1. Quizz 10
Definition 1.1. Let S be a nonempty set and f : S → R be a function. Suppose that f is bounded above1 We say that f attains its maximum on S if there exists s0 ∈ S such that f (s0) = sup f (S). In this case, f (s0) is called a maximum of f. Similarly, if f is bounded below, f attains its minimum on S if there exists s1∈ S such that f (s1) = inf f (S).
(1) Let S be a nonempty subset of R. Suppose that S is bounded above2. Show that sup S is an adherent point of S. Furthermore, if sup S 6∈ S, then sup S is an accu- mulation point of S.
Proof. Let M denotes sup S. When M ∈ S, M ∈ S. Now, let us consider the case when M 6∈ S. We will give two different proofs.
Proof I For > 0, M − is no longer an upper bound for S. There exists y ∈ S such that s > M − . Since M 6∈ S, y 6= M. We see that 0 < M − y < . This implies that y ∈ B0(M, ) ∩ S. We find that B0(M, ) ∩ S 6= ∅ for any > 0. This shows that M is an accumulation point of S and thus M ∈ S
Proof II For any n ≥ 1, we can find yn ∈ S such that yn > M − 1/n. Since M 6∈ S, we obtain a sequence (yn) in S such that
M − 1
n < yn< M for any n ≥ 1.
Since lim
n→∞
M − 1
n
= lim
n→∞M = M, by Sandwich principle, lim
n→∞yn = M. This shows that M is a limit point of S and hence M ∈ S.
Remark. When S is nonempty and bounded below, the above statement holds when we replace sup S by inf S.
(2) Let f : K → R be a continuous function on a sequentially compact metric space (K, d). Show that f attains its maximum and minimum on K.
Proof. Since K is sequentially compact and f : K → R is continuous, f (K) is sequentially compact in R. By Bolzano-Weierstrass Theorem, f (K) is closed and bounded in R. Since f (K) is bounded, sup f (K) and inf f (K) exists. By the previous exercise, sup f (K) and inf f (K) are both adherent to f (K). Since f (K) is closed, it contains all of its adherent points. Thus sup f (K) and inf f (K) both belong to f (K).
Remark. Choose sequences (xn) and (yn) in K such that limn→∞f (xn) = sup f (K) and limn→∞f (yn) = inf f (K). We can do this because sup f (K) and inf f (K) are adherent points of f (K).
1A function f : S → R is bounded above (below) if its range f(S) is bounded above (below).
2A nonempty subset S of R is bounded above if there exists U ∈ R such that x ≤ U for any x ∈ S. Such an U is called an upper bound for S. If S is bounded above, the supremum sup S of S is the smallest U so that U is an upper bound for S.
1
2
(3) Let f : K → R be a continuous function on a sequentially compact metric space.
Prove that there exist no sequences (xn) and (yn) in K satisfying the following properties.
(a) d(xn, yn) < 1/n for any n ≥ 1.
(b) There exists > 0 such that |f (xn) − f (yn)| ≥ for any n ≥ 1.
Proof. Since K is sequentially compact, we can choose a subsequence (xnj) of (xn) such that (xnj) is convergent to some x ∈ K. By sequentially compactness of K again, (ynj) has a subsequence (ynjk) convergent to some y in K. By properties, we have that for any k ≥ 1,
d(xnjk, ynjk) < 1 njk
≤ 1 jk ≤ 1
k and
|f (xn
jk) − f (ynjk)| ≥ .
By triangle inequality,
d(x, y) ≤ d(xnjk, x) + d(xnjk, ynjk) + d(ynjk, y)
for any k ≥ 1. Since (xnj) is convergent to x, its subsequence (xnjk) is also convergent to x. For any > 0, we can choose N0 ∈ N and N00 = [3/] + 1 and N000 ∈ N such that
d(xnjk, x) <
3 for k ≥ N0, d(xnjk, ynjk) <
3 for k ≥ N00, d(ynjk, y) <
3 for k ≥ N000.
Take N = max{N0, N00, N000}. For k ≥ N, d(x, y) < . We see that d(x, y) < for any > 0. Therefore d(x, y) = 0 and thus x = y.
(4) Let k : [0, 1] × [0, 1] → R be a continuous function.
(a) Prove that for any > 0 there exists δ> 0 such that
|k(x1, y1) − k(x2, y2)| <
whenever k(x1, y1) − (x2, y2)k < δ with (xi, yi) ∈ [0, 1] × [0, 1] for i = 1, 2.
Proof. Suppose not. We can find > 0 and choose a sequence (an) and a sequence (bn) in [0, 1] × [0, 1] such that kan− bnk < 1/n for any n ≥ 1 while
|k(an) − k(bn)| ≥ . Since [0, 1] × [0, 1] is closed and bounded, by Bolzano- Weierstrass Theorem, [0, 1] × [0, 1] is sequentially compact. Using Exercise 3, since k is continuous and [0, 1] × [0, 1] is compact, sequences (an) and (bn) do not exist.
Remark. Suppose not. Use (3) to prove the result by contradiction.
(b) Let f : [0, 1] → R be continuous. Define g : [0, 1] → R by g(x) =
Z 1 0
k(x, y)f (y)dy, x ∈ [0, 1].
3
Prove that g is also continuous on [0, 1]. Hint: consider g(x1) − g(x2) =
Z 1 0
(k(x1, y) − k(x2, y))f (y)dy and use (a).
Proof. When f = 0, the statement is obvious. Assume that f 6= 0. By (a), for any > 0, choose δ > 0 such that
|k(x1, y1) − k(x2, y2)| < kf k∞
whenever k(x1, y1) − (x2, y2)k < δ with (xi, yi) ∈ [0, 1] × [0, 1] for i = 1, 2. For this δ, if |x1− x2| < δ, then k(x1, y) − (x2, y)k < for any y ∈ [0, 1]. This implies that
|k(x1, y) − k(x2, y)| < kf k∞,
whenever |x1− x2| < δ for any y ∈ [0, 1]. Thus if |x1− x2| < δ,
|g(x1) − g(x2)| ≤ Z 1
0
|(k(x1, y) − k(x2, y))|f (y)|dy ≤ kf k∞·
kf k∞ = .
This implies that g is (uniformly) continuous.