• 沒有找到結果。

Prove that (N, dN) is complete if and only if N is a closed subset of M

N/A
N/A
Protected

Academic year: 2022

Share "Prove that (N, dN) is complete if and only if N is a closed subset of M"

Copied!
2
0
0

加載中.... (立即查看全文)

全文

(1)

1. hw 7

(1) Assuming that the Bolzano-Weierstrass Theorem is true for R1, prove that the Bolzano- Weierstrass Theorem is true for Rp for any p ∈ N by mathematical Induction. (Hint: Let (an) be a bounded sequence in Rp+1. Write an = (xn, yn) for all n ≥ 1 where (xn) is a sequence in Rp and (yn) is a sequence in R. Show that (xn) is bounded in Rp and (yn) is bounded in R.)

(2) Let (M, d) be a complete metric space and N be a subset of M. Let (N, dN) be the metric subspace of (M, d) associated with N. Prove that (N, dN) is complete if and only if N is a closed subset of M.

(3) In Calculus, we have learned the following important properties for convergent sequences:

Theorem 1.1. Let (an) and (bn) be sequence of real numbers and k be a real number.

Suppose that

n→∞lim an= a, lim

n→∞bn = b.

Then (a) lim

n→∞(an+ bn) = a + b;

(b) lim

n→∞(kan) = ka;

(c) lim

n→∞anbn= ab (d) lim

n→∞

an bn

= a

b if b ≥ 0.

In class, we also proved

Theorem 1.2. Let (an) be a sequence in Rp with an = (a1n, · · · , apn) for n ≥ 1. Then (an) is convergent in Rp if and only if (a1n), · · · , (apn) are convergent in R. In this case,

n→∞lim an=

n→∞lim a1n, · · · , lim

n→∞apn .

Let (xn) and (yn) be sequences in Rpsuch that limn→∞xn= x andlimn→∞yn= y. Use Theorem 1.1 and Theorem 1.2 to prove the following statements:

(a) lim

n→∞(xn+ yn) = x + y.

(b) lim

n→∞kxn= kx.

(c) lim

n→∞anxn = ax.

(d) lim

n→∞

xn bn

= x

b if b ≥ 0.

(e) lim

n→∞hxn, yni = hx, yi.

(4) Do project in Bartle’s Book. 8β (a), (b), (c) on page 61.

(5) Let 1 ≤ p < ∞ and lp(N) be the space of sequences a : N → R of real numbers such that P

n=1|a(n)|p< ∞. Define a function k · k : lp(N) → R by kak =

X

n=1

|a(n)|p

!1/p

.

(a) Prove that lp(N) is a vector space over R. (Prove that lp(N) is a vector subspace of the space of real valued functions F (N, R).)

(b) Prove that k · k is a norm on lp(N). This norm is denoted by k ·klp(N). (You need exercise (4)).

(c) Let us denote the norm space (lp(N), k · klp(N)) simply by lp(N). Prove that lp(N) is a Banach space over R.

(d) Let α be a real number and a : N → R be the function a(n) = 1/nα. Determine all α so that a ∈ lp(N).

(e) Prove or disprove that lp(N) ⊂ lq(N) for p > q.

(f) Prove that lp(N) ⊂ l(N) for all 1 ≤ p < ∞.

1

(2)

2

(6) Let a and b be two vectors in l2(N).

(a) Show that P

n=1a(n)b(n) is absolutely convergent in R. (Hint: use Cauchy-Schwarz inequality; see (4)).

(b) Define

ha, bi =

X

n=1

a(n)b(n).

Prove that h·, ·i defines an inner product on l2(N) so that ha, ai = kakl2(N). By Exercise (5) and this exercise, l2(N) is a Hilbert space over R.

(c) Two vectors v, w in an inner product space V is orthogonal if hv, wi = 0. For each i ∈ N, define ei: N → R by (ei)(n) = δin for all n ∈ N. Here δij is the number defined by

δij =

(1 if i = j 0 otherwise.

Prove that {ei : i ∈ N} is an orthonormal family of vectors in l2(N) (in other words, elements of {ei} are unit vectors and are orthogonal to each other) and compute the distance kei− ejkl2(N)for any i 6= j.

(d) Let S= {a ∈ l2(N) : kak = 1}. Prove that S is closed and bounded.

(e) Show that (ei) is a sequence in S and does not have any convergent subsequence in S. (This example tells us that in a metric space, a closed and bounded subset need not be sequentially compact).

(f) Let fi= ei−2ei+1for all i ≥ 1. Determine the orthogonal complement V of {fi : i ≥ 1}, i.e.

V = {a ∈ l2(N) : ha, fii = 0, for all i ≥ 1}.

(g) Let gi = ei − 3ei+1 for all i ≥ 1. Determine the orthogonal complement of W of {gi: i ≥ 1}, i.e.

W = {a ∈ l2(N) : ha, gii = 0, for all i ≥ 1}.

(h) Let hi = ei− 5ei+1+ 6ei+2 for all i ≥ 1. Determine the orthogonal complement Z of {hi: i ≥ 1}, i.e.

Z = {a ∈ l2(N) : ha, hii = 0, for all i ≥ 1}.

Prove or disprove that Z is the direct sum of V and W. In other words, is it true that V ∩ W = {0} and Z = V + W ?

(i) Let a be a vector in l2(N). For each n ≥ 1, we define sn= a(1)e1+a(2)e2+· · ·+a(n)en. Then (sn) is a sequence of vectors in l2(N). Show that (sn) is convergent to a in l2(N).

In this case, we denote

(1.1) a =

X

n=1

a(n)en.

The expression 1.1 for a is called the Fourier expansion of a with respect to the or- thonormal family {ei: i ∈ N}.

(j) Let (ak) be a sequence of vectors in l2(N). By (6i), we can write ak =

X

n=1

ak(n)en. Prove or disprove that lim

k→∞ak = a in l2(N) if and only if lim

k→∞ak(n) = a(n) for any n ≥ 1. (Compare to (3)).

(7) Let R be the subset of F (N, R) consisting of sequences a : N → R such that there exists N ∈ N so that a(n) = 0 for all n ≥ N.

(a) Prove that R is a vector subspace of lp(N) for all 1 ≤ p ≤ ∞.

(b) Prove that R is not a closed subset (subspace) of lp(N) for all 1 ≤ p ≤ ∞. (This implies that (R, k · kp) is not a Banach space, where kakp = kaklp(N) for a ∈ R by exercise (2).)

參考文獻

相關文件

The tensor product of M and N is the universal object in the category of bilinear maps on M × N.

Remark 2: Indicate the properties of limit given below when you use it in the test.. Show that (a n ) is convergent and evaluate

[r]

In fact, the statement is true if p &lt;

[r]

 (3) In Calculus, we have learned the following important properties for convergent

(3%) (c) Given an example shows that (a) may be false if E has a zero divisors. Find the invariant factors of A and φ and their minimal polynomial. Apply

(6) Prove that the following sequences have a convergent subsequence in the given spaces3. We use the Bolzano-Weierstrass Theorem to prove