Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Topology Midterm

Michael Lin
30 October 2016
158/12.

a. =:
Suppose that [a, b) and [b, c) have the order type of [0, 1). Then there are order preserving functions f : ([a, b) [0, 1) and g : [b, c) [0, 1). Define
f (x)/2
(x < b)
h : [a, c) [0, 1), h(x) =
. We find that h1 (y) =
(1 + g(x))/2 (x b)
(
1
f 1 (2y)
(y [0, 2 ))
. By inspection, h is order preserving because f and
1
g (2y 1) (y [ 21 , 1))
g are order preserving and if x1 < b x2 , then h(x1 ) = f (x1 )/2 < 21
(1 + g(x2 ))/2 = h(x2 ). Thus, [a, c) has the order type of [0, 1).
=:
Suppose that [a, c) has the order type of [0, 1). Then there is a bijective, order
preserving function f : [a, c) [0, 1). Define g : [a, b) [0, 1), where g(x) =
f (x)/f (b), and h : [b, c) [0, 1), where h(x) = (f (x) f (b))/(f (c) f (b)). By
inspection, these are order preserving (which implies injective), and surjective on
[0, 1), thus [a, b) and [b, c) have the order type of [0, 1).
a. =:
Suppose that each interval [xi , xi+1 ) has the order type of [0, 1). Thus, there is
a sequence of functions bijections {fi : [xi , xi+1 ) [0, 1)}. We can construct a
bijective function f : [x0 , b) [0, 1), where



1
1
1
+ fi (x)

x [xi , xi+1 ).
(1)
f (x) = 1
i+1
i+1 i+2
Essentially, f scales and maps in order each fi to the intervals [0, 1/2), [1/2, 1
(1/2 1/3)), [1 (1/2 1/3), 1 (1/3 1/4))..., which partition [0, 1). We can
see that f is order preserving becaus each fi is order preserving, and that f is
surjective and thus bijective. Thus, [x0 , b) has the order type of [0, 1).
=:
Suppose that [x0 , b) has the order type of [0, 1). Then by part a (forward direction), [x0 , x1 ) and [x1 , b) has the order type of [0, 1), since x0 < x1 < b. Also given
1

that [xi , xi+1 ) and [xi+1 , b) have the order type of [0, 1), [xi+1 , xi+2 ) and [xi+2 , b)
have the order type of [0, 1). By induction, [xi , xi+1 ) have the order type of [0, 1).
a. Since S is well ordered, there is a a1 , the immediate successor of a0 . Since every
point less than a1 0 lies in a0 [0, 1), [a0 0, a1 0) = [a0 0, a0 1), which
naturally has the order type of [0, 1), by only considering the second coordinate.
Thus the base case is proven.
Suppose that [a0 0, b 0) has the order type of [0, 1) for all b < a. Note that this
condition is equivalent to for all b Sa . By the properties of S , Sa is countable,
so there is a non-repeating sequence b1 , b2 , ... such that {bi } = Sa .
Suppose we can select a subsequence bi1 , bi2 , ... such that bik < bik+1 , infinitely.
Then sup{bik } = a, since sup{bi } = a. Then by part b, [a0 0, a 0) has the
order type of [0, 1).
Suppose we cant select a subsequence, and that there is a j such that for all i > j,
bj > bi . Then Sa has a maximal element, bk where 1 k j, and bk + 1 = a.
But [bk 0, a 0) = [bk 0, bk 1), which naturally has the order type of [0, 1).
By part a, [a0 0, a 0) has the order type of [0, 1).
By transfinite induction, the result is proved.
a. Let a1 b1 < a2 b2 be any two points in L. Then a1 < a2 or a1 = a2 , b1 < b2 .
If its the second case, we can take the path f (x) = a1 x, to join the two.
In the first case, since [a1 0, (a2 + 1) 0) has the order type of [0, 1), there is a
bijective function f : [0, 1) [a1 0, (a2 + 1) 0). The order-preserving property
of the function f makes it continuous, because fixing any x, y [a1 0, (a2 +1)0)
with x < y, if z (x, y), f 1 (z) (f 1 (x), f 1 (y)), and if a (f 1 (x), f 1 (y)),
then f (a) (x, y), so f 1 ((x, y)) = (f 1 (x), f 1 (y)), which is open. So we can
take the path from a1 b1 to a2 b2 to be f restricted to the closed interval
[f 1 (a1 b1 ), f 1 (a2 b2 )] [0, 1).
a. Take any point x L and note that since L has no maximal or minimal element,
we can always choose a1 b1 < x < a2 b2 . Since [a1 0, (a2 +1)0) has the order
type of [0, 1), there is a bijective function f : [0, 1) [a1 0, (a2 + 1) 0). We
already know that this is continutous, and its obvious from the order-preserving
property that the image of any open interval from [0,1) is an open interval in L.
Finally, we can restrict f and define h : (f 1 (a1 b1 ), f 1 (a2 b2 )) (a1 b1 , a2
b2 ), and noting that h is still a bicontinuous bijective function, and therefore a
homeomorphism. So x has a neighborhood homeomorphic to an open interval in
[0, 1) R.
S
a. Consider D = aS (a 0, a 1), an open set in L made up of uncountably
many disconnected and disjoint open intervals. Suppose there was an imbedding
f : L A Rn . Since f is a homeomorphism and thus bicontinuous, f (D) is an
open set in A made up of uncountably many connected components. But since
2

A Rn has a countable basis, this is a contradiction.


178/6.

a. Let x, y, x < y be any two points in C. Pick n suchSthat 3n > 1/(y  x). Thus
1+3k 2+3k
[0, 1]
y x > 1/3n , so x + 1/3n < y. But An+1 = An
,

S 1+3k 2+3k  S3n 1  3k 1+3k   2+3k 3+3k  k=00 3n+11 3n+1 2
4
3n+1 , 3n+1 = 3n+1 , 3n+1 3n+1 , 3n+1

,
k=0h 3n+1 , 3n+1 =i k=0
3n+1 3n+1i
h
3n+1 1 3n+1
3n+1 4 3n+1 2
... 3n+1 , 3n+1 3n+1 , 3n+1 = D, which is a union of disjoint closed
intervals of length at most 2/3n+1 < 1/3n . So x and y must belong to different
disjoint closed intervals in D, so they are in different connected subspaces in
C D. So the only connected subspaces in C are one point sets.
a. Since every An is the finite union of closed intervals (as we will show), An is closed.
Since arbitrary intersections of closed sets are closed, C is closed. Finally, since
[0, 1] is compact, C is compact because it is a closed subspace.
a. Just by computation and inspection: {An } = {[0, 1/3][2/3, 1], [0, 1/9][2/9, 3/9]
[6/9, 7/9] [8/9, 1], ...}. The end points lie of these closed intervals in C because
they never lie in a middle third, so they are never deleted in any An .
a. Given any point x C, and any -ball B of x, there exists an n such that 1/3n < .
Thus since x must lie in some closed interval of An with length 1/3n , there an
endpoint inside B, which is also inside C. So any neighborhood of x contains a
point of C so C has no isolated points.
a. Since R is Hausdorff, C is Hausdorff, and it is also non-empty, compact, and has
no limit points. By theorem 27.7, C is uncountable.

181/7.

a. We first prove that f is continuous. Given any x X,  > 0, d(x, y) < = 


d(f (x), f (y)) < , so f is continuous by the equivalent epsilon-delta criterion.
Thus f n (X) is compact for all n, since X is compact.
We know that f (X) X. Suppose f n+1 (X) f n (X). Since functions preserve
n+2
n+1
n+1
n
inclusions,
T fn (X) f (X). By induction, f (X) f (X) for all n. Let
A = n=1 f (X). Since A is the intersection of a nested sequence
T of nonempty
n+1
compact
sets,
A
is
nonempty.
Take
any
x

A,
and
so
f
(x)

(X) =
n=1 f
T n
n=1 f (X), so f (A) A.
We calculate that d(f 2 (x), f 2 (y)) d(f (x), f (y)) 2 d(x, y). By induction,
d(f n (x), f n (y)) n d(x, y). Thus A cannot have more than one point because, if
x, y A with d(x, y) =  > 0, But since x, y f n (X), There exists a, b X such
that f n (a) = x, f n (b) = y, so d(a, b) n d(x, y). But since n can be arbitrary,
this implies that X is not bounded, which cannot be true since X is compact.
Thus A has only one point, and f (A) A = {x} means f (x) = x. Finally any y
for which f (y) = y will obviously be in A, so the fixed point of f is unique.
a. We let A be as before, and let x A. We define a sequence xn such that x =
f n+1 (xn ), and define the sequence yn = f n (xn ), and this is a sequence in X, which
is compact, yn has a convergent subsequence with the limit a. Fix any N , and for
3

all  > 0, there is a sequence n1 , n2 , ... > N such that d(yni , a) = d(f ni (xni ), a) < .
Since yni f N (X), this means that a is a limit point of f N (X). Since f N (X)
is a metric space and therefore limit point compact, a f N (X) for all N . Thus
a A. Finally d(f n+1 (xn ), f (a)) = d(x, f (a)) < , for all  > 0, which means
that f (a) = x, which means that x f (A), or A f (A). Combining the result
from part a, A = f (A). Take any point in A, and since A is the intersection of
closed sets (compact iff closed in a Hausdorff space), A is compact, so fixing a
point x0 A, the function g(x) = d(x, x0 ) has a maximum in A by the continuity
of the metric function and extreme value theorem. Supposing max g =  > 0,
for all x, y A with x 6= y,  d(x, y) > d(f (x), f (y)) = max g f = max g, a
contradiction, so max g = 0. Thus A has only one point.
a. The derivative of f is 1 x, which is always non-negative on [0, 1]. Thus, minf =
f (0) = 0, and maxf = f (1) = 1/2, so f maps X into X.
According to the mean value theorem, there always exists c (x, y) such that
(x)|
(x)|
= |1 c|. Since 1 > c > 0, |f (y)f
=
f (y) f (x) = f 0 (c)(y x), or |f (y)f
|yx|
|yx|
|1 c| < 1, so f is a shrinking map.
In particular, there always exists c (0, x) such that f (x) f (0) = f 0 (c)(x 0),
(0)|
or |f (x)f
= |1 c| infc(0,x) |1 c| maxc[0,x] |1 c| |1 x|. Since x can
|x0|
be arbitrarily close to 0, f cannot be a contraction.
a. Again, we use the mean value theorem, that there always exists a c (x, y) such
(x)|
c
c
that |f (y)f
= |(1 + (c2 +1)
1/2 )/2| < |(1 + (c2 )1/2 )/2| 1, so f is a shrinking map.
|yx|
The map is not a contraction because f 0 (c) approaches abitrarily close to 1 as c
gets large, and we can choose x, y to be large which constrains the secant slope
to be arbitrarily close to 1. Finally, if f (x) = x, then [x + (x2 + 1)1/2 ]/2 = x, so
(x2 + 1)1/2 = x, so x2 + 1 = x2 , a contradiction.
194/9. Suppose that X Lindelof and A X is closed. Then take any open covering of A, and
add to this covering X \ A, an open set, which makes an open covering of X. Since X
is Lindelof, there is a countable subcovering which covers X and therefore A, so there
is a countable subcovering of A, and thus A is Lindelof.
Examine RL RL . The claim is that Q Q is a countable dense subset, since any basis
element [a, b) [c, d) inside any open neighborhod of any point in RL RL contains a
point in Q Q because the rationals are dense in R. But the Sorgenfrey plane L in R2L
is closed, because R2L L is open in the standard topology of R2 and thus is also open
in R2L (which is finer). But since the subspace topology of L is the discrete topology,
every subset of L is open and therefore every subset is also closed, so the only dense
subset of L is L which is uncountable.
194/15. Let A be set of continuous functions containing finitely many line segments with rational endpoints. A thus has the cardinality of Q2 Q4 Q6 ..., and the countable
union of countable sets is countable, so A is countable. Given any continuous function
4

f C(I, R), and any ball of radius  around f , we can find a g A that approximates f
better than  (thus f is a limit point of A) though the following: Since the set of continuous functions on a compact set is also uniformly continuous, there exists a such that
|x y| < |f (x) f (y)| < /5. Let N > 1/, and partition the [0, 1] interval into
N equal partitions of length 1/N , calling the endpoints x0 = 0, x1 = 1/N, ..., xN = 1.
Define g(xi ), such that g(xi ) Q and |g(xi ) f (xi )| < /5, and connect these points
with straight line segnments, noting that g A. Say we have any x, then xi x < xi+1
for some i. Because |xxi | < 1/N < , |f (x)f (xi )| < /5. By the triangle inequality,
|f (x) g(xi )| < 2/5. Since |xi xi+1 | = 1/N < , |f (xi ) f (xi+1 )| < /5. By the
triangle inequality, |g(xi ) g(xi+1 )| < 3/5. Since g is made of straight line segments,
|g(x) g(xi )| < 3/5. By the triangle inequality, |f (x) g(x)| < , so g is in the epsilon
ball, and so A is a dense countable subset of C(I, R).
195/16.

a. Consider the set of functions A that is a constant rational number on each interval
of a finite partition of [0, 1] the the following form: {[0, x1 ), [x1 , x2 ), ..., [xn2 , xn1 ), [xn1 , 1]},
where xi Q. For each n, the cardinality of the collection of such partitions is
that of Qn1 , and so the cardinality of constant rational functions on these n
intervals is that of Qn1+n = Q2n1 , a countable set. So the union A of the collections of such functions over n is a countable union of countable setes, so A is
countable.
Take any function f in RI , and any open neighborhood around Q
f , and note that
there is a basis element inside the neighborhood that is U = x[0,1] Ux where
Ux = R except for {Ux1 , ..., UxN } = {(a1 , b1 ), ...(an , bn )} for a finite N (we take
here xi < xi+1 ). Since Q is dense in R, for every xi , we can choose a yi Q such
that xi < yi < xi+1 , and we can partition [0, 1] using these yi . Next, we define a
function g such that g([yi1 , yi )) = ci , with ci Q and ci (ai , bi ). Now we can
see that g A and g U , so A is a countable dense subset of RI .
a. Fix an interval (a, b) in R, and define f : J P(D) by f () = D 1 ((a, b)).
Suppose f () = f (), then D 1 ((a, b)) = D 1 ((a, b)). Suppose that
by contradiction that 6= . Then consider the open set U in 1 ((a, b)) where
(U ) = (a, b) and (U ) = (b, b+2) and i (U ) = R otherwise, so U 6 1 ((a, b)).
Since D is dense, there is a d U D and so d
/ 1 ((a, b)), but this is a
contradiction, because d D 1 ((a, b)) = D 1 ((a, b)).
So f is injective, and we have an injection from a set with cardinality greater than
P(Z) + to a set with cardinality P(Z+ ) (because #D = #Z+ ), a contradiction.

195/17. Take the point (0, 0, 0, ...), suppose that we have a countable basis A1 , A2 , ... for this
point. Then we can construct a new open set U where i (U ) i (Ai ) by taking an
even smaller open interval around 0 for the ith box of U . Then U is a proper subset
of each Ai , so the countable basis is not actually a basis. So Q is not first countable.
Thus it is also not second countable.
Since Q = Q1 Q2 Q3 ..., the space has a countable number of points, any open
5

covering of a subset can have a countable subcovering by taking an open set for each
point in the subset, so it is Lindelof Also, the countable dense subset is the space itself,
so it is separable trivially.
199/3. Let X be an ordered set with the order topology. Suppose x U X with U open.
Then there is some a, b such that x (a, b) U .
If (a, b) is a one point set, then then we can take V = (a, b), and so V = (a, b), since
X {x} is always open in the order topology, so V U ,
Otherwise, if x is the minimal element of (a, b), we can take V = (a, c), where x < c < b,
and any element d such that d < x is not in the closure of V because we can take
(d 1, x), a neighborhood of d that has no intersection with V . So V (a, c] U . V
can be constructed similarly if x is the maximal element of (a, b).
If x is neither maximal or minimal, we can take V = (c, d), where a < c < x < d < b.
And every case, we show that there is a neighborhood V of X such that V U , which
by Lemma 31.1 (a) implies that X is regular.
200/9.

a. Take any irrational number x in [0, 1]. Since V is open in R2l , there is a basis
element [a, b) [c, d) contained by V that contains x x. But there always
exists a n such that n > 1/(b x) and n > 1/(d (x)). So, x + 1/n < b and
x + 1/n < d, and [x, x + 1/n) [x, x + 1/n) V . So x Kn . So every
irrational number is contained in Kn for some n, so the union of all Kn and the
rational numbers in [0, 1] is [0, 1].
a. Suppose that for all n, K n does not contain any open interval of R. This means
that the interior of K n is empty, because otherwise there would be an open set
and therefore an open interval in the interior.
S one point set (which is
SAlso, every
closed in [0,1]) has an empty interior. But n=1 K n qQ[0,1] {q} = [0, 1], which
has an interior, contradicting Exercise 5 of 27, which states that the union of a
countable collection of closed sets with empty interiors has an empty interior.
a. From part b., we have (a, b) K n . Take any point x (a, b), and x Kn or x is
a limit point of Kn with x
/ Kn . If it is the former case, x (x + ) V since
[x, x + 1/n) [x, x + 1/n) V . In the latter case, (a, b) is open, so take the
open interval (x , x) inside (a, b) such that 0 < <  and note that there is a
point y Kn that is also in (x , x), since this is a non-empty open interval is in
K n . Finally, note that y < x, and x < y + < y + , so x [y, y + ). Also, note
that x +  < y + , and y < x + < x + , so (x + ) [y, y + ).
So x (x + ) [y, y + ) [y, y + ) [y, y + 1/n) [y, y + 1/n) V .
a. Any neighborhood of q q contains a point of the form q (q + ), with
sufficiently small  < 1/n, which belongs to V , and thus q q is a limit point
of V .

You might also like