F (X, Y) Xy Is Continuous
F (X, Y) Xy Is Continuous
Exercises:
§18, 19
Show that f × g is continuous when A × C and B × D are given the product topologies.
That is, C is the set of sequences that are eventually equal to zero.
(a) Determine C when RN has the box topology.
(b) Determine C when RN has the product topology.
Q
5. Let {Xj | j ∈ J} be an indexed family Q of topological spaces. Let (xi )i∈I ⊂ j∈J Xj be a net; that is,
for each i in the directed set I, xi ∈ j∈J Xj is a J-tuple.
Q
(a) EquipQ j∈J Xj with the product topology and show that the net (xi )i∈I converges to some
x ∈ j∈J Xj if and only if for every j ∈ J the net (πj (xi ))i∈I converges to πj (x) in Xj .
Q
(b) Equip j∈J Xj with the box topology and prove one of the directions in the previous part is true
and show the other is false by finding a counterexample in RN .
6*. Let R have the standard topology and consider the functions f, g : R → R defined by
(
1 x∈Q
f (x) = ,
0 x∈R\Q
and (
1 n
m x ∈ Q with x = m for n ∈ Z and m ∈ N sharing no common factors
g(x) = .
0 x∈R\Q
Solutions:
(b) Let A := {x ∈ X | f (x) ≤ g(x)} and B := {x ∈ X | g(x) ≤ f (x)}. Then these are closed subsets
of X by the previous part (where we simply swap the roles of f and g for B), and moreover there
union is X: for all x ∈ X we must have either f (x) < g(x), f (x) = g(x), or f (x) > g(x) and
so A contains all those x satisfying the first two while B contains all those x satisfying the last
two. Observe that for x ∈ A we have h(x) = f (x), and for x ∈ B we have h(x) = g(x). Thus
h |A = f and h |B = g, which are continuous by assumption. Also note that for x ∈ A ∩ B we have
f (x) ≤ g(x) and g(x) ≤ f (x) which implies f (x) = g(x). Thus f |A∩B ≡|A∩B and so the pasting
lemma implies h is continuous.
4. (a) We claim that C is closed and hence equals its own closure. We will show RN \ C is open. Let
x = (xn )n∈N ∈ RN \ C. Consider A := {n ∈ N | xn 6= 0}, which is infinite by virtue of x 6∈ C. For
n ∈ A, define Un := (xn − |xn |, xn + |xn |) and note that 0 6∈ Un . For n ∈ N \ A, define Un := R.
Then then product Y
U := Un
n∈N
is a neighborhood of x in the box topology, but we claim it is disjoint from C. Indeed, for any
y = (yn )n∈N ∈ C we have yn 6= 0 for each n ∈ A. Since A is infinite, we see that y 6∈ C. We
have shown that every element of RN \ C has a neighborhood entirely contained in this set. Hence
RN \ C is open and therefore C is closed.
(b) We claim that C = RN . It suffices to show for any x = (xn )n∈N ∈ RN , all of its neighborhoods
intersect C. Let U be a neighborhood of x. Then there exists a sequence of open sets Un ⊂ R
with Un = R for all but finitely many n ∈ N satisfying
Y
x∈ Un ⊂ U
n∈N
(this is because such sets form a basis for the product topology). Consider the finite set A :=
{n ∈ N : Un 6= R}, and define y = (yn )n∈N ∈ RN by
(
xn if n ∈ A
yn := .
0 otherwise
Q
Since A is finite, we have y ∈ C and moreover y ∈ Un ⊂ U . Thus U ∩ C 6= ∅ and so x ∈ C.
Since x ∈ RN was arbitrary, we see that C = RN as claimed.
5. (a) (=⇒): Recall that we showed continuous functions map convergent nets to convergent nets. Since
the coordinate projections πj are continuous for each j ∈ J, it follows that (πj (xi ))i∈I converges
to πj (x).
(⇐=): Let U be a neighborhood of x. Then there exists an indexed collection of open sets
{Uj ⊂ Xj | j ∈ J} such that Uj = Xj for all but finitely many j ∈ J satisfying
Y
x∈ Uj ⊂ U.
j∈J
(b) The “=⇒” direction in the previous part is still true in the box topology: this topology is finer
than the product topology which implies the coordinate projections are still continuous in this
case and hence the same proof works. The other direction is false as the following counterexample
demonstrates.
Consider
x = (1, 1, 1, . . .) ∈ RN
and let xn ∈ RN be the sequence whose first n entries are 1 and the rest are zero. For each m ∈ N,
we have (
1 if m ≤ n
πm (xn ) = .
0 otherwise
Thus for all m ∈ N we have
lim πm (xn ) = 1 = πm (x).
n→∞
However, (xn )n∈N belongs to the set C from Exercise 4, while x 6∈ C. We showed in 4.(a) that C
is closed and thus this net cannot converge to a point outside of it. So (xn )n∈N does not converge
to x.
6. (a) Let x ∈ R and let U be a neighborhood of x. We must show U intersects Q and R \ Q. Recall
that there exists > 0 so that (x − , x + ) ⊂ U . By Exercise 6*.(c) on Homework 3, there
∈ Q with x − < z√< x + , and thus z ∈ U . We √
exists z √ also know there exists√ a w ∈ Q with
x − − 2 < w < x + − 2 and therefore √ x − < w + 2 < x
√ + . So w + 2 ∈ U and we
claim
√ w ∈ R \ Q. Indeed, otherwise w + 2 = q ∈ Q and hence 2 = q − w ∈ Q, contradicting
2 being irrational. Thus U also intersects R \ Q.
(b) Fix x ∈ R. If x ∈ Q, then V := ( 12 , 32 ) is a neigbhorhood of f (x) = 1. We claim there is no
neighborhood U of x satisfying f (U ) ⊂ V . Indeed, since R \ Q is dense we know there always
exists y ∈ U ∩ R \ Q, and thus f (y) = 0 6∈ V . If x ∈ R \ Q, then using V := (− 12 , 12 ), we can use
the density of Q to find y ∈ Q ∩ U for any neighborhood U of x and f (y) = 1 6∈ V . Thus in either
case f fails to be continuous at x ∈ R.
n 2
(c) Fix x ∈ Q, and suppose x = m for n ∈ Z and m ∈ N sharing no common factors. Thus V := (0, m )
1
is a neighborhood of g(x) = m . However, any neighborhood U of x contains some y ∈ R \ Q (by
the density of the irrationals), and thus g(y) = 0 6∈ V . Therefore g is not continuous at x.
(d) Fix x ∈ R \ Q. We will show g satisfies the − δ definition of continuity. Let > 0. Let M ∈ N
satisfy M ≥ 1 . Note that for m ∈ {1, . . . , M }, there are only finitely many n ∈ Z satisfying
n
m ∈ [x − 1, x + 1] (since this requires m(x − 1) ≤ n ≤ m(x + 1)). Thus the following is a finite
set: nn o
F := ∈ [x − 1, x + 1] | 1 ≤ m ≤ M, n ∈ Z .
m
Since F ⊂ Q, we know x 6∈ F and therefore
δ := min |x − y| > 0.
y∈F
We first claim δ < 1. Indeed, by Exercise 6*.(a) on Homework 3, there exists n ∈ Z with
n ≤ x < n + 1. Thus x − 1 < n and n + 1 ≤ x + 1, which means n, n + 1 ∈ F , and therefore
δ ≤ |x − n| = x − n < 1.
Now, we claim that if y ∈ R satisfies |x−y| < δ, then |g(x)−g(y)| < . Since g(x) = 0, if y ∈ R\Q
n
then |g(x) − g(y)| = |0 − 0| = 0 < . So now assume y ∈ Q, and say y = m for n ∈ Z and m ∈ N
with no common factors. If m > M , then we have
1 1 1
|g(x) − g(y)| = |0 − |= < ≤ ,
m m M
as needed. If m ≤ M , then since δ < 1 we have y ∈ [x − 1, x + 1] and hence y ∈ F . But then
|x − y| < δ ≤ |x − y| is a contradiction. Thus we cannot have m ≤ M and so in all cases we have
have show |g(x) − g(y)| < .