Homework 2 Solutions - Math 321,Spring 2015 (1) For each a ∈ [0, 1] consider fa ∈ B[0, 1] i.e. fa : [0, 1] → [0, 1] defined by ( 1 if t = a fa (t) = 0 if t = 6 a There are uncountably many such fa as [0, 1] is uncountable. Also if a 6= b then kfa − fb k∞ = 1 (if it is clear from context that we are using sup norm you may omit the ∞ symbol in the norm if you like.) Now consider all the balls B(fa , 12 ), a ∈ [0, 1] so if fb 6= fa then fb ∈ / B(fa , 21 ). Now if C ⊆ B[0, 1] is dense then for each fa there is a ga ∈ C such that kga −fa k∞ < 12 and if a 6= b then 1 >0 2 Hence for a 6= b, ga 6= gb so the dense subset C cannot be countable. kfb − ga k ≥ kfb − fa k − kfa − ga k ≥ 1 − (2) Claim If fn is uniformly cauchy in R (or any metric space) then it is uniformly convergent in R : Proof, for each x, {fn (x)}n≥1 is cauchy. So f (x) := limn→∞ fn (x) exists for every x. Now let > 0 we can find N, n, m > N implies |fm (x) − fn (x)| < for all x. Now with n.m > N we have that for all x, we have |fn (x) − f (x)| ≤ |fn (x) − fm (x)| + |fm (x) − f (x)| ≤ + |f (x) − fm (x)| Now let m goes to ∞ we have that if n > N then |fn (x) − f (x)| < 2 for all x.(We only take pointwise limit in m, we choose m large enough so that |f (x) − fm (x)| < . , this m may be different for different x. Hence we have uniform convergence. Now we will show that fn is uniformly cauchy on R. First since fn , fm are continuous, we can find δ1 , δ2 such that for any x ∈ R, |y − x| < δ1 =⇒ |fn (y) − fn (x)| < , |y − x| < δ2 =⇒ |fm (y) − fm (x)| < . Since Q is dense in R. Let δ = min δ1 , δ2 , choose xr ∈ Q (which may depend on x) such that |xr − x| < δ Also since fn is uniformly convergence on Q, there is N such that if n.m > N then |fn (x) − fm (x)| < 3 for all x ∈ Q. Combine these information, we have that for all n, m > N , |fn (x) − fm (x)| ≤ |fm (x) − fn (xr )| + |fn (xr ) − fm (xr )| + |fm (xr ) − fm (xr )| < 3 Note You may argue that the limiting f is continuous on R and use it to do this . This is true but not completely obvious. You may have to argue why it is (e.g. 1 2 problem 7.11 in textbook) Note Important: sup norm of f : X → R (or C) is defined as kf k∞ = sup |f (x)| x∈X (sometimes it is ∞ so is not defined). We should observe that this quantity depends only on f (and domain) , it does not depend on x so it is not necessary to write kf (x)k∞ . The statement like kfn (x) − fn (y)k∞ < or “kfn (x) − fm (x)k∞ < for all x” is confusing and sometimes is a misuse of notation. If you don’t like the notation you can also write in words e.g. you can either write kfn − fm k∞ < or |fn (x) − fn (x)| < for all x ∈ X. (3) 3a.1) (fn )n∈N is cauchy in B(X) means it is uniformly cauchy (since we are using sup norm) Hence as in the proof of the claim in Q2, f converges uniformly to f . Now we have to check that f ∈ B(X). To see this, fN is bounded means there is an M, |fN (x)| ≤ M for all x. Also there is and N such that n > N =⇒ kfN − f k < . Hence for all x, |f (x)| ≤ |f (x) − fN (x)| + |fN (x)| < M + so f ∈ B(X). 3a.2) Now we prove that limn→∞ kfn k∞ = kf k∞ This follows from Triangle inequality: |kfn k∞ − kf k∞ | ≤ kf − fn k∞ → 0(n → ∞.) And we have the result (or one can argue that kk∞ is continuous. 3a.3) To show that supn kfn k∞ < ∞.( The face that each fn ∈ B(X) only tells us that kfn k∞ exists as a finite number). First, there is N, ∀n ≥ N, kfn − f k∞ < 1 i.e. kfn k∞ ≤ kf k∞ + 1 for all n > N . Let M = max{kf k∞ + 1, kf1 k∞ , ..., , kfN k∞ }. Then kfn k∞ ≤ M for all n hence supn kfn k∞ ≤ M < ∞. PN 3b) Since TN = n=1 kgn k∞ is cauchy. Let > 0 there is an N such that for n, m > N , for all x ∈ X, m m X X gk (x) ≤ kgk k∞ < k=n+1 k=n+1 P This implies SN = N n=1 is a cauchy sequence in B(X) (sup norm). Using Q3a) we conclude that SN converges in B(X). The triangle inequalities is true for finite sum but by PQ 3a3) we can take P N → ∞ to get the required P triangle inequality. (4) 4a) | n≤N an sin nx| ≤ P |a | for all x ∈ R and n n≤N n≤N |an | < ∞. By Weierstrass M-test, we conclude that n≤N an sin nx converges uniformly on R. Same is true for P a n≤N n cos nx. 3 4b)If |x| > 0 then 1 + x2 > 1. Hence for x 6= 0 we can apply geometric series: ∞ ∞ X x2 1 x2 X x2 1 =1 = = 1 2 n 2 2 n 2 (1 + x ) 1 + x (1 + x ) 1 + x 1 − 2 1+x n=1 n=0 While at x = 0 the series obviously converges to 0. The limiting function is not continuous at x = 0 so the convergence cannot be uniform. Direct proof: limx→0+ (1 + x2 )N = 1 so we can find x ∈ (0, 1] such that (1 + x2 )N ≤ 2 so with this x, X X X x2 x2 x2 1 1 = = − ≥ 2 n 2 n 2 n 2 N (1 + x ) (1 + x ) (1 + x ) (1 + x ) 2 n≥1 n≥N +1 n≤N Hence the convergence can’t be uniform. (5) Write x ∈ [0, 1] in base 3 expansion x = 0.a1 a2 ... then we know that ai = 0, 2. Then it is not show that f (3n t) = a2n . Now pick any (x, y) ∈ [0, 1]2 whose base 2 expansion are x = 0.a0 a1 a2 ..., y = 0.b0 b1 b2 ... Let t ∈ [0, 1] has base 3 expansion 0.(2a0 )(2b0 )(2a1 )... then we can check that t ∈ ∆ and s(t) = (x, y). This shows [0, 1]2 ⊆ s(∆). On the other hand, ∞ ∞ X X −k−1 2k 2−k−1 = 1 2 f (3 t) ≤ 0≤ k=0 k=0 so 0 ≤ x(t) ≤ 1 and the same is true for y(t) hence s(∆) ⊆ [0, 1]2 .