Open and Closed Sets, Lecture Notes - Mathematics, Study notes of Calculus

Open and Closed Sets, Analogy with R, Examples, Prepositions, Compactness

Typology: Study notes

2010/2011

Uploaded on 10/08/2011

deville
deville ๐Ÿ‡บ๐Ÿ‡ธ

4.7

(23)

390 documents

1 / 5

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
Open and Closed Sets
Adrian Down 16779577
August 2, 2005
1 Analogy with R
Properties that make analysis on (a, b) very different than analysis on [a, b]:
i) Given xโˆˆ(a, b), โˆƒr > 0 such that (xโˆ’r, x +r)โІ(a, b). [a, b] does not
have this property. Take x=aโ‡’no r > 0 works
ii) Given (sn)โˆˆ[a, b] and limnโ†’โˆž sn=s, then sโˆˆ[a, b]. (a, b) does not
have this property. Consider sn=1
nโˆˆ(0,1). Then limnโ†’โˆž sn= 0 /โˆˆ(0,1).
Goal is to generalize these ideas to metric spaces.
2 Open sets
Definition: Given a metric space (S, d), a subset Eis open if
โˆ€xโˆˆE, โˆƒr > 0 such that Br(x)โІE(1)
Br(x) = {yโˆˆS|d(x, y)< r}(2)
In general, given any set E, a point xโˆˆEsatisfying (1) is called an interior
point of Eand E0={xโˆˆE|xis an interior point }.E0is called the interior
of E.
A set is open if and only if every point of Eis interior to E, or Eis open
if and only if E=E0. Can also be shown that E0is always open, for any set
E. Would have to consider a point in the interior, then find some positive
radius so that the open ball is contained in the interior. Does not follow
immediately from the definition.
1
pf3
pf4
pf5

Partial preview of the text

Download Open and Closed Sets, Lecture Notes - Mathematics and more Study notes Calculus in PDF only on Docsity!

Open and Closed Sets

Adrian Down 16779577

August 2, 2005

1 Analogy with R

Properties that make analysis on (a, b) very different than analysis on [a, b]: i) Given x โˆˆ (a, b), โˆƒr > 0 such that (x โˆ’ r, x + r) โІ (a, b). [a, b] does not have this property. Take x = a โ‡’ no r > 0 works ii) Given (sn) โˆˆ [a, b] and limnโ†’โˆž sn = s, then s โˆˆ [a, b]. (a, b) does not have this property. Consider sn = (^1) n โˆˆ (0, 1). Then limnโ†’โˆž sn = 0 โˆˆ/ (0, 1). Goal is to generalize these ideas to metric spaces.

2 Open sets

Definition: Given a metric space (S, d), a subset E is open if

โˆ€x โˆˆ E, โˆƒr > 0 such that Br(x) โІ E (1) Br(x) = {y โˆˆ S|d(x, y) < r} (2)

In general, given any set E, a point x โˆˆ E satisfying (1) is called an interior point of E and E^0 = {x โˆˆ E|x is an interior point }. E^0 is called the interior of E. A set is open if and only if every point of E is interior to E, or E is open if and only if E = E^0. Can also be shown that E^0 is always open, for any set E. Would have to consider a point in the interior, then find some positive radius so that the open ball is contained in the interior. Does not follow immediately from the definition.

3 Examples

  1. (a, b) is open in R. In general, any open ball Br(x) is open in (S, d).
  2. โˆ… is always open for any S.
  3. If d is the discrete metric, so d(x, y) = 1, โˆ€x 6 = y, then any E โІ S is open.

B 12 (x) = {x} โІ E, โˆ€x โˆˆ E

  1. [a, b] is not open in R. (1) fails at x = a. Given any r > 0, Br(a) = (aโˆ’r, a+r) which is not a subset of [a, b]. You can still compute the interior, and get that [a, b]^0 = (a, b).

4 Propositions

Proposition: if (Uฮฑ)ฮฑโˆˆA is any collection of open sets, then

U =

ฮฑโˆˆA

Uฮฑ

is open. Means you donโ€™t need a countable or finite union. Proof: let x โˆˆ U. Then x โˆˆ Uฮฑ for some

ฮฑ โˆˆ A โ‡’ โˆƒr > 0 such that Br(x) โІ Uฮฑ โІ U

So U is open. Proposition: if U 1 ,... , Un are open sets, then

U =

โ‹‚^ โˆž

i=

Ui

is open Proof: let x โˆˆ U. So x โˆˆ Ui, โˆ€i โ‡’ โˆƒri > 0 such that Bri (x) โІ Ui. Let r = min{r 1 ,... , rn} > 0 and

Br(x) โІ Bri (x)โˆ€i โІ Ui

So Br(x) โІ U The finiteness of the set of U โ€ฒs is used to get that the minimum of the radius is greater than 0.

d(xn, x) < (^1) n , โˆ€n โ‡’ (xn) converges to x, but x /โˆˆ E. So E is not closed, a contradiction. Assume S/E is open. Assume E is not closed โ‡’ โˆƒ(sn), sn โˆˆ E such that lim sn = s /โˆˆ E. s โˆˆ S/E โ‡’ โˆƒr > 0 such that Br(s) โІ S/E. โˆƒN such that n > N โ‡’ d(sn, s) < r, i.e. sn โˆˆ Br(s) โІ S/E, a contradiction.

6 Compactness

Closed is not the exact analogy to the closed interval, since we can make closed sets like [a, โˆž). Weโ€™re interested in the Bolzano Weirstrass property: given any sequence (sn) โˆˆ [a, b], [a, b] bounded โ‡’ โˆƒ a convergent subsequence (snk ) and limnโ†’โˆž snk = s โˆˆ [a, b] Definition: Given a metric space (S, d), a set E โІ S is compact if: given any sequence (sn) โˆˆ E, โˆƒ a convergent subsequence (snk ) and lim snk = s โˆˆ E. Notes:

  1. if E is compact then E is closed. Proof: if (sn) is convergent sequence in E, โˆƒ a convergent subsequence (snk ), lim snk = s โˆˆ E. But (sn) converging โ‡’ lim sn = lim snk = s โˆˆ E. So E = Eยฏ.
  2. Not every closed set is compact. For example, E = R is not compact, but [a, b] is compact. Theorem (Heine-Borel): In the space Rk^ (with the standard metric), a set E is compact if and only if E is closed and bounded. Proof: assume E is compact โ‡’ E is closed by the previous theorem. Assume for the sake of contradiction that E is unbounded. Fix x 0 โˆˆ S. So E/ โІ Bn(x 0 ) far any n โˆˆ N. So โˆ€n โˆˆ N, โˆƒxn โˆˆ E but d(xn, x 0 ) โ‰ฅ n. So E compact โ‡’ โˆƒ a convergent subsequence (xnj ) and d(xnj , x 0 ) โ‰ฅ nj , โˆ€j. However, (xnj ) convergent โ‡’ (xnj ) is bounded. Contradiction, since nj goes to โˆž as j goes to โˆž. Assume E is closed and bounded. Given any sequence (sn) โˆˆ E, E is bounded โ‡’ (sn) is a bounded sequence, so by the Bolzano-Weirstrass theorem (on Rk), โˆƒ a convergent subsequence (snk ) โˆˆ E and limkโ†’โˆž snk = s โˆˆ E because E is closed. So E is compact. Note that the Bolzano-Weirstrass theorem only applies to Rk. Thus the Heine-Borel theorem is false for metric spaces other than Rk.

Challenge: classify the compact sets in l^1 (N) and C[0, 1] = {f continuous on [0, 1]}, d(f, g) =

|f (x) โˆ’ d(x)|dx. These two spaces are similar. For C[0, 1], use the idea of an equicontinuous family of functions.