Supremum, Infimum

한상민·2023년 6월 22일

해석학

목록 보기
1/1

1. Definition : Supremum

ERE \subset \R and EE \ne \empty 이라고 가정하자.
(1) EE is bounded above     \iff aE, MR\forall a \in E, \exist\ M\in \R s.t. aMa\leq M.
MM is called upper-bound of EE.
(2) sRs\in\R is a supremum of EE     \iff ss is an upper-bound of EE and sMs\leq M for all upper-bounds of EE.
(다시 말해, upperbound들 중 가장 작은 개체가 supremum이다.)

1.1. Theorem : Approximation Property for Suprema

If EE has a (finite) supremum, then ϵ>0\forall\epsilon\gt0,  aE\exist\ a\in E s.t. supEϵ<asupEsupE-\epsilon<a\le supE.
pf)
Theorem이 False라고 가정하면, ϵ0,aE\exist\epsilon_0, \forall a \in E s.t. a(supEϵ0,supE]a\notin (supE-\epsilon_0, supE]이다. 그렇다면, s:=supEϵ0s:=supE-\epsilon_0EE의 upper-bound 라고 할 수 있다. 따라서 supEs=supEϵ0<supEsupE\leq s=supE-\epsilon_0<supE 라는 contradiction이 생기므로, Theorem은 True이다.

1.2. Completenes Axiom

If ERE\subset \R is non-empty and bounded above, then EE has a (finite) supremum.

1.3. Theorem

If EZE\subset\Z has a supremum, then supEEsup E\in E.
pf)
s:=supEs:=supE라고 정의하자. Approx. Property for Suprema에 의해, 우리는 x0Ex_0\in E s.t s1<x0s0s-1<x_0\le s_0 를 하나 선택할 수 있다.
Claim : s=x0s=x_0
If sx0s\ne x_0, choose x1Ex_1\in E s.t. x0<x1sx_0<x_1\le s (by Approx. Property for Suprema). Then, 0<x1x0sx0<10<x_1-x_0\le s-x_0<1 and hence x1x0(0,1)Zx_1-x_0\in (0,1) \cap\Z which is contradiction!
( Remark: There is no nZn\in\Z s.t 0<n<10<n<1. )

1.4. Theorem : Archimedes Principle

If a,bR,a>0a,b\in\R, a>0, then nN\exist n \in\N s.t b<nab<n*a.
pf)
Case-1 : b<ab<a. Set n=1. b<na=1a=ab<n*a=1*a=a
Case-2 : bab\ge a. Consider E={kN:kab}E=\{k\in\N:k*a\le b\}. Then 1EN1\in E\subseteq\N and EE is bounded above by ab\frac{a}{b}. By the Completenes Axiom and Thm 1.3, s:=supEs:=supE exist and sENs\in E\subseteq\N. Set n:=s+1n:=s+1, then nNn\in\N and nEn\notin E. Thus b<nab<n*a.

2. Definition : Infimum

ER\empty\ne E\subseteq\R 이라고 하자.
(1) EE is bounded below     \iff mR,aE,ma.\exist m\in\R, \forall a\in E, m\le a. mm is lower-bound of EE.
(2) tRt\in\R is an infimum of EE     \iff tt is a lower-bound of EE and m{x:x is lower bound of E},mt\forall m\in\{x:x\ is\ lower\ bound\ of\ E\}, m\le t. ( lower-bound 중 가장 큰 개체가 infimum이다. )
(3) EE is bounded     \iff EE is bounded both above and below.

2.1. Reflection

Reflection of EE란, ERE\subseteq\R에 대하여,
E={xR:x=a for some aE}-E = \{x\in\R :x=-a\ for\ some\ a\in E\}를 말한다.

2.2 Theorem : Reflection Principle

For ER\empty\ne E\subseteq\R, EE has an infimum     \iff E-E has a supremum.
In this case, inf(E)=sup(E)inf(E)=-sup(-E)
pf)
(1) "\Leftarrow" : Suppose that E-E has a sup(E)=:s.sup(-E)=:s. Then s-s is a lower-bound of E.E. Let mm be an any lower-bound of EE. Since sms\le -m, ms.m\le-s. Thus, s-s is the infimum of E.E.
(2) "\Rightarrow": By reflection, E-E is bounded above. By Completenes Axiom, E-E has a supremum.

3. Theroem : Well-ordering Principle

If EN\empty\ne E\subseteq\N, then EE has a least element xx i.e. xE.x\in E.
pf)
0 is a lower-bound of E.E. By reflection, Completenes Axiom and Theorem 1.3, E-E has a sup(E)=:sE.sup(-E)=:s\in -E. By Reflection Principle, EE has an inf(E)=sEinf(E)=-s\in E which is the least element.

0개의 댓글