







Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
A proof that the space x of all real sequences of the form x = (x1, x2, x3, ..., xn, 0, 0, ...) with xi ∈ r and whose terms are zero from some point on, is not complete. The document also describes the completion of x as the space c0 of all real sequences (xi) such that xi → 0 as i → ∞. The proof that x is dense in c0 and c0 is complete, as well as a proof of a theorem about uniformly continuous functions and dense subsets.
Typology: Assignments
1 / 13
This page cannot be seen from the preview
Don't miss anything!








Analysis Solutions: Problem Set 3 Math 201A, Fall 2006
Problem 1. Let X be the space of all real sequences of the form
x = (x 1 , x 2 , x 3 ,... , xn, 0 , 0 ,.. .) , xi ∈ R
whose terms are zero from some point on. Define
‖x‖∞ = max i∈N
|xi|.
(a) Show that (X, ‖ · ‖∞) is a normed linear space.
(b) Show that X is not complete.
(c) Give a description of the completion of X as a space of sequences.
Solution.
λx = (λxi) , x + y = (xi + yi).
The 0-vector is the sequence (0, 0 , 0 ,.. .).
‖λx‖∞ = max i∈N |λxi| = |λ| max i∈N |xi| = |λ| ‖x‖∞;
and
‖x + y‖∞ = max i∈N
|xi + yi|
≤ max i∈N {|xi| + |yi|}
≤ max i∈N |xi| + max i∈N |yi|
≤ ‖x‖∞ + ‖y‖∞.
xn =
n
For m > n, we have
‖xm − xn‖∞ =
n + 1
which implies that the sequence is Cauchy.
so (xn) does not converge to y as n → ∞. Therefore, the sequence (xn) has no limit in X, and X is not complete.
xn = (y 1 , y 2 ,... , yn, 0 , 0.. .) ∈ X.
Since yi → 0 as i → ∞, we have
‖xn − y‖∞ = max i>n |yi| → 0 as n → ∞.
Thus, every point in c 0 is a limit of a sequence in X, so X is dense in c 0.
|xn,i − xm,i| ≤ ‖xn − xm‖∞,
so (xn,i)∞ n=1 is a Cauchy sequence in R. Since R is complete, there exists yi ∈ R such that xn,i → yi as n → ∞. Let y = (yi)∞ i=1. We will show that xn → y with respect to ‖ · ‖∞ and y ∈ c 0.
Problem 2. Suppose that (X, dX ) and (Y, dY ) are metric spaces and (Y, dY ) is complete. If D is a dense subset of X and f : D → Y is uniformly continuous on D, prove that there exists a unique continuous function F : X → Y such that F |D = f.
Solution.
dY (f (an), f (a′ n)) → 0 as n → ∞.
It follows that lim n→∞ f (an) = lim n→∞ f (a′ n),
so the value F (x) is independent of the choice of sequence converging to x.
d(aN , x) <
δ 3
, dY (f (aN ), F (x)) <
d(a′ N , x′) <
δ 3
, dY (f (a′ N ), F (x′)) <
Then
dX (aN , a′ N ) ≤ dX (aN , x) + dX (x, x′) + dX (x′, a′ N ).
Hence, if dX (x, x′) < δ/3, then dX (aN , a′ N ) < δ, and
dY (F (x), F (x′)) ≤ dY (F (x), f (aN )) + dY (f (aN ), f (a′ N )) +dY (f (a′ N ), F (x′)) < .
It follows that F is uniformly continuous on X
r =
i=k
ripi^ with k, ri ∈ Z and 0 ≤ ri ≤ p − 1 (1)
in which the ri are periodic functions of i.^1 For example,
1 1 − p
i=
pi^ ∈ Q.
It then follows that the partial sums of a series with non-periodic co- efficients cannot converge to any rational number. For example, the series (rn) defined by
rn =
∑^ n
i=
pi!^ = p + p^2 + p^6 +... + pn!
is Cauchy in Qp but does not converge to any r ∈ Q.
dQ×Q ((r, s), (r′, s′)) = d(r, r′) + d(s, s′).
We will temporarily use the notation +(r, s) = r + s for the addition function. (^1) To obtain this algorithm, write
m =
∑^ M
i=
mipi, n =
∑^ N
i=
nipi, r =
∑
i≥k
ripi
where 0 ≤ mi, ni, ri ≤ p − 1 and m 0 , n 0 6 = 0, multiply the series in the equation nr = mpk, carry multiples of powers of p to the succeeding terms so that all coefficients of pi^ are between 0 and p − 1, and equate coefficients of pi. One finds that ri is determined by the previous N coefficients {ri− 1 ,... , ri−N }. Since there are only finitely many such sequences, a sequence must eventually repeat, and then the coefficients ri will repeat.
d (+(r, s), +(r′, s′)) = |r + s − (r′^ + s′)|p = |r − r′^ + s − s′)|p
≤ max
|r − r′|p , |s − s′)|p
≤ |r − r′|p + |s − s′)|p ≤ dQ×Q ((r, s), (r′, s′)) ,
which proves that + : Q × Q → R is uniformly continuous.
r = 1 + 0p + 0p^2 +... , s = (p − 1) + (p − 1)p + (p − 1)p^2 +... ,
then r + s = 0, so s is the additive inverse of 1 in Qp, meaning that the sum of the series for s is −1.
Remark. Elements of Qp are called p-adic numbers, important in algebraic number theory. Multiplication of rational numbers also extends continuously from Q to Qp, so Qp is a complete field. Analysis on (Qp, |·|p) is an ultrametric analog of the more familiar analysis on the Euclidean real line (R, | · |).
is dense in X.
Solution.
δn =
0 if sn = tn, 1 if sn 6 = tn, n =
0 if tn = rn, 1 if tn 6 = rn, ηn =
0 if sn = rn, 1 if sn 6 = rn,
then ηn ≤ δn + n, so d(s, r) ≤ d(s, t) + d(t, r).
sk = (s 1 ,k, s 2 ,k,... , sn,k,.. .).
For any N ∈ N, there exists K ∈ N such that d(sj , sk) < 1 / 2 N^ for all j, k > K, and hence sn,j = sn,k for all n ≤ N. Thus, the terms in the sequences are eventually the same. We denote the eventual common value of (sn,k)∞ k=1 by sn, and define
s = (s 1 , s 2 ,... , sn,.. .).
Then for k > K we have
d(sk, s) ≤
n=N +
2 n
≤
It follows that sk → s as k → ∞, so X is complete.
s = (s 1 , s 2 , s 3 ,... , sN , sN +1, sN +2,.. .)
then the open ball B(s) contains all sequences of the form
t = (s 1 ,... , sN , tN +1, tN +2.. .) where tn ∈ { 0 , 1 } for n > N.
There are 2N^ initial sequences (s 1 ,... , sN ) ∈ { 0 , 1 }N^ , so X is covered by 2N^ balls of radius , and X is totally bounded.
UN = {s ∈ X | sN = 0} , VN = {s ∈ X | sN = 1}.
Then UN and VN are open. For example, suppose s ∈ UN. If d(s, t) < 1 / 2 N^ then tN = sN = 0, so t ∈ UN , meaning that B 1 / 2 N (s) ⊂ UN , so UN is open. (It follows that X = UN ∪ VN is the disjoint union of non-empty open sets so it is not connected.)
sk = (s 1 ,k, s 2 ,k,... , sn,k,.. .) ∈ X
by sn,k = sn if n 6 = k and sn,k 6 = sn if n = k. Thus, sk differs from s only in the kth term. Then sk 6 = s and d(sk, s) = 1/ 2 k. Hence sk → s as k → ∞, and X is perfect.
sn =
0 if an = 0, 1 if an = 2.
tn 6 = sn for n ≥ N + 1. Then d(s, t) = 1/ 2 N^ < , so t ∈ U. On the other hand,
σN^ (s) = (sN +1, sN +2,.. .), σN^ (t) = (tN +1, tN +2,.. .)
differ in every term, so
d
σN^ (s), σN^ (s
= 1 > δ.
s = (0, 1 , 0 , 0 , 0 , 1 , 1 , 0 , 1 , 1 , 0 , 0 , 0 , 0 , 0 , 1 , 0 , 1 , 0 , 0 , 1 , 1 , 1 , 0 , 0 ,.. .).
Then for every t ∈ X and every N ∈ N there exists an n ∈ N such that σn(s) and t have the same first N terms. Hence, d(σn(s), t) ≤ 1 / 2 N^ , which proves that the orbit of s under σ is dense in X.
Remark. The metric space X is an example of a symbol space. Such symbol spaces arise in the study of chaotic dynamical systems, and the representa- tion of chaotic dynamical systems by shift maps on symbol spaces is called symbolic dynamics. In this context, the property in (e) is called ‘sensitive dependence on initial conditions.’ For example, consider the logistic map Fμ : Λ ⊂ [0, 1] → [0, 1] defined by Fμ(x) = μx(1 − x). For μ > 4 the map Fμ has an invariant Cantor set Λ of points that remain in [0, 1] under all iterates F (^) μn. One can prove that there is a homeomorphism φ : Λ → X such that Fμ = φ−^1 ◦ σ ◦ φ is topologically conjugate to the shift map σ.