. For a time-like P R-warped product N T ×f N ⊥ in the para-
Kähler (h + p)-plane P h+p , a similar classification result as Theorem 16.8
holds as well (cf. Remark 6.4 of [Chen and Munteanu (2012)]).
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 409
Chapter 17
The following result can be found in page 43 of [Yano and Kon (1983)].
409
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 410
φX = P X + F X, (17.8)
P ξ = F ξ = 0, (17.9)
∇X ξ = −P X, σ(X, ξ) = −F X. (17.10)
φζ = tζ + f ζ, (17.11)
where tζ and f ζ are the tangential and the normal components of φζ,
respectively.
By applying Proposition 5.6 we know that the Riemann curvature tensor
of a submanifold M in a Sasakian space form M̃ (c) of constant φ-sectional
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 411
curvature c satisfies
c + 3
R(X, Y )Z = g(Y, Z)X − g(X, Z)Y
4
c − 1
− η(Y )η(Z)X − η(X)η(Z)Y
4 (17.12)
+ g(Y, Z)η(X)ξ − g(X, Z)η(Y )ξ − g(P Y, Z)P X
+ g(P X, Z)P Y + 2g(P X, Y )P Z
+ Aσ(Z,Y ) X − Aσ(Z,X) Y
for vectors X, Y, Z ∈ T M . Moreover, we know that the equations of Gauss
and Codazzi of M in M̃ (c) are given respectively by
R(X, Y ; Z, W ) = R̃(X, Y ; Z, W ) + hσ(X, W ), σ(Y, Z)i
(17.13)
− hσ(Y, W ), σ(X, Z)i
and
¯ X σ)(Y, Z) − (∇
(∇ ¯ Y σ)(X, Z)
c−1 (17.14)
= {g(P Y, Z)F X − g(P X, Z)F Y − 2g(P X, Y )F Z} .
4
Definition 17.3. The second fundamental form σ of M in M̃ (c) is said to
satisfy the classical Codazzi equation if
¯ X σ)(Y, Z) = (∇
(∇ ¯ Y σ)(X, Z) (17.15)
holds for all X, Y, Z ∈ T M.
Remark 17.1. Theorem 17.1 was proved by modifying the proof of The-
orem 10.1, using (17.12) instead of (2.21). For the details, see [Matsumoto
and Mihai (2002)].
such that
φ(Hx (M )) = Hx (M ), φ(Hx⊥ (M )) ⊂ Tx⊥ M.
Remark 17.2. Theorem 17.3 was proved in [Munteanu (2005)]. This result
is due to [Hasegawa and Mihai (2003)] under an additional condition that
ξ is tangent to N1⊥ .
Analogous to Theorem 12.4, the next result from [Hasegawa and Mihai
(2003)] determines the smallest codimension for contact CR-warped prod-
ucts in a standard Sasakian sphere.
For strictly proper contact CR-products we have the following two re-
sults from [Munteanu (2005)].
Example 17.2. [Munteanu (2005)] Let R2α+1 (−3) denote the standard
Sasakian space form of constant φ-sectional curvature −3. Denote by S β (1)
the unit hypersphere in the Euclidean space Eβ+1 .
For a positive integer a ≤ α, let us consider the map
ϕ : R2α+1 (−3) × S β (1) → R2aα+2β+1 (−3) (17.35)
defined by
ϕ(x1 , y1 , . . . , xα , yα , z, w0 , w1 , . . . , wβ )
= w0 x1 , w0 y1 , . . . , wβ x1 , wβ y1 , . . . , w0 xa ,
w0 ya , . . . , wβ xa , wβ ya , xa+1 , ya+1 , . . . , xα , yα , z ,
where w02 + w12 + . . . + wp2 = 1.
It is direct to verify that ϕ is an isometric immersion of the warped
product R2α+1 (−3) ×f S β (1) intoqthe Sasakian space form R2aα+2β+1 (−3)
Ph
with the warped function f = 12 2 2
i=1 (xi + yi ) such that R
2α+1
(−3) is
β
immersed as an invariant submanifold and S as an anti-invariant subman-
ifold in R2aα+2β+1 (−3).
It is direct to verify that ϕ : R2α+1 (−3) × S β (1) → R2aα+2β+1 (−3)
satisfies the equality: ||σ||2 = 2β |∇ ln f |2 − ∆ ln f + 1 .
where dVT and vol(N T ) are the volume element and the volume of N T ,
respectively.
The equality sign of (17.45) holds identically if and only if
(a) The warping function f is constant;
(b) (N T , gT ) is isometric to S 2α+1 (1) and it is isometrically immersed in
S 2m+1 as a totally geodesic invariant submanifold, and
(c) (N ⊥ , f 2 g⊥ ) is isometric to an open portion of the sphere S β (1) of
constant sectional curvature one and it is isometrically immersed in
S 2m+1 (1) as a totally geodesic C-totally real submanifold.
Proof. Since N T is compact, (17.36) and Hopf’s lemma imply
Z Z
||σ||2 dVT ≥ 2β |∇(ln f )|2 dVT
N T ×{q} NT (17.46)
+ 2β(2α + 1)vol(N T )
for each q ∈ N ⊥ . Obviously, inequality (17.46) implies (17.45). Clearly,
the equality sign of (17.45) holds if and only if we have
Proof. Follows from the fact that g([X, φX], ξ) 6= 0 for X ∈ Dθ1 .
Given a semi-slant submanifold M in a Sasakian manifold M̃ , we denote
by Pi the projection on the distribution Dθi for i = 1, 2. Let us put
Ti = Pi ◦ P,
where P is defined by (17.8). Then we have
φX = φP1 X + P P2 X + F P2 X, X ∈ T M.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 426
Chapter 18
Let Rm be the standard real vector m-space. For an ordered pair (p, q)
of points in Rm , let −→
pq denote the vector q − p ∈ Rm . The vector space
Rm together with the mapping Rm × Rm → Rm : (p, q) 7→ − →
pq is called an
affine m-space.
Let x(t), a ≤ t ≤ b, be a smooth curve in Rm and let Y be a vector field
on Rm . If we take an affine coordinate system {x1 , . . . , xm } and write
m
X ∂
Y = Yi and x(t) = (x1 (t), . . . , xm (t)),
i=1
∂xi
(1) ∇f X Y = f ∇X Y ,
(2) ∇X (f Y ) = X(f )Y + f ∇X Y .
427
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 428
for X, Y ∈ X(N ), where φ∗ (∇′X Y ) and −φ∗ (SX) are the tangential com-
ponents of ∇X φ∗ (Y ) and ∇X ξ, respectively.
The induced affine connection ∇′ , the (1, 1)-tensor S, the 1-form τ , and
the symmetric (0, 2)-tensor h are called the induced affine connection, the
affine shape operator, the torsion form, and the affine fundamental form
(relative to the transversal vector field ξ), respectively.
The transversal vector field ξ of φ : N → Rn+1 is called a relative
normalization (or equiaffine) if τ = 0 holds identically, i.e., ∇X ξ is tangent
to N for each X ∈ T N .
Notice that ∇′ and h depend upon the choice of transverse vector field
ξ. We consider only those hypersurfaces for which h is non-degenerate.
Interestingly, this is a property of the hypersurface N and not depend upon
the choice of transverse vector field ξ. When h is non-degenerate, it defines
a pseudo-Riemannian metric on N , called the relative metric. Throughout
this chapter, we assume that the affine fundamental form h is definite.
For tangent vectors X1 , . . . , Xn of N let (hij ) be the n × n matrix given
by hij = h(Xi , Xj ). Define an affine volume element v on N by
If we put
(Z
p )
s
f ′2 + 1
ψ(s, t) = f (s) exp ds sinh t,
0 f
(Z p ) Z s ! (18.30)
s
f ′2 + 1 f ds
f (s) exp ds cosh t, p ,
0 f 0 f′ − f ′2 + 1
then ψ satisfies
p
f f ′′ + f ′2 + 1 f ′ + f ′2 + 1
ψss = p ψs + ξ, ψst = ψt ,
f f ′2 + 1 f
p
ψtt = f f ′2 + 1 − f ′ ψs + f 2 ξ, ξ = (0, 0, 1).
Thus ψ is a realization of I ×f R as a graph surface in R3 with the metric
ds2 + f 2 (s)dt2 as the Calabi metric h and ξ as the Calabi normal.
−
Next, assume that a is a positive real number. Let Na,f be the warped
n−1 2
product manifold I ×f H (−a ) with the warped product metric
n−1
Y
2 2
ds + f (s) du2 + cosh (au2 )du3 + · · · +
2 2 2 2
cosh (auj )dun . (18.31)
2
j=2
−
Consider the immersion φ of Na,f into Rn+1 given by
(Z p )
s
f ′2 + a2
φ(s, u1 , . . . , un ) = f (s) exp ds sinh(au2 ),
0 f
n−1 n
!
Y Y
sinh(au3 ) cosh(au2 ), . . . , sinh(aun ) cosh(auj ), cosh(auj ), 0
j=2 j=2
Z s p
1 ′
− 2 0, . . . , 0, f ′2 2
f + a + f ds .
a 0
Then a direct computation yields
f f ′′ + f ′2 + a2
φss = p φs + ξ, ξ = (0, . . . , 0, 1),
f f ′2 + a2
p
f ′ + f ′2 + a2
φsuj = φuj ,
f
φui uj = a tanh(aui )φuj , 2 ≤ i < j ≤ n,
(18.32)
p j−1
Y
φuj uj ′2 2 ′ 2
= f f + a φs − f f φs + f ξ cosh2 (aui )
i=2
j−1 j−1
!
X sinh(2auk ) Y 2
−a cosh (aui ) φuk ,
2
k=2 i=k+1
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 434
φ+ +
ui uj = −a tan(aui )φuj , 2 ≤ i < j ≤ n,
p j−1
Y
φ+
uj uj
′2 2 + ′ + 2
= f f − a φs − f f φs + f ξ cos2 (aui )
i=2
j−1
X j−1
sin(2auk ) Y
+a cos2 (aui )φ+
uk , j = 2, . . . , n,
2
k=2 i=k+1
(a) if f ′ (s)2 > f 2 (s) − a2 , then I ×f H n−1 (−a2 ) can be realized as a cen-
troaffine hypersurface in Rn+1 ;
(b) if f ′ (s)2 > f (s)2 , then I ×f En−1 can be realized as a centroaffine
hypersurface in Rn+1 ;
(c) if f ′ (s)2 > f (s)2 + a2 , then I ×f S n−1 (a2 ) can be realized as a graph
hypersurface in Rn+1 .
Proof. Without loss of generality, we may assume that the open interval
− +
I contains 0. Let Na,f , Nfo and Na,f be the warped products defined as
before. Suppose that f satisfies f (s)2 > f 2 (s)−a2 . Consider the immersion
′
−
η − of Na,f into Rn+1 given by
(Z p )
s
− f ′2 − f 2 + a2
η (s, u2 , . . . , un ) = f (s) exp ds
0 f
n−1 n
!
Y Y
× sinh(au2 ), . . . , sinh(aun ) cosh(auj ), cosh(auj ), 0
j=2 j=2
Z !!
s
f ds
+ 0, . . . , 0, exp p .
0 f′ − f ′2 − f 2 + a2
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 436
− f ′2 + f f ′′ − 2f 2 + a2 −
ηss = p ηs + η − ,
f f ′2 − f 2 + a2
p
− f ′ + f ′2 − f 2 + a2 −
ηsuj = ηuj , j = 2, . . . , n,
f
ηu−i uj = a tanh(aui )ηu−j , 2 ≤ i < j ≤ n,
(18.38)
p X
j−1
− ′2 2 2 − ′ − 2 −
ηuj uj = f f − f + a ηs − f f ηs + f η cosh2 (aui )
i=2
j−1 j−1
!
X sinh(2auk ) Y
−a cosh2 (aui ) φ−
uk , j = 2, . . . , n,
2
k=2 i=k+1
−
which implies that η is a centroaffine hypersurface with (18.31) as the
−
centroaffine metric. Hence η − is a realization of Na,f as a centroaffine
hypersurface with (18.31) as the centroaffine metric.
If f satisfies f ′ (s)2 > f 2 (s), we consider the immersion
Z s p ′2 !
f − f2
η(s, u2 , . . . , un ) = f (s) exp ds 1, u2 , . . . , un , 0
0 f
Z s !!
f ds
+ 0, . . . , 0, exp p .
0 f ′ − f ′2 − f 2
A direct computation shows that η satisfies (18.38) with a = 0. Hence η
defines a centroaffine hypersurface with (18.33) as its centroaffine metric.
Thus η is a realization of Nfo as a centroaffine hypersurface with the warped
product metric as its centroaffine metric.
If f ′ (s)2 > f 2 (s) + a2 , we consider
(Z p )
s
+ f ′2 − f 2 − a2
η (s, u2 , . . . , un ) = f (s) exp ds
0 f
n−1
Y Y n
× sin(au2 ), . . . , sin(aun ) cos(auj ), cos(auj ), 0
j=2 j=2
Z !!
s
f ds
+ 0, . . . , 0, exp p .
0 f′ − f ′2 − f 2 − a2
+
Then η + is a realization of Na,f as a centroaffine hypersurface with the
warped product metric as its centroaffine metric.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 437
The following examples illustrate that all of the results obtained in this
section are best possible.
Example 18.1. Let M = N1 ×cos s N2 be the warped product of the open
interval N1 = (−π, π) and an open portion N2 of the unit (n − 1)-sphere
S n−1 (1) equipped with the warped product metric:
n−1
!
Y
2 2 2 2 2 2 2
h = ds + cos s du2 + cos u2 du3 + · · · + cos uj dun . (18.52)
j=2
Consider the immersion of N into the affine (n + 1)-space defined by
n−1 n
!
Y Y
2
sin s, sin u2 cos s, . . . , sin un cos s cos uj , cos s cos uj .
j=2 j=2
Then N is a centroaffine elliptic hypersurface whose centroaffine metric is
(18.52) and it satisfies T # = 0. Moreover, the warping function f = cos s
satisfies (∆f )/f = 1 = εn1 . Hence, this centroaffine hypersurface satisfies
the equality case of (18.39) identically. Consequently, the estimate given
in Theorem 18.1 is optimal for centroaffine elliptic hypersurfaces.
Example 18.2. Let N = R×cosh s H n−1 (−1) be the warped product of the
real line and the unit hyperbolic space H n−1 (−1) equipped with warped
product metric:
n−1
!
Y
2 2 2 2 2 2 2
h = ds + cosh s du2 + cosh u2 du3 + · · · + cosh uj dun . (18.53)
j=2
Consider the immersion of N into the affine (n + 1)-space defined by
n−1 n
!
Y Y
sinh s, sinh u2 cosh s, . . . , sinh un cosh s cosh2 uj , cosh s cosh uj .
j=2 j=2
Then N is a centroaffine hyperbolic hypersurface whose centroaffine metric
is (18.53) and it satisfies T # = 0. Moreover, f = cosh s satisfies
∆f
= −1 = εn1 .
f
Thus, this centroaffine hypersurface satisfies the equality case of (18.39)
identically. Consequently, the estimate given in Theorem 18.1 is optimal
for centroaffine hyperbolic hypersurfaces as well.
Remark 18.2. Example 18.1 shows that conditions “∆f ≤ 0” in Corollary
18.1 and “harmonicity” in Corollary 18.3 are both necessary. Example 18.1
implies that condition “N1 is a compact Riemannian manifold” in Corollary
18.4 is necessary. Example 18.2 shows that condition “(∆f )/f < − dim N1 ”
in Corollary 18.2 is sharp.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 442
For warped product graph hypersurfaces we have the following results from
[Chen (2005c)].
Remark 18.3. Example 18.3 implies that the condition “∆f < 0” given
in Corollary 18.5 is optimal.
∂ ∂
where φt = φ∗ ( ∂t ), φx = φ∗ ( ∂x ) are tangent vector fields of M .
If we choose the transversal vector field ξ to be φ, then it follows
from (18.3) and (18.59) that the induced affine fundamental form h on
the Robertson-Walker spacetime M 2 (0, f ) is given by
h = −dt2 + f 2 (t)dx2 , (18.60)
2
which is exactly the Lorentzian metric (18.57) on M (0, f ). This is done by
comparing the component of ξ from (18.3) and of φ from (18.59). Thus the
embedding defined by (18.58) is a realization of M 2 (0, f ) as a centroaffine
hypersurface. This proves statement (a).
(ii) Consider the Robertson-Walker spacetime M n (1, f ) = I ×f S n−1 with
the warped product Lorentzian metric:
( n−1
)
Y
2 2 2 2 2 2 2
g = −dt + f (t) du2 + cos u2 du3 + · · · + cos uj dun . (18.61)
j=2
n n+1
Define η : M (1, f ) → R by
√
Rt f (f ′ + f ′ 2 +f 2 +1)
dt
η(t, u2 , . . . , un ) = 0, . . . , 0, e 0 f 2 +1
!
R t √f ′2 +f 2 +1 n−1
Y n
Y
dt
+ f (t)e 0 f sin u2 , . . . , sin un cos uj , cos uj , 0 .
j=2 j=2
By a long direct computation we obtain
f ′2 + f f ′′ + 2f 2 + 1
ηtt = p ηt − η,
f f ′2 + f 2 + 1
p
f ′ + f ′2 + f 2 + 1
ηtuj = ηuj , j = 2, . . . , n,
f
ηui uj = −(tan ui )ηuj , 2 ≤ i < j ≤ n,
j−1
! j−1
!
Y f + f3 Y
2 2 2
ηuj uj = f cos ui η − p cos ui ηt
i=2 f ′ + f ′2 + f 2 + 1 i=2
j−1 j−1
!
X sin(2uk ) Y 2
+ cos ui ηuk , 2 ≤ j ≤ n.
2
k=2 i=k+1
Comparing this with (18.3) in the same way as above shows that with ξ = η
the induced affine fundamental form on the Robertson-Walker spacetime
M n (1, f ) is given by
( n−1
)
Y
2 2 2 2 2 2 2
h = −dt + f (t) du2 + cos u2 du3 + · · · + cos uj dun ,
j=2
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 446
!
R t √f ′ 2 +f 2 −1 n−1
Y n
Y
dt
+ f (t)e 0 f sinh u2 , . . . , sinh un cosh uj , cosh uj , 0 .
j=2 j=2
After comparing (18.3) with (18.62) in a same way as before shows that
with ξ = ψ the induced affine fundamental form on the manifold M n (−1, f )
is given by
n−1
Y
h = −dt2 + f 2 (t) du22 + cosh2 u2 du23 + · · · + cosh2 uj du2n ,
j=2
n
which is exactly the Lorentzian metric on M (−1, f ). Therefore we obtain a
realization of the Robertson-Walker spacetime M n (−1, f ) as a centroaffine
hypersurface. This proves statement (c).
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 447
f f ′′ + f ′2 + 1
ψtt = p ψt − ξ,
f f ′2 + 1
p
f ′ + f ′2 + 1 (18.66)
ψtu = ψu ,
f
p
ψuu = f f ′ − f ′2 + 1 ψt + f 2 ξ
with ξ = (0, 0, 1). It follows from (18.3) and (18.66) that the induced Calabi
metric h is exactly the Lorentzian metric g = −dt2 + f 2 (t)du2 . This shows
that every 2-dimensional Robertson-Walker spacetime can be realized as a
graph surface in R3 . This proves statement (a).
(B) Consider the Robertson-Walker spacetime M n (1, f ) = I ×f S n−1
equipped with Lorentzian metric:
( n−1
)
Y
2 2 2 2 2 2 2
g = −dt + f (t) du2 + cos u2 du3 + · · · + cos uj dun . (18.67)
j=2
Rt p
{ f ′ 2 + 1/f }dt
+ f e− 0 sin u2 , sin u3 cos u2 , . . . ,
n−1 n
!
Y Y
sin un cos uj , cos uj , 0
j=2 j=2
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 449
ζ(t, u2 , . . . , un ) = u2 , . . . , un ,
(18.70)
n
X Z t Z t
dt f (t)
u2j + , dt .
j=2 0 f (t)f ′ (t) 0 2f ′ (t)
It follows from (18.3) and (18.70) that the induced Calabi metric h via
ζ is exactly the Lorentzian metric (18.69) on M n (0, f ). Consequently, ζ
is a realization of M n (0, f ) as a graph hypersurface in Rn+1 . This proves
statement (b.2).
(C) Finally, assume that f ′2 (t) > 1 holds for each t ∈ I. Let us consider
the Robertson-Walker spacetime
M n (−1, f ) = I ×f H n−1
with the Lorentzian(metric: )
n−1
Y
g = −dt2 + f 2 (t) du22 + cosh2 u2 du23 + · · · + cosh2 uj du2n . (18.72)
j=2
Define φ : M n (−1, f ) → Rn+1 by !
Z t p
′
φ(t, u2 , . . . , un ) = 0, . . . , 0, f ′2
f − 1 − f dt
0
f
+ Rt p sinh u2 , sinh u3 cosh u2 , . . . ,
{ f ′2 − 1/f }dt
e 0
n−1 n
!
Y Y
sinh un cosh uj , cosh uj , 0 .
j=2 j=2
Then a straightforward long computation yields
1 − f f ′′ − f ′2
φtt = p φt − ξ,
f f ′2 − 1
p
f ′ − f ′2 − 1
φtuj = φuj ,
f
φui uj = (tanh ui )φuj ,
( ) j−1 (18.73)
f Y
φuj uj = p φt + f 2 ξ cosh2 ui
f ′ − f ′2 − 1 i=2
j−1
( j−1
)
X sinh 2uk Y
2
− cosh ui φuk ,
2
k=2 i=k+1
for 2 ≤ i < j ≤ n with ξ = (0, . . . , 0, 1).
It follows from (18.3) and (18.73) that the induced Calabi metric h on
the Robertson-Walker spacetime M n (−1, f ) via the map φ is exactly the
Lorentzian metric given in (18.72) on M n (−1, f ). Therefore φ gives rise to
a realization of M n (−1, f ) as a graph hypersurface in Rn+1 . This proves
statement (b.3) of the theorem. Consequently, we complete the proof of
the theorem.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 451
Bibliography
451
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 452
Bibliography 453
Bibliography 455
Bibliography 457
Bibliography 459
Bibliography 461
Dillen, F., Opozda, B., Verstraelen, L. and Vrancken, L. (1987). On totally real
3-dimensional submanifolds of the nearly Kaehler 6-sphere, Proc. Amer.
Math. Soc. 99, 741–749.
Dillen, F., Opozda, B., Verstraelen, L. and Vrancken, L. (1988). On totally real
surfaces of the nearly Kaehler 6-sphere, Geom. Dedicata 27, 325–334.
Dillen, F., Verstraelen, L. and Vrancken, L. (1990). Classification of totally real
3-dimensional submanifolds of S 6 (1) with K ≥ 1/16, J. Math. Soc. Japan
42, 565–584.
Dillen, F. and Vrancken, L. (1994). Calabi-type composition of affine spheres,
Differential Geom. Appl. 4, 303–328.
Dillen, F. and Vrancken, L. (1996). Totally real submanifolds in S 6 (1) satisfying
Chen’s equality, Trans. Amer. Math. Soc. 348, 1633–1646.
Dobarro, F. and Lami Dozo, E. (1987). Scalar curvature and warped products of
Riemann manifolds, Trans. Amer. Math. Soc. 303, 161–168.
Dobarro, F. and Ünal, B. (2005). Curvature of multiply warped products, J.
Geom. Phys. 55, 75–106.
Dobarro, F. and Ünal, B. (2007). About curvature, conformal metrics and warped
products, J. Phys. A 40, no. 46, 13907–13930.
do Carmo, M. and Dajczer, M. (1983) Rotation hypersurfaces in spaces of con-
stant curvature, Trans. Amer. Math. Soc. 277, 685–709.
Dragomir, S. and Ornea, L. (1998). Locally conformal Kähler geometry, (Progress
in Mathematics, 155. Birkhäuser Boston, Inc., Boston, MA.)
Duggal, K. L. and Sahin, B. (2010). Differential geometry of lightlike submani-
folds, (Birkhäuser Verlag, Basel).
Duggal, K. L. and Sharma, R. (1999). Symmetries of spacetimes and Riemannian
manifolds, Kluwer Academic Publishers, Dordrecht, 1999.
Dwivedi, M. K. and Kim, J.-S. (2012). Chen-Tripathi inequality for warped prod-
uct submanifolds of S-space forms, An. Stiint. Univ. Al. I. Cuza Iasi. Mat.
(N.S.) 58, 195–208.
Ejiri, N. (1981). Totally real submanifolds in a 6-sphere, Proc. Amer. Math. Soc.
83, 759–763.
Erbacher, J. (1971). Reduction of the codimension of an isometric immersion, J.
Differential Geom. 5, 333–340.
Eschenburg, J.-H. and Tribuzy, R. (1993). Existence and uniqueness of maps into
affine homogeneous spaces, Rend. Sem. Mat. Univ. Padova, 89, 11–18.
Escobales, R. (1975). Riemannian submersions with totally geodesic fibers, J.
Differential Geom. 10, 253–276.
Escobales, R. (1978). Riemannian submersions from complex projective spaces,
J. Differential Geom. 13, 93–107.
Etayo, F. (1998). On quasi-slant submanifolds of an almost Hermitian manifold,
Publ. Math. Debrecen 53, 217–223.
Faghfouri, M. and Ghaffarzadeh, N. (2015). On doubly warped product subman-
ifolds of generalized (κ, µ)-space forms, Afr. Mat. 26, 1443–1455.
Faghfouri, M. and Majidi, A. (2015). On doubly warped product immersions. J.
Geom. 106, 243–254.
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 462
Bibliography 463
Bibliography 465
Bibliography 467
Bibliography 469
Bibliography 471
General Index
Cn s∗ , 79 φ-section, 100
A, 28 σ, 26
CHsn (4c), CH n (4c), 118 σ-submanifold, 364
CPsn (4c), CP n (4c), 118 r
σij , 28
D, 28 τ , 14
F , 125 | · |, 2
H, 29 h, 180
H(M ), 415 p, 180
Hsk (c), 16 Cm+1
∗ , 270
H P , 384 Cm+1
1∗ , 271
P , 125 C∗ , 233
Pcr (B), 387 Rm , 427
Q(u, v), 12 R∗ , 271
R, 10 Em ∗ , 234
Ric, 14 En s, 4
Ssk (c), 16 R2n+1 , 108
Sh , 32 A, 103
#S(n), 326 F(M ), 4
∆, 17 O, 361
Ω, 116 S(n), 326
Ωαβ , 46 X(M ), 5
¯ 31
∇σ, Åπ , 103
¯ j σ, 31
∇
Ăπ , 103 Adapted pseudo-orthonormal frame,
δ-invariant, 326 127
R2n+1 (−3), 108 Affine connection, 5
LC, 4 Affine space, 427
L, 48 Affine volume element, 428
ω i , 46 almost complex structure
ωαβ , 46 compatible, 141
k · k, 2 Almost Hermitian manifold
473
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 474
Author Index
481
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 482
19, 20, 30, 40, 58, 65, 71, 72, 74, 96, 460
78–80, 89, 96, 99, 119, 124, 127, Dillen, F., vii, xx, 65, 78, 212, 213,
129–131, 133, 134, 136, 140, 141, 337, 355, 357, 358, 367, 369, 381,
143, 144, 146–148, 151, 153, 158, 382, 431, 455, 458, 460, 461
162, 165, 170, 171, 173, 174, 177– Dimitric, I., xxiv
179, 188, 190–194, 196, 199, 201– Djorić, M., 369, 452
204, 206, 207, 209, 212, 213, 215, do Carmo, M., 216, 217, 221, 461
217, 222, 225, 229, 231, 235, 238, Dobarro, F., 60, 461
247, 252–254, 258, 259, 275, 282, Dombrowski, P., xiv, xx
283, 296, 299–301, 308, 315, 316, Dragomir, S., 174, 201, 254–256,
318, 321, 324, 326, 328, 329, 331– 453, 461
333, 336–338, 340, 341, 344–351, Duggal, K. L., 110, 230, 453, 461
354, 355, 357, 358, 366, 367, 381, Dwivedi, M. K., 461
382, 393, 397–400, 402, 406–408,
432, 435, 437, 439, 440, 442–444, Easley, K. L., 452
447, 451–459, 471, 472 Ehrlich, P. E., 452
Chen, C.-H., 459 Einstein, A., ix, xi, xvi, xviii, xx,
Chern, S. S., v, vii–ix, xi, 360, 459 xxiii, 1, 14, 15, 23, 47, 52, 54, 55,
Cherubini, C., 98, 459 78, 92, 96–98, 164, 325, 359, 360
Chi, A. Y., 374, 471 Ejiri, N., 366, 367, 369, 461
Cho, K. S., 472 Ellis, G, F. R., 98, 463
Chojnacka-Dulas, J., 460 Erbacher, J., 36, 221, 461
Christoffel, E. B., 6, 460 Eschenburg, J.-H., 34, 129, 153, 390,
Clarke, C. J. S., 26, 460 461
Codazzi, D., 31, 39, 70, 87, 88, 90, Escobales, R., 106, 107, 461
127, 128, 134, 138, 139, 153, 164, Etayo, F., 171, 175, 461
184, 189, 203–205, 221, 278, 283, Ezentas, R., 452, 466
304, 305, 368, 379, 390, 391, 396,
411, 418, 421, 429 Faghfouri, M., 71, 214, 461
Colares, A. G., 94, 451 Fernández, L. M., 425, 454
Crittenden, R. J., 34 Fernández, M., 425, 454
Fernández-Lápez, M., 73, 77, 78,
Dajczer, M., 216, 217, 221, 332, 460, 462
461 Fladt, K., xi
de Lima, H. F., 94, 451 Florit, L. A., 332, 460
de Rham, G., 61, 460 Freudenthal, H., vii, x, xi
de Sitter, W., 16, 81 Friedmann, A., xx, 81
De, A., 453 Friedrich, T., 361, 462
De, U. C., 96, 460, 465
Deahna, F., 460 Günes, R., 372, 373, 469
Decruyenaere, F., 78, 460 Gödel, K., viii
Defever, F., 460 Gadea, P. M., 385, 387, 462
Deng, S., 384, 460, 463 Garay, O. J., xxiv, 173, 452, 458
Deprez, J., 65 Garcı́a-Rı́o, E., 58, 59, 73, 74, 77,
Deshmukh, S., 341, 344–348, 451 78, 453, 462
Deszcz, R., xii, xiii, xvii, xviii, xx, Gauduchon, P., 282, 453
March 21, 2017 15:26 Diff. Geometry of Warped Product Manifolds-10419 ws-book9x6 page 483
Tazawa, Y., 144, 162, 177, 178, 459, 147, 148, 150, 163, 185, 188, 197,
470 203, 229, 230, 239, 243, 249, 250,
Thakur, M., 462 301, 302, 307, 352, 368, 371, 372,
Thurston, W. P., xv, xvi, xix, 1 379, 380, 390, 410, 472
Tits, J. H., xiv Wettstein, B., 34, 153, 472
Tiwari, S. K., 469 Weyl, H., xii, xvii–xix, 58, 255, 472
Tojeiro, R., 460, 470 Wilking, B., 107, 472
Tribuzy, R., 34, 129, 153, 390, 461 Wolf, J., 360, 472
Tripathi, M. M., 471 Wong, B. R., 466, 471
Tsukamoto, Y., 359, 471 Woodward, L. M., 366, 453
Wu, B. Q., 206, 459
Ucci, J., 106, 471 Wylie, W., 463
Udagawa, S., 471
Uddin, S., 247, 252–254, 372, 374, Yüksel Perkas, S., 472
413, 414, 452, 459, 463, 464, 466, Yano, K., v, xix, 19, 206, 359, 409,
471 453, 459, 472
Unal, B., 60, 73, 77, 78, 212, 461, Yoon, D. W., 464, 472
462, 469, 471
Urbano, F., 133, 454 Zafindratafa, G., 460
Zhang, P., 472
Vázquez-Lorenzo, R., 58, 59, 74, 453 Zhong, H. H., 369, 472
Vaisman, I., 471 Zhong, Z., 472
Valli, G., 164, 469
Van der Veken, J., xxiv, 89, 337,
458, 459
Vanhecke, L., 228, 364, 459, 471, 472
Verheyen, P., 188, 459, 462
Veronese, G., 124
Verstraelen, L., vii, xi, xvii, xx, xxi,
xxiv, 19, 78, 80, 357, 358, 367,
369, 381, 382, 455, 458, 460–462,
467
Vilcu, G.-E., 254, 470, 472
Vitório, F., 466
Vlachos, T., 460
Vrancken, L., 78, 151, 153, 337, 355,
357, 358, 366, 367, 369, 381, 382,
431, 452, 453, 458–461, 468