🎉 从330转为WashU帮助手册啦!

Math4111
模块
Math4111 L13

Lecture 13

Review

Consider the metric space X=RX=\mathbb{R} (with the usual metric d(x,y)=xyd(x,y)=|x-y|). Let E=(0,1)E=(0,1).

  1. Find several examples of sets YRY\subset \mathbb{R} such that EYE\subset Y and EE is closed in YY.
    Example:

    1. Y=EY=E, EE is closed in YY.
      We can prove this using normal ways, or Theorem 2.23 EE is closed in Y    EcY\iff E^c is open in YY
      Y    Ec=ϕY\iff E^c=\phi and it's open.
    2. Y=R\{0,1}Y=\mathbb{R}\backslash\{0,1\}
      Y\E=(,0)(1,)Y\backslash E=(-\infty,0)\cup (1,\infty) Theorem 2.30 EYXE\subset Y\subset X, EE is open in Y    Y\iff G\exists G open in XX such that GY=EG\cap Y=E
      GY=Y\EG\cap Y=Y\backslash E And we know Y\EY\backslash E is open in YY. By Theorem 2.23 EE is closed in YY.
  2. If YY is as in part 1, we can conclude that EE is closed and bounded in yy. Part of Theorem 2.41 says: "If a set is closed and bounded, then it is compact." Why doesn't that theorem apply here.
    The set is not closed in Rk\mathbb{R}^k.

New stuffs

Connected sets

Definition 2.45

A,BXA,B\subset X, we say AA and BB are separated in XX if AB=ϕA\cup \overline{B}=\phi and AB=ϕ\overline{A}\cup B=\phi

  • EXE\subset X disconnected in XX if \exists nonempty separated A,BXA,B\subset X such that E=ABE=A\cup B
  • EXE\subset X is connected in XX if it is not disconnected.

Example 2.46

(0,1),(1,2)(0,1),(1,2) are separated [so (0,1)(1,2)(0,1)\cup (1,2) is disconnected]

[0,1],(1,2)[0,1],(1,2) are not separated [so [0,1](1,2)={1}[0,1]\cup (1,2)=\{1\}] So this doesn't tell us where [0,1](1,2)[0,1]\cup (1,2) is connected or not.

Theorem 2.47

Suppose ERE\subset \mathbb{R}

EE is connected     (x,y,z)\iff \forall (x,y,z) with x,yE,x<z<yx,y\in E,x<z<y such that zEz\in E.

By negating, this is equivalent to

EE is disconnected     (x,y,z)\iff \exists (x,y,z) with x,yE,x<z<yx,y\in E,x<z<y such that zEz\notin E.

Proof:

    \impliedby

Suppose (x,y,z)\exists (x,y,z) with x,yE,x<z<yx,y\in E,x<z<y such that zEz\notin E.

Let A=(,z)E,B=(z,)EA=(-\infty,z)\cap E,B=(z,\infty)\cap E

Lemma: EF    EFE\subset F\implies \overline{E}\subset \overline{F}

Since A(,z),A(,z]A\subset (-\infty,z), \overline{A}\subset (-\infty,z]

Since AB=ϕ\overline{A}\cap B=\phi, similarly, AB=ϕA\cap \overline{B}=\phi. So A,BA,B are separated. Also they are non empty (xA,yBx\in A,y\in B) and E=ABE=A\cap B. So EE is disconnected.

    \implies

Suppose \exists nonempty separated A,BXA,B\subset X such that E=ABE=A\cup B.

Our goal is to find x,yE,x<z<yx,y\in E,x<z<y such that zEz\notin E.

Aϕ,Bϕ    xA,yBA\neq \phi,B\neq \phi\implies \exists x\in A,y\in B

Without loss of generality, assume x<yx<y.

Let w=sup(A[x,y])w=sup(A\cup [x,y]), wA[x,y]w\in \overline{A\cup [x,y]} (by Theorem 2.28) This implies wAw\subset \overline{A} and w[x,y]w\in [x,y]

Since A,BA,B are separated, wBw\notin B (AB=ϕ\overline{A}\cup B=\phi).

Since yBy\in B, w[x,y)w\in [x,y)

Consider 2 cases,

Case 1. wAw\notin A.

let z=wz=w, and x,y,zx,y,z satisfy the desired properties

Case 2. wAw\in A

Since A,BA,B are separated, wBw\notin \overline{B} (AB=ϕA\cup \overline{B}=\phi).

Thus, r>0\exists r>0 such that (wr,w+r)B=ϕ(w-r,w+r)\cap B=\phi.

Let z=w+r2z=w+\frac{r}{2}, then x,y,zx,y,z satisfy the desired properties.

EOP

Chapter 3: Numerical Sequences and Series

Numerical Sequences

Notations

Rudin use {pn}\{p_n\} to denote a sequence p1,p2p_1,p_2.

To avoid confusion with sets, we use (pn)n=1(p_n)_{n=1}^\infty or (pn)(p_n)

Definition 3.1

Let (X,d)(X,d) be a metric space. Let (pn)(p_n) be a sequence in XX.

Let pXp\in X. We say (px)(p_x) converges to pp if ε>0,NN\forall \varepsilon>0,\exists N\in\mathbb{N} such that nN\forall n\geq N, d(pn,p)<εd(p_n,p)<\varepsilon. (pnBε(p)p_n\in B_\varepsilon (p))

Notation limnpn=p\lim_{n\to \infty} p_n=p, pnpp_n\to p

We say (pn)(p_n) converges if pX\exists p\in X such that pnpp_n\to p.

i.e. pX\exists p\in X such that ε>0,NN\forall\varepsilon>0,\exists N\in\mathbb{N} such that nN,d(pn,p)<ε\forall n\geq N,d(p_n,p)<\varepsilon

We say (pn)(p_n) diverges if (pn)(p_n) doesn't converge.

i.e. pX\forall p\in X, pnpp_n\cancel{\to} p

i.e. pX\forall p\in X such that ε>0,NN\exists \varepsilon>0,\forall N\in\mathbb{N} such that nN,d(pn,p)ε\exists n\geq N,d(p_n,p)\geq\varepsilon

Definition 3.2

We say a sequence (pn)(p_n) is bounded if xX\exists x\in X, r>0\forall r>0 such that nN,pnBr(x)\forall n\in \mathbb{N},p_n\in B_r(x)

Example:

X=CX=\mathbb{C}, sn=1ns_n=\frac{1}{n}

Then sn0s_n\to 0 i.e. ε>0NN\forall \varepsilon>0 \exists N\in \mathbb{N} such that nN\forall n\geq N, sn0<ε|s_n-0|<\varepsilon.

Proof:

Let ε>0\varepsilon >0 (arbitrary)

Let NNN\in \mathbb{N} be greater than 1ε\frac{1}{\varepsilon} (by Archimedean property) e.g. N=1ε+1N=\frac{1}{\varepsilon}+1 (we choose NN)

Let nNn\geq N (arbitrary)

Then snq=1n1Nε|s_n-q|=\frac{1}{n}\leq \frac{1}{N}\leq \varepsilon

EOP