You are on page 1of 38

Point Set Topology Homework

Quinton Aboud
Spring 2015

These are the assigned problems (out of Topology by Munkres) and the solutions I turned in. Only one
problem was graded from each assignment, so there are probably a lot of errors that went unnoticed.

1 January 21, 2015

1; 5: Let A be a nonempty collection of sets. Determine the truth of each of the following statements
and their converses:
S
1. x AA A = x A for at least one A A.
S
2. x AA A = x A for every A A.
T
3. x AA A = x A for at least one A A.
T
4. x AA A = x A for every A A.

(a) True. This is because this is the definition of the arbitrary


S union.
The converse: x A for at least one A A = x AA A, is also true because this is the definition of
the arbitrary union. S
(b) False. Let A0 A and let x A0 , but x 6 B for all B A {A0 }. Then x AA A and x is not in
each A A (unless A has only one element). S
The converse: x A for every A A = x AA A, is true. This is because if x is in every A A, then
surely it is in at least one A A.
T
(c) True. This is because if x AA A, then x is in every T A A, then surely it is in at least one A A.
The converse: x A for at least one A A = x AA A, is false. Consider when we have the case
specified in part (b). Then x is in Tat least one A A, namely x A0 . So the antecedent is true. However,
since x is not in every A A, x 6 AA A, so the consequent is false.
(d) True. This is because this is the definition of the
T arbitrary intersection of sets.
The converse: x A for every A A = x AA A, is true because this is again, the definition and
definitions are if and only if statements.

1; 6: Write the contrapositive of each statement in Exercise 5.

S
(a) x 6 A for every A A = x 6 AA S A.
(b) x 6 A for at least one A A =
T x
6 AA A.
(c) x 6 A for every A A = x 6 AA A. T
(d) x 6 A for at least one A A = x 6 AA A.

1
1; 8: If a set A has two elements, show that P(A) has four elements. How many elements does
P(A) have if A has one element. Three elements? No elements? Why is P(A) called the power set
of A?

We generalize the result first.


Theorem 1. Suppose a set A has n 0 elements. Then P(A) has 2n elements.
Proof. First note that When A = , then there is one possible subset, the empty set itself. This is because
the empty set is a subset of every set, including itself. We induct on n. When n = 1, we get that the subsets
of A are and A. So P(A) has two elements. Now suppose that our theorem is true for all natural numbers
m < n. Consider the set A that has n elements. Then let a A. So A {a} has n 1 elements. So by
hypothesis, P(A {a}) has 2n1 elements. This means that there are 2n1 distinct subsets of A that do
not contain the element a. Every other subset of A will have to contain a. So if we take the union of each
of these 2n1 subsets with {a}, we will have all possible subsets of A that contain a. By counting, there are
2n1 subsets that contain a and 2n1 that do not contain a. Therefore, the total number of subsets of A is
2n1 + 2n1 = 2(2n1 )
= 2n1+1
= 2n ,
as desired. Our claim follows.
Thus, when A has 2 elements P(A) has 22 = 4; when A has 3 elements, P(A) has 23 = 8. The cases when
A has 1 and no elements were addressed above.
P(A) is called the power set of A because its size is given by a formula (namely |P(A)| = 2|A| ) that is
exponential.

2 January 23, 2015

2; 2: Let f : A B and let Ai A and Bi B for i = 0 and i = 1. Show that f 1 preserves


inclusions, unions, intersections, and differences of sets:

1. B0 B1 = f 1 (B0 ) f 1 (B1 ).
2. f 1 (B0 B1 ) = f 1 (B0 ) f 1 (B1 ).
3. f 1 (B0 B1 ) = f 1 (B0 ) f 1 (B1 ).

4. f 1 (B0 B1 ) = f 1 (B0 ) f 1 (B1 ).


Show that f preserves inclusions and unions only.
5. A0 A1 = f (A0 ) f (A1 ).
6. f (A0 A1 ) = f (A0 ) f (A1 ).

7. f (A0 A1 ) f (A0 ) f (A1 ); show that equality holds if f is injective.


8. f (A0 ) f (A1 ) f (A0 A1 ); show that equality holds if f is injective.

(a) Suppose x f 1 (B0 ). Then there exists x0 B0 s.t. f (x) = x0 . Then since B0 B1 , x0 B1 , i.e.
f (x) B1 . Thus, x f 1 (B1 ).

2
(b) Suppose x f 1 (B0 B1 ). Then f (x) B0 or f (x) B1 . Without loss of generality, say f (x) B0 .
Then x f 1 (B0 ) and so x f 1 (B0 ) f 1 (B1 ). On the other hand, suppose x f 1 (B0 ) f 1 (B1 ).
Then x f 1 (B0 ) or x f 1 (B1 ). W.L.O.G., say x f 1 (B0 ). Then f (x) B0 and so f (x) B0 B1 ,
i.e. x f 1 (B0 B1 ). Thus f 1 (B0 B1 ) = f 1 (B0 ) f 1 (B1 ).
(c) Suppose x f 1 (B0 B1 ). Then f (x) B0 B1 , so f (x) B0 and f (x) B1 . Then x f 1 (B0 ) and
x f 1 (B1 ). Thus, x f 1 (B0 ) f 1 (B1 ). On the other hand, suppose x f 1 (B0 ) f 1 (B1 ). Then
x f 1 (B0 ) and x f 1 (B1 ). So f (x) B0 and f (x) B1 . Thus f (x) B0 B1 . Hence, x f 1 (B0 B1 ).
Therefore, f 1 (B0 B1 ) = f 1 (B0 ) f 1 (B1 ).
(d) Suppose x f 1 (B0 B1 . Then f (x) B0 B1 , i.e. f (x) B0 and f (x) 6 B1 . So x f 1 (B0 )
and x 6 f 1 (B1 ), and x f 1 (B0 ) f 1 (B1 ). On the other hand, suppose x f 1 (B0 ) f 1 (B1 ), then
x f 1 (B0 ) and x 6 f 1 (B1 ). Then f (x) B0 B1 , so x f 1 (B0 B1 ). Therefore, f 1 (B0 B1 ) =
f 1 (B0 ) f 1 (B1 ).
(e) Suppose y f (A0 ). Then there exists x A0 s.t. f (x) = y. So x A1 , thus y f (A1 ). So
f (A0 ) f (A1 ).
(f ) Suppose y f (A0 A1 ). Then there exists x A0 A1 . W.L.O.G. say x A0 . Then y f (A0 ) and
y f (A0 ) f (A1 ). On the other hand, suppose y f (A0 ) f (A1 ). W.L.O.G., say y f (A0 ). Then there
exists x A0 s.t. f (x) = y. Then x A0 A1 , so f (x) = y f (a0 A1 ) and so f (A0 A1 ) = f (A0 ) f (A1 ).

(g) Suppose y f (A0 A1 ). Then there exists x A0 A1 s.t. f (x) = y. So x A0 and y f (A0 ).
Similarly, x A1 so y f (A1 ). Thus, y f (A0 ) f (A1 ) and f (A0 A1 ) f (A0 ) f (A1 ).
Now suppose f is injective. If y f (A0 ) f (A1 ), then y f (A0 ) and y f (A1 ). Since f is injective, there
exists a unique x in both A0 and A1 s.t. f (x) = y. Thus, x A0 A1 , so f (x) = y f (A0 A1 ).
(h) Suppose y f (A0 ) f (A1 ). Then y f (A0 ) and y 6 f (A1 ). So there exists x A0 s.t. f (x) = y and,
for all x0 A1 we have f (x0 ) 6= y. So x A0 A1 . Thus f (x) = y f (A0 A1 ).
Now suppose f is injective. Then if y f (A0 A1 ), there exists a unique (since f is injective) x A0 A1 s.t.
f (x) = y. So x A0 and x 6 A1 . Thus f (x) = y f (A0 ) and f (x) = y 6 f (A1 ) and so y f (A0 ) f (A1 ).

2; 3: Show (b), (c), (f), and (g) of Exercise 2 hold for arbitrary unions and intersections.

(b) Suppose x f 1 ( bB B). Then


S S
S f (x) BB B. W.L.O.G., let B0 B be s.t. f (x) B0 . Then
1 1
x f (B0 ), and x f (B0 ) ( BB{Bo } f 1 (B)) = BB f 1 (B). On the other hand, suppose
S

x BB f 1 (B). Then let f 1 (B0 ) be s.t. x f 1 (B0 ). Then f (x) B0 , so f (x) B0 ( BB{B0 } B =
S S
1
( BB B). Hence, f 1 ( BB B) = BB f 1 (B).
S S S S
BB B, and thus x f

(c) Suppose x f 1 B for all B B. Thus x f 1 (B) for


T T
T ( BB B). Then f (x) BB B. So f (x) T
1
all B B, i.e. x BB f (B). On the other hand, suppose x BB f 1 (B). 1
T Then x f (B) for all
1
T
B B. So f (x) B for all B B. Thus, f (x) BB B. Therefore, x f ( BB B).
S S
(f ) Suppose y f ( AA A). Then there exists S x AA A s.t. S f (x) = y. W.L.O.G., let A0 A be s.t.
x A0 . Then y f (A0 ), and so y f (A0 ) ( AA{A0 } f (A)) = AA f (A). On the other hand, suppose
S
y AASf (A). W.L.O.G., S let f (A0 ) be s.t. y f (A0S). Then there existsS x A0 s.t. Sf (x) = y. So
x A0 ( AA{A0 } A = AA A. Thus, y = f (x) f ( AA A). Hence, AA f (A) = f ( AA A).
T T
(g) Suppose y f ( AA A). Then there exists x T AA A s.t. f (x) = y. So x A for each A A. Thus,
f (x) f (A) for each A A. Therefore, y = f (x) AA f (A).
T
Now suppose that f is injective. Suppose y AA f (A). Then y f (A) for all A T A. Then since f
is injective, there
T exists a unique x that is in each A A s.t. f (x) = y. Then x AA A. Therefore,
y = f (x) f ( AA A).

3
3 January 28, 2015

3; 1: Define two points (x0 , y0 ) and (x1 , y1 ) of the plane to be equivalent if y0 x20 = y1 x21 . Check
that this is an equivalence relation and describe the equivalence classes.

Call this relation R.


1. (Reflexive): Let (a, b) be any point in the plane. Then b a2 = b a2 . So for all (a, b) we have
(a, b)R(a, b).
2. (Symmetric): Suppose (a, b)R(c, d). Then b a2 = d c2 . So d c2 = b a2 and therefore (c, d)R(a, b).

3. (Transitive): Suppose (a, b)R(c, d) and (c, d)R(e, f ). Then b a2 = d c2 and d c2 = f e2 . So by


substitution, we get b a2 = f e2 , and therefore (a, b)R(e, f ).
Thus R is an equivalence relation.
The equivalence classes are parabolas. That is to say, two points are equivalent under R if and only if they lie
on the same parabola given by the general equation y = x2 + k. This can be seen by considering two points
(a, b), (c, d) that both satisfy y = x2 + k. Then b = a2 + k and d = c2 + k, so b a2 = k and d c2 = k. Thus,
b a2 = d c2 and (a, b)R(c, d). On the other hand, if we have (a, b)R(c, d), and (a, b) satisfies y = x2 + k,
we know b a2 = d c2 . Also, b = a2 + k. Subtracting a2 , we get k = b a2 = d c2 . Thus d = c2 + k and
(c, d) satisfies y = x2 + k.

3; 11: (a) Show that an element in an ordered set has at most one immediate successor and at
most one immediate predecessor. (b) Show that a subset of an ordered set has at most one smallest
element and at most one largest element.

(a) Let S be an ordered set. Suppose b S is an immediate successor (predecessor) of a S. Then


{x S | a < x < b} ({x S | b < x < a}) = . Assume that b0 S is also an immediate successor
(predecessor) of a and b0 6= b. Then by comparability, b0 < b or b < b0 .
1. In the first case, we have a < b0 < b (b0 < b < a) so b0 {x S | a < x < b} =
6 (b {x S | b0 < x <
0
6 ), which contradicts that b (b ) is an immediate successor (predecessor) of a.
a} =
2. In the second case, we have a < b < b0 (b < b0 < a) so b {x S | a < x < b0 } 6= (b0 {x S | b <
x < a} =6 ), which contradicts that b0 (b) is an immediate successor (predecessor) of a.
Thus, b = b0 and the immediate successor (predecessor) is unique if it exists.
(b) Suppose A S. Suppose A has a largest (smallest) element. Then there exists s (i) A s.t. for all
x A, we have x < s (i < x). Suppose s0 (i0 ) is another largest (smallest) element of A that is not equal to
s (i). Then by comparability, we have s0 < s (i0 < i) or s < s0 (i < i0 ).
1. In the first case, we have s (i0 ) A and s0 < s (i0 < i) which contradicts that s0 (i) is a largest
(smallest) element of A.

2. In the second case, we have s0 (i) A and s < s0 (i < i0 ) which contradicts that s (i0 ) is the largest
(smallest) element of A.
Thus, s = s0 (i = i0 ) and we conclude the largest (smallest) element of a subset A S, is unique if it exists.

4
3; 12: Let Z+ denote the set of positive integers. Consider the following relations on Z+ Z+ :
1. The dictionary order.
2. (x0 , y0 ) < (x1 , y1 ) if either x0 y0 < x1 y1 , or x0 y0 = x1 y1 and y0 < y1 .

3. (x0 , y0 ) < (x1 , y1 ) if either x0 + y0 < x1 + y1 , or x0 + y0 = x1 + y1 and y0 < y1 .


(i) In these order relations, which elements have immediate predecessors? (ii) Does the set have a
smallest element? Show that all three order types are different.

1. (a) In the dictionary order, all elements (a, b) will have an immediate predecessor if and only if b > 1.
Given (a, b) Z+ Z+ , with b > 1, the set {x | (a, b 1) < x < (a, b)} = . Otherwise, we have
(a, b 1) < (a, b0 ) < (a, b). But this implies b 1 < b0 < b in the integers, which is clearly false.
So (a, b 1) will be the immediate predecessor.
Now suppose (a, b) Z+ Z+ where b = 1. Then the only elements less than (a, b) have the form
(a0 , b0 ) where a0 < a. So let a0 = a 1. Then for every b0 Z+ , we have {x | (a 1, b0 ) < x <
(a, b)} = 6 because (a 1, b0 ) < (a 1, b0 + 1) < (a, b). Thus, (a, b) has no immediate predecessor.
(b) This set does have a smallest element. It is (1, 1). This is because if (a, b) < (1, 1), then a < 1
and b < 1 which is impossible in Z+ .
2. We start by noting that in this ordering, the first relation is the same as y1 x1 < y0 x0 . Intuitively,
each point lies on a unique line yi = xi + ki of slope 1. So the ordering becomes (x0 , y0 ) < (x1 , y1 ) if
k1 < k0 or y0 < y1 (i.e. the point is higher).
(a) A point (a, b) has an immediate predecessor if and only if b > 1. To see why, given (a, b) with
b > 1, then {x | (a 1, b 1) < x < (a, b)} = . Otherwise, we have (a 1, b 1) < (a0 , b0 ) < (a, b).
So a 1 b + 1 = a b < a0 b0 or a b = a0 b0 and b < b0 . But also, a0 b0 < a b or b0 < b,
which is clearly impossible. Thus, (a 1, b 1) is the immediate predecessor of (a, b).
On the other hand, suppose (a, b) has b = 1. Then elements (a0 , b0 ) smaller than (a, b) have
a0 b0 < a b, because there is no x < 1 in Z+ . So the set {x | (a0 , b0 ) < x < (a, b)} = 6 since
(a0 , b0 ) < (a0 +1, b0 +1), and also, a0 +1(b0 +1) = a0 b0 < ab so (a0 , b0 ) < (a0 +1, b0 +1) < (a, b).
Thus, (a, b) has no immediate predecessor.
(b) The set does not have a least element. Suppose it did, called (a, b). Then for all (x, y), we have
a b < x y. However, the element (a, b + 1) has a (b + 1) < a b, which contradicts that it
was the smallest. Thus, no such element exists.
3. This is a similar situation as in part (b) except that we consider the points on the unique line yi =
xi + ki of slope 1. Then the ordering becomes (x0 , y0 ) < (x1 , y1 ) if k0 < k1 or y0 < y1 , (similarly,
larger if the point is higher).

(a) Every point that is not (1, 1) has an immediate predecessor. Suppose (a, b) 6= (a, 1). Then
(a + 1, b 1) and a + 1 + b 1 = a + b, but b 1 < b. Also, if we had (a + 1, b 1) < (a0 , b0 ) < (a, b),
then a0 + b0 = a + b and b 1 < b0 < b which is impossible. Thus, (a + 1, b 1) is the immediate
predecessor of (a, b). If b = 1, then (1, a 1) is the immediate predecessor. This is because if we
had (1, a 1) < (a0 , b0 ) < (a, 1), then a0 + b0 < a + 1, or b0 < 1 (provided a0 + b0 = a + 1). So
a0 + b0 < a + 1, which gives a0 + b0 = a. But at the same time a < a0 + b0 , or a 1 < b0 (provided
a = a0 + b0 ), which then gives a < b0 + 1. So a0 + b0 < b0 + 1, which gives that a0 < 1. Thus,
(1, a 1) is the immediate predecessor of (a, 1).
(b) Yes, the smallest element is (1, 1) because 2 is the smallest sum of two positive integers.

5
Since the ordering in part (a) admits a smallest element, but some elements do not have immediate prede-
cessors, it cannot be of the same order type as that in part (b) which has no smallest element, or that in
part (c) because every element has an immediate predecessor. Furthermore, the order in part (b) cannot be
the same order type as that in part (c) because some elements from part (b) have no immediate predecessor
while all in part (c) do have them.

3; 15: Assume that the real line has the least upper bound property.

1. Show that the sets


(i) [0, 1] = {x | 0 x 1},
(ii) [0, 1) = {x | 0 x < 1}
have the least upper bound property.

2. (i) Does [0, 1] [0, 1] in the dictionary order have the least upper bound property? (ii) What
about [0, 1] [0, 1)? (iii) What about [0, 1) [0, 1]?

1. (a) Let =6 A [0, 1]. Then A is bounded above by 1 [0, 1]. Since A [0, 1] R, then there is a
least upper bound 1 and so A. Thus, [0, 1] has the least upper bound property.
(b) Let =
6 A [0, 1) have an upper bound b in [0, 1). Since A [0, 1) R, it has a least upper
bound b < 1 and [0, 1). Therefore, [0, 1) has the least upper bound property.

2. (a) Yes it does. Consider a non-empty subset A [0, 1] [0, 1] that is bounded above by (b0 , b1 )
[0, 1] [0, 1]. Then define A0 to be the set of first coordinates in A, and A1 to be the second
coordinates. Then A0 is a non-empty subset of [0, 1] so it has a least upper bound 0 [0, 1] with
0 b0 . So if 0 6 A0 and a0 < 0 , then for all a1 [0, 1] we have (0 , a1 ) < (, a1 ). Further,
(0 , 0) < (0 , a1 ) for all a1 (0, 1]. Therefore, (0 , 0) is the least upper bound of A. If 0 A0 ,
then A1 has a least upper bound 1 b1 , and given any a1 < 1 we have (0 , a1 ) < (0 , 1 ).
Thus (0 , 1 ) is the least upper bound.
(b) Does not. Consider the subset A = {0} [0, 1). Then A [0, 1] [0, 1), it is non-empty, and is
bounded above by (1, 0) [0, 1] [0, 1). But A has no least upper bound. To see why, suppose
there were, call it (0 , 1 ) [0, 1] [0, 1). Then 0 = 0 otherwise, 20 < 0 so ( 20 , 1 ) < (0 , 1 ).
Thus 0 = 0. But then we have that for all a1 A1 , we have (0, a1 ) < (0, 1 ). This implies that
there is a largest element (namely 1 ) in [0, 1), which is false. Thus, [0, 1] [0, 1) does not have
the least upper bound property.
(c) Suppose A [0, 1) [0, 1] that is non-empty and bounded above by (b0 , b1 ) [0, 1) [0, 1].
Then define A0 to be the set of first coordinates in A, and A1 to be the second coordinates. So
A0 [0, 1) and is bounded above by b0 . So A0 has a least upper bound 0 . If 0 A0 , then A1
has a least upper bound 1 b1 , and given any a1 < 1 we have (0 , a1 ) < (0 , 1 ). So (0 , 1 )
is the least upper bound. Assuming 0 6 A0 , we have that (a0 , a1 ) (0 , 0) (0 , a1 ) for all
a0 < 0 (assuming for the time that 0 6= 0). Thus (0 , 0) is the least upper bound of A. If
0 = 0, then A = {0} A1 . Since b1 [0, 1], there is a least upper bound for A1 called 1 [0, 1].
So (0, a1 ) (0, 1 ), for all (0, a1 ) A. Thus, (0, 1 ) is the least upper bound for A.

4 February 2, 2015

6
5; 4: Let m, n Z+ . Let X 6= .
1. If m n, find an injective map f : X m X n .

2. Find a bijective map g : X m X n X m+n .


3. Find an injective map h : X n X .
4. Find a bijective map k : X n X X .

5. Find a bijective map l : X X X .


6. If A B, find an injective map m : (A )n B .

1. Proof. Define the map f : X m X n where each xm 7 xn and (bearing in mind Si = {1, ..., i 1} -
a section), 
q : q Sm+1
(x1 , ..., xn ) = xn : Sn+1 X, xn (q) =
0 :q/ Sm+1 .
This is to say, the image of f is the set of n-tuples where the m + 1 j n values are zero (the
imbedding mapping). Then this map is injective, because if xm =6 x0m , i.e.

(x1 , x2 , ..., xm ) 6= (x01 , x02 , ..., x0m ),

then
f (xm ) = (x1 , x2 ..., xm , 0m+1 , ..., 0m ) 6= (x01 , x02 , ..., x0m , 0m+1 , ..., 0m ) = f (x0 m ).

2. Proof. Define the map g : X m X n X m+n where each

(am , bn ) = ((a1 , ..., am ), (b1 , ..., bn )) 7 (a1 , ..., am , b1 , ..., bn ).

Then, g is a bijection.
(a) We first show it is an injection. Suppose g((am , bn )) = g((cm , dn )). Then

(a1 , ..., am , b1 , ..., bn ) = (c1 , ..., cm , d1 , ..., dn ).

Thus, ai ci = 0 and bj dj = 0 for all 1 i m and 1 j n respectively. But this is the


same as saying that am = cm and bn = dn . Thus, (am , bn ) = (cm , dn ).
(b) It is surjective. Given any k m+n X m+n , then we can write

k m+n = (k1 , .., km , km+1 , ..., km+n ).

Then if we let
k m = (k1 , ..., km ) X m and k n = (km+1 , ..., km+n ) X n ,
then g((k m , k n )) = k m+n , as desired.

3. Proof. We define h : X n X where each xn 7 x and



q : q Sn+1
(x1 , ..., xn , ...) = x : Z+ X, s.t. x (q) =
0 :q / Sn+1 .

7
So x = (x1 , ..., xn , 0, 0, ...). Then h is injective. This is because if h(x ) = h(x0 ), then
(x1 , ..., xn , 0, 0, ...) = (x01 , ..., x0n , 0, 0, ...),
so xi x0i = 0 for all 1 i n and since 0 0 = 0 always, we have the exact condition necessary to
say that x = x0 . So our claim follows.
4. Proof. Define the map k : X n X X where each
(an , b ) 7 (a1 , ..., an , b1 , b2 , ...).
Then this is a bijection.
(a) We check injectivity. Suppose k((an , b )) = k((cn , d )). Then
(a1 , ..., an , b1 , b2 , ...) = (c1 , ..., cn , d1 , d2 , ...),
and
(a1 c1 , ..., an cn , b1 d1 , b2 d2 , ...) = (0, ..., 0, 0, 0, ...).
Thus, ai = ci for 1 i n and bj = dj for all j Z+ . Thus, an = cn and b = d .
(b) We check surjectivity. Suppse z X . Then write
z = (z1 , ..., zn , zn+1 , zn+2 , ...).
Define z n = (z1 , ..., zn ) X n and z 0 = (zn+1 , zn+2 , ...) X . Then k((z n , z 0 )) = z , as desired.

5. Proof. Define the map l : X X X where each


(a , b ) = ((a1 , a2 , ...), (b1 , b2 , ...)) 7 (a1 , b1 , a2 , b2 , ...).
Then this map is a bijection.
(a) We check that it is an injection. Suppose l((a , b )) = l((c , d )). Then
(a1 , b1 , a2 , b2 , ...) = (c1 , d1 , c2 , d2 , ...),
and
(a1 c1 , b1 d1 , a2 c2 , b2 d2 , ...) = (0, 0, 0, 0, ...).
Thus, ai = ci and bi = di for all i Z+ . So a = c and b = d .
(b) We check surjectivity. Suppose z X . Then
z = (z1 , z2 , z3 , z4 , ...).
We then define a = (z1 , z3 , ...) X and b = (z2 , z4 , ...) X . Then l((a , b )) = z as
desired.

6. Proof. Suppose A B. Then definie the map m : (A )n B where


(a1 , ..., an ) = ((a11 , a12 , ...), ..., (an1 , an2 , ...)) 7 (a11 , ..., an1 , a12 , ..., an2 , ...).
Then this map is injective. This is because if m((a1 , ..., an )) = m((b1 , ..., bn )), then
(a11 , ..., an1 , a12 , ..., an2 , ...) = (b11 , ..., bn1 , b12 , ..., bn2 , ...),
so aij = bij for all 1 i n and j Z+ . Thus, we have that ai = bi for all 1 i n. Therefore,
(a1 , ..., an ) = (b1 , ..., bn ), as desired.

8
5; 5: Which of the following subsets of R can be expressed as the cartesian product of subsets of
R?
1. A1 = {x | xi is an integer for all i}.
2. A2 = {x | xi i for all i}.

3. A3 = {x | xi is an integer for all i 100}.


4. A4 = {x | x2 = x3 }.

1. This can be because A1 = (Z) and Z R. This is becaues if x (Z) , then each xi Z, so x A1 .
Further, if x A1 , then each xi Z, thus x (Z) .
This can be. Define (for any i R) Pi = {t R | t i}. Then Pi R and A2 =
2. Q Q P1 P2 ... =
P
iZ+ i . This is because if a A 2 , then for each ai i. So ai P i and thus a iZ+ Pi . On the
Q
other hand, if a iZ+ Pi , then each ai i for all i Z+ . Thus, a A2 .

3. This can be. This is because A3 = (R)99 (Z) . To see why consider if a A3 . Then each of the
first 99 components are in R and after that, the components are in Z. Thus, a (R)99 (Z) . On the
other hand, suppose a (R)99 (Z) . Then each ai Z for all i 100. Thus, a A3 .
Q
4. This cannot be expressed as the cartesian product. Suppose it could. Then A4 = iZ+ Bi , where each
Bi R. Then if b2 = Q b3 , we must have B2 = B3 = {b2 }. But if b2 6= h R, then (b1 , h, h, b4 , ...) A4
and (b1 , h, h, b4 , ...) 6 iZ+ Bi . A contradiction. Thus, this set cannot.

6; 3: Let X be the two-element set {0, 1}. Find a bijective correspondence between X and a
proper subset of itself.

Let X = {0, 1}. Define f : X {x X | x2 = 0} := X 0 where

z 7 (z1 , 0, z2 , z3 , ...).

Then X 0 is clearly a proper subset of X . We now show that f is a bijection.


1. (Injective) Suppose f (a) = f (b). Then

(a1 , 0, a2 , ...) = (b1 , 0, b2 , ...).

So
(a1 b1 , 0 0, a2 b2 , ...) = (0, 0, 0, ...),
and so ai = bi for all i Z+ . Thus, a = (a1 , a2 , ...) = (b1 , b2 , ...) = b.
2. (Surjective) Suppose z X 0 . Then
z = (z1 , 0, z3 , z4 , ...).
So we define z 0 = (z1 , z3 , z4 , ...). Then it is the case that f (z 0 ) = z as desired.

6; 5: If A B is finite, does it follow that A and B are finite?

Yes it does, provided A, B 6= . This is because if (W.L.O.G.) A = and B = Z+ , then Z+ = .

9
Proof. Suppse A B is finite and A, B 6= . Then there exists a bijection f : A B {1, ..., n} for some
n Z+ . If B = , the result is trivial. So assume B 6= . Then there exists some element b B. We define
the set A0 = A {b}. Then A0 sits as a proper subset of A B and is in bijection with A. [This is because
g : A A0 s.t. a 7 (a, b) is a bijection. This can be seen because the map g 0 : A0 A where (a, b) 7 a is
a two sided inverse to g.] So we know a subset of a finite set is itself finite. So A0 is finite and therefore A is
finite.
The exact same argument can be made with B so our claim follows.

6; 6:

1. Let A = {1, ..., n}. Show there is a bijection of P(A) with the cartesian product X n , where X
is the two-element set X = {0, 1}.
2. Show that if A is finite P(A) is finite.

1. We induct on n. When n = 1 we have P(A) = {, {1}}. So letting 7 0 and {1} 7 1, our claim
follows.
Suppose our claim is true for A0 = {1, ..., n 1}. Then there is a bijective function g : X n1 P(A0 ).
Then for any x X n , we define x to be the restriction of x to A0 A (since each x X n is a
function from {1, ..., n} to X. Intuitively, we are just dropping the last component of x). Then for
A = {1, ..., n}, we define the function f : X n P(A) where

g(x) if xn = 0
x 7
g(x) {n} if xn = 1.

Then this is a bijection.


(a) (Injective) This is because if f (x) = f (y), we have

g(x) = g(y) if xn = 0
g(x) {n} = g(y) {n} if xn = 1.

In both cases, we can conclude that g(x) = g(y) and since g is a bijection, x = y.
(b) (Surjective) Furthermore, if S P(A), then n / S or n S. In the first case, we have S P(A0 )
n1
so there is an a X s.t. g(a) = S since g is a bijection. Therefore, f (a..) = S where we define
a.. to be the element of X n that shares the first n 1 components with a and the last comonent
is zero.
In the second case, we can write S = S 0 {n}. Then S 0 P(A0 ) and we are in the same situation
as above. So f is surjective.

Therefore, our claim follows.


2. Since A is finite, it is in bijection with {1, ..., n} for some n Z+ . Then P(A) is in bijection with
X n as defined as above. We know finite cartesian products of finite sets are finite, so X n is a finite
cartesian product of a finite set and is therefore finite. Thus, P(A) is finite, as desired.

5 February 6, 2015

10
7; 3: Let X be the two-element set {0, 1}. Show there is a bijective correspondence between the set
P(Z+ ) and the cartesian product X .

Define the map


f : P(Z+ ) X by S 7 (x1 , x2 , x3 , ...) = x
by the following rule: If q S, then the qth component of x (i.e. xq ) is equal to one. If q
/ S, then xq = 0.
This is a well definied bijection. Suppose S 6= T - elements of P(Z+ ). Then one has an element the other
does not. W.L.O.G. say that s S and s / T . Then the sth component of f (S) will be 1 while that of
f (T ) will be 0. Thus, f (S) 6= f (T ). On the other hand, suppose x X . Then define S P(Z+ ) as the
following: (viewing tuples as functions) for each q Z+ , if x(q) = 1, then q S and if x(q) = 0, then q
/ S.
Then f (S) = x by construction. Thus f is a bijection.

7; 6: We say that two sets A and B have the same cardinality if there is a bijection of A with B.
1. Show that if B A and if there is an injection

f : A B,

then A and B have the same cardinality.


[Hint: Define A1 = A and B1 = B, and for n > 1, An = f (An1 ) and Bn = f (Bn1 ). Note
that A1 B1 A2 B2 .... Define a bijection h : A B by

f (x) if x An Bn for some n,
h(x) =
x otherwise.]

2. Prove if there are injections f : A C and g : C A, then A and C have the same cardinality.

f
1. Suppose B A and there is an injection A B. Define A1 = A and B1 = B, and for n > 1,
An = f (An1 ) and Bn = f (Bn1 ). Note that A1 B1 A2 B2 .... Then define h : A B by

f (x) if x An Bn for some n,
h(x) =
x otherwise.

So h is a bijection. Define the set [


S= An B n .
nZ+

(a) (Injective) Suppose x1 6= x2 . Then if x1 and x2 are in S, we have h(x1 ) = f (x1 ) and h(x2 ) = f (x2 ).
Since f is injective, we have that f (x1 ) 6= f (x2 ) so h(x1 ) 6= h(x2 ). Similarly, if x1 and x2 are not
in S, we have h(x1 ) = x1 and h(x2 ) = x2 , and h(x1 ) 6= h(x2 ).

Now suppose (W.L.O.G.) that x1 S and x2 / S. So x1 An Bn for some n. So h(x1 ) =


f (x1 ) f (An ) f (Bn ) (by homework 2, problem 2).
Now suppose for the sake of contradiction that h(x2 ) = h(x1 ) i.e. x2 = f (x1 ). Then x2 = f (x1 )
f (An ) f (Bn ). A contradiction since x2
/ S. Thus, h(x1 ) 6= h(x2 ).

(b) (Surjective) Suppose y B. If y / S, then h(y) = y. If y S, then let n be the smallest index
s.t. y An and y / Bn . So n > 1 since y B1 . Thus, y f (An1 ). So there exists an a An1
s.t. f (a) = y. Then a
/ Bn1 otherwise f (a) = y Bn . So a An1 Bn1 and so h(a) = y.

11
2. Suppose there are injections f : A C and g : C A. Then we define f 0 : A f (A). Then
f 0 is a bijection trivially and f (A) C. So we look for a injective function k : C f (A). Define
k(b) = f (g(b)). This is injective because if b 6= b0 , then g(b) 6= g(b0 ) since g is injective. Then
f (g(b)) 6= f (g(b0 )) since f is injective. Thus, k(b) 6= k(b0 ). So by part (a), we have that C has the same
cardinality as f (A). Since f (A) is in bijection with A, we know C has the same cardinality as A.

7; 8: Let X = {0, 1}. Let B be the set of countable subsets of X . Show that X and B have the
same cardinality.

1. We define the map f : X B by f (x) = {x}. This is obviously well definied and this is an injection
because if x1 6= x2 , then clearly {x1 } =
6 {x2 }.
2. Then by the Axiom of Choice, there is a function
[
c:B B s.t. c(B) B.
BB

Given any B B we call x1 = c(B). Then xi = c(B {x1 , x2 , ..., xi1 }), where each xj (1 j < i)
has been defined.
Then we define the function
g : B X s.t. B 7 x.
If B is infinite, then let the odd components of x be zero and if B is finite with n elements, let the first
n, odd, comonents be 1, and after that the rest 0.
(a) In the case of the first comonent being zero, we order the elements of B according to the choice
function, i.e., let xi = c(B {x1 , x2 , ..., xi1 }). Then we can arrange the strings as such:
x1 = (x1.1 , x1.2 , x1.3 , ...)
x2 = (x2.1 , x2.2 , x2.3 , ...)
x3 = (x3.1 , x3.2 , x3.3 , ...)
...
and use a diagonalization on the even places to let
x = (0, x1.1 , 0, x2.1 , 0, x1.2 , 0, x3.1 , 0, x2.2 , 0, x1.3 , 0, x4.1 , 0, x3.2 , ...).
Then this will define an injection because given two subsets B 6= C, they cannot have the same
image because W.L.O.G. some bi / C. So the components of f (B) that are mapped to by bi will
be different than the components of f (C).
(b) In the case when B is finite, we have that the first n odd elements uniquely determine the size of
B. We then map the n - first components to the first n even components of x, and the n - second
components to the second n even components, i.e.
x = (1, x1.1 , 1, x2.1 , ..., 1, xn.1 , 0, x1.2 , 0, x2.2 , ...).
Then this is injective as well. Given any two subsets B 6= C, if the cardinalty of B differs from
that of C, then W.L.O.G., say |C| < |B|. Then in f (B), the |B|th odd component will be 1 but
will be 0 in f (C). If they are the same cardinality, but say b
/ C, then some component mapped
to by b in f (B) will be different from f (C).
Thus, the map is injective, as well as painstakingly well-definied. Since we have injections from X
into B and from B into X , by the Schroeder-Bernstein Theorem |X | = |B|.

12
6 February 9, 2015

9; 3: Suppose that A is a set and that {fn }nZ+ is a given family of injective functions: fn :
{1, ..., n} A. Show that A is infinite. Can you define an injective function f : Z+ A without
using the A.o.C.?

Suppose A were not infinite. There there exists a bijection g : A {1, ..., q} for some m Z+ . Then
g fq+1 : A A is an injection into a proper subset of itself. Thus, A is not finite.
Define the function f : Z+ A recursively. Let f (1) = f1 (1). Then let f (p) = fp (q) s.t. q is the smallest
integer that makes fp (q) / {f (1), f (2), ..., f (q 1)}. Then this function is well defined because each fn is
injective. Furthermore, if p 6= q, then W.L.O.G. say p < q, and so f (q) / {f (1), ..., f (p), ..., f (q 1)}. Thus,
f (p) 6= f (q).

9; 7: Let A and B be two non-empty sets. If there is an injection of B into A, but no injection of
A into B, we say that A has greater cardinality than B.
1. Conclude from theorem 9.1 that every uncountable set has greater cardinality than Z+ .

2. Show that if A has greater cardinality than B, and B has greater cardinality than C, then A
has greater cardinality than C.
3. Find a sequence A1 , A2 , ... of infinite sets s.t. for each n Z+ , the set An+1 has greater
cardinality than An .
4. Find a set that for every n has cardinality greater than An .

1. Suppose S is an uncountable set. Then it is, in particular, infinite and by Theorem 9.1 (and secretly
the A.o.C.), there exists an injection g : Z+ S. However, since S is uncountable, there is no bijection
between Z+ and S. Thus, there is no injection from S to Z+ , or by Schroeder-Bernstein, we would
have that S is countable. So S has greater cardinality than Z+ .
2. Suppose Ahas greater cardinality than B, and B has greater cardinality than C. Then there exist
f g f g
injections B A and C B, but no injections the other direction. Then C A is an injection
(since c1 6= c2 = g(c1 ) 6= g(c2 ) = f (g(c1 )) 6= f (g(c2 ))). However, there is no injection from
A to C because if there was, then |A| = |C| and so there would be a bijection F : A C and then
g F : A B would be an injection but there was assumed to be no injection from A to B. Thus, A
has greater cardinality than C.

3. For any set A, recursively define P n (A) as follows: Let P 0 (A) = A, P 1 (A) = P(A). Then let
P n (A) = P(P n1 (A)). Then by Theorem 7.8, each An+1 has greater cardinality than An .
4. For each n, define [
An = P( P i (A)).
0in

Then for each n, An has cardinality greater than An .

13
7 February 11, 2015

10; 2:

1. Show that in a well-ordered set, every element except the largest (if one exists) has an immediate
successor.
2. Find a set in which every element has an immediate successor that is not well-ordered.

1. Suppse A is a well-ordered set with order relation <. Then suppose a A is not the largest element of
A (if one exists). Then we look at the set {x A | a < x} = S A, which is non-empty by assumption.
Then S has a least element. Call it b. Thus, a < b and if we examine the set {x A | a < x < b}, we
assert that this set is empty. If it were not, then there would be an element c s.t. a < c < b. This gives
that b was not the smallest element of S. Therefore, {x A | a < x < b} = . So b is the immediate
successor of a.

2. The integers under the standard order are not well ordered, but each element has an immediate
successor.

10; 7: Let J be a well-ordered set. A subset J0 of J is said to be inductive if for every J,

(S J0 ) = J0 .

Theorem (The Principle of Transfinite Induction): If J is a well-ordered set and J0 is an inductive


subset of J, then J0 = J.

Suppose J is a well-ordered set and J0 is an inductive subset of J. If J0 6= J, then J J0 6= . So by well


ordering, this has a smallest element, call it j. Then Sj J0 because every element s Sj has s < j. So
s/ J J0 since j was assumed to be the smallest, and therefore s J0 . We then see that j J0 because
J0 is an inductive subset of J. This contradicts our assumption that j J J0 . Thus, we conclude that
J = J0 .

14
10; 10: Theorem: Let J and C be well-ordered sets; assume that there is no surjective function
mapping a section of J onto C. Then there exists a unique function h : J C satisfying the equation

() h(x) = smallest[C h(Sx )]

for each x J, where Sx is the section of J by x.


Show:

1. If h and k map sections of J, or all of J, into C and satisfy () for all x in their respective
domains, show that h(x) = k(x) for all x in both domains.
2. If there exists a function h : S C satisfying (), show that there exists a function k :
S {} C satisfying ().
3. If K J and for all K there exists a function h : S C satisfying (), show that there
exists a function [
k: S C
K

satisfying ().
4. Show by transfinite induction that for every J, there exists a unique function h : S C
satisfying (). [Hint: If has an immediate predecessor , then S = S {}. If not, S is
the union of all S with < .]
5. Prove the theorem.

1. Suppose h and k map sections of J, or all of J, into C and satisfy () for all x in their respective
domains. Then let j0 be the smallest element in J. Suppose h and k are defined on sections of J,
S and S , respectively. Define the set D to be the elements that are in the domain of h and the
domain of k. Then j0 D so h(j0 ) = smallest[C] = k(j0 ). Then h(x) = k(x) for all x in both domains
otherwise the set of x in both domains s.t. h(x) 6= k(x) is non-empty. So this has a minimal element
x0 . Then h(x0 ) = smallest[C h(Sx0 )] = smallest[C k(Sx0 )] = k(x0 ), a contradiction.

[Note: The middle equality holds because for all x Sx0 we have h(x) = k(x) since x0 was assumed
to be the minimal element for which h(x) 6= k(x).]

Thus, our claim follows.


2. Suppose there exists a function h : S C satisfying (), then there exists a function k : S {} C
satisfying (). This function is given by extending h as follows:

h(x) if x S
k(x) =
smallest[C h(S )] if x = .

This is satisfies () because the assumption that there is no surjection from a section of J to C.

3. If K J and for all K there exists a function h : S C satisfying (), then there exists a
function [
k: S C
K

satisfying (). This function is given by k(x) = smallest[C hx (Sx )]. This satisfies () because
smallest[C hx (Sx )] = smallest[C k(Sx )].

15
4. Let A0 be the set of s.t. there exists a unique function h : S C satisfying (). Then suppose
Sb A0 .
(a) If b has an immediate predecessor a, then Sb = Sa {a}. Then, since Sa Sb , there exists a
function k : Sa C and by part (b) we know there exists a function k 0 : Sa {a} = Sb C
satisfying () as well. Thus, b A0 .
(b) If b does not have an immediate predecessor, then
[
Sb = Sa .
a<b

Then since each Sa Sb , we S know there exits functions ha : Sa C satisfying (). So by part
(c), there exists a function l : a<b Sa = Sb C that satisfies (). Thus, Sb A0 .
We conclude that A0 = J.
5. Such a function h exists. This is because by part (d),
S for each section S of J, there exists a function
h : S C. So by part (c), there is a function k : K S C satisfying ().
S
(a) Suppose J does not have a largest element. Then J = K S , and our claim follows with
h : J C where h = k as definied above.
S
(b) If J does have a largest element , then J = [ K S ] {}. By part (b) there exists a function
h : J C satisfying ().
Thus, such a function exists.

This function is unique. This is because if h and h0 satisfied the above construction, then by part (a),
h = h0 at all points of the domain. Thus, the function is unique.

8 February 16, 2015

11; 5: Show that Zorns Lemma implies the following:


Lemma 1. (Kuratowski) Let A be a collection of sets. Suppose that for every subcollection B of A
that is simply ordered by proper inclusion, the union of the elements of B belong to A. Then A has
an element that is properly contained in no other element of A.

Suppose Zorns lemma and that every subcollection B of A that is simply ordered by proper inclusion, the
union of the elements of B belong to A. [
So the union of all elements of B is an upper bound for each B because for any B B, B B0 , and this
B0 B
union is assumed to be in A. Since this is true for all such B, so we may apply Zorns lemma to conclude
that A has a maximal element M . Thus, no A A has M A, by definition of a maximal element. Thus,
our claim follows.

11; 6: A collection A of subsets of a set X is said to be of finite type provided that a subset B of
X belongs to A iff every finite subset of B belongs to A. Show that the Kuratowski lemma implies
the following:

Lemma 2. (Turkey, 1940) Let A be a collection of sets. If A is of finite type, then A has an element
that is properly contained in no other element of A.

16
Suppose A is a collection of sets of X and A is of finte type. Then say B A. Let B A be the collection
finite subsets of B. Then by construction B is simply ordered by proper inclusion, so we want to show
of all [
that B A, because if this is the case, then by Kuratowskis lemma, A has an element that is properly
BB
contained in no[ther element of A. [
We show that B A. Let Bf be a finite subset of B with cardinalty n, then each bi Bf (1 i n)
BB BB
is contained in some B B. However, since B is simply ordered by inclusion, Bf B0 for some B0 B.
[ then B0 A and therefore, all finite subsets of B0 belong to A as well, in particular, Bf A. Thus,
So
B A. So by Kuratowskis lemma, we have that A has an element that is properly contained in no
BB
other element of A, as desired.

11; 7: Show that the Turkey lemma implies the Hausdorff maximum principle. [Hint: If is a
strict partial order on X, let A be the collection of all subsets of X that are simply ordered by .
Show that A is of finite type.]

Let X be a set and let be a strict partial ordering on X. Let A be the collection of all subsets of X that
are simply ordered by . Then we claim that A is of finite type.
This is because if A A, then A X and A is simply ordered by . Therefore, any finite subset Af of A
is also simply ordered by , and so Af A.
On the other hand, suppose A X has the property that every finite subset Af of A belongs to A. Then
we claim that A is itself simply ordered by . This is because given any two elements a1 , a2 A, then the
set {a1 , a2 } is finite and is therefore in A. So it must be simply ordered. Thus a1 and a2 are comparable.
Therefore, A is simply ordered by , and A A.
So A is finite type. So by the Turkey lemma, A has an element M that is properly contained in no other
element of A. Thus, M is a maximal simply ordered subset of X, as desired.

9 February 18, 2015

13; 2: Consider the nine topologies on the set X = {a, b, c} indicated in Example 1 of 12. Compare
them; i.e., for each pair of topologies, determine whether they are comparable, and if so, which is the
finer.

We denote the nine topologies as if they were entries aij in a 3 3 matrix A. For sake of brevity, we omit
redundant mention of comparability, i.e., if we have alreay stated that T1 is comparable T2 , we will not state
that T2 is comparable to T1 , when we are addressing T2 later. If no mention is made, then it is to be assumed
that the topologies are not comparable.
1. (a11 and a33 ) a11 is the trivial topology. It is comparable to every other topology since they all have
to contain it. Further, it is the coarsest topology. Similarly a33 is the descrete topology and it is
comparable to every other topology since they all are contained in it. So it is the finest topology.
2. (a12 ) This topology is comparable to a31 and a32 . a12 is finer than a31 and coarser than a32 .
3. (a13 ) This topology is comparable to a21 , a23 , and a31 . a13 is finer than a21 and a31 but coarser than
a23 .
4. (a21 ) This topology is comparable to a23 and a32 . It is coarser than both.

17
5. (a22 ) This topology is not comparable with any of the later topologies.
6. (a23 ) This topology is comparable with a31 and it is finer than a31 .
7. (a31 ) This topology is comparable and coarser than a32 .

13; 4:
T is a topology on X. Is T a topology
T S
1. If {T } is a family of topologies on X, show that
on X?
2. Let {T } be a family of topologies on X. Show that there is a unique smallest topology on X
containing all the collections of T , and a unique largest topology contained in all T .
3. If X = {a, b, c}, let

T1 = {, X, {a}, {a, b}} and T2 = {, X, {a}, {a, c}}.

Find the smallest topology containing T1 and T2 , and the largest topology contained in T1 and
T2 .

1. Suppose {T } is a family of topologies on X. Then we claim T := T is a topology on X.


T

(a) Since T for every , we have T. Similarly, since X T for every , we have that X T.
\n n
\
(b) Suppose Ai T for 1 i n. Then Ai T for every , and thus, Ai T.
i=1 i=1
(c) Suppose {A } is an arbitrary family of open sets in T. S
Then for all and , we have that A T .
Thus, we have that A T for every , and thus A T.
S

Hence, T is a topology on X.
Note that T is not a topology on X. Consider the topologies a21 and a22 from the previous problem.
S
a21 a22 is not a topology because {a}, {b} a21 a22 but {a} {b} = {a, b}
/ a21 a22 .
2. Let {T } be a family of topologies on X.

(a) The unique largest topology contained in all T is T := T . This is because if there were a
T
larger topology T 0 , then we would have S T 0 and S
/ T. Then since T 0 T for every , we
have that S T , and thus, S T, . So T is the unique largest topology containing all T .
(b) The Sunique smallest topology containing all TSis the topology T generated by the subbasis
B = T (which is indeed a valid subbasis since B = X for B B). This is because if there
n
\
were a smaller topology T 0 , then there would be an open set S T and S
/ T 0 . So S = Ai ,
i=1
n
\
where Ai B. So each Ai T 0 . So then Ai = S T , . Thus, T is the smallest topology
0

i=1
containing all the T .

3. (a) By part (b), the smallest topology containing T1 and T2 is the topology generated by the subbasis
B = {, X, {a}, {a, b}, {a, c}}. This is given by the collection of all unions of finite intersections.
However, these just give back B. So B is the smallest topology containing T1 and T2 (this is the
same topology as a13 from the first problem, up to a relabeling).

18
(b) By part (b) again, the largest topology T that is contained in T1 and T2 is given by T1 T2 . So
T = {, X, {a}}.

13; 5: Show that if A is a basis for a topology on X, then the topology generated by A equals the
intersection of all topologies on X that contain A. Prove the same if A is a subbasis.

T the topology generated by A as T and denote the intersection of all topologies on X that contain A
Denote
as T .
1. Suppose A is a basis for a topology T on X. Each element of A is in T, so the intersection of all
topologies containing A is clearly a subset of T.
Now suppose S TT. Then for each xT S, there existsTAx A s.t. x Ax and Ax S. So S = Ax .
S
Since each Ax T , we have S T . Thus, T = T .
2. Supoose A is a subbasis for a topology T. Then each element of A is in T, so T T.
T
n n
S \ \ \
Now suppose S T. Then S = ( Ai ) . Then since each Ai T , we have that T ,
T
Ai
i=1 i=1
n
S \
T is a topology. Finally, we also know that ( Ai ) T . Thus, T = T .
T T T
since
i=1

10 February 23, 2015

16; 4: A map f : X Y is said to be an open map if for every open set U of X, the set f (U ) is
open in Y . Show that 1 : X Y X and 2 : X Y Y are open maps.

[
Suppose Z is open in in X Y . Then Z = (U V ) where each U is open in X and V is open in Y .
[ [
So 1 (Z) = U (resp. 2 (Z) = V ). Thus 1 (resp. 2 ) is an open map.

16; 8: If L is a straight line in the plane, describe the topology L inherits as a subspace of R` R
and as a subspace of R` R` . In each case it is a familiar topology.

I will assume that the sets R` R and R` R` have the product topology. The line (as a set) is given by
L = {(x, y) R R | y = mx + b} for some slope m and y-intecept b.

1. We first consider R` R. Suppose the line is not vertical. If B is a basis for the topology on R` R,
then the basis B0 for the subspace topology on L is given by B0 = {B R` R | B B}. Given any
element B of B, B has the form [a, b) (c, d). There are 3 possibilites for B 0 = B L:

(a) The first is that the intersection is empty.


(b) Second, there exists some k (c, d) s.t. the point (a, k) L. In this case, B L = {(p, q)
L | (a, k) (p, q) < (b0 , d0 )} (where b0 = b or d0 = d or d0 = c). Since L will be ordered
by a dictionary ordering, we can order it according to its first component and B L will be a
half-open interval and L will have the lower limit topology.

19
(c) Third, for no k (c, d), do we have (a, k) L. In this case, B L = {(p, q) L | (a0 , c0 ) <
(p, q) < (b0 , d0 )} (where a < a0 and c0 = c or d, and b0 = b or d0 = d or d0 = c). So B L will
be a open interval.
So, this topology T on L has elements that are open intervals and half-open intervals, and we can
achieve any interval of this form. Clearly this topology is finer than the lower limit topology. Further,
since the lower limit topology is strictly finer than the standard topology, we see that open intervals
are open in the lower limit topology. So this topology is coarser than the lower limit topology. Hence,
we have the lower limit topology on L.
When L is vertical, then its intersection with basis elements clearly give it the standard topology since
L will be ordered by the second component.

2. Now consider R` R` . If the slope of L is greater than or equal to zero, then by (almost) the same
analysis as above we see that L will have the lower limit topology. [Almost, because we never run into
the case of having open intervals being basis elements. So the argument is a simpler version of the
above.]
If the slope of L is less than zero, then given any point (x, y) L, choose (arbitrarily) x0 > x and
y0 > y and form the basis element B = [x, x0 ) [y, y0 ) B. Then B L = {(x, y)}. Therefore, we
have the discrete topology on L.

16; 10: Let I = [0, 1]. Compare the product topology on I I, the dictionary topology on I I,
and the topology I I inherits as a subspace of R R in the dictionary order topology.

Denote these three topologies Tprod , Tdict , and Tsub . Consider the element B = {(x, y) | ( 12 , 12 ) < (x, y) <
( 12 , 1)} of the basis for Tdict . Then for any (x0 , y0 ) B, there is no neighborhood about (x0 , y0 ) in Tprod that
is properly contained in B. This is because any set properly contained in B has the form { 21 } V , where
V ( 12 , 1). Since { 12 } is not open in I, we conclude that no open set in Tprod can have the required form
and Tprod 6 Tdict . However, given an arbitrary element B of the basis for Tdict , for every (x0 , y0 ), there is
a vertical interval S = {(x0 , y) | y1 < y < y2 } that is properly contained in B and is open in R R. So
S (I I) is a element of the basis for Tsub and this contains (x0 , y0 ) and S (I I) B. Thus, Tsub Tdict .
Now consider the element B 0 = ( 41 , 14 ) ( 12 , 1] of the basis for Tprod . Then ( 12 , 1) B 0 but there is no
neighborhood in Tdict that contains ( 21 , 1) and is properly contained in B 0 . This is because any element V of
Tdict that contains ( 21 , 1) must contain a point (x0 , y0 ) > ( 21 , 1). Thus, (x0 , 0) would also have to be in V , so
V 6 B 0 . We conclude that Tdict 6 Tprod . However, given an arbitrary element B 0 of the basis for Tprod , and
an element (s0 , t0 ) B 0 , as was the case above, there is a vertical interval S 0 that is properly contained in
B 0 but that is also open in R R. So S 0 (I I) is a element of the basis for Tsub and this contains (s0 , t0 )
and S 0 (I I) B 0 . Thus, Tsub Tprod .
Since Tprod and Tdict are not comparable but are both coarser than Tsub , we conclude that Tsub is strictly
finer than both.

11 February 25, 2015

20
17; 8: Let A, B, and A denote subsets of a space X. Determine whether the following equations
hold; if an equality fails, determine whether one of the inclusions or holds.
1. A B = A B.
\ \
2. A = A .

3. A B = A B.

1. Suppose x A B. Then given any neighborhood U of x, U intersects (A B). So there exists


x0 U (A B) and so x0 U A and x0 U B. Thus, U intersects A so x A, and U intersects
B so x B. Therefore, x A B and A B A B.
On the other hand, working in R with the standard topology, let A = (a, b) be an open interval, let
c > b. Then = = (a, b) (b, c) 6= (a, b) (b, c) = [a, b] [b, c] = {b}.
\ \
2. Suppose x A . Then given any neighborhood U of x, U intersects A . So U intersects A for
\
each . Thus, x A for each . Therefore, x A .
The reverse implication does not hold as the above is a special case which does not hold.
3. Suppose x A B. Then x A and x / B. So for every neighborhood U of x U B is a
neighborhood of x. So A intersects U B and therefore A intersects U B (since B B). That is to
say =
6 A (U B) = U (A B). Thus, any neighborhood of x intersects A B, so x A B.
The reverse does not hold. Working in R with the standard topology, let A = (a, b) be an open interval,
let c (a, b) and d > b. Then consider that [a, c] = (a, b) (c, d) 6= (a, b) (c, d) = [a, c).

17; 13: Show that if X is Hausdorff iff the diagonal = {x x | x X} is closed in X X.

( = ) Suppose X is Hausdorff. Then we aim to show c is open.


c = {x y | x, y X, x 6= y}. So given any point a b c , a 6= b so there exists Ua , Ub - neighborhoods
about
[ a and b s.t. Ua Ub = . Then Ua Ub is open in X X. So a b Ua Ub . Since U V =
(Ua Ub ) c is open (since each Ua is disjoint from Ub ), we have
[ [
c = {a b} (Ua Ub ) c .
abc
c c
Thus, U V = and is open and so is closed.
( = ) Suppose is closed in X X. Then its compliment c is open. So for every a, b X where a 6= b,
we have a b c . So there is a neighborhood Ua Ub c about a b where Ua , Ub are open in X. So
Ua Ub = and Ua is an open neighborhood about a in X and similarly, Ub is an open neighborhood about
b in X. Thus, X is Hausdorff.

1
17; 14: In the finite compliment topology on R, to what point or points does the sequence xn = n
converge?

In R with the finite compliment topology, this sequence converges to every point. Consider x R. Every
open set that contains x will contain all but finitely many points. In particular every open neighborhood of
x will contain all but finitely many sequence elements, because the sequence elements are all distinct. Thus,
our sequence converges to x.

21
12 March 9, 2015

19; 7: Let R be the smallest subset of R consisting of all sequences that are eventually zero,
i.e. all sequences (x1 , x2 , ...) s.t. xi 6= 0 for only finitely many values of i. What is the closure of R
in R in the box and product topologies? Justify your answer.

1. In the box topology, the closure of R (in R ) is itself. This is because if ~x R , then for every
neighborhood Ux of ~x, Ux R 6= . We claim that this means ~x R . Suppose it was not. Then
xi 6= 0 for infinitely many values of i. However, we could then say that for each xi 6= 0, we can
Yfind
an interval Ui = (xi , xi + ) s.t. 0
/ Ui . When xj = 0, define Uj = (1, 1). Then let B = U
(where runs through the is and js in the appropriate order) will be an open neighborhood about ~x
that has infinitely many components bound away from zero. So nothing in B can ever be eventually
zero, and therefore, B R = . So we conclude that R R . Since the reverse is true, we get
that R = R .
2. In the product topology, the closure of R is R . Let ~x R . Then every open neighborhood Ux of
~x has R as all but finitely many components and therefore, every neighborhood is eventually R so
it contains a sequence which is eventually zero. So Ux R 6= . Thus, ~x R .

20; 1:

1. In Rn , define d0 (~x, ~y ) = |x1 y1 | + ... + |xn yn |. Show that d0 is a metric that induces the
usual topology of Rn . Sketch the basis elements under d0 when n = 2.
2. More generally, given p 1, define
" n
# p1
X
d0 (~x, ~y ) = |xi yi |p
i=1

for ~x, ~y Rn . Assume that d0 is a metric. Show that it induces the usual topology on Rn .

Denote the metric topology with respect to a metric d as Td .


Then by theorem 20.3, we know that the usual topology on Rn is the same as the topology T induced by
the square metric on Rn . So we use this topology.
1. Let ~x Rn and . Then let = . So for all ~y Bd0 (~x, ), we have d0 (~x, ~y ) = |x1 y1 |+...+|xn yn | < .
So we have (~x, ~y ) = max{|x1 y1 |, ..., |xn yn |} |x1 y1 | + ... + |xn yn | < = , so ~y B (~x, ).
Thus, by lemma 20.2, Td0 T .

On the other hand, let ~x Rn and  be given. Then let = n . Then for all ~y B (~x, ), we have
(~x, ~y ) = max{|x1 y1 |, ..., |xn yn |} < n . So d0 (~x, ~y ) = |x1 y1 | + ... + |xn yn | n max{|x1
y1 |, ..., |xn yn |} < n n = . So ~y Bd0 (~x, ). By lemma 20.2, again, T Td0 .
So we showed that Td0 = T and since T is equivalent to the (usual) product topology on Rn , we have
that Td0 induces the usual topology of Rn .

2. Let ~x Rn and . Then let = 1 . Then for all ~y B (~x, ), we have
np
1 1
d0 (~x, ~y ) = (|x1 y1 |p + ... + |xn yn |p ) p [n (max{|x1 y1 |, ..., |xn yn |})p ] p

22
1 1 1 1 
= n p max{|x1 y1 |, ..., |xn yn |}p ) p = n p (~x, ~y ) < n p 1 = .
np
Thus, ~y Bd0 (~x, ) and so T Td0 .

On the other hand, let ~x Rn and . Then let = . Then for all ~y Bd0 (~x, ), we have
n
X
(~x, ~y )p = (max{|xi yi |}ni=1 )p |xi yi |p = d0 (~x, ~y )p < p = p .
i=1

This implies (~x, ~y ) <  (because the value of the metric is always positive and the function that takes
the p-th root is strictly monotone increasing on the non-negative reals). Thus, ~y B (~x, ) and so
Td0 T . By the same argument as in part (a), we have that d0 induces the usual topology on Rn .

20; 2: Show that R R in the dictionary order topology is metrizable.

Define the function d : (R R) (R R) R by



1 if x1 6= y1
d((x1 , x2 ), (y1 , y2 )) =
d(x2 , y2 ) = min{|x2 y2 |, 1} if x1 = y1 .

Then we claim this is a metric on R R in the dictionary order topology, and further, that the metric
topology induced by d on R R is the dictionary order topology.

1. (a) Since absolute values are always non-negative, d(~x, ~y ) 0. If d(~x, ~y ) = 0, then x1 = y1 and then
d(x2 , y2 ) = 0, so x2 = y2 and ~x = ~y .
(b) Given ~x, ~y , if x1 6= y1 then d(~x, ~y ) = 1 = d(~y , ~x). If x1 = y1 , and |x2 y2 | 1, the we get the
same result. If |x2 y2 | < 1, then we have d(~x, ~y ) = |x2 y2 | = |y2 x2 | = d(~y , ~x).
(c) Given ~x, ~y , ~z, if d(~x, ~z) = 1, then d(~x, ~y ) is at most 1, so d(~x, ~y ) 1 + d(~z, ~y ) = d(~x, ~z) + d(~z, ~y ). If
both d(~x, ~z) < 1 and d(~z, ~y ) < 1, then x1 = z1 = y1 . Since d has the triangle inequality, it follows
that d(~x, ~y ) = d(x2 , y2 ) d(x2 , z2 ) + d(z2 , y2 ) = d(~x, ~z) + d(~z, ~y ).

2. We claim that this topology Td is the dictionary order topology Tdict .


Given any basis element of Tdict , (~x, ~y ) := I, and any ~a I, if a1 6= x1 or y1 , then Bd (~a, 21 ) is an  = 21 -
ball about ~a that is contained in (~x, ~y ). Now consider if a1 = x1 or y1 . Without loss of generality, say
that a1 = x1 . Take  = min{|x2 a2 |, 1}. Then Bd (~a, ) is an -ball about ~a that is contained in (~x, ~y ).
So we have Td Tdict .
On the other hand, a basis element for Td is an -ball about a point ~x, i.e. Bd (~x, ). Then if  1,
we have that for every ~y R R, ~y Bd (~x, ), so ((x1 1, 0), (x1 + 1, 0)) is a basis element of
Tdict that contains ~x and is contained in Bd (~x, ). If  < 1, then for all ~y Bd (~x, ), we have
d(~x, ~y ) < 1 = x1 = y1 and |x2 y2 | < 1. Therefore,

((x1 , x2 |x2 y2 |), (x1 , x2 + |x2 y2 |)) = (~a, ~b) Tdict and ~x (~a, ~b) Bd (~x, ).

So Tdict Td and we have that Td = Tdict as desired.

13 March 11, 2015

23
18; 2: Suppose f : X Y is continuous. If x is a limit point of the subset A of X, is it necessarily
true that f (x) is a limit point of f (A)?

No. Consider the function f : R R that sends x 7 0 identically. Then for any A R, we have f (A) = {0}.
So if a is a limit point of A, then f (a) = 0 and 0 is not limit point of f (A). This is because any neighborhood
of 0 only intersects f (A) at itself.

18; 7:
1. Suppose that f : R R is continuous from the right, i.e.

lim f (x) = f (a),


xa+

for each a R. Show that f is continuous when considered as a function from R` to R.


2. Can you conjecture what functions f : R R are continuous when considered as maps from
(i.) R to R` ? As maps from (ii.) R` to R` ?

1. Suppose that f has the aforementioned property. Let V be open in R and let x0 f 1 (V ) (if
f 1 (V ) = , then is open so were good. We assume that f 1 (V ) 6= ). Then f (x0 ) V . Since
we have the continuous from the right property, we have that for every  > 0, there exists > 0
s.t. (since our limit is actually equal to f (x0 )) x [x0 , x0 + ) = f (x) (f (x0 ) , f (x0 ) + ).
Let this  be such that (f (x0 ) , f (x0 ) + ) V . Then f (x0 ) f ([x0 , x0 + )) V and thus,
x0 [x0 , x0 + f 1 (V ). Thus, f 1 (V ) is open in R` and we ahve our claim.
2. I can conjecture.
(a) Conjecture: The continuous functions functions f : R R` are constant functions.
This is because constant functions are continuous. Further, if a function is not constant, then we
know that there must be an interval where it is monotone which gives that the preimage will be
a half-open interval in R, which is not open.
(b) Conjecture: f must be continuous from the right and must be contant or increasing on neigh-
borhoods U of the domain on which f : R R is continuous.
It must be continuous from the right becuase of part (a) paired with R` being strictly finer than
R. It must not be decreasing on any neighborhood U on which f : R R is continuous because
if it were, then there would be a restriction on which the function was monotonically decreasing
and an open set in this image would have a preimage of the form (a, b] which is not open in R` .

18; 13: Let A X; let f : A Y be continuous; let Y be Hausdorff. Show that if f may be
extended to a continuous function g : A Y, then g is uniquely determined by f .

Let A X; let f : A Y be continuous and Y be Hausdorff. Suppose f may be extended to continuous


functions g : A Y and g 0 : A Y. Say for some z A we have g(z) 6= g 0 (z). Then since Y is
Hausdorff, there exist neighborhoods U, V of g(z), g 0 (z) respectively s.t. U V = . Then g 1 (U ) is an open
neighborhood of z. Similarly, g 01 (V ) is an open neighborhood of z as well. So g 1 (U ) g 01 (V ) 6= is an
open neighborhood of z (this is not empty because, in particular, z g 1 (U )g 01 (V )). So g 1 (U )g 01 (V )
intersects A in a point called a. Then a g 1 (U ) and a g 01 (V ), so g(a) U and g 0 (a) V . However,
since a A, we have g(a) = g 0 (a) = f (a) and so g(a) V and therefore, U V 6= , a contradiction. Thus,
g = g 0 and our claim follows.

24
14 March 25, 2015

23; S
2: Let {An } be a sequence of connected subspaces of X, s.t. An An+1 6= for all n. Show
that An is connected.

We first induct on n. S
The base case is trivial because A1 was already given to be connected. So we suppose that An1 is
connected. Then given the connected subspaces A1 , ..., An1 , An with Ai Ai+1 6= for all 1 i < n, then
n1
[ n2
[
we know by hypothesis that P := Ai = An1 is connected. Then since An1 An 6= , then P and
i=1 i=1
n
[
An share a point in common. So P An is connected. Thus, Ai is connected for all n Z+ .
i=1
Now if the index is finite, we are done. However, if it is infinite we do the following.
n
[
[
Define Pn := Ai . Then we consider the union Pn := R. Then this is a union of connected sets that
i=1 n=1
all have P1 in common (which is non-empty). Thus, R is connected and our claim follows.

23; 7: Is the space R` connected? Justify your answer.

R` is not connected. We give a separation.


The sets U and V given by [ [
U := [n, 0) and V := [0, m)
nZ+ mZ+

are open in R` . These sets are not empty because the real numbers are not empty. Additionally U V = R`
because for any x R` , we know x is negative or non-negative. If the former, then by Archimedian property
of the reals, x [n, 0) for some n Z+ ; and, if x is non-negative then, again, by the Archimedian property
we have that x [0, m) for some m Z+ .
Lastly, U V = because of the trichotomy of the reals (comparing each number to 0).
So R` is not connected.

23; 8: Determine whether or not R is connected in the uniform topology.

R is not connected in the uniform topology.


Consider the sets U := {~x R | ~x is bounded.} and V := {~x R | ~x is unbounded.}. Then it is clear (cf.
example that shows that the box topology is not connected) that U V = R and U V = . Then we
show that U and V are open.
Suppose ~y U , then there is an M R s.t. |yi | M for all i Z+ . Then consider the ball B (~y , 12 ). If
~z B (~y , 21 ), then sup{d(yi , zi ) | i Z+ } 12 which means that for all i Z+ , we have |zi yi | < 21 , and
therefore |zi | |zi yi | + |yi | < 12 + M . Thus, ~z U . Thus, B (~y , 12 ) U and U is open.
Now we show V is open. Suppose w ~ V , then for every M R, there is some N Z+ s.t. if n > N
we have |wn | > M . Then if ~t B (w, ~ 12 ), then sup{d(ti , wi ) | i Z+ } 12 . This gives that for all
1
i Z+ , we have |wi ti | < 2 . Given any M R we can find an N Z+ s.t. for n > N we have
M 12 < |wn | = |wn tn + tn | |wn tn | + |tn | < 12 + |tn |. Then for i > N we have M < |ti | and ~t V .
~ 12 ) V . Thus, V is open.
So B (w,
Since we have established a separation, R is not connected in the uniform topology.

25
15 March 27, 2015

24; 1:

1. Show that no two of the spaces (0, 1), (0, 1], and [0, 1] are homeomorphic.
[Hint: What happens if you remove a point from each of these spaces?]
2. Suppose that there exist imbeddings f : X Y and g : Y X. Show by means of an example
that X and Y need not be homeomorphic.

3. Show Rn and R are not homeomorphic of n > 1.

1. If you remove a point q from (0, 1), then 0 < q < 1, so (0, q) and (q, 1) form a separation of (0, 1) {q}.
However, if you remove the point r = 1 from (0, 1] and [0, 1], you get (0, 1) and [0, 1) respectively,
which are connected sets. Thus, (0, 1) is not homeomorphic to (0, 1] or [0, 1].
Furthermore, you can remove two points s0 = 0 and s1 = 1 from [0, 1] and get (0, 1) - a connected
set. However, any two points you remove from (0, 1] will never leave you with a connected set. This
is because once you remove r = 1 from (0, 1], you get (0, 1); if you remove another point q, you have
0 < q < 1 and (0, q) and (q, 1) forms a separation of (0, 1] {r, q}. Thus, (0, 1] is not homeomorphic
to [0, 1].

2. Consider (0, 1) and [0, 1]. Then, as shown above, these are not homeomorphic. However, the identity
t+1
mapping imbeds (0, 1) into [0, 1]. We can also imbed [0, 1] in (0, 1) by (t) = .
3
3. Suppose n > 1. Then Rn {~0} := Rn0 is connected, but R{0} is not. Rn0 is connected for an two points
in Rn0 , if the line connecting them passes through the origin, then we can choose a non-origin point p
and connect each point to p, i.e. Rn0 is path connected. This can only be done because, geometrically,
there is more than one direction to go. For the same reason, R10 is not connected. So R and Rn are
not homeomorphic for n > 1.

24; 2: Let f : S 1 R be a continuous map. Show there exists a point x of S 1 s.t. f (x) = f (x).

S 1 is a connected space and R is an ordered topology. Then define g : S 1 R by g(x) = f (x) f (x).
Then g is continuous because f is continuous. Observe that g(x) = g(x) so that we have two points
x, x S 1 and (W.L.O.G.) g(x) = g(x) 0 g(x). Thus, by the intermediate value theorem, we have
that there exists z S 1 s.t. g(z) = 0, i.e. f (z) f (z) = 0 f (z) = f (z) as desired.

24; 5: Consider the following sets in the dictionary order. Which are linear continua?
1. Z+ [0, 1)
2. [0, 1) Z+
3. [0, 1) [0, 1]

4. [0, 1] [0, 1)

26
1. This is a linear continuum.
This is homeomorphic with R1 , given by (a r) = a + r with inverse 1 (x) = bxc x bxc. It is
clear that is continuous, since sums are continuous. However, given any open interval

I = {a r Z+ [0, 1) : (a1 r1 ) < (a r) < (a2 r2 )},

(1 )1 (I) = (a1 + r1 , a2 + r2 ) - an open set in R1 .


Since R1 inherits all the properties of a linear continuum from R, and R1 is homeomorphic with
Z+ [0, 1), we know this is a linear continuum as well.
2. This is not a linear continuum because ( 21 1) < ( 21 2) but there is no number between them since
there is no integer between 1 and 2.
3. This is a linear continuum. This is because if (a1 r1 ) < (a2 r2 ), then either a1 < a2 , or a1 = a2
and r1 < r2 . In the first case, we know R is a linear continuum, so there is a (a1 , a2 ). Then
(a1 r1 ) < (a 0) < (a2 r2 ). If we are in the second case, then we know that there is r (r1 , r2 ) so
(a1 r1 ) < a1 r) < (a1 r2 ).
We also know that this has the least upper bound property by homework 3 (3; 15(b)).
4. This is not a linear continuum. By homework 3 (3; 15(b)), we know that this set does not have the
least upper bound property in the dictionary order topology.

16 March 30, 2015

26; 1:
1. Let T and T 0 be two topologies on the set X; suppose that T 0 T. What does compactness of
X under one of these topologies imply about compactness under the other?
2. Show that if X is compact Hausdorff under both T and T 0 , then either T and T 0 are equal or
they are not comparable.

1. Suppose X is compact under T 0 . Then let A = {A } be an open cover of X in T. Then A T T 0 .


Then since X is compact in T 0 there is a finite subcover called T 0 A0 = {A1 , ..., An }. Then A0 A
T. So X is compact under T.
Counterexample the of converse: Consider R under the discrete topology: Rd ; and R under the standard
topology. Then the interval [0, 1] is compact in the standard topology on R. However, we claim it
is not compact in the discrete topology. This is evident because we can take the open cover to be
{{x } Rd : x [0, 1]}. Then this covers [0, 1], but if we remove any set, it will fail to cover [0, 1].
It follows from this and the fact that the interval [0, 1] has infinite elements, that there is no finite
subcover. Thus, [0, 1] is not compact in the discrete topology.
2. Suppose that X is compact and Hausdorff under both T and T 0 , and that T and T 0 are comparable.
Say, w.l.o.g., T 0 T. Then let V be open in (X, T 0 ). Then we know that X V is closed under T 0
and so it is compact. Then since it is compact under T 0 , it is compact under T. Since X is Hausdorff
under T as well, we know compact subspaces of compact sets are closed, so X V is closed in (X, T).
Thus, V is open in (X, T). Therefore, T = T 0 .

27
26; 5: Let A and B be disjoint compact subspaces of the Hausdorff space X. Show that there exists
disjoint open sets U and V that contain A and B.

Let a A. Since X is Hausdorff, for each b B, there exists open U and V about a and b , respectively,
that are disjoint. So the set {V } will be an open cover of B. Since B is compact, there is a finite subcover
Vi := {Vi }mi=1 . However, this set depends on a , so Vi = Vi, . So we can do this for each a to get the set
{U } that will be an open cover of A, and a collection of open covers {Vi, } of B. Then since A is compact,
there is a finite subcover, {Uj }nj=1 with its finite set of corresponding B-covers, {Vi,j }nj=1 . Then for each
Vi,j , we define !
m
[
0
Vj = Vi .
i=1 j

So for each j, Vj0 will contain B, since it is a union of a cover. Further, Vj0 Uj = , for each j, by construction.
Then
n
\
V := Vj0
j=1

will still contain B and will be disjoint from all Uj . Thus, if we define
n
[
U := Uj ,
j=1

then U will clearly be open, and disjoint from V . We also have that V will be open because it is a finite
intersection of open sets. Lastly, by construction, U and V contain A and B respectively.

26; 9: Generalize the tube lemma as follows:


Theorem 2. Let A and B be subspaces of X and Y , respectively; let N be an open set in X Y
containing A B. If A and B are compact, then there exist open sets U and V in X and Y ,
respectively, s.t.
A B U V N.

Suppose A and B are compact subspaces of X and Y , respectively. Let N be an open set in X Y
containing A B. Then fix a A. So for each b B, there is a point a b N . Since N is
open, there is a basic open set U, V, containing a b . Then {U, V, }I covers a B.
m
\
Since B is compact, there exist a finite subcover {U,i V,i }m
i=1 . Define U := U,i , which will be
i=1
m
[
open in X, and each U V = U V,i N will contain a B. Then {U V }J will be an
i=1
open cover of A B and {U } will be an open cover of A. Since A is compact, there is a finite subcover
n
\ n
[
{Uj }nj=1 . So {Uj Vj }nj=1 will be a finite subcover of A B. Then define V := Vj and U := Uj . So
j=1 j=1
n
[ n
[
(Uj V ) = ( Uj ) V := U V .
j=1 j=1

17 April 6, 2015

28
27; 1: Prove that if X is an ordered set in which every closed interval is compact, then X has the
least upper bound property.

Suppose X is an ordered set in which every closed interval is compact. Let A X be bound from above
by a X. If we consider [x, a] for any x A, we get a collection {[x, a] : x A} of closed sets which by
hypothesis are compact. Then the intersection T of these sets will be a closed and compact
T set. Further, the
intersection will be non-empty because a [x, a]. Then there is some c A s.t. [x, a] = [c, a]. Then we
claim this c = sup A. It is clear that c is an upper bound.
Suppose c0 c is an upper bound. Then [c0 , a] [c, a]. SoTthere is y A [c0 , a] (if not, then c c0 ), and
thus y = c0 or c0 wouldnt be an upper bound. Therefore, [x, a] = [c0 , a] and c0 = c.
So c X and X has the least upper bound property.

27; 3: Recall that RK denotes R in the K-topology.


1. Show that [0, 1] is not compact as a subspace of RK .
2. Show that RK is connected. [Hint: (, 0) and (0, ) inherit their usual topologies as sub-
spaces of RK .]

3. Show that RK is not path connected.

1. Consider the open cover given by C = {(1, 2) K} {( n+2


1
, n1 )}nZ+ {( 34 , 54 )}.
1 1
Removing any one of the sets in C will mean leaving a point uncovered. Removing any of the ( , )
n+2 n
1
will result in the point n+1 not being covered. If you remove (1, 2) K, you will not cover 0. If you
3 5
remove ( , ) you will not cover 1. Thus, it has no finite subcover which means [0, 1] is not compact.
4 4
2. (, 0) and (0, ) as subspaces of R and RK are the same. Any basis element (a, b) of (, 0) or
(0, ) in R is itself a basis element in RK . Given a basis element B of RK , we can fit a neighborhood
(r, s) in between any deleted points of K, so we have established the hint.
Then (, 0) and (0, ) are connected in both topologies. Then closing them in RK , we get (, 0]
and [0, ), which are still connected. Since they share a point, their union is connected and our claim
follows.
3. Suppose it was path connected. Then in particular, [0, 1] would be path connected. So we would have
a continuous fucntion f : [0, 1] RK s.t. f (0) = 0 and f (1) = 1. So we have that the image of a
compact set under a continuous mapping is compact. But this contradicts part (a).

27; 4: Show that a connected metric space having more than one point is uncountable.

Suppose X is a connected metric space with metric d. Then suppose there exist a, b X s.t. a 6= b. Then
say d(a, b) = c R. Then if we define D : X [0, c], by D(x) = d(x, b), we have that D is continuous and
so by the intermediate value theorem, it achives every value in (0, c) R which is uncountable. Thus, X is
also uncountable.

29
18 April 13, 2015

30; 5:
1. Show that every metrizable space with a countable dense subset has a countable basis.
2. Show that every metrizable Lindelof space has a countable basis.

1. Suppose that X is a metric space with metric d and that A is a countable dense subset of A.
[
Let Bx be the collection of balls {Bd (x, 1/n)}nZ+ about x X. Then we claim B := Ba will be
aA
a basis for X. First, B is a subset of the basis for the metric topology. Then, given any x X and
open basis element Bd (x, r) about x. Then there exists and N Z+ s.t. 1/r < N = 1/N < r.
So Bd (x, 1/4N ) Bd (x, 1/N ) Bd (x, r). Then by density, a Bd (x, 1/4N ) A 6= . Observe
x Br (a, 3/4N ) Bd (x, 1/N ) Bd (x, r). Thus, B is a basis for X and since it is the countable union
of countable sets, it is itself countable.
2. Suppose X is a metric space with metric d and it is Lindelof. Then let (as in the hint for problem
[ section) An be the countable covering of X by balls of radius 1/n. Then we claim that
4 of this
B := An is a basis for the metric topology on X.
nZ+

Suppose x X and Bd (x, r) is a basic element about x. Then we know that there is an N Z+ s.t.
r > 1/N and so Bd (x, 1/N ) Bd (x, r). We also know that (by construction) for every n Z+ , there
exists an element an X s.t. x Bd (an , 1/n). So for n = 4N , we have that there is an a4N X s.t.
x Bd (a4N , 1/4N ). Observe that x Bd (a4N , 1/4N ) Bd (x, 1/N ) Bd (x, r). Thus, B is a basis for
the metric topology on X and since it is the countable union of countable sets, it is countable and our
claim follows.

31; 2:
Show that if X is normal, every pair of disjoint closed sets have neighborhoods whose closures are
disjoint.

Suppose X is normal and A, B are a pair of disjoint closed sets. Then there exist disjoint open sets UA , UB
that contian A, B respectively. But there also exist open sets VA , VB that contain A, B respectively s.t.
V A UA and V B UB . Thus, V A V B = since UA and UB are disjoint.

31; 3:
Show that every order topology is regular.

Let X be a space in the order topology. Suppose x X with x (a, b) - a basic open set.

1. Suppose x has an immediate successor s and an immediate predecessor p. Then V := (, s)(p, ) =


{x} (a, b) is open. Observe, V = V so V (a, b) and our claim follows.
2. Suppose x has an immediate successor s but does not have an immediate predecessor. Then if x
is the smallest element of X, then V := [x, s) = {x} (which is open) and we are back in the first
case. So assume that x is not the smallest element. Then there exists some a < < x, so that
V := (, s) = (, x] is open and V = [, x] (a, b).

30
3. When x has an immediate predecessor but no immediate successor, then we make the same (but
symmetric) argument as above.
4. Suppose x has no immediate successor or predecessor. Then there exists a < < x and x < < b. So
x V := (, ) which is open and V = [, ] (a, b) as desired.

So every order topology is regular.

19 April 15, 2015

29; 2: Let {X } be an indexed family of nonempty spaces.


Q
1. Show that if X is locally compact, then each X is locally compact and X is compact for
all but finitely many .
2. Prove the converse, assuming the Tychonoff Theorem.

Q
1. Suppose that X is locally compact. Then for the Q index , we aim to show that X is locally
compact. Let xa Xa . Then
Q there is a sequence ~
x in X Q
that has xa in the Xa - component. Then
there is an open
Q set U X that contains ~
x . Then since X , is locally compact, there is an open
set V U X about ~x s.t. V U and V is compact.
[Note that the previous implies that X is compact for all but finitely many because V is compact,
which means that (V ) = X is compact for all but finitely many .]
Then a (V ) Xa is open since the projection function is an open function. Further, a (V ) is compact
because (we claim) a (V ) = a (V ).
To see why consider x a (V ) and open Ox containing x, then Ox intersects a (V ). Then a1 (Ox ) is
open and contains every sequence that contains Ox in the a-th component. Thus, it intersects V . So
x a (V ). Now if x a (V ), then let Px be an open set containing x. Then a1 (Px ) is an open set
that contains every sequence that is Px in the a-th component. So it intersects V . Then Px intersects
a (V ). Thus, x a (V ).

2. Suppose
Q X is locally compact for every and is compact for all but finitely many . Then let
~x X . All but finitely many components of ~x are in compact spaces and the rest are in locally
compact spaces. So we may consider ~x in two parts, the part in the product of compact spaces, and
the finite part that is in the product of locally compact spaces. Then by Tychonoffs theorem, the
infinite product of compact spaces is compact.
We now show that the product of two locally compact spaces is locally compact.
Suppose X and Y are locally compact. Then say xy X Y with basic neighborhood U1 U2 . Then
1 (U1 U2 ) = U which is an open neighborhood of x. So there is an open V1 whose closure is compact
and V 1 U1 . Then 11 (V1 ) = V1 U2 . Similarly, we can find the analogous 21 (V2 ) = U1 V2 .
Then 11 (V1 ) 21 (V2 ) = V1 V2 U1 U2 . Further, if C is an open cover of V1 V2 , then we have
that {1 (C )} and {2 (C )} are open covers of V 1 and V 2 respectively (with C C). So each has a
finite subcover called {1 (Ci )} and {2 (Cj )} respectively. Then {Ci } {Cj } will be a finite subcover
of V1 V2 .
So by induction, we get that finite products of locally compact spaces is locally compact. Then we
know that every compact space is locally compact, thus the product of the two parts give a locally
compact space as desired.

31
32; 4: Show that every regular Lindelof space is normal.

Let X be regular and Lindelof. If A and B are disjoint closed sets in X. Then for each point a A, there
is a neighborhood Ua such that U a B = . Then {Ua }aI (X A) is an open covering of X, so there
is a countable subcovering, which gives a countable subcovering of A called {Ui }iZ+ . Then we can do the
same thing to get a countable covering of B called {Vi }iZ+ . Then as in the proof of theorem 32.1, we let
i
[ i
[
Ui0 = Ui Vj and Vi0 = Vi Uj.
j=1 j=1

So that each Ui0 and Vi0 are open and {Ui0 } (resp. {Vi0 }) covers A (resp. B). Then the open sets
[ [
U0 = Ui0 and V 0 = Vi0
iZ+ iZ+

are disjoint because if x U 0 V 0 , then x Ui0 Vj0 . Then w.l.o.g. say i j. Then x Ui0 but at the same
time, x Vj0 = x / Uk0 for all k j = x / Ui0 . Therefore, we have our claim.

32; 6: A space X is said to be completely normal if every subspace of X is normal. Show that X is
completely normal iff for every pair A, B of separated sets in X (i.e. sets such that AB = AB = ),
there exist disjoint open sets containing them. [Hint: If X is completely normal, consider X (AB)].

( = ) Let Y be a subspace of X. Then if A is closed in Y , and V is an open set (in Y ) that contains it,
then V = U Y where U is open in X. So U is an open set that contains A. So there exits open W X
s.t. W U . Then W Y is open in Y and W Y V .
( = ) Suppose X is completly normal. Then consider two separated sets A, B. Let Y = X (A B). So
Y A, B is open and a subspace of X, and therefore normal. So Y = (X A) (X B). Then in Y ,
A B = . So there exists open U, V Y s.t. U Y = , and A U , B V . Since we have that Y is open
in X, then U, V are open in X as well. Therefore, we have disjoint open sets containing A and B.

20 April 20, 2015

33; 5: Prove:
Theorem (Strong form of the Urysohn lemma): Let X be a normal space. There is a continuous
function f : X [0, 1] s.t. f (x) = 0 for x A, and f (x) = 1 for x B, and 0 < f (x) < 1 otherwise,
A and B are disjoint closed G sets in X.

Let X be a normal space.


( = ) Suppose there is a continuous function f : X [0, 1] s.t. f (x) = 0 for x A, and f (x) = 1
for x B, and 0 < f (x) < 1 otherwise. Then we claim that f 1 ({0}) = A. This is clear because
x A f (x) = 0 x f 1 ({0}) by assumption. Similarly f 1 ({1}) = B. So we have that
f 1 ((0, 1]) = X A which implies that A is closed. Similarly we get that B is closed. Then we observe that
f 1 ([0, 1/2)) and f 1 ((1/2, 1]) are two disjoint open sets that contain A and B respectively. Therefore, A
and B are disjoint.

32
Now consider the collection of sets {[0, 1/n)}nZ+ . The intersection of this collection is {0}. This induces
the collection {f 1 ([0, 1/n))}nZ+ of open sets in X. Then by continuity of f , we have
\ \
f 1 ([0, 1/n)) = f 1 ( [0, 1/n)) = f 1 ({0}) = A.
nZ+ nZ+

So we have that A is G . The same argument holds for B.


( = )
Lemma: A is a closed G set in X = there exists a function f : X [0, 1] s.t. f (x) = 0 for x A and
f (x) > 0 for x
/A.
T
Proof : Suppose A is a closed G set in X. So if A = nZ+ An , then we may w.l.o.g. consider the An as
being ordered by proper inclusion where An+1 An , by normality. By Urysohns lemma, there is a sequence
of continuous function f1 , f2 , ... defined as follows. We let

f1 : X [1/2, 1] s.t. f1 (A2 ) = 1/2 and f1 (X A1 ) = 1.

Then
f2 : X [1/3, 1/2] where f2 (A3 ) = 1/3 and f2 (X A2 ) = 1/2
The general term will be:

fn : X [1/(n + 1), 1/n] s.t. fn (An+1 ) = 1/(n + 1) and fn (X An ) = 1/n.

So we now define f : X R as follows:



0 if x A
f (x) = 1 if x X A1
fn (x) if x An An+1

This will be well defined because the An are ordered by proper inclusion. This is continuous because each fn
is continuous and fn agrees with fn+1 on theT boundary of An+1 , which is the only place where a potential
discontinuity may occour. Then since A = nZ+ An , this function will have the desired properties. 
So we find functions f, g : X [0, 1] s.t. f vanishes precicely on A and g vanishes precicely on B. Then we
define the function
g(x)
F : X [0, 1] s.t. F (x) = 1 .
f (x) + g(x)
Then F (x) = 1 precicely on B and F (x) = 0 precicely on A and 0 < F (x) < 1 for all other x, as desired.

33; 6: A space X is said to be perfectly normal if X is normal and if every closed set in X is a G
set in X.
a) Show that every metrizable space is perfectly normal.

b) Show that a perfectly normal space is completely normal. For this reason the condition of
perfect normality is sometimes called the T6 axiom.
[Hint: Let A and B be separated sets in X. Choose continuous functions f, g : X [0, 1] that
vanish precisely on A and B, respectively. Consider the function f g.]
c) There is a familiar space that is completely normal but not perfectly normal. What is it?

33
a) We know that every metric space is normal, so we show that every closed set in X isSa G set. Let A
be a closed set in X with metric d. Then for each n Z+T, define the open set
T Cn := aA Bd (a, 1/n).
Then Cn is an open set that contains A. However, A = nZ+ Cn , because nZ+ Bd (a, 1/n) = {a}.

b) Suppose X is perfectly normal. Then let A, B be disjoint closed sets in Y . Then let A and B be
separated sets. So A, B are closed, G sets in X. Then there are continuous functions f, g : X [0, 1]
that vanish precicely on A, B respectively. So f g := h is continuous and h : X R. Then
h1 ((, 0)) A and h1 ((0, )) B which are open and disjoint. Thus, X is completely normal.
c) S .

34; 7: A space X is locally metrizable if each point x X has a neighborhood that is metrizable in
the subspace topology. Show that a compact Hausdorff space X is metrizable if it is locally metrizable.
[Hint: Show that X is a finite union of open subspaces, each of which has a countable basis.]

Suppose that X is a compact Hausdorff space that is locally metrizable. Since X is compact Hausdorff, it is
normal. Then for each point x X, there is a neighborhood Ux that is metrizable in the subspace topology.
Also, Ux is locally compact. So there is some neighborhood Vx of x s.t. V x Ux is compact. So V x has
a countable basis, which means that Vx ( V x ) does as well. Then the set {Vx }xX is a cover of X, so by
compactness, has a finite subcover {Vxi }ni=1 . Then X is the finite union of these open subspaces, each of
which has a countable basis. Thus, X is a regular space with a countable basis. So by Urysohns metrization
theorem, X is metrizable.

21 April 24, 2015

37; 1: Let X be a space. Let D be a collection of subsets of X that is maximal w.r.t. the finite
intersection property.
a) Show that x D for every D D iff every neighborhood of x belongs to D. Which implication
uses maximality of D?
b) Let D D. Show that if A D, then A D.
T
c) Show that if X satisfies the T1 axiom, there is at most one point belonging to DD D.

Suppose D is a collection of subsets of X that is maximal w.r.t. the FIP. So we may apply Lemma 37.21 .

a) ( = ) Suppose x D for every D D. Then every neighborhood U of x intersects every D D.


Therefore, U D.
(This used maximality so that we could apply Lemma 37.2.)
( = ) Suppose that every neighborhood U of x belongs to D. Then this means that U D D for
every D D. Thus, U D 6= (since D is maximal) and so x D.
1 Lemma 37.2. Let X be a set; let D be a collection of subsets of X that is maximal w.r.t. the FIP. Then:
a) Any finite intersection of elements of D is an element of D.
b) If A is a subset of X that intersects every element of D, then A is an element of D.

34
b) Suppose A D D. Then since D intersects every element of D, so does A. Therefore A D.
T
c) Suppose that X is T1 and there is more than one point in DD D := D0 . Say there are 2 and
D0 = {a, b}. Then if either {a} or {b} is in D (w.l.o.g. say it is {a}), then so is X {a} since this
is a neighborhood of b. So (X {a}) {a} = which contradicts that D has the finite intersection
property.
So assume that neither {a} nor {b} is in D. Then this means thatthere are some sets in D that do not
contain w.l.o.g. {a}. Then we have that D {D {a} : D D} ) D also has the finite intersection
property. This contradicts the maximality of D. Therefore, we conclude that there cannot be more
than one point in D0 .

Alans counterexample to the above:

Consider the set Z+ with the finite compliment topology. Thenthere exists a set that has the finite
intersection property, since {Z+ } itself has this property. So we take a maximal collection D that has
the finite intersection property.
Then D has only infinite sets. This means that the closure of every set is Z+ and so our claim is not
true.

The proof that we gave above works for the modified statement
\
D,
DD

which is what (I think) was intended.

37; 2: A collection A of subsets of X has the countable intersection property if every countable
intersection of elements of A is nonempty. Show that X is a Lindelof space iff for every collection A
of subsets of X having the countable intersection property,
\
A 6= .
AA

Let A be any collection of open subsets of X, and let C = {X A | A A}.


X is Lindelof. If A is an open cover of X, then there exists a countable subcover {An }nZ+ .
If there is no countable subcollection {An }nZ+ of A that covers X, then A does notTcover X. If
there is no countable subcollection {Cn }nZ+ of C that has trivial intersection (i.e. nZ+ Cn 6= ), then
C does not haveTtrivial intersection (i.e.
T
T CC C 6= ). If every countable subcollection {Cn }nZ+
intersects (i.e. nZ+ Cn 6= ), then CC C 6= .
Since we let A be any collection of open sets, we can make the C C anything we want. In particular, we
can make them the closures of subsets of X. So the last statement
T is equivalent to For every collection A
of subsets of X having the countable intersection property, AA A 6= . as desired.

22 April 29, 2015

35
51; 2: Given spaces X, Y , let [X, Y ] denote the set of homotopy classes of maps of X into Y .
a) Let I = [0, 1]. Show that for any X, the set [X, I] has a single element.

b) Show that if Y is path connected, then the set [I, Y ] has a single element.

a) We can, w.l.o.g. let I = [0, 1] because we can scale this continuously to any other interval. Then we
aim to show that any continuous function f : X [0, 1] is homotopic to the zero map. We define the
family ft : X [0, 1] with t [0, 1]. This induces a map

F : X [0, 1] [0, 1] which is given by F (x, t) = ft (x) = tf (x).

Then this F is continuous because tf (x) is the product of two continuous functions and finally, F (x, 0) =
0f (x) = 0 and F (x, 1) = f (x).
b) Suppose Y is path connectd. Then any continuous function f : I Y will be a path in Y . We may
assume w.l.o.g. that I = [0, 1]. Let f be a path in Y . Then if x0 and x1 are the initial and final points
respectively, then we want to continuously shrink f . At time t [0, 1], our movie sees us walking
along f then at position x = t on our path f , we stop and stand in place for the remaining time.
Explicitly, this is given by the family ft where

f (x) if x [0, t]
ft (x) =
f (t) if x [t, 1]

which is continuous by the pasting lemma. Thus, this gives that a path connected space is homotopic
to a constant. Then if we are given any two paths g0 , g1 : [0, 1] Y , we can shrink both to their initial
points, and then connect the initial points via path connectivity. Thus, our claim follows.

52; 4: Let A X; suppose r : X A is a continuous map s.t. r(a) = a for each a A. (The
map r is called a retraction of X onto A.) If a0 A, show that

r : 1 (X, a0 ) 1 (A, a0 )

is surjective.

Let I = [0, 1]. Suppose [g : I A] 1 (A, a0 ). Then we are looking for an [f : I X] 1 (X, a0 ), s.t.
r ([f ]) = [r f ] = [g]. So we want to find f s.t. the following diagram to commutes:

f
I X
r
g
A

Since A X and r |A = iA , we simply take f : I X where f (t) = g(t). Then r f = g trivially. So we


have that [r f ] = [g] and therefore, r ([f ]) = [g] and r is surjective as desired.

36
52; 6: Show that if X is path connected, the homomorphism induced by a continuous map is
indepenedent of base point, up to isomorphisms of the groups involved. More precisely, let h : X Y
be continuous, with h(x0 ) = y0 and h(x1 ) = y1 . Let be a path in X from x0 to x1 , and let = h.
Show that
(hx0 ) = (hx1 ) .
This expresses the fact that the following diagram of maps commutes.
(hx0 )
1 (X, x0 ) 1 (Y, y0 )

(hx1 )
1 (X, x1 ) 1 (Y, y1 )

Let [p] 1 (X, x0 ). Then

((hx0 ) ([p])) = ([hx0 p]) = [] [hx0 p] [].

On the other hand

(hx1 ) (([p])) = (hx1 ) ([] [p] []) = (hx1 ) ([ p ]) = [h ( p )]

[(h ) (h p) (h )] = [h ] [hx0 p] [h ] = [] [hx0 p] [].


Thus, the maps are equal and so the diagram commutes and we have our claim.

23 May 1, 2015

54; 2: In defining the map F in the proof of Lemma 54.2, why were we so carful about the order
in which we considered the small rectangles?

The reason that we were so careful when defining the map was that we needed to make sure that we were
lifting ourselves to the same slice of the covering space at each step. The only way to do this was to rely
on the previously defined portions of the map, which we knew (inductively) were on the correct slice. The
order that was chosen in the proof was very important because only by backtracking (after finishing the,
w.l.o.g., 1st row) to define F next on the 2 1 block was that we are matching the start of the next stage
(i.e. the second stage which corresponds to the second row) with being on the same slice as the start of
the first row, which (inductively, again) guarantees that we define the entire I I to be on the correct slice.
This is to avoid the possibility of accidentially going to a different slice, which would tear (of sorts) our
homotopy.

54; 3: Let p : E B be a covering map. Let and be paths in B with (1) = (0); let and
be liftings of them such that (1) = (0). Show that is a lifting of .

37
Since we have that s final point corresponds to s initial point, we can take the product of these paths to
get that is a path in B from (0) to (1). Then we are given that we can lift and to a slice of E
s.t. the lifted paths agree on their corresponding final and initial points (as in B).
We then lift to - a path in a slice V of the cover. Using the final point of in place of e0 , we can lift
to s.t. (0) = (1) = e0 . Then since this slice is homeomoprhic with B, and since and share a point,
they are in the same slice and, again, because this slice V is homeomorphic to B, we have that is a
lifting of , as desired.

54; 5: Consider the covering map p p : R R S 1 S 1 of Example 4 of 53. Consider the


path
f (t) = (cos(2t), sin(2t) (cos(4t), sin(4t)
in S S . Sketch what f looks like when S 1 S 1 is identified with the doughnut surface D. Find
1 1

a lifting f of f to R R, and sketch it.

Recall that p(t) = (cos 2t, sin 2t).

38

You might also like