COMPACTNESS

If C ⊂ Rn is closed and bounded, then by B-W it is sequentially compact:

any sequence of points in C has a subsequence converging to a point in C
Conversely, any sequentially compact C ⊂ Rn is closed and bounded.
This is true very generally (metric spaces, for instance.) But closed bounded
sets are sequentially compact only in finite dimensions.
If F : C → Rn is continuous (where C is compact: closed and bounded),
then f (C) is compact in Rn . As a corollary, continuous functions on compact
sets are bounded, and achieve their max and min values.
Another corollary is that a continuous map on a compact set K ⊂ Rn
is closed: maps closed subsets of K to closed subsets of Rn . In particu-
lar, a bijective continuous map f : K → f (K) ⊂ Rm is automatically a
homeomorphism.
The same is not true for preimages: we say a continuous map f : Rn →
Rm is proper if f −1 (C) is compact, whenever C ⊂ Rm is compact. This is
equivalent to: f (xn ) → ∞, for any sequence xn → ∞.
A continuous map F : C → Rm defined in a compact subset C ⊂ Rn
is uniformly continuous on C. This is easily proved by contradiction, using
sequential compactness.
A decreasing sequence Kn ⊃ Kn+1 of non-empty compact sets (in Rn )
has non-empty intersection (Cantor’s theorem). This follows from sequential
compactness: take xn ∈ Kn and consider a convergent subsequence.
Application. A monotone sequence fn : K → R of continuous functions
on a compact set (fn ≤ fn+1 on K, say) which converges pointwise to
f : K → R, converges uniformly to f , provided f is continuous. (Dini’s
theorem.)
For the proof, given  > 0, consider Kn = {x ∈ K; |fn (x) − f (x)| ≥
}. Since the sequence (fn ) is monotone, this is a decreasing sequence of
compact sets (Kn ⊃ Kn+1 ), yet their intersection is empty (by pointwise
convergence). Thus one of the Kn is empty, which is uniform convergence.
Heine-Borel theorem. A set A ⊂ Rn is compact if, and only if, any open
cover {Uλ }λ∈Λ of C admits a finite subcovering.
Proof. Let A ⊂ Rn satisfy this condition. Considering the cover
{B1 (x)}x∈A of A (by open balls of radius one) shows A is bounded. If A
is not closed, let (xn ) be a sequence in A, converging to a point x0 not in

1
A. Then the open sets B̄1/n (x0 )c (complements of the closed balls at x0
with radius 1/n) cover A. Extracting a finite subcover, we find N so that
||x − x0 || > 1/N for all x ∈ A, contradiction.
Conversely, assume A is closed and bounded, and let (Uλ )λ∈Λ be an open
covering of A. By Lindelöf’s theorem (see below) we find a countable open
subcovering (Uλi )i≥1 of A. Let Kn = A ∩ (Uλ1 ∪ . . . Uλn )c . Then (Kn ) is a
decreasing sequence of compact sets, with empty intersection. By Cantor’s
theorem, some KN must be empty. Then (Uλn )N n=1 is a finite subcovering
of A.
Lindelöf ’s theorem. Let A ⊂ Rn . Any open covering of A admits a
countable subcovering.
Proof. Let (Uλ )λ∈Λ be an open covering of A. Let E = (xi )i≥1 a count-
able dense subset of A. Consider the countable set B of open balls Bri (xi )
with center at a point xi ∈ E, rational radius ri > 0, and contained in some
Uλ . We claim the sets (Bri (xi ))i≥1 in the collection B cover A.
Let x ∈ A. Then x ∈ Uλ for some λ, and since Uλ is open we may find
r > 0 so that B2r (x) ⊂ Uλ . Since E is dense in A, we find xi ∈ E and a
rational ri > 0 so that ri < r and ||x − xi || < ri , and it is easy to see that
Bri (xi ) ⊂ B2r (x) ⊂ Uλ , so Bri (xi ) occurs in B and x ∈ Bri (xi ).
Now take an enumeration (Bi )i∈N of B and for each i ≥ 1 choose λi with
Bi ⊂ Uλi to conclude.
Application. f : A → Rn is locally Lipschitz if for each x ∈ A we may find
an open ball Brx (x), so that f is Lipschitz in this ball (with some Lipschitz
constant Mx ):

∀y, z ∈ Brx (x)||f (y) − f (z)|| ≤ Mx ||y − z||.

Exercise. If f : I → R is differentiable on the interval I ⊂ R, with f 0

continuous on I, then f is locally Lipshitz on I. (This follows from the
Mean Value Theorem.)
Proposition. If K ⊂ Rn is compact and f : K → Rm is locally Lipschitz
on K, then f is Lipschitz on K.
The proof of this fact requires the notion of Lebesgue number of a pair
A, U, where U = {Uλ }λ∈Λ is an open covering of a subset A ⊂ Rn : a positive
number δ > 0 so that any x, y ∈ A with ||x − y|| < δ are contained in a
single set Uλ of the cover. Not every open cover of a set has a Lebesgue
number: for instance the cover U1 = (−∞, 0), U2 = (0, ∞) of A = R \ {0}
doesn’t have one (prove it.)

2
 Prop. Every open cover of a compact set A ⊂ Rn has a Lebesgue number.
(Otherwise we may find sequences (xk ), (yk ) with ||xk − yk || ≤ 1/k, but no
single Uλ contains both xk and yk (for any k); taking subsequential limits
leads to a contradiction.)
Given this fact, the proof that locally Lipschitz implies Lipschitz on each
compact set follows easily.

EXTENSION of continuous functions.

Exercise: Let A, B ⊂ Rn be closed and disjoint. Define f : Rn → [0, 1]
via:
d(x, A)
f (x) = .
d(x, A) + d(x, B)
Then f is continuous, takes the value 0 exactly on A, and the value 1 exactly
on B. (In Topology a function with these properties would be called a strict
Urysohn function for the pair A, B.)

Tietze’s extension theorem for Rn : Let f : A → R be continuous and

bounded, where A ⊂ Rn is closed: |f (x)| ≤ M in A. Then there exists a
continuous extension f¯ : Rn → R of f , with |f¯| ≤ M on Rn .
Proof Consider the sets:

B = {x ∈ A; f (x) ≥ M/3}; C = {x ∈ A; f (x) ≤ −M/3},

both closed in Rn and disjoint. Assume first they are both non-empty.
Let g1 : Rn → [−M/3, M/3] be equal to M/3 in B, and to −M/3 in C
(Urysohn’s lemma.) Then |f (x) − g1 (x)| ≤ 2M/3 on A. (If either B or C is
empty, we may find a constant g1 so that this inequality holds in A.)
Let f1 = f − g1 . Repeating the argument with f1 replacing f and
M1 = 2M/3 replacing M , we find g2 : Rn → [−M1 /3, M1 /3] continuous,
satisfying on A:

|f (x) − (g1 (x) + g2 (x))| ≤ (2/3)2 M, |g2 (x)| ≤ (1/3)M1 = (1/3)(2/3)M.

Inductively we find a sequence gn (x) ∈ Cb (Rn ) satisfying on A:

n
X
|f (x) − gi (x)| ≤ (2/3)n M, |gn (x)| ≤ (1/3)(2/3)n−1 M ( in Rn ).
i=1

Thus f¯(x) = ∞
P
i=1 gi (x) converges to f pointwise in A. By the Weierstrass
M-test and the second inequality, convergence is uniform in A, and we have

3
in Rn : X
|f¯(x)| ≤ (M/3) (2/3)i−1 = M.
i≥1

Remark: It is easy to see the theorem fails if we don’t assume A is closed,

even if f is bounded on A (example?) On the other hand, the result does
extend to the case the function is unbounded (if A is closed).
Proposition. Let A ⊂ Rn be closed, f : A → R continuous. Then f
admits a continuous extension f¯ : Rn → R.
Proof. Since R is homeomorphic to (−1, 1), we may assume f takes
values in (−1, 1) (for example, consider f ◦ φ−1 , where φ(x) = tan( 2x
π ) is a
homeomorphism from (−1, 1) to R.)
We know there is continuous extension g : Rn → [−1, 1], g = f on
A. Consider the closed set: B = g −1 ({−1, 1}), and let h : Rn → [0, 1]
(continuous) be a Urysohn function: h ≡ 1 on A, h ≡ 0 on B. Consider the
continuous function f¯ : Rn → [−1, 1]:

f¯(x) = h(x)g(x).

Note f¯ = f on A. On the other hand, f¯ = ±1 at a point only if h = 1

and g = ±1 at that point; this never happens, since h = 0 at the points
where g = ±1. Hence in fact f¯ takes values in the open interval (−1, 1), and
extends f continuously to all of Rn .
Corollary. If A ⊂ Rn is any non-empty set and f : A → R is uniformly
continuous, then f admits a continuous extension to all of Rn . (Exercise.)

CONVEX FUNCTIONS
Let K ⊂ Rn be a convex set. A function f : K → R is convex if:

∀x, y ∈ K, t ∈ [0, 1] : f (tx + (1 − t)y) ≤ tf (x) + (1 − t)f (y).

(Strictly convex if the inequality is strict for t ∈ (0, 1).)

The epigraph of f is the subset of Rn+1 :

Ef = {(x, y) ∈ K × R; y ≥ f (x)}.

Exercise: (i) For any subset K ⊂ Rn and any function f : K → R, Ef is

closed in Rn+1 if and only if f is lower semicontinuous on K. This means:

(∀x0 ∈ K)f (x0 ) ≤ lim inf f (xn ) if xn → x0 , xn ∈ K.

n

(ii) f is convex on K if and only if Ef is a convex subset of Rn+1 .

4
 Given c ∈ R, the sublevel set Kc of f defined by c is:

Kc = {x ∈ K; f (x) ≤ c}.

Exercise. (i) In general (K not nec. convex) Kc is closed (or empty) for
each c ∈ R if, and only if, f is lower semicontinuous.
(ii) If K is convex and f is convex on K, each nonempty sub level set of
f is convex.
The converse is not true: for instance, if f : R → R is an increasing
function, each sublevel set is either empty, a half-infinite interval or the
whole line (hence convex); but f itself need not be convex (say, f (x) = x3 .)
Exercise. Three-slopes lemma for convex functions on the real line.
mf (a, b) = f (b)−f
b−a
(a)
.

a < b < c ⇒ mf (a, b) ≤ mf (a, c) ≤ mf (b, c).

Exercise. Convex functions defined on an open interval are continuous

(but this is false on other types of intervals.)
Hint: Let x0 ∈ I (open interval) and choose δ > 0 so that [x0 −δ, x0 +δ] ⊂
I. Let M = max{f (x0 − δ), f (x0 + δ)}. Assume first x0 < x < x0 + δ and
write:
1 t x − x0
x = (1 − t)x0 + tx, x0 = x+ (x0 − δ), where t = .
1+t 1+t δ
Then, by convexity:
1 t
f (x) ≤ (1 − t)f (x0 ) + tf (x0 + δ), f (x0 ) ≤ f (x) + f (x0 − δ).
1+t 1+t
These inequalities lead respectively to:
x − x0 x − x0
f (x) − f (x0 ) ≤ (M − f (x0 )), (f (x0 ) − M ) ≤ f (x) − f (x0 ),
δ δ
and we introduce absolute values to combine them into one estimate:
|x − x0 |
|f (x) − f (x0 )| ≤ |M − f (x0 )|.
δ
This clearly implies continuity at x0 . The case x0 − δ < x < x0 is analogous.
Exercise. If K ⊂ Rn is convex open, and f : K → R is convex, then f is
continuous on K.

5
 Hint: Given x0 ∈ K, choose δ > 0 so that the closed n-cube Q = Qδ (x0 )
of side length 2δ and center x0 is contained in K (using the fact K is open).
Let M be the maximum value of f on the finite set V of vertices of Q. Since
Q is the convex hull of V , the sublevel set KM is convex and V ⊂ KM , it
follows that Q ⊂ KM , or f (x) ≤ M for x ∈ Q.
The closed ball B = {x; ||x − x0 || ≤ δ} is contained in Q. For u on the
unit sphere, consider the closed line segment in B:

σu = {x ∈ B; x = x0 + tu; −δ ≤ t ≤ δ}.

Applying the one-dimensional estimate to this segment we find, for x ∈ σu :

||x − x0 ||
|f (x) − f (x0 )| ≤ |M − f (x0 )|.
δ
Therefore this holds for all x ∈ B, implying continuity at x0 .

(Reference: [Fleming], p.110.)

Convexity and differentiability in one dimension.

We consider f : I → R convex, where I ⊂ R is an open interval.
For h > 0, denote by mf (x, x + h) = f (x+h)−f
h
(x)
the slope of the (ori-
ented) secant to the graph of f from x to x + h. The three-slopes lemma
implies this is monotone increasing in h; thus its limit as h → 0+ exists, the
right-derivative of f at x:

f+0 (x) = lim mf (x, x + h) = inf mf (x, x + h).

h→0+ h>0

Similarly, the left-derivative exists for each x ∈ I:

f−0 (x) = lim mf (x − k, x) = sup mf (x − k, x).

k→0+ k>0

And since mf (x − k, x) ≤ mf (x, x + h) for each h, k > 0, it follows that

f−0 (x) ≤ f+0 (x), for each x ∈ I. Also, if x < y are points in I and h > 0, k > 0
are chosen so that x + h = y − k, we have:

f−0 (x) ≤ f+0 (x) ≤ mf (x, x + h) ≤ mf (y − k, y) ≤ f−0 (y) ≤ f+0 (y).

6
We see that f−0 , f+0 are both monotone increasing (nondecreasing) in I, with
f+0 (x) ≤ f−0 (y) if x < y.
And now it is easy to see that, if a ∈ I is a point of continuity of f−0 ,
then f−0 (a) = f+0 (a):

f−0 (a) ≤ f+0 (a) ≤ lim f−0 (y) = f−0 (a).

y→a+

Thus f is differentiable at a. Being monotone, f−0 is continuous at all but a

countable set of points in I. We conclude:

f convex in I ⊂ R ⇒ f differentiable in the complement of a countable set D ⊂ I.