Professional Documents
Culture Documents
PDF An Invitation To Alexandrov Geometry Cat 0 Spaces Stephanie Alexander Ebook Full Chapter
PDF An Invitation To Alexandrov Geometry Cat 0 Spaces Stephanie Alexander Ebook Full Chapter
https://textbookfull.com/product/alexandrov-geometry-preliminary-
version-no-1-alexander-s/
https://textbookfull.com/product/an-invitation-to-geomathematics-
willi-freeden/
https://textbookfull.com/product/an-invitation-to-health-dianne-
r-hales/
https://textbookfull.com/product/an-invitation-to-feminist-
ethics-hilde-lindemann/
An Invitation to Computational Homotopy 1st Edition
Graham Ellis
https://textbookfull.com/product/an-invitation-to-computational-
homotopy-1st-edition-graham-ellis/
https://textbookfull.com/product/an-invitation-to-model-
theory-1st-edition-jonathan-kirby/
https://textbookfull.com/product/spaces-an-introduction-to-real-
analysis-tom-lindstrom/
https://textbookfull.com/product/essentials-of-meteorology-an-
invitation-to-the-atmosphere-c-donald-ahrens/
https://textbookfull.com/product/geometry-and-martingales-in-
banach-spaces-1st-edition-wojbor-a-woyczynski/
SPRINGER BRIEFS IN MATHEMATICS
Stephanie Alexander
Vitali Kapovitch
Anton Petrunin
An Invitation
to Alexandrov
Geometry
CAT(0) Spaces
SpringerBriefs in Mathematics
Series Editors
Nicola Bellomo, Torino, Italy
Michele Benzi, Pisa, Italy
Palle Jorgensen, Iowa City, USA
Tatsien Li, Shanghai, China
Roderick Melnik, Waterloo, Canada
Otmar Scherzer, Linz, Austria
Benjamin Steinberg, New York City, USA
Lothar Reichel, Kent, USA
Yuri Tschinkel, New York City, USA
George Yin, Detroit, USA
Ping Zhang, Kalamazoo, USA
Anton Petrunin
An Invitation to Alexandrov
Geometry
CAT(0) Spaces
123
Stephanie Alexander Anton Petrunin
Department of Mathematics Department of Mathematics
University of Illinois Pennsylvania State University
Urbana, IL, USA University Park, PA, USA
Vitali Kapovitch
Department of Mathematics
University of Toronto
Toronto, ON, Canada
Mathematics Subject Classification (2010): 53C23, 53C20, 53C45, 53C70, 97G10, 51F99, 51K10
This Springer imprint is published by the registered company Springer Nature Switzerland AG.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
The idea that the essence of curvature lies in a condition on quadruples of points
apparently originated with Abraham Wald. It is found in his publication on
“coordinate-free differential geometry” [66] written under the supervision of Karl
v
vi Preface
Menger; the story of this discovery can be found in [43]. In 1941, similar definitions
were rediscovered independently by Alexandr Danilovich Alexandrov; see [7]. In
Alexandrov’s work the first fruitful applications of this approach were given.
Mainly:
• Alexandrov’s embedding theorem—metrics of nonnegative curvature on the
sphere, and only they, are isometric to closed convex surfaces in Euclidean
3-space.
• Gluing theorem, which tells when the sphere obtained by gluing of two disks
along their boundaries, has nonnegative curvature in the sense of Alexandrov.
These two results together gave a very intuitive geometric tool for studying
embeddings and bending of surfaces in Euclidean space, and changed this subject
dramatically. They formed the foundation of the branch of geometry now called
Alexandrov geometry.
The study of spaces with curvature bounded above started later. The first paper
on the subject was written by Alexandrov; it appeared in 1951; see [8]. It was based
on the work of Herbert Busemann, who studied spaces satisfying a weaker con-
dition [24].
Yurii Grigorievich Reshetnyak proved fundamental results about general spaces
with curvature bounded above, the most important of which is his gluing theorem.
An equally important theorem is the Hadamard–Cartan theorem (globalization
theorem). These theorems and their history are discussed in chapters 2 and 3.
Surfaces with upper curvature bounds were studied extensively in the 50s and
60s, and are by now well understood; see the survey [57] and the references therein.
Alexandrov geometry can use “back to Euclid” as a slogan. Alexandrov spaces are
defined via axioms similar to those given by Euclid, but certain equalities are
changed to inequalities. Depending on the sign of the inequalities, we get
Alexandrov spaces with curvature bounded above or curvature bounded below.
The definitions of the two classes of spaces are similar, but their properties and
known applications are quite different.
Consider the space M4 of all isometry classes of 4-point metric spaces. Each
element in M4 can be described by 6 numbers—the distances between all 6 pairs of
its points, say ‘i;j for 1 6 i \ j 6 4 modulo permutations of the index set (1, 2, 3, 4).
These 6 numbers are subject to 12 triangle inequalities; that is,
holds for all i, j and k, where we assume that ‘j;i ¼ ‘i;j and ‘i;i ¼ 0.
Preface vii
M4
0.0.1. Exercise. Prove the latter statement.
One of the components will be denoted by P 4 and the other by N 4 . Here P and
N stand for positive and negative curvature because spheres have no quadruples of
type N 4 and hyperbolic space has no quadruples of type P 4 .
A metric space, with length metric, that has no quadruples of points of type P 4
or N 4 respectively is called an Alexandrov space with nonpositive or nonnegative
curvature, respectively.
Here is an exercise, solving which would force the reader to rebuild a consid-
erable part of Alexandrov geometry.
0.0.2. Advanced exercise. Assume X is a complete metric space with length
metric, containing only quadruples of type E 4 . Show that X is isometric to a convex
set in a Hilbert space.
In fact, it might be helpful to spend some time thinking about this exercise
before proceeding.
In the definition above, instead of Euclidean space one can take hyperbolic space
of curvature 1. In this case, one obtains the definition of spaces with curvature
bounded above or below by 1.
To define spaces with curvature bounded above or below by 1, one has to take
the unit 3-sphere and specify that only the quadruples of points such that each of the
four triangles has perimeter less than 2 p are checked. The latter condition could
be considered as a part of the spherical triangle inequality.
We want to thank David Berg, Richard Bishop, Yurii Burago, Maxime Fortier
Bourque, Sergei Ivanov, Michael Kapovich, Bernd Kirchheim, Bruce Kleiner,
Nikolai Kosovsky, Greg Kuperberg, Nina Lebedeva, John Lott, Alexander Lytchak,
Dmitri Panov, Stephan Stadler, Wilderich Tuschmann, and Sergio Zamora Barrera
for a number of discussions and suggestions.
We thank the mathematical institutions where we worked on this material,
including BIRS, MFO, Henri Poincaré Institute, University of Colone, Max Planck
Institute for Mathematics.
The first author was partially supported by the Simons Foundation grant
#209053. The second author was partially supported by a Discovery grant from
NSERC and by the Simons Foundation grant #390117. The third author was par-
tially supported by the NSF grant DMS 1309340 and the Simons Foundation
#584781.
ix
Contents
1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Metric spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Geodesics, triangles, and hinges . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Length spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Model angles and triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.6 Angles and the first variation . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.7 Space of directions and tangent space . . . . . . . . . . . . . . . . . . . . . 11
1.8 Hausdorff convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.9 Gromov–Hausdorff convergence . . . . . . . . . . . . . . . . . . . . . . . . . 14
2 Gluing theorem and billiards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1 The 4-point condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Thin triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Reshetnyak’s gluing theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Reshetnyak puff pastry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5 Wide corners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.6 Billiards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3 Globalization and asphericity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1 Locally CAT spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Space of local geodesic paths . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Globalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Polyhedral spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.5 Flag complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.6 Cubical complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.7 Exotic aspherical manifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.8 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
xi
xii Contents
4 Subsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.1 Motivating examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 Two-convexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 Sets with smooth boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4 Open plane sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.5 Shefel’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.6 Polyhedral case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.7 Two-convex hulls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.8 Proof of Shefel’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.9 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Appendix: Semisolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Chapter 1
Preliminaries
In this chapter we fix some conventions and recall the main definitions. The chapter
may be used as a quick reference when reading the book.
To learn background in metric geometry, the reader may consult the book of
Dmitri Burago, Yuri Burago, and Sergei Ivanov [20].
The distance between two points x and y in a metric space X will be denoted by
|x − y| or |x − y|X . The latter notation is used if we need to emphasize that the
distance is taken in the space X .
The function
dist x : y → |x − y|
are called, respectively, the open and the closed balls of radius R with center x.
Again, if we need to emphasize that these balls are taken in the metric space X ,
we write
B(x, R)X and B[x, R]X .
A metric space X is called proper if all closed bounded sets in X are compact.
This condition is equivalent to each of the following statements:
1. For some (and therefore any) point p ∈ X and any R < ∞, the closed ball
B[ p, R] ⊂ X is compact.
2. The function dist p : X → R is proper for some (and therefore any) point p ∈ X ;
that is, for any compact set K ⊂ R, its inverse image { x ∈ X : | p − x|X ∈ K }
is compact.
1.1.1 . Exercise. Let K be a compact metric space and
f:K →K
1.2 Constructions
Product space. Given two metric spaces U and V, the product space U × V is
defined as the set of all pairs (u, v) where u ∈ U and v ∈ V with the metric defined
by formula
|(u 1 , v 1 ) − (u 2 , v 2 )|U ×V = |u 1 − u 2 |2U + |v 1 − v 2 |2V .
Cone. The cone V = Cone U over a metric space U is defined as the metric space
whose underlying set consists of equivalence classes in [0, ∞) × U with the equiv-
alence relation “∼” given by (0, p) ∼ (0, q) for any points p, q ∈ U, and whose
metric is given by the cosine rule
|( p, s) − (q, t)|V = s 2 + t 2 − 2 · s · t · cos α,
are isometric.
Geodesic. Let X be a metric space and I be a real interval. A globally isometric map
γ : I → X is called a geodesic1 ; in other words, γ : I → X is a geodesic if
|γ(s) − γ(t)|X = |s − t|
In general, a geodesic between p and q need not exist and if it exists, it need not
be unique. However, once we write [ pq] we mean that we have made a choice of
geodesic.
A metric space is called geodesic if any pair of its points can be joined by a
geodesic.
A geodesic path is a geodesic with constant-speed parametrization by [0, 1]. Given
a geodesic [ pq], we denote by path[ pq] the corresponding geodesic path; that is,
parametrization of a local geodesic by the unit interval [0, 1] is called a local geodesic
path.
Triangle. For a triple of points p, q, r ∈ X , a choice of a triple of geodesics
([qr ], [r p], [ pq]) will be called a triangle; we will use the short notation
[ pqr ] = ([qr ], [r p], [ pq]).
Again, given a triple p, q, r ∈ X there may be no triangle [ pqr ] simply because
one of the pairs of these points cannot be joined by a geodesic. Also, many different
triangles with these vertices may exist, any of which can be denoted by [ pqr ].
However, if we write [ pqr ], it means that we have made a choice of such a triangle;
that is, we have fixed a choice of the geodesics [qr ], [r p], and [ pq].
The value
| p − q| + |q − r | + |r − p|
A curve is defined as a continuous map from a real interval to a space. If the real
interval is [0, 1], then the curve is called a path.
Directly from the definition, it follows that if a path α : [0, 1] → X connects two
points x and y (that is, if α(0) = x and α(1) = y), then
length α |x − y|.
1.4 Length spaces 5
length α < |x − y| + ε,
then X is called a length space and the metric on X is called a length metric.
If f : X̃ → X is a covering, then a length metric on X can be lifted to X̃ by
declaring
lengthX̃ γ = lengthX ( f ◦ γ)
for any curve γ in X̃ . The space X̃ with this metric is called the metric cover of X .
Note that any geodesic space is a length space. As can be seen from the following
example, the converse does not hold.
1.4.3 . Exercise. Give an example of a complete length space for which no pair of
distinct points can be joined by a geodesic.
|x − z| = |y − z| = 1
2
· |x − y|.
|x − z|, |y − z| 1
2
· |x − y| + ε.
2 Note that while this notation slightly conflicts with the previously defined notation for distance on
a general metric space, we will usually work with ambient length spaces where the meaning will
be unambiguous.
6 1 Preliminaries
Proof. Let X be a locally compact length space. Given x ∈ X , denote by ρ(x) the
supremum of all R > 0 such that the closed ball B[x, R] is compact. Since X is
locally compact,
It is sufficient to show that ρ(x) = ∞ for some (and therefore any) point x ∈ X .
Indeed, X is a length space; therefore for any ε > 0, the set B[x, ρ(x) − ε] is a
compact ε-net in B. Since B is closed and hence complete, it must be compact.
Next we claim that
➍ |ρ(x) − ρ(y)| |x − y|X for any x, y ∈ X ; in particular ρ : X → R is a con-
tinuous function.
Indeed, assume the contrary; that is, ρ(x) + |x − y| < ρ(y) for some x, y ∈ X .
Then B[x, ρ(x) + ε] is a closed subset of B[y, ρ(y)] for some ε > 0. Then compact-
ness of B[y, ρ(y)] implies compactness of B[x, ρ(x) + ε], a contradiction.
Set ε = min { ρ(y) : y ∈ B }; the minimum is defined since B is compact. From
➋, we have ε > 0.
Choose a finite 10ε -net {a1 , a2 , . . . , an } in B. The union W of the closed balls
B[ai , ε] is compact. Clearly B[x, ρ(x) + 10ε ] ⊂ W . Therefore B[x, ρ(x) + 10ε ] is
compact, a contradiction.
1.4.7 . Exercise. Construct a geodesic space that is locally compact, but whose
completion is neither geodesic nor locally compact.
Let X be a metric space and p, q, r ∈ X . Let us define the model triangle [ p̃q̃ r̃ ]
(briefly, [ p̃q̃ r̃ ] = ˜ ( pqr )E2 ) to be a triangle in the plane E2 with the same side
lengths; that is,
In the same way we can define the hyperbolic and the spherical model triangles
˜ ( pqr )H2 , ˜ ( pqr )S2 in the hyperbolic plane H2 and the unit sphere S2 . In the latter
case the model triangle is said to be defined if in addition
| p − q| + |q − r | + |r − p| < 2 · π.
In this case the model triangle again exists and is unique up to an isometry
of S2 .
If [ p̃q̃ r̃ ] = ˜ ( pqr )E2 and | p − q|, | p − r | > 0, the angle measure of [ p̃q̃ r̃ ] at p̃
q
will be called the model angle of the triple p, q, r and will be denoted by ˜ ( p r )E 2 .
q q
In the same way we define ˜ ( p r )H2 and ˜ ( p r )S2 ; in the latter case we assume in
addition that the model triangle ˜ ( pqr )S2 is defined.
q
We may use the notation ˜ ( p r ) if it is evident which of the model spaces H2 , E2 ,
or S is meant.
2
8 1 Preliminaries
x
1.5.1 . Alexandrov’s lemma. Let p, x, y, z be distinct points
in a metric space such that z ∈ ]x y[. Then the following expr-
essions for the Euclidean model angles have the same sign: z
p p
(a)
˜ (x y ) −
˜ (x z ), y
p p
˜ (z x ) +
(b) ˜ (z y ) − π.
Moreover, p
˜ ( p xy)
˜ ( p zx) +
˜ ( p zy),
with equality if and only if the expressions in (a) and (b) vanish.
The same holds for the hyperbolic and spherical model angles, but in the latter
case one has to assume in addition that
| p − z| + | p − y| + |x − y| < 2 · π.
Proof. Consider the model triangle [x̃ p̃ z̃] = ˜ (x pz). Take a point ỹ on the extension
of [x̃ z̃] beyond z̃ so that |x̃ − ỹ| = |x − y| (and therefore |x̃ − z̃| = |x − z|).
Since increasing the opposite side in a plane triangle
increases the corresponding angle, the following expre-
ỹ
ssions have the same sign:
p̃ p
(i) [x̃ ỹ ] −
˜ (x y ),
z˜
(ii) | p̃ − ỹ| − | p − y|,
p̃ p
(iii) [z̃ ỹ ] −
˜ (z y ).
x̃
Since p˜
p̃ p̃
[x̃ ỹ ] = [x̃ z̃ ] =
˜ (x zp)
and
p̃
[z̃ ỹ ] = π − [z̃ x̃p̃ ] = π −
˜ (z xp),
Therefore
˜ ( p zx) +
˜ ( p zy) = [ p̃ z̃x̃ ] + [ p̃ z̃ỹ ] =
= [ p̃ x̃ỹ ]
˜ ( p xy).
➊ [ p xy ] == lim
˜ ( p x̄ȳ)E2 ,
def
x̄, ȳ→ p
where x̄ ∈ ] px] and ȳ ∈ ] py]. (The angle [ p xy ] is defined if the limit exists.)
The value under the limit can be calculated from the cosine law:
1.6.1 . Lemma. For any three points p, x, y in a metric space the following inequal-
ities
|˜( p xy)S2 −
˜ ( p xy)E2 | | p − x| · | p − y|,
➋
|˜( p xy)H2 −
˜ ( p xy)E2 | | p − x| · | p − y|
˜ ( p xy)H2
˜ ( p xy)E2
˜ ( p xy)S2 .
Therefore
˜ ( p xy)S2 −
0 ˜ ( p xy)H2
˜ ( p xy)S2 +
˜ (x yp)S2 +
˜ (y xp)S2 −
˜ ( p xy)H2 −
˜ (x yp)H2 −
˜ (y xp)H2 =
= area ˜ ( px y)S2 + area ˜ ( px y)H2 .
10 1 Preliminaries
0 area ˜ ( px y)H2
area ˜ ( px y)S2
| p − x| · | p − y|.
the number such that γ̃2 (s) ∈ [γ̃1 (t) γ̃3 (τ )]. Clearly s max{t, τ }, so t, τ , s may
be taken sufficiently small.
Proof. Take a sufficiently small ε > 0. For all sufficiently small t > 0, we have
|geod[q p] (t/ε) − geod[q x] (t)| t
ε
· 1 + ε2 − 2 · ε · cos α + o(t)
t
ε
− t · cos α + t · ε.
for any fixed ε > 0 and all sufficiently small t. Hence the result.
Let X be a metric space with defined angles for all hinges. Fix a point p ∈ X .
Consider the set S p of all nontrivial geodesics that start at p. By 1.6.2, the triangle
inequality holds for on S p , so (S p , ) forms a pseudometric space; that is,
satisfies all the conditions of a metric on S p , except that the angle between distinct
geodesics might vanish.
The metric space corresponding to (S p , ) is called the space of geodesic direc-
tions at p, denoted by p or p X . Elements of p are called geodesic directions
at p. Each geodesic direction is formed by an equivalence class of geodesics in S p
for the equivalence relation
[ px] ∼ [ py] ⇐⇒ [ p xy ] = 0.
12 1 Preliminaries
|α(ε) − β(ε)|X
➊ |α − β|T p = lim
ε→0 ε
for every x ∈ X .
In this case, the sequence of closed sets (An )∞
n=1 is said to converge in the sense
of Hausdorff.
Example. Let Dn be the disk in the coordinate plane with center (0, n) and radius n.
Then Dn converges to the upper half-plane as n → ∞.
1.8.3 . First selection theorem. Let X be a proper metric space and (An )∞ n=1 be a
sequence of closed sets in X . Assume that for some (and therefore any) point x ∈ X ,
the sequence dist An (x) is bounded. Then the sequence (An )∞ n=1 has a convergent
subsequence in the sense of Hausdorff.
for any y.
Assume the contrary; that is,
for some z ∈ X and R > 0. Then for any sufficiently large n, there is a point z n ∈ An
such that |x − z n | R. Since X is proper, we can pass to a partial limit z ∞ of z n as
n → ∞.
It is clear that f (z ∞ ) = 0; that is, z ∞ ∈ A∞ . (Note that this implies that A∞ = ∅).
On the other hand,
14 1 Preliminaries
a contradiction.
On the other hand, since f is 1-Lipschitz, dist A∞ (y) f (y). Therefore
X1
X2
...
X∞
For another metric ρ —also guess it from the diagram—the limit space X∞ is
isometric to the real line.
X1
X2
...
X∞
Proof. From the main assumption in the theorem, in each space Xn there is a sequence
of points z i,n ∈ Xn such that the following condition holds for a fixed sequence of
integers M1 < M2 < . . .
• |z i,n − xn |Xn k + 1 if i Mk ,
• the points z 1,n , . . . , z Mk ,n form an k1 -net in B[xn , k]Xn .
Passing to a subsequence, we can assume that the sequence
Let Ŵ be the metric space corresponding to W; that is, points in Ŵ are equivalence
classes in W for the relation ∼, where wi ∼ w j if and only if |wi − w j |W = 0, and
where
satisfying definitions 1.9.1 and 1.9.2. The metric ρ can be constructed as the maximal
compatible metric such that
ρ(z i,n , wi ) m1
for any n Nm and i < Im for a suitable choice of two sequences (Im ) and (Nm )
with I1 = N1 = 1.
In this chapter we define CAT(κ) spaces and give the first application, to billiards.
Here “CAT” is an acronym for Cartan, Alexandrov, and Toponogov. It was
coined by Mikhael Gromov in 1987. Originally, Alexandrov called these spaces
“Rκ domain”; this term is still in use.
Riemannian manifolds with nonpositive sectional curvature provide a motivating
example. Specifically, a Riemannian manifold has nonpositive sectional curvature if
and only if each point admits a CAT(0) neighborhood.
1 That is, if
| p − x| + | p − y| + |x − y| 2 · π or |q − x| + |q − y| + |x − y| 2 · π.
We can do the same for the model plane of curvature κ; that is, a sphere if κ > 0,
Euclidean plane if κ = 0 and Lobachevsky plane if κ < 0. In this case we arrive at the
definition of CAT(κ) comparison. However in these notes we will mostly consider
CAT(0) comparison and occasionally CAT(1) comparison; so, if you see CAT(κ),
you can assume that κ is 0 or 1.
If all quadruples in a metric space X satisfy CAT(κ) comparison, then we say
that the space X is CAT(κ). (Note that CAT(κ) is an adjective.)
In order to check CAT(κ) comparison, it is sufficient to know the 6 distances
between all pairs of points in the quadruple. This observation implies the following.
2.1.2 . Exercise. Let V be a metric space and U = Cone V. Show that U is CAT(0)
if and only if V is CAT(1).
Analogously, if U = Susp V, then U is CAT(1) if and only if V is CAT(1).
2.1.3 . Exercise. Assume U and V are CAT(0) spaces. Show that the product space
U × V is CAT(0).
2.1.4 . Exercise. Show that any complete length CAT(0) space is geodesic.
The inheritance lemma 2.2.9 proved below plays a central role in the theory of CAT(κ)
spaces. It will lead to two fundamental constructions: patchwork globalization (3.3.2)
and Reshetnyak gluing (2.3.1), which in turn are used to prove the globalization
theorem (3.3.1).
Recall that a triangle [x 1 x 2 x 3 ] in a space X is a triple of minimizing geodesics
˜ 1 x 2 x 3 )E 2
[x x ], [x 2 x 3 ], and [x 3 x 1 ]. Consider the model triangle [x̃ 1 x̃ 2 x̃ 3 ] = (x
1 2
z̃ ∈ [x̃ i x̃ j ] to the corresponding point z ∈ [x i x j ]; that is, z is the point such that
|x̃ i − z̃| = |x i − z|, and therefore |x̃ j − z̃| = |x j − z|.
In the same way, the natural map can be defined for the spherical model triangle
˜ 1 x 2 x 3 )S2 .
(x
2.2 Thin triangles 19
(3) In Nehemiah xii. 22 we find the significant phrase “to the reign
of Darius, the Persian.” Now as long as the Persian empire stood
such a description would have no point when written by a Jewish
writer. For two hundred years down to 332 b.c., when Syria and
Phoenicia fell into the hands of Alexander the Great, the rulers of
Judea were all “Persians.” But from 332 b.c. onward Greek
monarchs were the rulers of Palestine, and nothing is more natural
than that a Jewish chronicler writing under their rule should refer to a
king of the older régime as “the Persian.”
(4) Further, in Nehemiah xii. 26, 47 the phrase “in the days of
Nehemiah” occurs, implying that for the writer Nehemiah belonged to
the past, but, as one cannot say how near or how distant a past, the
point carries little weight.
(5) Again, in Nehemiah xii. 10, 11 and 22, 23, a list of high-priests
is given, concluding with the name of Jaddua, whom the Chronicler
evidently (and correctly, compare Josephus, Antiquities of the Jews
xi. 7, 8) knew to have been the high-priest about 332 b.c., at the end
of the reign of Darius (Darius III, Codomannus), when the Persian
Empire collapsed before the attack of Alexander the Great.
(B) The character of Chronicles has already been referred to, but
in a different connection (§ 2, pp. xvi f.). Here the point to notice is
that throughout the entire work the whole system of law and ritual
found in the Pentateuch is presupposed as existing in its final form.
This system, which may conveniently be described as Priestly (P) in
distinction from the earlier system to which the name Deuteronomic
(D) is applied, and the still earlier standpoint represented by the
Jahvistic and Elohistic writers (J and E), may have been of slow
growth, and no doubt embodies features of law and ritual which are
of relatively high antiquity. But there is overwhelming evidence to
prove that, as an organised and completed system, it cannot be
dated earlier than the period of Nehemiah (circa 425 b.c.). Now in
Chronicles not only is this final code in force; it has evidently been so
long and so firmly established that the Chronicler did not know, or at
least did not believe, that any other earlier system had once ruled
the practice of Israel. He belonged to a period when the
development of the Pentateuch was no more remembered, and
when its origin—in all completeness—had come to be ascribed with
absolute confidence to the remote past, in accordance with that
religious instinct which we have described above on p. xiv.
Manifestly, a considerable lapse of time after Nehemiah’s period
must be allowed for that conviction to have become established.
(C) The late date of Chronicles is finally put beyond all doubt by
the linguistic peculiarities of the book. Excluding, of course, the
passages drawn from earlier Scriptures, the Hebrew of Chronicles is
of such a character that it is impossible to assign anything but a late
post-exilic date for its composition. In every aspect of language—
grammar and syntax and vocabulary—the diction exhibits the
unmistakable characteristics of late Hebrew. It lies beyond the scope
of the present volume to give details of the Hebrew, and reference
may be made to the edition of Chronicles by Curtis and Madsen
(International Critical Commentary), pp. 27 ff., where a list of 136
such peculiarities is given.
1 Samuel xxxi.
1 Kings iii. 4‒14; v.‒vii. (in part); viii.‒x.; xi. 41‒xii. 24; xiv. 21‒xv.
24 (in part); xxii. (in part).
2 Kings viii. 17‒29; xi., xii.; xiv. 1‒22; xv., xvi. (in part); xxi.‒xxiv.
(in part).
Ezra i. 1‒3.
(2) More important and difficult is the problem of the source of the
new material in Chronicles. Nearly one-half of the two books of
Chronicles is material otherwise unknown to us, and not to be
regarded as mere ornamental amplification of the passages drawn
from canonical sources. Rather it is precisely these new parts which
give colour to the whole work, and there can be no doubt that the
Chronicler must have dwelt with special fondness on just these
passages. The question is, Can we discern or infer sources from
which these independent chapters and paragraphs have been
derived, or is the Chronicler himself their only source and origin?
(12) The vision of Isaiah the prophet the son of Amoz in the
books of the kings of Judah and Israel (2 Chronicles
xxxii. 34).
(13) ? The history of Hozai (literally the seers) (2 Chronicles
xxxiii. 19).
(1) A Book of the Kings of Judah and Israel (cited for the
reigns of Asa, Amaziah, Ahaz, and Hezekiah; 2
Chronicles xvi. 11, xxv. 26, xxviii. 26, xxxii. 32).
Compare (12) above.
On the one side is Torrey (Ezra Studies, 1910) who argues that
the Chronicler had no source at all other than the canonical books—
all else was the product of his imaginative skill. He describes this
supposed midrashic history of Judah and Israel as “a phantom
source, of which the internal evidence is absolutely lacking, and the
external evidence is limited to the Chronicler’s transparent parading
of authorities.” The strength of Torrey’s contention lies in the fact that
almost all the additional matter in Chronicles is written in one and the
same distinctive style. That style has certain unmistakable
peculiarities. Thus Driver in the Encyclopedia Britannica s.v.
Chronicles, col. 772, writes, “It is not merely that the style of the
Chronicler presents characteristically late linguistic novelties ... but it
has also a number of special mannerisms.... So constant are [these
marks] that there is hardly a single sentence, not excerpted from
Samuel or Kings, in which they are not discernible.” On the other
side we have to consider the attitude adopted in the commentaries of
Benzinger (1901) and Kittel (1902), following up a suggestion made
by Büchler in 1899. These scholars not only believe that non-
canonical sources supplied much of the new material of Chronicles,
but they have attempted to analyse that material minutely into
various contributory elements. According to their view the Chronicler
was essentially a compiler, following his sources closely and
showing such little independence as he exercised chiefly in those
verses and passages where the affairs and interests of the Levites