Professional Documents
Culture Documents
Merged
Merged
Topology
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-1: Introduction to definition
of Topological spaces
To get a proper notion of continuity in abstract case one needs a notion of nearness,
so that one can say about continuous transformation.
The idea is that if one geometric object can be continuously transformed into another,
then the two objects are to be viewed topologically equivalent.
For example, an open interval of length one can be stretched to an open interval of length
two, a circle and a square are topologically equivalent, one can be continuously trans-
formed into another by radial projection.
On the other hand, a closed interval is topologically distinct from a circle or square. In
fact, when a point is removed from a circle it remains is still connected but if a point
lying between 0 and 1 remover from a closed interval it produces two different pieces.
The term used to describe two geometric objects are topologically equivalent is home-
omorphism.
Thus a circle and a square are homeomorphic. Concretely, a radial projection from a
circle C to a square S with the same center point, produces a homeomorphism between
C and S.
One of the basic problems of Topology is to determine when two given geometric objects
are non homeomorphic, for example whether the letters M, N, H, B are non homeomor-
phic.
This can be quite difficult in general. Our first goal will be to define exactly what the
‘geometric objects’ are that one studies in Topology. These are called topological spaces.
2
Roughly speaking continuous function from one topological space to another is that
which preserve nearness.
Beyond ε − δ definition a nice definition of a continuous function in terms of open
sets:
To see that this definition corresponds to the intuitive notion of continuity, let us
examine, what would happen if this condition fails. There would then be an open set O
for which f −1 (O) is not open. This means there would be a point x0 ∈ f −1 (O) for which
there is no interval (a, b) containing x0 and contained in f −1 (O).
This is equivalent to saying there would be points x arbitrarily close to x0 those are
in the complement of f −1 (O). For x to be in the complement of f −1 (O) means that f (x)
is not in O. On the other hand, x0 was in f −1 (O) so f (x0 ) is in O. Since O was assumed
to be open, there is an interval (c, d) about f (x0 ) that is contained in O. The points
f (x), those are not in O are therefore not in (c, d), so they remain at least a fixed positive
distance from f (x0 ). A reasonable interpretation of discontinuity of f at x0 would be
that there are points x arbitrarily close to x0 for which f (x) stays at least a fixed positive
distance say ε away from f (x0 ). Let O be the open set (f (x0 ) − ε, f (x0 ) + ε). Then
f −1 (O) contains x0 but it does not contain any points x for which f (x) is not in O, and
we are assuming there are such points x arbitrarily close to x0 , so f −1 (O) is not open
since it does not contain all points in some interval (a, b) about x0 .
In trying to find a satisfactory definition of a topological space we shall have two aims
in mind.
1. The definition should be general enough to allow a wide range of different structures
as spaces. We would like to consider a finite, discrete set of points as a space, or equally
a whole uncountable continuum of points such as the real line; our nice geometrical sur-
faces should qualify under the definition, and also sets of functions such as the set of
continuous complex-valued functions defined on the unit circle in the complex plane.
3
2. The definition will be so that, we would be able to perform simple constructions with
our spaces, such as taking the cartesian product of two spaces, or identifying some of the
points of a space in order to form a new one. For example in the construction of Mobius
strip we take a rectangular piece of rubber sheets and then identify two opposite sides in
different direction.
The definition of a space should contain enough information so that we can define the
notion of continuity for functions between spaces. It is really this second consideration
which leads to the abstract definition given below.
The most known example of topological space is R with usual open sets as the open
sets of the topology of R.
Example 1. Let X be a nonempty set and O = P(X). Then it is easy to observe that
(X, O) is a topological space called discrete space also denote by Xd .
Example 2. Let X be a nonempty set and O = {∅, X}. Then it is topological space
called indiscrete space.
Example 3. Let X = {a, b, c} and O = {∅, X, {a}, {b}, {a, b}}. Then observed that O is
a topology on X, so that (X, O) is a topological space.
4
Proof. ∅ ∈ τcf by default and X ∈ τcf is obvious.
Now let A1 , A2 ∈ τcf . Then
X \ (A1 ∩ A2 ) = (X \ A1 ) ∪ (X \ A2 )
Example 5. Let X be an uncountable set and let τco = {A ⊂ X : X\A is countable or finite}.
Then (X, τco ) is a topological space, usually called cocountable space.
Example 6. Let X be a plane. Let O consist of ∅, X and all open disks with centre at
the origin. Then clearly O is topology on X.
Definition 3. Let X = {0, 1, 2, ..., n, .....} = N ∪ {0}. For A ⊂ X, let |A ∩ [1, n]| denotes
the cardinality of the set A ∩ [1, n].
T
Theorem 1. Define τ = {A : 0 ∈
/ A or 0 ∈ A and lim |A [1,n]|
= 1}. Then τ is a
n→∞ n
topology on X.
Proof. Clearly ∅ ∈ τ .
Observe that lim |X∩[1,n]| = 1 so that X ∈ τ . If A, B ∈ τ , that is lim |A∩[1,n]| =1
n→∞ n n→∞ n
c c
and lim |B∩[1,n]| = 1. Then lim |A ∩[1,n]| = 0 and lim |B ∩[1,n]| = 0. This implies
n→∞ n n→∞ n n→∞ n
c c
that lim |(A ∪Bn)∩[1,n]| = 0. Since X \(A ∩ B) = (X \ A)∪(X \ B) we have the desired
n→∞
result. Hence τ is closed under finite intersection. That τ is closed under arbitrary union
is easy.
5
Proof. (1) is obvious.
(2) Let {Fα : α ∈ Γ} be a collection of closed sets. Then by definition {X \ Fα : α ∈ Γ}
is a collection of open sets. Using elementary set theory the result follows.
Like elementary real analysis we can introduce the notions of Interior, limit points,
closure, boundary points for arbitrary topological spaces.
The set of all limit points of a set A is called derived set of A, By the closure of A
we mean the set A along with its limit points, and denoted by A. The set of all interior
points of A called interior of A and denoted by A◦ .
Question 1. What will be the interior, closure of the following sets : (a) {(x, y)|1 <
x2 + y 2 ≤ 2}
(b) R2 with both axes removed
6
Proof. (1) follows easily. For part (2) let x ∈ A ∪ B and U be a neighborhood of x. Then
U ∩ (A ∪ B) 6= ∅. This means that either U ∩ A 6= ∅ or U ∩ B 6= ∅. This implies that
x ∈ A ∪ B, so that A ∪ B ⊆ A ∪ B.
Conversely let x ∈ A ∪ B and U be a neighborhood of x. Then either U ∩ A 6= ∅ or
U ∩ B 6= ∅, that is U ∩ (A ∪ B) 6= ∅ so that x ∈ A ∪ B. This gives the reverse inequality.
Proof. (1) If x ∈ A◦ then there exists some open Bx such that x ∈ Bx ⊂ A. We have
Bx ⊂ A◦ . In fact for each y ∈ Bx , Bx is an open set contained in A and containing y.
Now let O be an open set contained in A. Then for each x ∈ O there exists some open
Bx such that x ∈ Bx ⊂ A, i.e. each element of O is an interior point of A so that O ⊂ A◦ .
(2) Let x ∈ X \ A. Then there exists an open set O containing x which misses A.
We claim that O also misses A. If not so then there exists some y ∈ A \ A such that
y ∈ O. This implies that O meets A, which is a contradiction. Therefore A is closed. To
prove that A is the smallest closed set containing A, let K be a closed set containing A.
If possible let there exists some y ∈ A \ K. Then there exists an open set O containing
y, which misses A, which contradicts the fact that y ∈ A.
7
We can define convergency of sequence in topological spaces analogues to metric space.
In the following we will examine the property which is responsible for unique limit of
any convergent sequence in Ru .
Proof. Let x ∈ X and y 6= x. Then there exist disjoint open sets U and V containing
x and y respectively. Therefore y is not a limit point of the set {x}. This implies that
every singletoned is closed and therefore every finite set is closed.
Note that cofinite space is not Hausdorff. In fact if x and y are two distinct points U
and V are two disjoint opensets containing x and y respectively, then X \ U and X \ U
are finite. But then (X \ U ) ∪ (X \ V ) = X \ (U ∩ V ). But the left hand side being finite
and the right hand side being whole X as U ∩ V = ∅ we get a contradiction.
One can observe that for the cofinite space every sequence converges to every point.
In this position we like to introduce the new two definitions. Here we prove that Haus-
dorfness is responsible for unique limit of any convergent sequence.
Proposition 2. In a Hausdorff space any sequence can have at most one limit.
Proof. Let x ∈ X and y 6= x. Then there exist disjoint open sets U and V containing
x and y respectively. Now if a sequence (xn ) converges to x then it is not possible to
converge to y, as except finitely many points, all the points of the sequence are in U .
8
CHAPTER 1
Topology
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-2: Base of Topological
spaces
Since open sets are defined in terms of open intervals many arguments with open sets
in R reduce to looking at what happens with open intervals. A similar statement holds
for Rn with open balls in place of open intervals. In each case arbitrary open sets are
unions of the special open sets given by open intervals, and balls. Generalizing this idea
we introduce the following definition.
Theorem 1. Let X be a non empty set and B be a basis. Let τ (B) be the collection
defined as follows: U ∈ τ (B) if for each x ∈ U there exists some B ∈ B with the property
x ∈ B ⊂ U . Then τ (B) is a topology on X.
2
Most known example is that R with usual topology has the following two bases:
1. {(a, b) : a, b ∈ R}
2. {(a, b) : a, b ∈ Q}.
Another way of describing the topology generated by a basis is given in the following
lemma:
Lemma 1. Let X be a set; let B be a basis on X. Then τ (B), the topology generated by
B equals the collection of all unions of elements of B.
Observation 1. The above theorem shows that a topology on a set is all possible unions
of members of base.
In the above Theorem we mentioned how to from topology from a basis. The following
is one way of obtaining a basis for a given topology. We shall use it frequently.
Proof. The first condition, is obvious. For the second condition, suppose B1 and B2 are
elements of B and x ∈ B1 ∩ B2 . Since B1 ∩ B2 is an open set there exists some B3 ∈ B
such that x ∈ B3 ⊂ B1 ∩ B2 .
The topology generated by B equals to the topology of X, is left as an exercise.
We have already seen that both the following sets generates the usual topology on R.
1. {(a, b) : a, b ∈ R}
2. {(a, b) : a, b ∈ Q}. But when topologies are given by bases, it is useful to have a
criterion in terms of the bases for determining whether one topology is finer than another.
Proposition 2. Let B and B1 for the topologies τ1 and τ2 on a set X. Then the following
conditions are equivalent.
3
(1) τ2 is finer than τ1 .
(2) For each x ∈ X and each basis element B ∈ B containing x, there is a basis element
B2 ∈ B2 such that x ∈ B2 ⊂ B1 .
Example 1. 1. One of the most beautiful example of the above Theorem is that the set
{(a, b) : a, b ∈ R} and {(a, b) : a, b ∈ Q} both generates the same topology R. It happens
as Q is dense in R.
2. Another beautiful application of the above Theorem is that open rectangles in R2 and
open discs in R2 generates the same topology on R2 . In fact if we take an open disk D
and a point x in D then we can inscribed a rectangle in D containing x. Similarly the
other. Therefore the topologies generated by them are equivalent.
4
Therefore the collection B generates some topology on R, which is known as lower
limit topology and denoted by Rl .
An interesting point to observe that, Rl has a basis each of whose member is closed
as well as open. In fact each set [a, b) is also closed in Rl . In fact if x 6∈ [a, b) then either
x < a or x ≥ b. In the first case we can choose r ∈ (x, a) sot that [x, a) misses [a, b) and
in the second case [x, x + 1) misses [a, b). Therefore x is not a limit point of [a, b) and
hence [a, b) is a closed set.
Another important observation about Rl is that each open interval in Rl is open, that
means that the topology of Rl is larger than the usual topology.
is a basis that generates a topology different from the lower limit topology on R.
Example 4. Consider the set K = { n1 : n ∈ N} and the subsets of the form (a, b) \ K.
The collection
B1 = {(a, b) ⊂ R : a, b ∈ R} ∪ {(a, b) \ K ⊂ R : a, b ∈ R}
A partially ordered set is a pair (X, ≤), where X is a set and ≤ is a relation on X such
that: (i) x ≤ x for all x; (ii) if x ≤ y and y ≤ z, then x ≤ z; (iii) if x ≤ y and y ≤ x, then
x = y. Let us call a partially ordered set linearly ordered if whenever x and y are in X,
either x ≤ y or y ≤ x. (For example, X = R or any of its subsets is linearly ordered.) (a)
If X is a partially ordered set and S is the collection of all sets having the form of either
5
{y : y ≤ x and y 6= x} or {y : x ≤ y and y 6= x}, we can show that S is a subbase for a
topology on X. This is called the order topology on X. (b) Show that if (X, ≤) is linearly
ordered, then the order topology satisfies the Hausdorff property. Can you find another
condition on the ordering such that the order topology has the Hausdorff property? (c)
When a and b are elements of a partially ordered space, let (a, b) = {x ∈ X : a < x < b}.
If (X, ≤) is linearly ordered, show that (a, b) = {x ∈ X : a < x < b} is a base of the
order topology.
As an example of this property consider the set of open balls in a metric space.
For a ∈ Z and K ∈ N, we define a + KZ = {a + Kn : n ∈ Z}. We say that A ⊂ Z is
open if for each point a ∈ A, there is a number K ∈ N such that {a + Kn : n ∈ Z} ⊂ A.
In other words, a subset of Z is open if each of its points is contained in an arithmetic
progression belonging to the set. Obviously the sets {a + Kn : n ∈ Z} are open. So the
arithmetic progressions form a basis for the topology. Surprisingly, this basic sets are
6
closed also because the complement of a + KZ = {a + Kn : n ∈ Z} is the union of other
arithmetic progressions with the same difference.
There are many proofs that there exist an infinity of primes. Using this topology we
can prove that the infiniteness of primes.
Proof. Since the sets {i, i + d, i + 2d, ........}, i = 1, 2, ...., d are open, pairwise disjoint
and cover the whole N, it follows that each of them is closed. In particular, for each
prime number p the set {p, 2p, 3p, .......} is closed. All together, the set of the form
{p, 2p, 3p, ......} cover N − {1}. Hence if the set of prime numbers were finite, then the
set {1} would be open. However, it is not a union of arithmetic progressions.
Proof. Let τ (S) be the intersection of all topologies containing S; such topology exists,
since P(X) is one such. So clearly τ (S) is a topology. It evidently satisfies the re-
quirements of ‘unique’ and ‘smallest’. To verify the members of τ (S) are as described,
S
note that since S ⊂ τ (S) so τ (S) must contain all the sets listed. Conversely, since
T α
distributes over , the sets listed actually do form a topology containing S, and which
therefore contains τ (S).
Remark 1. The construction of a topology from a subbasis loses some control over the
open sets; they build up from the finite intersections of the Sα ’s rather than from the Sα
themselves.
7
CHAPTER 1
Topology
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-3: New spaces from old one
In this chapter we deal with constructing new spaces from old ones. This topic which
seems simple enough at first glance, but turns out to be a source of many examples of
topological spaces.
Given a topology O on a space X and a subset A ⊂ X, we would like to use the
topology on X to define a topology OA on A. This can be done just taking the trace of
open sets in X on A. Define a set O ⊂ A to be in OA if there exists an open set O0 in O
such that O = A ∩ O0 . Let us first prove that this is in fact a topology.
∅ = A ∩ ∅ and A = A ∩ X.
The fact that it is closed under finite intersection and arbitrary union follows from
the following equations
!
[ [
(Uα ∩ A) = Uα ∩ A.
α∈J α∈J
The topology OA on A is called the subspace topology, and A with this topology is
called a subspace of X. For example, if we take X to be R2 with its usual topology, then
every subset of R2 becomes a topological space. In particular, geometric figures such
as circles and polygons can now be viewed as topological spaces. Likewise, geometric
figures in R3 such as spheres and polyhedra become topological spaces, with the subspace
topology from the usual topology on R3 .
2
In case the space X is a metric space, any subset A ⊂ X becomes a metric space
by restricting the metric X × X −→ R to A × A , since the three defining properties of
ametric obviously still hold for the restricted distance function. The following Proposition
gives some strong evidence that the subspace topology is a natural topology to use on
subsets.
Proof. Observe first that for a ball Br (x) in X, the intersection A ∩ Br (x) consists of all
points in A of distance less than r from x, so this is a ball in A regarded as a metric
space in itself. For a collection of such balls Br (x) we have
The left side of this equation is a typical open set in A with the subspace topology,
and the right side is a typical open set in the metric topology, so the two topologies
coincide.
Question 1. Give an example of a topological space to establish that open or closed for
the subspace Y in the above lemma can not be removed.
Consider the closed interval [0, 1] in Ru . (0, 1] is open in [0, 1] in the subspace topology
but not open in Ru .
The following Theorem show that closures behave nicely with respect to subspaces:
3
Proof. For a point y ∈ Y to be a limit point of A in X means that every open set O in
X that contains y meets A. Since A ⊂ Y , this is equivalent to O ∩ Y meeting A, or in
other words, that every open set in Y containing y meets A.
Question 2. Show by an example that analogues statement of the above theorem is not
true for interiors.
Exercises
(1) Prove that subspae topology on Z induced from R is discrete.
Theorem 2. Let X be a linearly ordered topological space and let Y be a subset of X that
is convex in X. Then the order topology on Y is the same as the topology on Y inherits
as a subspace of X.
Question 3. Give an example to show that convexity can not be removed in the above
Theorem.
Consider Z in R.
If L is a straight line in the plane, describe the topology L inherits as a subspace
of Rl × R and as a subspace of Rl × Rl . In each case it is a familiar topology.
4
If we take trace of any open disk we get a an open arc, on the circle which are open sets
of the circle. Mathematically it can be expressed as {e2πiθ : 0 < θ ≤ 1} and denoted by
S 1.
Figure 8 is another interesting topological space, which has more importance in alge-
braic topology.
Assigning a standard sort of topology to the cartesian product of spaces is a tool of
constructing new topological spaces from old. For example, the Euclidean plane is the
product space of the real numbers (with usual topology) with itself and Euclidean n-
space is the product of the real numbers n times.
5
Our objective now is to define a topology on the cartesian product of topological
spaces in some natural and useful way.
If X and Y are topological spaces, we can define a topology on X × Y by saying that
a basis consists of the subsets U × V as U ranges over open sets in X and V ranges over
open sets in Y .
In the next Theorem we shall show that it is sufficient to consider basis elements also.
Theorem 3. Let X and Y be topological spaces with bases B and C. Then the collection
B × C = {U × V : U ∈ B, V ∈ C}
More generally one can define the productX = X1 × . . . × Xn to consist of all ordered
n-tuples (x1 , . . . , xn ) with xi ∈ Xi for each i. A basis for the product topology on
X = X1 × . . . × Xn consists of all products U1 × . . . × Un as each Ui ranges over open sets
in Xi , or just over a basis for the topology on Xi .
Thus Rn with its usual topology is also describable as the product of n copies of R,
with basis the open ‘boxes’ (a1 , b1 ) × . . . × (an , bn ).
6
CHAPTER 1
Topology
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-4: Introduction to
Continuity
Recall that a function f : R → R is continuous if f −1 (O) is open in R for each open set
O in R. We take this as formal definition of continuous function for arbitrary topological
spaces.
Definition of continuous function may be described using closed sets as well, which
can be shown from the following Prposition.
The following proposition shows that continuous maps behave well under formation
of closure.
2
Proof. (a) ⇔ (b) If x is a limit point of A then f (x) is a limit point of f (A) (verify).
Hence f (c`X A) ⊂ c`Y f (A).
Conversely let K be a closed set in Y and A = f −1 (K). Thenf (c`X A) ⊂ c`Y f (A) =
c`Y f ((f −1 (K)). This implies that f (c`X A) ⊂ c`Y K = K, so that A ⊂ c`X A ⊂ f −1 (K) =
A. Hence c`X A = f −1 (K), that is f −1 (K) is closed and therefore f is continuous.
(a) ⇔ (c) The necessary part is obvious. Let V be a non empty open set of Y and
choose x ∈ f −1 (V ). This implies that f (x) ∈ V . Then by the given hypothesis there
exists a neighborhood U of x such that f (U ) ⊂ V , so that x ∈ U ⊂ f −1 (V ). Hence x is
an interior point and hence f is continuous.
Question 2. Let τ1 and τ2 be two topology on X. Prove that the identity map id :
(X, τ1 ) → (X, τ2 ) is contnuous if and only if τ2 ⊂ τ1 .
Lemma 1. Let (Xα , τα )α∈I be a collection of topological spaces, indexed by the set I,
and we have maps fα : X → Xα , α ∈ I. Then there is a smallest topology τ on X for
which the maps fα are continuous.
Proof. Clearly discrete topology makes all fα continuous. A topology σ on X makes all
−1 (U ) : U ∈ τ , α ∈ I}.
the fα continuous if and only if it contains A = {fα α
The following fact again shows that information about base is sufficient for any topo-
logical spaces, whose easy proof is left as exercise.
3
Proof. Let O be an open set of Y and B be a basis for the topology of Y . Then O = ∪α Bα ,
where Bα ∈ B. Hence f −1 (O) = f −1 (∪α Bα ) = ∪α f −1 Bα . Since each Bα is open in Y
and open sets are closed under union we have that f −1 (O) is also open.
The following lemma shows that continuous maps behave well under composition.
Proof. To prove that it is sufficient to observe that for any open set U in Z, (g ◦f )−1 (U ) =
f −1 (g −1 (U )).
Proof. Obvious.
Theorem 4. If X is a topological space then C(X), the set of all real valued continuous
functions form a lattice ordered ring.
(f + g)(x) = f (x) + g(x) for all x ∈ X and f · g(x) = f (x) · g(x) for all x ∈ X
are continuous. Hence C(X) becomes a ring under pointwise addition and multiplication.
f ≤ g iff f (x) ≤ g(x) for all x ∈ X makes C(X) a lattice ordered ring.
Proof. That C(X) is a vectorspace follows from the above Theorem. Now if we define
kf k = sup{f (x) : x ∈ X}
4
Theorem 6. Let X = A ∪ B, where A and B are closed sets in X. Let f : A → Y and
g : B → Y be continuous. If f (x) = g(x) for all x ∈ A ∩ B, then there exists a continuous
function h : X → Y such that h(x) = f (x) for all x ∈ A and h(x) = g(x) for all x ∈ B.
Proof. Let C be a closed subset of Y . Then h−1 (C) = f −1 (C) ∪ g −1 (C). Since f is
continuous, f −1 (C) is closed in A and, therefore, closed in X. Similarly, g −1 (C) is closed
in B and therefore closed in X. Their union h−1 (C) is thus closed in X.
Example 2. (a) Give an example of a continuous function which is neither open nor
closed.
(b) Give an example of a open map which is not continuous.
(c) Give an example of a closed map which is not continuous.
Example 3. Continuous but neither open nor closed The mapping p : [0, 1) → S 1 defined
by p(x) = e2πix is a continuous bijection, which is neither open nor closed.
Proof.
5
Example 7. Let (X, τ ) be a topological space and (Y, d) be a metric space. If f, g : X → Y
are continuous then the set
{x ∈ X : f (x) = g(x)} is closed. To see this, let E={x ∈ X : f (x) = g(x)}. We show
that the complement of E is open.
Let y ∈ X − E. Then f (y) 6= g(y). We can find open sets U and V such that
f (y) ∈ U ,g(y) ∈ V and U V = φ. Now f −1 (U ) is open, as is g −1 (V ), so y ∈
T
Note that in the above example if Y is taken Hausdorff space then the result also
remains valid.
Proof. If possible let there exists a point x ∈ X − A such that f (x) 6= g(x). Then there
exist disjoint open sets Uf (x) and Vg(x) containing f (x) and g(x) respectively. Then there
exist an open set U containing x in X such that f (U ) ⊂ Uf (x) and f (U ) ⊂ Ug(x) . But
this gives us a contradiction as f and g agree on A and U meets A non trivially.
6
CHAPTER 1
Topology
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-5: Homeomorphism
In studying group theory, metric spaces we have observed structure preserving map-
pings such as isomorphism, isometry. In this module we will discuss structure preserving
mappings of topological spaces.
Example 2. Now we can observe that any two same type of intervals are homeomorphic.
Without loss of generality let us consider open intervals (0, 1) and (3, 4). The mapping
f : (0, 1) → (3, 5), defined by f (x) = ax + b gives a homeomorphism. We have just to
choose a, b suitable real numbers.
2
Example 3. This example shows that R and (0, ∞) are homeomorphic.
Proof. Let f : R → (0, ∞) defined by f (x) = ex . The following diagram clearly shows
that f is a homeomorphism.
Example 4. This example shows that unit square and unit circle are homeomorphic.
3
Example 7. The mapping p : [0, 1) → S 1 defined by p(x) = e2πix is a continuous
bijection, which is not a homeomorphism.
Let S n denote the n-dimensional sphere {x ∈ Rn+1 : kxk = 1} taken with the subspace
topology. We claim that removing a single point from S n gives a space homeomorphic
to Rn . Which point we remove is irrelevant because we can rotate any point of S n into
any other; for convenience we choose to remove the point N = (0, 0, . . . , 0, 1). Now the
set of points of Rn+1 which have zero as their final coordinate, when given the induced
4
Definition 2. A property say P of a topological space is said to be topological property if
when ever two topological spaces X and Y are homeomorphic and one posses the property
then the other will posses the property.
Example 8. Let X = R and let d be the usual metric on R. Let Y = (0, 1) (the open
interval) and let ρ be the usual metric on (0, 1). Then X and Y are homeomorphic as
topological spaces, but (X, d) is complete and (Y, ρ) is not. So the completeness is not a
topological property.
The following result will be needed in the study of manifold. By a disc we shall mean
any space homeomorphic to the closed unit disc D in R2 . If A is a disc, and if h : A → D
is a homeomorphism, then h−1 (S 1 ) is called the boundary of A and is written ∂A.
5
Theorem 3. Any homeomorphism from the boundary of a disc to itself can be extended
to a homeomorphism of the whole disc.
We have already discussed about finite product topology. One can observe easily that
each projection map is continuous. Let us examine the following example.
Example 9. If we view points in the unit circle S1 in R2 as angles θ, then polar coordi-
nates give a homeomorphism f : S1 ×(0, ∞) → R2 \{0} defined by f (θ, r) = (rcosθ, rsinθ).
This is one-to-one and onto since each point in R2 , other than the origin has unique polar
coordinates (θ, r). To see that f is a homeomorphism, just observe that it takes a basic
open set U × V , (where U is an open interval (θ0 , θ1 ) and V is an open interval (r0 , r1 ))
to an open polar rectangle and such rectangles form a basis for the topology on R2 \ {0},
as a subspace of R2 . By restricting f to a product S1 × [a, b] for 0 < a < b we obtain
a homeomorphism from this product to a closed annulus in R2 , the region between two
concentric circles.
sider [0, 1] and R. This following Theorem shows that we cann’t hope a Theorem like
Cantor Bernstein Theorem.
6
CHAPTER 1
Topology
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-6: Product topology
We have already known how to form the finite product of topological spaces. Now we
wish to generalize this for arbitrary product.
Wer want to mention here that the Axiom of Choice is precisely the statement that
Q
if Xi 6= ∅ for all i, then i∈I Xi 6= ∅. Idea behind it is that if we can make well ordered
the set I then arbitrary product of non empty sets is non empty.
Next we want to give a topology on the infinite product which will come from the topol-
ogy of factor spaces. Let us first examine that what happens if we proceed like finite case.
The above mentioned property is so natural, one should expect that in any successful
generalization of finite product this property will carry over.
Let us try to proceed as the finite case and see what happens. Given topological spaces
{Xn : n ∈ N} let
Y
B={ Ui : Ui is an open set in Xi }.
n∈N
2
Then it can be easily shown that the collection B is a basis. One might think to consider
this collection as a basis for infinite product. But this definition will not be a successful
generalization as can be seen from the following example.
Example 1. Let us consider R with usual topology and consider the topology constructed
as above on RN . Consider the mapping f : R → RN . If the above definition is a successful
generalization then f : R → RN defined by f (x) = (x, x, . . .) should be continuous as the
finite case. But if we consider the open set
1 1 1 1
U = (−1, 1) × (− , ) × . . . (− , ) . . . .
2 2 n n
Then f −1 (U ) = 1 1
T
n∈N (− n , n ) = {0}, which is not open. Hence f is not continuous
This example shows that the straight forward generalization from finite product to
infinite product is not a successful generalization.
Si = {πi−1 (Ui ) : i ∈ I}
S
and S = i∈I Si .
Q
It can be easily verified that S is a subbase for some topology on i∈I Xi . The topol-
Q
ogy generated by S will be called product topology on i∈I Xi .
If we view in the coordinate wise pattern, a typical subbasic element will be of the
form
Y
Ui , where Ui = Xi for all i except a single coordinate .
i∈I
As we know that we get basis from subbasis taking all possible finite intersection, here
a typical basis element will be of the form i∈F πi−1 (Ui ) where F is a finite subset of I
T
and Ui is an open set in Xi . Coordinate wise a typical subbasic element will be of the
form
Y
Ui , where Ui = Xi for all except finitely many i’s .
i∈I
As usual taking basic open sets in the factor spaces in place of open sets also give rise to
the product topology.
3
The following Theorem is a very useful one.
Q
Proposition 2. Let {Xi : i ∈ I} be a collection of topological spaces and i∈I Xi be
product space.
Then projection πi onto Xi is continuous and open map for each i.
Proof. Write basic open sets of the product in coordinate form. The proof will be clear.
The projection maps need not be closed. In fact The projection map π : R2 → R is
not closed map. Consider {(x, x1 ) : x > 0}.
Q
For the converse let x = (xi )i∈I be a point in i∈I Ai . We claim that for any index
i ∈ I xi ∈ Ai . Let Vj be an arbitrary open set of Xj containing xj . Since πj−1 (Vj ) is
Q Q
open in i∈I Xi in product topology, it contains a point y = (yi )i∈I in i∈I Ai . Then yj
belongs to Vj ∩ Aj . It follows that xj ∈ Aj .
The following Theorem shows that product topology is the successful generalization
of finite product topology.
Q
Theorem 1. Let f : X → a∈I Xa be given by the equation f (x) = (fa (x))a∈I . Where
Q
fa : X → Xa for each a. Let a∈I Xa have the product topology. Then the function f is
continuous if and only if each function fa is continuous.
Q
Proof. Let f : X → a∈I Xa be continuous. The function fb is in fact equal with the
composition map πb ◦ f . So being the composite of two continuous functions, it is con-
tinuous.
4
Converse follows by the equality f −1 (πb−1 (Ub )) = fb−1 (Ub ).
Q
Theorem 2. The product topology is the weak topology on Xi for which each projection
πj is continuous.
Proof. If τ is any topology on the product in which each projection is continuous, then
for each j, if Gj is open in Xj , so that πj−1 (Gj ) ∈ τ . Consequently, the members of
a subbasis for the product topology all belong to τ and hence the product topology is
contained in τ .
Q
Theorem 3. Let Aα be a subspace of Xα for each α ∈ J. Then α∈J Aα is a subspace
Q
of α∈J Xα .
Q Q
Proof. Let us consider a basic open set α∈J Uα of α∈J Aα . Then there exists a finite
set {α1 , α2 , . . . , αk } ⊂ J such that for all α ∈ J \{α1 , α2 , . . . , αk } we have Uα = Aα . Since
each Aα is a subspace of Xα there exist open sets Oα in Xα such that Uα = Oα ∩Aα for each
Q
α ∈ J. Let us consider the set α∈J Oα where Oα = Xα for all α ∈ J \ {α1 , α2 , . . . , αk }.
Q T Q Q
Then we can see that α∈J Oα α∈J Aα = α∈J Uα . This proves the result.
We end this section with the following definition, which will be required in future.
5
Definition 4. Let {fα : X → Xα : α ∈ I} be a collection of continuous maps. Then the
Q
evaluation map e : X → Xα induced by the collection {fα : X → Xα } is defined by
[e(x)]α = fα (x) for each x ∈ X.
6
CHAPTER 2
Countability axioms
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-1: Metrizable spaces
One of the most important topological spaces are metrizable topological spaces. Topol-
ogy of such spaces has to be defined in terms of a metric on the set. Topologies given in
this way lie at the heart of modern analysis. In this section, we shall define the metric
topology and shall give a number of examples.
Proposition 1. If d is a metric on the set X, then the collection of all r-balls Bd (x, r),
for x ∈ X and r > 0, is a basis for a topology on X, called the metric topology induced
by d.
Proof. The first condition for a basis is trivial, since x ∈ Bd (x, r) for any r > 0.
First we claim that if y is a point of the basis element Bd (x, r), then there is a basis element
Bd (y, s) centered at y that is contained in Bd (x, r). Define s to be the positive number
r − d(x, y). Then Bd (y, s) ⊂ Bd (x, r). In fact if z ∈ Bd (y, s), then d(y, z) < r − d(x, y)
from which we conclude that
Now we shall be able to prove the second condition for a basis, let B1 and B2 be two
basis elements and let y ∈ B1 ∩ B2 . We have just shown that, we can choose positive
numbers r1 and r2 so that B(y, r1 ) ⊂ B1 and B(y, r2 ) ⊂ B2 . Letting 0 < r < min{r1 , r2 }
we conclude that B(y, r) ⊂ B1 ∩ B2 .
2
Now we can introduce the following definition.
Proof. In the proof first we show that ρ is in fact a metric and then we show that these
two metrics generate same topology.
The first two conditions are obvious. To check the triangle inequality if ρ(x, y) or
ρ(y, z) are greater or equal to 1 then ρ(x, z) ≤ ρ(x, y) + ρ(y, z).
So let both ρ(x, y) or ρ(y, z) are strictly less than 1. Then we have
3
Definition 4. Let us define d on Rn as follows :
p
d(x, y) = (x1 − y1 )2 + (x2 − y2 )2 + . . . + (xn − yn )2 .
In the following Theorem we show that two topologies on a set determined by metrics
can be compared in terms of open balls. This is a particular form of a Theorem we
already told for bases.
Lemma 1. Let d and ρ be two metrics on the set X; let τ1 and τ2 be two topologies
induced by d and ρ respectively. Then τ1 is finer than τ2 if and only if for each x in X
and each r > 0, there exists s > 0 such that
Theorem 1. The topologies on Rn induced by the Euclidean metric d, the square metric
ρ and the product topology are equivalent.
Bd (x, r) ⊂ Bρ (x, r)
r
Bρ (x, √ ) ⊂ Bd (x, r).
n
for all x and r. It follows that the topologies generated by these two metrics are the
same.
Next we show that the product topology is the same as that generated by the metric ρ.
4
First let B be a basic open set in the product topology, that means product of open
intevals and let
x ∈ B = (a1 , b1 ) × (a1 , b1 ) × . . . × (an , bn ).
Then for i there exist εi such that (xi − ε, xi + ε) ⊂ (ai , bi ). Let us choose ε =
min{ε1 , ε2 , . . . , εn }. Then Bρ (x, ε) ⊂ B. As a result, the ρ-topology is finer than the
product topology.
Conversely, let Bρ (x, ε) be a basis element for the ρ-topology. Given the element
y ∈ Bρ (x, ε), we need to find a basis element B for the product topology such that
y ∈ B ⊂ Bρ (x, ε). We just take B = ni=1 (yi − ε, Yi + ε).
Q
But this equation does not always make sense, for the series in question need not converge.
Similarly, one can attempt to generalize the square metric ρ to Rω by defining
If we replace the original metric d in a metric space (X, d) by its bounded counterpart
d(x, y) = min{d(x, y), 1}, then the sup metric makes sense; it gives a metric on countable
product, called the uniform metric. In this lecture let us concentrate on Rω .
where d is the standard bounded metric on R. It is easy to check that ρ is indeed a metric;
it is called the uniform metric on Rω , and the topology it induces is called the uniform
topology.
The relation between uniform topology and the product topology is the following:
5
Theorem 2. The uniform topology on Rω is finer than the product topology.
Q
Proof. Let n Un be a basic open set in the product topology containing a point x =
(xn )n . Let n1 , n2 , . . . , nn be the indices for which Un 6= R. Then for Then for each i,
choose ε > 0 so that the ε-balls centered at xni in the d metric is contained in Uni . This
we can do because Uni is open in R. Let ε = min{ε1 , ε2 , . . . , εn }; then the ε-ball centered
at x in the ρ metric is contained in n Un . For if z is a point of Rω such that ρ(x, z) < ε,
Q
Q
then d(xn , zn ) < ε for all n, so that z ∈ n Un . It follows that the uniform topology is
finer than the product topology.
Theorem 3. Let d(a, b) = min{|a − b|, 1} be the standard bounded metric on R. If x and
y are two points of Rω , define
d(xi , yi )
ρ(x, y) = sup .
i
Proof. Let us first proof that ρ is a metric. First two conditions are clear. Let us prove
the triangle inequality.
Therefore
d(xi , zi )
sup ≤ ρ(x, y) + ρ(y, z).
i
Now let U be open in the metric topology and let x ∈ U . We shall find an open set
V in the product topology such that x ∈ V ⊂ U .
Choose an r-ball Bρ (x, r) lying in U . Then choose a positive integer N such that
1
N
< r. Finally, let V = (x1 − r, x1 + r) × (x2 − r, x2 + r) × . . . × (xN − r, xN + r) × R × R . . .
be a basic open set in the product topology. We claim that V ⊂ Bρ (x, r): Given any
y ∈ Rω ,
d(xi , zi ) 1
≤ for i ≥ N.
i N
6
Therefore
d(x1 , z1 ) d(x2 , z2 ) d(xN , zN ) 1
ρ(x, y) ≤ max , ,..., , .
1 2 N N
If y is in V , this expression is less than r, so that V ⊂ Bρ (x, r) as desired.
Q
Conversely let n Un be a basic open set in the product topology. Let a1 , a2 , . . . , an be
the indices for which Ua 6= R. Given x ∈ U , choose an interval (xi −εi , xi +εi ) in R lying in
Ui for i = a1 , a2 , . . . , an ; choose each εi ≤ 1. Then define ε = min{ εii : i = a1 , a2 , . . . , an }.
We claim that x ∈ Bρ (x, ε) ⊂ U .
Let y be a point of Bρ (x, ε). Then for all i,
d(xi , yi )
≤ ρ(x, y) < ε.
i
εi
Q
Finally i = a1 , a2 , . . . , an , ε ≤ i
so that d(xi , yi ) ≤ εi ≤ 1. Therefore y ∈ n Un .
7
CHAPTER 3
Separation axioms
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-2: Separation Axioms,
Normality
Proof. Let X be a normal space and Y be a closed subspace of X. Then any two disjoint
closed sets in Y is also closed in X. Then normality of X can be used to prove the
normality of Y .
In the following example we show that normality is not even finitely productive prop-
erty.
2
subset of the set D, as follows :
f (A) = D ∩ UA ∅ A L
f (∅) = ∅ .
f (L) = D
Next as D is countable and L has cardinality of the continuum, there exist a bijection
ϕ : P(D) → L. Then ϕ ◦ f : P(L) → L is an injective mapping, which is a contradiction.
Proof. We assert that every interval of the form (x, y] is open in a well-ordered set X. In
fact if X has a largest element and y is that element, (x, y] is just a basis element about
y. If y is not the largest element of X, then (x, y] equals the open set (x, y) , where y
is the immediate successor of y. Now let A and B be disjoint closed sets in X, assume
for the moment that neither A nor B contains the smallest element a0 of X. For each
a ∈ A, there exists a basis element about a disjoint from B it contains some interval of
the form (x, a], as a is not the smallest element of X. Choose, for each a ∈ A, such an
interval (xa , a] disjoint from B.
Similarly, for each b ∈ B, choose an interval (yb , b] disjoint from A. The sets
[ [
U= (xa , a] and V = (yb , b].
a∈A b∈B
are open sets containing A and B, respectively; we claim that they are disjoint. In
fact if z ∈ U ∩ V . Then z ∈ (xa , a] ∩ (yb , b] for some a ∈ A and some b ∈ B. Assume that
a < b. Then if a ≤ yb , the two intervals are disjoint, while if a > yb , we have a ∈ (yb , b],
contrary to the fact that (yb , b] is disjoint from A. A similar contradiction occurs if b < a.
3
Finally, assume that A and B are disjoint closed sets in X and A contains the smallest
element a0 of X. The set {a0 } is both open and closed in X. Then by above there exist
disjoint open sets U and V containing the closed sets A − {a0 } and B, respectively. Then
U ∪ {a0 } and V are disjoint open sets containing A and B, respectively.
In the coming discussions we shall see that Linearly ordered topological spaces satisfies
more stronger separation property called completely normal.
One of the useful properties of the set N of positive integers is the fact that each of its
nonempty subsets has a smallest element. This property leads to the following definition.
Theorem 3. Every nonempty finite ordered set has the order type of a section {1, 2, . . . , n}
of N, so it is well-ordered.
For any well ordered set X and given α ∈ X, let Sα let us denote the set Sα = {β ∈
X : β < α}. It is called the section of X by α.
Theorem 5. There exists a well-ordered set A having a largest element Ω, such that the
section SΩ of A by Ω is uncountable but every other section of A is countable.
4
Theorem 6. If A is a countable subset of SΩ , then A has an upper bound in SΩ .
With these few properties of ordinal numbers, which will be useful for us for further
discussions, we end this module.
5
CHAPTER 3
Countability axioms
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-1: Properties of normal
spaces
We have already proved that Rl × Rl is not normal. In the following we shall provide
another example to show product of normal spaces may not be normal. This example is
taken from J. R. Munkres.
Example 1. Consider the well-ordered set SΩ + 1, in the order topology, and consider
the subset SΩ , in the subspace topology (which is the same as the order topology). Both
spaces are normal, as linearly ordered topological spaces are normal. We wish to prove
that (SΩ + 1) × SΩ is not normal.
First, we consider the space (SΩ + 1) × (SΩ + 1), and its “diagonal” 4 = {(x, x) : x ∈
SΩ + 1}. Because SΩ + 1 is Hausdorff, 4 is closed in (SΩ + 1) × (SΩ + 1). If U and V
are disjoint neighborhoods of x and y, respectively, then U × V is a neighborhood of x × y
that does not intersect 4.
Therefore, in the subspace SΩ × (SΩ + 1), the set
is closed. Similarly the set B = SΩ × {Ω} is closed in SΩ × (SΩ + 1). The sets A and B
are disjoint.
Assuming there exist disjoint open sets U and V of SΩ × (SΩ + 1) containing A and
B, respectively, we shall derive a contradiction.
Given x ∈ SΩ , consider the vertical slice x × (SΩ + 1). We assert that there is some point
β with x < β < Ω such that (x, β) lies outside U . In fact if U is contained all points
(x, β) for x < β < Ω then the top point (x, Ω) of the slice would be a limit point of U ,
2
which it is not because V is an open set disjoint from U containing this top point
Choose β(x) to be such a point; just to be definite, let β(x) be the smallest element of
S( Ω) such that x < β(x) < Ω and (x, β(x)) lies outside U .
Let us define a sequence of points of S( Ω) as follows: Let x1 be any point of S( Ω). Let
x2 = β(x1 ), and in general,xn+1 = β(xn ). We have
x1 < x2 < . . . ,
in the product space. Now we have a contradiction, for the point (b, b) lies in the set
A, which is contained in the open set U ; and U contains none of the points (xn , β(xn ).
Now we shall enter to some deeper study of normal spaces. Two important Theorem to
note here are Urysohn Lemma ant Tietz extension lemma. The first one gives information
about rich source of continuous functions and the second one tells about extension of
continuous functions. Let us start with following Theorem
Proof. Let X be a regular space with a countable basis say B = {Bn ; n ∈ N}. Let A and
B be disjoint closed subsets of X. The idea of the proof is to cover A with some basic open
sets and to cover B with some basic open sets and then to to eliminate the overlapped
parts. Each point x of A has a neighborhood U not intersecting B. Using regularity,
choose a neighborhood V of x whose closure lies in U ; we can do these operations taking
elements from B. By choosing such a basis element for each x in A, we construct a
countable covering of A by open sets whose closures do not intersect B. Since this
covering of A is countable, we can index it with the positive integers; let us denote it by
{Un }. As an immediate application of the above Theorem we can say that real line with
usual topology is normal.
3
Similarly, choose a countable collection {Vn } of open sets covering B, such that each
S S
set Vn is disjoint from A. The sets U = n Un and V = n Vn are open sets containing
A and B. Given n ∈ N, define
! !
[ [
Un0 = Un \ Vn and Vn0 = Vn \ Un .
n n
Then each Un0 and Vn0 are open sets. It is easy to observe that {Un0 } is a cover of A and
{Vn0 } is a cover of B.
The collection {Un0 } is a cover of A because each x in A belongs to Un for some n, and
x belongs to none of the sets V i . Similarly, the collection {Vn0 } covers B. Now let us set
[ [
U0 = Un0 and V 0 = Vn0 .
n n
It remains to show that U 0 and V 0 are disjoint. For if x ∈ U 0 ∩ V 0 then x ∈ Uj0 ∩ Vk0
for some j and k. Suppose that j < k. It follows from the definition of U 0 that x ∈ Uj ,
and since j ≤ k it follows from the definition of Vk0 that x 6∈ U j . A similar contradiction
arises if j ≥ k.
Quite same construction like above Theorem proves the following Theorem.
Theorem 3. A space is completely normal iff every pair of separated subsets can be
separated by neighborhoods.
4
Conversely take a set Y ⊂ X and two disjoint subsets A, B ⊂ Y closed in Y . AX ∩ B =
AX ∩ Y ∩ B = AY ∩ B = ∅ . Similarly, B X ∩ A = ∅. Therefore, A and B can be separated
by neighborhoods in X and their intersections with Y separate A and B in Y .
In the following we shall examine the completely normality of the some spaces.
Proof. If a set is well-ordered then every element has a successor and {(a, x]} is a basis
at x where a < x or a = −∞. Therefore, for a pair of separated sets we can cover
each set with such neighborhoods that do not intersect the other set. Moreover, the
neighborhoods belonging to one set do not intersect the neighborhoods belonging to the
other set.
Proof. Indeed, the proof of the fact that Rl is Completely normal, is extremely similar to
the previous example: both use the fact that there is a basis at x with sets of the form
[x, a). This shows that every point in one closed set has such a neighborhood that does
not intersect the other set. Then coverings by such basis sets are disjoint automatically.
So, we obtain the disjoint open neighborhoods for any pair of sets such that neither one
contains limit points of the other one.
Question 2. Does every regular space with a countable basis is completely normal?
Every regular second-countable space is normal. Also every its subspace is also regular
and second-countable. Therefore, every regular second-countable space is completely
normal.
5
CHAPTER 3
Separation Axioms
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-4: Urysohn’s Lemma
Theorem 1 (Urysohn’s Lemma). A space X is normal if and only if for any two disjoint
closed sets E and F there exists a continuous functions f : X :→ R such that
f (E) = 0 and f (F ) = 0.
This is a deep Theorem, both from the point of view that its proof, which involves
really original idea, also from the point of view of its application.
E ⊂ U 1 ⊂ U 1 ⊂ V.
2 2
By successive application of normality criteria gives that there exist open sets U 1 and U 3
4 4
such that
E ⊂ U 1 ⊂ U 1 ⊂ U 1 ⊂ U 1 ⊂ U 3 ⊂ U 3 ⊂ V.
4 4 2 2 4 4
Continuing this manner for each dyadic rational number r ∈ (0, 1), an open set Ur
such that
2
E ⊂ Ur , 0 < r < 1,
Ur ⊂ V, 0 < r < 1.
0 if x ∈ Ur for all r > 0
f (x) =
sup{r : x 6∈ Ur }
Proof. Suppose there exists a continuous function f : X → [0, 1] such that f (x) = 0
for x ∈ A, and f (x) > 0 for x 6∈ A. Then A = f −1 (0) must be closed. Now A =
T −1 1 1
nf (− n , n ) and each f −1 (− n1 , n1 ) is an open set. Hence A is a Gδ set also.
Conversely let A be a closed Gδ set. Then there exists a sequence (Un ) of open sets such
T
that A = n Un . Then for each n there exists a continuous function fn which vanishes
3
on A and equal to 1 on X − Un . Now take
X |fn |
f= .
n
2n
Proof. We know that any regular second countable space X is normal so that we can
invoke Urysoh’s lemma. Let B = {Bn : n ∈ N} be a countable base. Then for any
m, n ∈ N if Bm ⊂ Bn there exists some f ∈ C ∗ (X) such that f (Bm ) = 0 and f (Bn ) = 1.
In this way we get a countable collention of functions say {fn : n ∈ N} which does the
above job. We define e : X → RN to be πn (e(x)) = fn (x). We shall prove that e is an
embedding. Continuity follows from the fact that each factor map fn is continuous. If
x 6= y then there exists Bm Bn ∈ B such that x ∈ Bm , y 6∈ Bn and Bm ⊂ Bn . So there
exists some fk such that fk (x) = 0 and fk (y) = 1.
It’s remain to show that e : X → e(X) is open. Let U be open in X, then we need
to show that set e(U ) is open in e(X). Let v ∈ e(U ). We have to find an open set W
of e(X) such that v ∈ W ⊂ e(U ). Let u be the point of U such that e(u) = v. Choose
an index N for which fN (u) > 0 and fN (X − U ) = {0}. Take the open ray (0, +∞) in
−1
R, and let V be the open set πN ((0, +∞)) of RN . We claim that v ∈ W ⊂ e(U ) First,
v ∈ W because πN (u) = πN (e(u)) = fN (u) > 0. Second, W ⊂ e(U ). For if z ∈ W , then
z = e(x) for some x ∈ X, and πN (z) ∈ (0, +∞). Since πN (z) = πN (e(x)) = fN (x), and
fN vanishes outside U the point x must be in U . Then z = e(x) is in e(U ) as desired.
Thus e is an imbedding of X in RN .
Definition 1. If E and F be two disjoint closed sets in a space X and there exists a
continuous functions f : X :→ R such that f (E) = 0 and f (F ) = 0. We say that E and
F can be separated by a continuous function.
The Urysohn lemma says that if every pair of disjoint closed sets in X can be separated
by disjoint open sets, then each such pair can be separated by a continuous function. The
converse is trivial, for if
f (E) = 0 and f (F ) = 0.
4
is the function, then f −1 [0, 13 ) and and f −1 ( 32 , 0] are disjoint open sets containing A and
B, respectively.
This fact leads to the analogue question for regular spaces, that is whether regularity
is equivalent with the fact that points and closed sets can be separated by continuous
functions. Unfortunately this does not hold in general, an example will be provided latter.
This fact leads to the following definition.
Definition 2. A space X is said to be Completely Regular if all finite sets are closed and
for any x ∈ X and a closed sets F not containing x, there exists a continuous functions
f : X :→ R such that
f (x) = 0 and f (F ) = 0.
This is quite clear that Completely regular spaces are regular and normal spaces are
Completely Regular.
5
CHAPTER 3
Separation axioms
BY
Dr. Dibyendu De
Associate Professor
Department of Mathematics
University Of Kalyani
West Bengal, India
E-mail : dibyendude@gmail.com
1
Module-5: Tietze Extension
Theorem
Let X be the closed unit interval and A = (0, 1). If we consider the function by
1
f (x) = x
defined on (0, 1), then f cannot be extended to the closed interval [0, 1]. In fact
any continuous function defined on [0, 1] must be bounded. Tietze extension theorem is
one of the immediate consequence of the Urysohn lemma which deals with the problem
of extending a continuous real-valued function that is defined on a subspace of a space
X to a continuous function defined on all of X. This theorem is important in many of
the applications of topology.
Proof. The idea of the proof is approximation. We shall construct a sequence of continu-
ous functions defined on the entire space X, such that the sequence converges uniformly,
and such that the restriction of each function to C approximates f . Then the limit
function will be continuous, and its restriction to C will equal f .
Let X be a normal space, C a closed subset, and f : C → R a continuous map. Let
us first consider the bounded case, i.e. say |f (x)| ≤ M for all x ∈ C.
M
Let A1 = {x ∈ C : f (x) ≥ 3
} and B1 = {x ∈ C : f (x) ≥ − M3 }. Then A1 and B1
are obviously disjoint, and they are both closed subsets of C. But C is closed in X, and
therefore A1 and B1 must be closed in X. By Urysohn’s lemma we can find a continuous
map g1 : X → [− M3 , M3 ] which takes the value M
3
on A1 and − M3 on B1 and which takes
values in (− M3 , M3 ) on X − (A1 ∪ B1 ). Notice that |f (x) − g1 (x)| ≤ 2M
3
on C.
Now consider the function f − g1 and let A2 consist of those points of C for which
2M
|f (x)−g1 (x)| ≥ 9
, and B2 those points for which |f (x)−g1 (x)| ≤ − 2M
9
. Again applying
2
Uryshon’s lemma a second time we can find a map g2 : X → [− 2M
9
, 2M
9
] which takes the
2M
value 9
on A2 and − 2M
9
on B2 , and values in (− 2M
9
, 2M
9
) on the remaining points of X.
4M
If we compute f (x) − g1 (x) − g2 (x), we see that |f (x) − g1 (x) − g2 (x)| ≤ 9
on C.
n−1 M n−1 M
By repeating this process we can construct a sequence of maps gn : X → [− 2 3n
,2 9
]
which satisfy:
2n M
(a) |f (x) − g1 (x) − g2 (x) − . . . gn (x)| ≤ 3n
on C; and
2n−1 M
(b) |gn (x)| < 3n
on X − C.
We now define
∞
X
g(x) = gn (x)
n=1
for all x in X. The convergency of the infinite series follows from the comparison
theorem of calculus, it converges by comparison with the geometric series
∞ n−1
1X 2
.
3 n=1 3
To show that g is continuous, we must show that the sequence sn of partial sums converges
to g uniformly. This fact follows at once from the “Weierstrass M-test”.
Finally we show that g(x) = f (x) for x ∈ C. g(x) is by definition the limit of the
infinite sequence sn (x) of partial sums. Since
n n
X 2
|f (x) − gi (x)| = |f (x) − sn (x)| ≤
i=1
3
for all x in C, it follows that sn (x) → f (x) for all x ∈ C. Therefore, f and g agree on C.
3
The following corollary can be obtained easily.
Though we have used Urysohn Lemma in the proof Tietz extension Theorem, but if
a space satisfies Tietz extension Theorem, then it satisfies Urysohn Lemma.
Proposition 1. Show that the Tietze extension theorem implies the Urysohn lemma.
Proof. We need to use pasting lemma and then to apply Tietze extension theorem.
Definition 1. A space Y is said to have the universal extension property if for any given
normal space X, any closed subset A of X, and any continuous function f : A → Y ,
there exists an extension of f to a continuous map of X into Y .
x
Example 1. S1 is a retract of R2 −{0}. The map r : R2 −{0} → S1 defined by r(x) = kxk
is a retraction.
4
Proposition 3. If X has the universal extension property, then X is an absolute retract.