2nd April 2005

Munkres §18
Ex. 18.1 (Morten Poulsen). Recall the ε-δ-definition of continuity: A function f : R → R is
said to be continuous if
∀ a ∈ R ∀ ε ∈ R+ ∃ δ ∈ R+ ∀x ∈ R : |x − a| < δ ⇒ |f (x) − f (a)| < ε.
Let T be the standard topology on R generated by the open intervals.
Theorem 1. For functions f : R → R the following are equivalent:
(i) ∀ a ∈ R ∀ ε ∈ R+ ∃ δ ∈ R+ ∀ x ∈ R : |x − a| < δ ⇒ |f (x) − f (a)| < ε.
(ii) ∀ a ∈ R ∀ ε ∈ R+ ∃ δ ∈ R+ : f ((a − δ, a + δ)) ⊂ (f (a) − ε, f (a) + ε).
(iii) ∀ U ∈ T : f −1 (U ) ∈ T .
Proof. “(i) ⇔ (ii)”: Clear.
“(ii) ⇒ (iii)”: Let U ∈ T . If a ∈ f −1 (U ) then f (a) ∈ U . Since U is open there exists
ε ∈ R+ such that (f (a) − ε, f (a) + ε) ⊂ U . By assumption there exists δa ∈ R+ such that
a + δa )) ⊂ (f (a) − ε, f (a) + ε), hence (a − δa , a + δa ) ⊂ f −1 (U ). It follows that
f ((a − δa , S
−1
f (U ) = a∈f −1 (U ) (a − δa , a + δa ), i.e. open.
“(iii) ⇒ (ii)”: Let a ∈ R. Given ε ∈ R+ then f −1 ((f (a) − ε, f (a) + ε)) is open and contains
a. Hence there exists δ ∈ R+ such that (a − δ, a + δ) ⊂ f −1 ((f (a) − ε, f (a) + ε)). It follows that
f ((a − δ, a + δ)) ⊂ (f (a) − ε, f (a) + ε). 
Ex. 18.2. Let f : R → {0} be the constant map. Then 2004 is a limit point of R but f (2004) = 0
is not a limit point of f (R) = {0}. (The question is if f (A0 ) ⊂ f (A)0 in general. The answer is
no: In the above example R0 = R so that f (R0 ) = f (R) = {0} but f (R)0 = ∅.)
Ex. 18.6 (Morten Poulsen).
Claim 2. The map f : R → R, defined by

x, x ∈ R − Q
f (x) =
0, x ∈ Q,
is continuous only at 0.
Proof. Since |f (x)| ≤ |x| for all x it follows that f is continuous at 0.
Let x0 ∈ R − {0}. Since limx→x0 ,x∈Q f (x) = 0 and limx→x0 ,x∈R−Q f (x) = x it follows that f
is not continuous at x0 . 
Ex. 18.7 (Morten Poulsen).
(a). The following lemma describes the continuous maps Rl → R.
Lemma 3. Let f : R → R. The following are equivalent:
(i) f : R → R is right continuous.
(ii) ∀ x ∈ R ∀ ε ∈ R+ ∃ δ ∈ R+ : f ([x, x + δ)) ⊂ (f (x) − ε, f (x) + ε).
(iii) For each x ∈ R and each interval (a, b) containing f (x) there exists an interval [c, d)
containing x such that f ([c, d)) ⊂ (a, b).
(iv) f : Rl → R is continuous.
Proof. ”(i) ⇔ (ii)”: By definition.
”(ii) ⇒ (iii)”: Let x ∈ R. Assume f (x) ∈ (a, b). Set ε = min{f (x) − a, b − f (x)} ∈ R+ . Then
(f (x)−ε, f (x)+ε) ⊂ (a, b). By (ii) there exists δ ∈ R+ such that f ([x, x+δ)) ⊂ (f (x)−ε, f (x)+ε),
hence [c, d) = [x, x + δ) does the trick.
”(iii) ⇒ (ii)”: Let x ∈ R and ε ∈ R+ . By (iii) there exists [c, d) such that x ∈ [c, d) and
f ([c, d)) ⊂ (f (x) − ε, f (x) + ε). Set δ = d − x ∈ R+ . Then [x, x + δ) = [x, d) ⊂ [c, d), hence
f ([x, x + δ)) ⊂ (x − ε, x + ε).
”(iii) ⇔ (iv)”: Clear. 
1
2

(b). The continuous maps from R to Rl are the constant maps, c.f. Ex. 25.1.
A map f : Rl → Rl is continuous if and only if for any x and ε > 0 there exists δ > 0 such
that f ([x, x + δ)) ⊂ [f (x), f (x) + ε). Hence f : Rl → Rl is continuous if and only if f : R → R is
right continuous and increasing.
Ex. 18.8. Let Y be an ordered set. Give Y × Y the product topology. Consider the set
∆− = {y1 × y2 | y1 > y2 }
of points below the diagonal. Let (y1 , y2 ) ∈ ∆− so that y1 > y2 . If y2 is the immediate predecessor
of y1 then
y1 × y2 ∈ [y1 , ∞) × (−∞, y2 ) = (y2 , ∞) × (−∞, y1 ) ⊂ ∆−
and if y1 > y > y2 for some y ∈ Y then
y1 × y2 ∈ (y, ∞) × (−∞, y) ⊂ ∆−
This shows that ∆− is open.
∆ f ×g
(a). Since the map (f, g) : X −→ X × X −−−→ Y × Y is continuous, the preimage
(f, g)−1 (∆− ) = {x ∈ X | f (x) > g(x)}
is open and the complement {x ∈ X | f (x) ≤ g(x)} is closed.
(b). The map (
f (x) f (x) ≤ g(x)
min{f, g}(x) =
g(x) f (x) ≥ g(x)
is continuous according to [1, Thm 18.3].
Q
Ex. 18.10. Let (fj : Xj → Yj )j∈J be an indexed family of continuous maps. Define fj :
Q Q Q Q
Xj → Yj to be the map that takes (xj ) ∈ Xj to (fj (xj )) ∈ Yj . The commutative
diagram
Q
fj Q
/ Yj
Q
Xj
πk πk
 
Xk / Yk
fk
Q Q Q Q
shows that πk ◦ fj = fk ◦ πk is continuous for all k ∈ J. Thus fj : Xj → Yj is continuous
[1, Thm 18.4, Thm 19.6].
Ex. 18.13. Let f, g : X → Y be two continuous maps between topological spaces where the
codomain, Y , is Hausdorff. The equalizer
Eq(f, g) = {x ∈ X | f (x) = g(x)} = (f, g)−1 (∆)
is then a closed subset of X for it is the preimage under the continuous map (f, g) : X → Y × Y
of the diagonal ∆ = {(y, y) ∈ Y × Y | y ∈ Y } which is closed since Y is Hausdorff [1, Ex 17.13].
(This is [1, Ex 31.5].)
It follows that if f and g agree on the subset A ⊂ X then they also agree on A for
A ⊂ Eq(f, g) =⇒ A ⊂ Eq(f, g)
In particular, if f and g agree on a dense subset of X, they are equal: Any continuous map into
a Hausdorff space is determined by its values on a dense subset.

References
[1] James R. Munkres, Topology. Second edition, Prentice-Hall Inc., Englewood Cliffs, N.J., 2000. MR 57 #4063