Notes on point set topology, Fall 2010

Stephan Stolz
September 3, 2010
Contents
1 Pointset Topology 1
1.1 Metric spaces and topological spaces . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Constructions with topological spaces . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Properties of topological spaces . . . . . . . . . . . . . . . . . . . . . . . . . 9
1 Pointset Topology
1.1 Metric spaces and topological spaces
We recall that a map f : R
m
R
n
between Euclidean spaces is continuous if and only if
x X > 0 > 0 y X d(x, y) < d(f(x), f(y)) < , (1.1)
where d(x, y) = |x y| R
0
is the distance of two points x, y in some Euclidean space.
Example 1.2. (Examples of continuous maps.)
1. The addition map a: R
2
R, x = (x
1
, x
2
) x
1
+x
2
;
2. The multiplication map m: R
2
R, x = (x
1
, x
2
) x
1
x
2
;
The proofs that these maps are continuous are simple estimates that you probably remember
from calculus. Since the continuity of all the maps well look at in these notes is proved by
expressing them in terms of the maps a and m, we include the proofs of continuity of a and
m for completeness.
Proof. To prove that the addition map a is continuous, suppose x = (x
1
, x
2
) R
2
and > 0
are given. We claim that for := /2 and y = (y
1
, y
2
) R
2
with d(x, y) < we have
d(a(x), a(y)) < and hence a is a continuous function. To prove the claim, we note that
d(x, y) =

|x
1
y
1
|
2
+|x
2
y
2
|
2
and hence |x
1
y
1
| d(x, y), |x
1
y
1
| d(x, y). It follows that
d(a(x), a(y)) = |a(x) a(y)| = |x
1
+x
2
y
1
y
2
| |x
1
y
1
| +|x
2
y
2
| 2d(x, y) < 2 = .
1
To prove that the multiplication map m is continuous, we claim that for
:= min{1, /(|x
1
| +|x
2
| + 1)}
and y = (y
1
, y
2
) R
2
with d(x, y) < we have d(m(x), m(y)) < and hence m is a
continuous function. The claim follows from the following estimates:
d(m(y), m(x)) = |y
1
y
2
x
1
x
2
| = |y
1
y
2
x
1
y
2
+x
1
y
2
x
1
x
2
|
|y
1
y
2
x
1
y
2
| +|x
1
y
2
x
1
x
2
| = |y
1
x
1
||y
2
| +|x
1
||y
2
x
2
|
d(x, y)(|y
2
| +|x
1
|) d(x, y)(|x
2
| +|y
2
x
2
| +|x
1
|)
d(x, y)(|x
1
| +|x
2
| + 1) < (|x
1
| +|x
2
| + 1)
Lemma 1.3. The function d: R
n
R
n
R
0
has the following properties:
1. d(x, y) = 0 if and only if x = y;
2. d(x, y) = d(y, x) (symmetry);
3. d(x, y) d(x, z) +d(z, y) (triangle inequality)
Denition 1.4. A metric space is a set X equipped with a map
d: X X R
0
with properties (1)-(3) above. A map f : X Y between metric spaces X, Y is
an isometry if d(f(x), f(y)) = d(x, y) for all x, y X;
continuous if condition (1.1) is satised.
Two metric spaces X, Y are isometric (resp. homeomorphic) if there are isometries (resp.
continuous maps) f : X Y and g : Y X which are inverses of each other.
Example 1.5. An important class of examples of metric spaces are subsets of R
n
. Here are
particular examples we will be talking about during the semester:
1. The n-disk D
n
:= {x R
n
| |x| 1} R
n
, and more generally, the n-disk of radius
r D
n
r
:= {x R
n
| |x| r}. We note that D
2
r
is homeomorphic to D
2
for all r, but
D
2
r
is isometric to D
2
if and only if r = 1. (To see that D
n
r
is not isometric to D
n
s
we
note if a metric space X is isometric to a metric space Y , then diam(X) = diam(Y ),
where diam(X), the diameter of X is dened by diam(X) := sup{d(x, y) | x, y X}
R
0
{}. It is easy to see that diam(D
n
r
) = 2r.)
2. The n-sphere S
n
:= {x R
n+1
| |x| = 1} R
n+1
.
3. The torus T = {v R
3
| d(v, C) = r} for 0 < r < 1. Here C = {(x, y, 0) | x
2
+ y
2
=
1} R
3
is the standard circle, and d(x, C) = inf
yC
d(x, y) is the distance between x
and C.
2
4. The general linear group
GL
n
(R) = {vector space isomorphisms f : R
n
R
n
}
{(v
1
, . . . , v
n
) | v
i
R
n
, det(v
1
, . . . , v
n
) = 0}
= {invertible n n-matrices} R
n
R
n

n
= R
n
2
Here the bijection sends f : R
n
R
n
to (f(e
1
), . . . , f(e
n
)), where {e
i
} is the standard
basis of R
n
.
5. The special linear group
SL
n
(R) = {(v
1
, . . . , v
n
) | v
i
R
n
, det(v
1
, . . . , v
n
) = 1} R
n
2
6. The orthogonal group
O(n) = {linear isometries f : R
n
R
n
}
= {(v
1
, . . . , v
n
) | v
i
R
n
, v
i
s are orthonormal} R
n
2
We recall that a collection of vectors v
i
R
n
is orthonormal if |v
i
| = 1 for all i, and v
i
is perpendicular to v
j
for i = j.
7. The special orthogonal group
SO(n) = {(v
1
, . . . , v
n
) O(n) | det(v
1
, . . . , v
n
) = 1} R
n
2
8. The Stiefel manifold
V
k
(R
n
) = {linear isometries f : R
k
R
n
}
= {(v
1
, . . . , v
k
) | v
i
R
n
, v
i
s are orthonormal} R
kn
Example 1.6. The following maps between metric spaces are continuous. While it is pos-
sible to prove their continuity using the denition of continuity, it will be much simpler to
prove their continuity by building these maps using compositions and products from the
continuous maps a and m of Example 1.2. We will do this below in Lemma 1.17.
1. Every polynomial function f : R
n
R is continuous. We recall that a polynomial
function is of the form f(x
1
, . . . , x
n
) =

i
1
,...,i
n
a
i
1
,...,i
n
x
i
1
1
x
i
n
n
for a
i
1
,...,i
n
R.
2. Let M
nn
(R) = R
n
2
be the set of n n matrices. Then the map
M
nn
(R) M
nn
(R) M
nn
(R) (A, B) AB
given by matrix multiplication is continuous. Here we use the fact that a map to the
product M
nn
(R) = R
n
2
= R R is continuous if and only if each component
map is continuous (see Lemma 1.16), and each matrix entry of AB is a polynomial
and hence a continuous function of the matrix entries of A and B. Restricting to the
invertible matrices GL
n
(R) M
nn
(R), we see that the multiplication map
GL
n
(R) GL
n
(R) GL
n
(R)
is continuous. The same holds for the subgroups SO(n) O(n) GL
n
(R).
3
3. The map GL
n
(R) GL
n
(R), A A
1
is continuous (this is a homework problem).
The same statement follows for the subgroups of GL
n
(R).
Denition 1.7. Let X be a metric space. A subset U X is open if for every point x U
there is some > 0 such that B

(x) U. Here B

(x) = {y X | d(y, x) < } is the ball of

radius around x.
Lemma 1.8. A map f : X Y between metric space is continuous if and only if f
1
(V ) is
an open subset of X for every open subset V Y .
Proof: homework
Lemma 1.9. Let X be a metric space, and U be the collection of open subsets of X. Then
U has the following properties:
1. X and belong to U.
2. The union of a collection in U belongs to U.
3. The intersection of a nite collection of subsets in U belongs to U.
Denition 1.10. A topology on a set X is a collection U of subsets of X satisfying the
properties of the previous lemma. A topological space is a pair (X, U) consisting of a set X
and a topology U on X. If (X, U) is a topological space, a subset U X is open if U belongs
to U; it is closed if its complement X \ U belongs to U.
Let (X, U), (Y, V) be topological spaces. A map f : X Y is continuous if and only if
f
1
(U) V for every U U. It is easy to see that any composition of continuous maps is
continuous.
Examples of topological spaces.
1. Let X be a metric space. Then U = {open subsets of X} is a topology on X, the
metric topology.
2. Let X be a set. Then U = {all subsets of X} is a topology, the discrete topology. We
note that any map f : X Y to a topological space Y is continuous. We will see later
that the only continuous maps R
n
X are the constant maps.
3. Let X be a set. Then U = {, X} is a topology, the indiscrete topology.
1.2 Constructions with topological spaces
The subspace topology. Let X be a topological space, and A X a subset. Then
U = {A U | U
open
X}
is a topology on A called the subspace topology.
Lemma 1.11. Let X be a metric space and A X. Then the metric topology on A agrees
with the subspace topology on A (as a subset of X equipped with the metric topology).
4
Lemma 1.12. Let X, Y be topological spaces and let A be a subset of X equipped with the
subspace topology. Then the inclusion map i : A X is continuous and a map f : Y A
is continuous if and only if the composition i f : Y X is continuous.
Basis for a topology. Sometimes it is convenient to dene a topology U on a set X by rst
describing a smaller collection B of subsets of X, and then dening U to be those subsets of
X that can be written as unions of subsets belonging to B. Weve done this already when
dening the metric topology: Let X be a metric space and let B be the collection of subsets
of X of the form B

(x) := {y X | d(y, x) < } (we call these subsets balls in X). A subset
of X is open (in the sense of Denition 1.7) if and only if it is a union of balls in X.
Lemma 1.13. Let B be a collection of subsets of a set X satisfying the following conditions
1. Every point x X belongs to some subset B B.
2. If B
1
, B
2
B, then for every x B
1
B
2
there is some B B with x B and
B B
1
B
2
.
Then U := {unions of subsets belonging to B} is a topology on X.
If the above conditions are satised, the collection B is called a basis for the topology U
or B generates the topology U. It is easy to check that the collection of balls in a metric
space satises the above conditions and hence the collection of open subsets is a topology as
claimed by Lemma 1.9.
The Product topology
Denition 1.14. The product topology on the Cartesian product X Y = {(x, y) | x
X, y Y } of topological spaces X, Y is the topology with basis
B = {U V | U
open
X, V
open
Y }
The collection B obviously satises property (1) of a basis; property (2) holds since (U
V ) (U

) = (U U

) (V V

). We note that the collection B is not a topology since

the union of U V and U

is typically not a Cartesian product (e.g., draw a picture for

the case where X = Y = R and U, U

, V, V

are open intervals).

Lemma 1.15. The product topology on R
m
R
n
(with each factor equipped with the metric
topology) agrees with the metric topology on R
m+n
= R
m
R
n
.
Proof: homework.
Lemma 1.16. Let X, Y
1
, Y
2
be topological spaces. Then the projection maps p
i
: Y
1
Y
2
Y
i
is continuous and a map f : X Y
1
Y
2
is continuous if and only if the component maps
X
f
//
Y
1
Y
2
p
i
//
Y
i
are continuous for i = 1, 2.
Proof: homework
5
Lemma 1.17. 1. Let A R
n
and let f, g : A R be continuous maps. Then f + g
and f g continuous maps from A to R. If g(x) = 0 for all x A, then also f/g is
continuous.
2. Any polynomial function f : R
n
R is continuous.
3. The multiplication map GL
n
(R) GL
n
(R) GL
n
(R) is continuous.
Proof. To prove part (1) we note that the map f +g : A R can be factored in the form
A
fg
R R
a
R
The map f g is continuous by Lemma 1.16 since its component maps f, g are continuous;
the map a is continuous by Example 1.2, and hence the composition f + g is continuous.
The argument for f g is the same, with a replaced by m. To prove that f/g is continuous,
we factor it in the form
A
fg
//
R R

p
1
(Ip
2
)
//
R R

m
//
R,
where p
1
(resp. p
2
) is the projection to the rst (resp. second) factor of RR

and I : R

is the inversion map x x

1
. By Lemma 1.16 the p
i
s are continuous, in calculus
we learned that I is continuous, and hence again by Lemma 1.16 the map p
1
(I p
2
) is
continuous.
To prove part (2), we note that the constant map R
n
R, x = (x
1
, . . . , x
n
) a is
obviously continuous, and that the projection map p
i
: R
n
R, x = (x
1
, . . . , x
n
) x
i
is continuous by Lemma 1.16. Hence by part (1) of this lemma, the monomial function
x ax
i
1
1
x
i
n
n
is continuous. Any polynomial function is a sum of monomial functions and
hence continuous.
For the proof of (3), let M
nn
(R) = R
n
2
be the set of n n matrices and let
: M
nn
(R) M
nn
(R) M
nn
(R) (A, B) AB
be the map given by matrix multiplication. By Lemma 1.16 the map is continuous if and
only if the composition
M
nn
(R) M
nn
(R)

M
nn
(R)
p
ij
R
is continuous for all 1 i, j n, where p
ij
is the projection map that sends a matrix A to
its entry A
ij
R. Since the p
ij
((A, B)) = (A B)
ij
is a polynomial in the entries of the
matrices A and B, this is a continuous map by part (2) and hence is continuous.
Restricting to invertible matrices, we obtain the multiplication map

|
: GL
n
(R) GL
n
(R) GL
n
(R)
that we want to show is continuous. We will argue that in general if f : X Y is a
continuous map with f(A) B for subsets A X, B Y , then the restriction f
|A
: A B
is continuous. To prove this, consider the commutative diagram
A
f
|A
//
i

B
j

X
f
//
Y
6
where i, j are the obvious inclusion maps. These inclusion maps are continuous w.r.t. the
subspace topology on A, B by Lemma 1.12. The continuity of f and i implies the continuity
of f i = j f
|A
which again by Lemma 1.12 implies the continuity of f
|A
.
Quotient topology. Let X be a topological space, let be an equivalence relation on X,
let X/ be the set of equivalence classes and let
p: X X/ x [x]
be the projection map that sends a point x X to its equivalence class [x]. The quotient
topology on X/ is the collection of subsets U = {U X/ | p
1
(U) is an open subset of X}.
The set X/ equipped with the quotient topology is called the quotient space.
Lemma 1.18. The projection map p: X X/ is continuous and a map f : X/ Y to
a topological space Y is continuous if and only if the composition pf : X Y is continuous.
Example 1.19. 1. Let A be a subset of a topological space X. Dene a equivalence
relation on X by x y if x = y or x, y A. We use the notation X/A for the
quotient space X/ .
(a) We claim that the quotient space [1, +1]/{1} is homeomorphic to S
1
via the
map f : [1, +1]/{1} S
1
given by [t] e
it
. Here we use that a continuous
bijection f : X Y from a compact space to a Hausdor space is a homeomor-
phism.
(b) More generally, D
n
/S
n1
is homeomorphic to S
n
. (proof: homework)
2. quotients of the square by various equivalence relations gives: torus, Klein bottle, real
projective plane D
2
/ = S
2
/ . We can obtain a surface of genus 2 from an 8-gon
with suitable boundary identications (rst redraw 8-gon as a union of squares with a
corner chipped o; identifying boundaries on each square leads to punctured torus).
3. The real projective space
RP
n
:= {1-dimensional subspaces of R
n+1
} = S
n
/v v
Homework: RP
1
S
1
; RP
3
SO(3)
4. The complex projective space
CP
n
:= {1-dimensional subspaces of C
n+1
} = S
2n+1
/v zv, z S
1
homework: CP
1
S
2
5. The Grassmann manifold G
k
(R
n+k
) := {k-dimensional subspaces of R
n+k
}. There is a
surjective map
V
k
(R
n+k
) = {isometries f : R
k
R
n+k
} G
k
(R
n+k
) f im(f)
Two isometries f, f

have the same image if and only if there is some isometry g : R

k

R
k
such that f

= f g. In other words, we get a bijection V

k
(R
n+k
)/ G
k
(R
n+k
)
if we dene an equivalence relation on the Stiefel manifold by f f

if and only if
there is some isometry g : R
k
R
k
such that f

= f g. This the quotient topology

on V
k
(R
n+k
)/ then gives G
k
(R
n+k
) a topology (note that for k = 1, V
k
(R
n+k
) = S
n
,
and this agrees with how we put a topology on the projective space RP
n
= G
1
(R
n+1
).
7
6. If X is a topological space and a group H acts X (say from the right via XH X,
(x, h) xh; requirement: (xh)h

= x(hh

) for x X, h, h

H). The group action

denes an equivalence relation on X via x

x if and only if there is some h H

such that x

= xh. Equivalence classes are called the orbits of the action; the quotient
space X/ is the orbit space, denoted X/H.
(a) G
k
(R
n+k
) = V
k
(R
n+k
)/O(k)
(b) homogeneous spaces G/H for topological groups G. Explanation: a topological
group is a group G equipped with a topology such that the multiplication map
GG G and the inversion map G G, g g
1
are continuous. A subgroup
H G act on G via the multiplication map G H G, (g, h) gh. The
orbit space is denoted G/H (or H\G if we use the corresponding left H-action
on G), and is called homogeneous space. Warning: there is dierence between the
homogeneous space G/H and the quotient space of G obtained by collapsing the
subspace H to a point (Example 1.19 (1)), which we also would denote by G/H
(unfortunately, both notations are standard; fortunately, it is usually clear from
the context which version of G/H we are talking about, since the homogeneous
space makes only sense if H is a subgroup of a topological group G).
We want to show that many topological spaces weve discussed so far are actually homo-
geneous spaces. To do that we use the following result.
Proposition 1.20. (Recognition principle for homogeneous spaces) Let G be a com-
pact topological group that acts continuously and transitively on a topological space X. Then
X is homeomorphic to the homogeneous space G/H where H = {g G | gx
0
= x
0
} is the
isotropy subgroup of some point x
0
X.
Proof. Let
f : G/H X be dened by [g] gx
0
This map is surjective by the transitivity assumption; it is injective since if gx
0
= g

x
0
, then
x
0
= g
1
g

x
0
and hence h := g
1
g

belongs to the isotropy subgroup H. This implies g

= gh,
and hence [g

] = [g] G/H.
To show that f is continuous it suces to show that the composition f p: G X,
g gx
0
is continuous. To see this, we factor f p in the form
G = G{x
0
} GX

X
where is the action map
Examples of homogeneous spaces.
1. spheres S
n
O(n + 1)/O(n)
(take the action O(n + 1) S
n
S
n
, (f, v) f(v) and x
0
= (0, . . . , 0, 1) S
n
)
2. Stiefel manifold V
k
(R
n+k
) O(n +k)/O(n)
(take the action O(n +k) V
k
(R
n+k
) V
k
(R
n+k
), (g, f) g f and x
0
: R
k
R
n+k
,
v (0, v)).
3. Grassmann manifold G
k
(R
n+k
) O(n +k)/O(n) O(k) (homework problem).
8
1.3 Properties of topological spaces
Denition 1.21. Let X be a topological space, x
i
X, i = 1, 2, . . . a sequence in X and
x X. Then x is the limit of the x
i
s if for all open subsets U X containing x there is
some N such that x
i
U for all i N.
Caveat: If X is a topological space with the indiscrete topology, every point is the limit
of every sequence. The limit is unique if the topological space has the following property:
Denition 1.22. A topological space X is Hausdor if for every x, y X, x = y, there are
disjoint open subsets U, V X with x U, y V .
Note: if X is a metric space, then the metric topology on X is Hausdor (since for x = y
and = d(x, y)/2, the balls B

(x), B

(y) are disjoint open subsets).

Warning: The notion of Cauchy sequences can be dened in metric spaces, but not in
general for topological spaces (even when they are Hausdor).
Lemma 1.23. Let X be a topological space and A a closed subspace of X. If x
n
A is a
sequence with limit x, then x A.
Proof. Assume x / A. Then x is a point in the open subset X \ A and hence by the
denition of limit, all but nitely many elements x
n
must belong to X\A, contradicting our
assumptions.
Denition 1.24. An open cover of a topological space X is a collection of open subsets of
X whose union is X. If for every open cover of X there is a nite subcollection which also
covers X, then X is called compact.
Some books (like Munkres Topology) refer to open covers as open coverings, while newer
books (and wikipedia) seem to prefer to above terminology, probably for the same reasons
as me: to avoid confusions with covering spaces, a notion well introduce soon.
Now well prove some useful properties of compact spaces and maps between them, which
will lead to the important Corollaries ?? and 1.27.
Lemma 1.25. If f : X Y is a continuous map and X is compact, then the image f(X)
is compact.
In particular, if X is compact, then any quotient space X/ is compact, since the
projection map X X/ is continuous with image X/ .
Proof. To show that f(X) is compact assume that {U
a
}, a A is an open cover of the
subspace f(X). Then each U
a
is of the form U
a
= V
a
f(X) for some open subset V
a
Y .
Then {f
1
(V
a
)}, a A is an open cover of X. Since X is compact, there is a nite subset
A

of A such that {f
1
(V
a
)}, a A

is a cover of X. This implies that {U

a
}, a A

is a
nite cover of f(X), and hence f(X) is compact.
Lemma 1.26. 1. If K is a closed subspace of a compact space X, then K is compact.
2. If K is compact subspace of a Hausdor space X, then K is closed.
9
Proof. To prove (1), assume that {U
a
}, a A is an open covering of K. Since the U
a
s
are open w.r.t. the subspace topology of K, there are open subsets V
a
of X such that
U
a
= V
a
K. Then the V
a
s together with the open subset X \ K form an open covering of
X. The compactness of X implies that there is a nite subset A

A such that the subsets

V
a
for a A

, together with X \ K still cover X. It follows that U

a
, a A

is a nite cover
of K, showing that K is compact.
The proof of part (2) is a homework problem.
Corollary 1.27. If f : X Y is a continuous bijection with X compact and Y Hausdor,
then f is a homeomorphism.
Proof. We need to show that the map g : Y X inverse to f is continuous, i.e., that
g
1
(U) = f(U) is an open subset of Y for any open subset U of X. Equivalently (by passing
to complements), it suces to show that g
1
(C) = f(C) is a closed subset of Y for any
closed subset C of C.
Now the assumption that X is compact implies that the closed subset C X is compact
by part (1) of Lemma 1.26 and hence f(C) Y is compact by Lemma 1.25. The assumption
that Y is Hausdor then implies by part (2) of Lemma 1.26 that f(C) is closed.
Lemma 1.28. Let K be a compact subset of R
n
. Then K is bounded, meaning that there
is some r > 0 such that K is contained in the open ball B
r
(0) := {x R
n
| d(x, 0) < r}.
Proof. The collection B
r
(0) K, r (0, ), is an open cover of K. By compactness, K is
covered by a nite number of these balls; if R is the maximum of the radii of these nitely
many balls, this implies K B
R
(0) as desired.
Corollary 1.29. If f : X R is a continuous function on a compact space X, then f has
a maximum and a minimum.
Proof. K = f(X) is a compact subset of R. Hence K is bounded, and thus K has an inmum
a := inf K R and a supremum b := sup K R. The inmum (resp. supremum) of K is the
limit of a sequence of elements in K; since K is closed (by Lemma 1.26 (2)), the limit points
a and b belong to K by Lemma 1.23. In other words, there are elements x
min
, x
max
X
with f(x
min
) = a f(x) for all x X and f(x
max
) = b f(x) for all x X.
In order to use Corollaries 1.27 and 1.29, we need to be able to show that topological
spaces we are interested in, are in fact compact. Note that this is quite dicult just working
from the denition of compactness: you need to ensure that every open cover has a nite
subcover. That sounds like a lot of work...
Fortunately, there is a very simple classical characterization of compact subspaces of
Euclidean spaces:
Theorem 1.30. (Heine-Borel Theorem) A subspace X R
n
is compact if and only if
X is closed and bounded.
We note that weve already proved that if K R
n
is compact, then K is a closed subset
of R
n
(Lemma 1.26(2)), and K is bounded (Lemma 1.28).
There two important ingredients to the proof of the converse, namely the following two
results:
10
Lemma 1.31. A closed interval [a, b] is compact.
This lemma has a short proof that can be found in any pointset topology book, e.g.,
[Mu].
Theorem 1.32. If X
1
, . . . , X
n
are compact topological spaces, then their product X
1
X
n
is compact.
For a proof see e.g. [Mu, Ch. 3, Thm. 5.7]. The statement is true more generally for a
product of innitely many compact space (as discussed in [Mu, p. 113], the correct denition
of the product topology for innite products requires some care), and this result is called
Tychonos Theorem, see [Mu, Ch. 5, Thm. 1.1].
Proof of the Heine-Borel Theorem. Let K R
n
be closed and bounded, say K B
r
(0).
We note that B
r
(0) is contained in the n-fold product
P := [r, r] [r, r] R
n
which is compact by Theorem 1.32. So K is a closed subset of P and hence compact by
Lemma 1.26(1).
References
[Mu] Munkres, James R. Topology: a rst course, Prentice-Hall, Inc., Englewood Clis,
N.J., 1975. xvi+413 pp.
11