가측 공간 문서 원본 보기
←
가측 공간
둘러보기로 이동
검색으로 이동
문서 편집 권한이 없습니다. 다음 이유를 확인해주세요:
요청한 명령은 다음 권한을 가진 사용자에게 제한됩니다:
사용자
.
문서의 원본을 보거나 복사할 수 있습니다.
{{위키데이터 속성 추적}} [[측도론]]에서, '''가측 공간'''(可測空間, {{llang|en|measurable space}})은 '''가측 집합'''(可測集合, {{llang|en|measurable set}})이라는 특별한 [[부분 집합]]들의 족이 부여된 [[집합]]이다. 가측 집합들은 가산 [[합집합]] · 가산 [[교집합]] · [[여집합]]에 대하여 닫혀 있다. 측도론에서는 모든 집합들에 적절한 [[측도]]를 부여하는 것이 불가능하므로 흔히 사용되는 특정 집합들을 골라야 하며, 가측 공간의 개념은 이러한 선택을 공리화하여 얻는다. 두 가측 공간 사이의 자연스러운 사상은 '''[[가측 함수]]'''라고 한다. == 정의 == 임의의 집합 <math>X</math>에 대하여, 그 [[멱집합]] <math>\mathcal P(X)</math>는 [[완비 불 대수]]이며, 특히 [[시그마 대수]]이다. '''가측 공간''' <math>(X,\Sigma)</math>은 다음과 같은 데이터로 구성된다. * <math>X</math>는 [[집합]]이다. * <math>\Sigma\subseteq\mathcal P(X)</math>는 [[시그마 대수]] <math>\mathcal P(X)</math>의 부분 [[시그마 대수]]이다. 즉, 다음 두 조건이 성립한다. ** ([[여집합]]에 대한 닫힘) 모든 <math>S\in\Sigma</math>에 대하여, <math>X\setminus S\in\Sigma</math>이다. ** (가산 [[합집합]]에 대한 닫힘) 가산 부분 집합 <math>\mathcal S\subseteq\Sigma</math> (<math>|\mathcal S|\le\aleph_0</math>)에 대하여, <math>\textstyle\bigcup\mathcal S\in\Sigma</math>이다. (특히, 만약 <math>\mathcal S=\varnothing</math>일 경우 <math>\textstyle\bigcup\varnothing=\varnothing\in\Sigma</math>이다.) <math>\Sigma</math>의 원소를 <math>X</math>의 '''가측 집합'''이라고 한다. 가측 공간들과 [[가측 함수]]들은 [[구체적 범주]] <math>\operatorname{Measble}</math>를 이룬다. === 분리 가측 공간 === 가측 공간 <math>(X,\Sigma)</math>가 다음 조건을 만족시킨다면 '''분리 가측 공간'''(分離可測空間, {{llang|en|separated measurable space}})이라고 한다. * <math>f\colon X\to\mathcal P(\Sigma)</math>, <math>f\colon x\mapsto\{S\in\Sigma\colon x\in S\}</math>는 [[단사 함수]]이다. 즉, 임의의 두 점 <math>x,y\in X</math>에 대하여, 만약 <math>x\ne y</math>라면, <math>x\in S\not\ni y</math>인 <math>S\in\Sigma</math>가 존재한다. == 성질 == 임의의 가측 공간 <math>X</math>에서, [[공집합]] <math>\varnothing</math>과 전체 집합 <math>X</math>은 항상 가측 집합이다. === 연산에 대한 닫힘 === 집합 <math>X</math> 위의 1개 이상의 (유한 또는 무한 개의) 임의의 가측 공간 구조들 <math>\{\Sigma_i\}</math>이 주어졌을 때, 그 [[교집합]] :<math>\bigcap_{i\in I}\Sigma_i</math> 역시 <math>X</math> 위의 가측 공간 구조이다.<ref name="Tao" />{{rp|Exercise 1.4.13}} (그러나 이는 [[합집합]]에 대하여 성립하지 않는다.) 따라서, 임의의 집합족 <math>\mathcal F\subseteq\mathcal P(X)</math>에 대해, <math>\mathcal F</math>의 원소들을 가측 집합으로 하는 가장 엉성한 가측 공간 구조가 존재한다.<ref name="Tao" />{{rp|Definition 1.4.14}} 구체적으로, 이는 <math>\mathcal F</math>를 포함하는 가측 공간 구조들의 [[교집합]]이다. 이를 <math>\sigma(\mathcal F)</math>로 표기하자. 따라서, 주어진 집합 <math>X</math> 위의 가측 공간 구조들의 족은 [[완비 격자]]를 이룬다. === 크기 === 가측 공간 <math>(X,\Sigma)</math>의 가측 집합의 수는 항상 <math>2^n</math> 꼴의 양의 정수이거나, 아니면 <math>2^{\aleph_0}</math> 이상이다. 특히, 가측 공간은 가산 무한 개의 가측 집합을 가질 수 없다.<ref>{{웹 인용|url=http://www.yaronhadad.com/why-arent-there-infinitely-countable-sigma-algebras/|제목=Why aren't there infinitely countable sigma-algebras?|이름=Yaron|성=Hadad|날짜=2012-09-06|언어=en}}</ref> <div class="mw-collapsible mw-collapsed toccolours"> '''증명:''' <div class="mw-collapsible-content"> <math>(X,\Sigma)</math>가 무한 개의 가측 집합들을 갖는 가측 공간이라고 하자. 그렇다면, 다음과 같은 집합렬을 재귀적으로 고른다. # <math>X\setminus(S_0\cup\cdots\cup S_i)</math>가 무한 개의 가측 부분 집합들을 갖는다고 가정한다. # <math>X\setminus(S_0\cup\cdots\cup S_i)</math>에서, 공집합이나 전체 집합이 아닌 임의의 가측 부분 집합 <math>\varnothing\ne E_{i+1}\subsetneq X\setminus(S_0\cup\cdots\cup S_i)</math>을 고른다. # <math>X\setminus(S_0\cup\cdots\cup S_i)</math>는 무한 개의 가측 부분 집합들을 가지므로, <math>\{S\in\Sigma\colon S\subseteq E_{i+1}\}</math> 또는 <math>\{S\in\Sigma\colon S\subseteq X\setminus(S_0\cup\cdots\cup S_i\cup E_{i+1})\}</math> 가운데 적어도 하나가 무한 집합이며, 다음과 같이 정의한다면 <math>X\setminus(S_0\cup\cdots\cup S_{i+1})</math> 역시 무한 개의 가측 부분 집합들을 갖는다. #: <math>S_{i+1}=\begin{cases} E_{i+1}&|\{S\in\Sigma\colon S\subseteq E_{i+1}\}|<\aleph_0\\ X\setminus(S_0\cap\cdots\cup S_i\cup E_{i+1})&|\{S\in\Sigma\colon S\subseteq E_{i+1}\}|\ge\aleph_0 \end{cases}</math> 따라서, <math>S_0,S_1,\dots\in\Sigma</math>는 가산 무한 개의 서로소 가측 집합들을 이루며, :<math>\left\{\bigcup_{i\in I}S_i\colon I\subseteq\mathbb N\right\}\subseteq\Sigma</math> 는 크기 <math>2^{\aleph_0}</math>의, <math>\Sigma</math>의 부분 집합을 이룬다. 따라서 <math>|\Sigma|\ge 2^{\aleph_0}</math>이다. </div> </div> 집합 <math>X</math>의 집합족 <math>\mathcal A\subseteq\mathcal P(X)</math>에 대하여, <math>\mathcal A</math>로부터 생성되는 가측 공간 구조 <math>\sigma(\mathcal A)</math>의 크기의 상계는 다음과 같다.<ref name="Tao">{{서적 인용 |url=https://terrytao.files.wordpress.com/2012/12/gsm-126-tao5-measure-book.pdf |형식=PDF |성=Tao |이름=Terence |저자링크=테런스 타오 |제목=An introduction to measure theory |언어=en |총서=Graduate Studies in Mathematics |권=126 |출판사=American Mathematical Society |날짜=2011 |isbn=978-0-8218-6919-2 |zbl=1231.28001 }}</ref>{{rp|Exercise 1.4.16}} :<math>|\sigma(\mathcal A)|\le|\mathcal A|^{\aleph_0}</math> 이는 <math>\sigma(\mathcal A)</math>를 [[초한 귀납법]]으로 구성할 때 <math>\omega_1</math>번의 단계로 끝나기 때문이다. (여기서 <math>\omega_1</math>은 최소의 비가산 [[순서수]]이다.) === 딘킨 π-λ 정리와 단조류 정리 === 집합 <math>X</math> 속의 집합족 <math>\mathcal A\subseteq\mathcal P(X)</math>에 대하여 다음 조건들을 정의하자. * (π) <math>\mathcal A</math>는 2항 [[교집합]]에 대하여 닫혀 있다. 즉, 임의의 <math>A,B\in\mathcal A</math>에 대하여, <math>A\cap B\in\mathcal A</math>이다. 이를 만족시키는 집합족을 '''π계'''({{llang|en|π-system}})라고 한다. * (λ) <math>\mathcal A</math>는 [[여집합]]에 대하여 닫혀 있으며, 가산 개의 서로소 집합들의 합집합에 대하여 닫혀 있다. (특히, 0개의 서로소 집합들의 합집합은 [[공집합]]이므로, <math>\varnothing\in\mathcal A</math>이다.) 이를 만족시키는 집합족을 '''λ계'''({{llang|en|λ-system}})라고 한다. * (π′) <math>\mathcal P(X)</math>의 부분 [[불 대수]]를 이룬다. 즉, [[여집합]] · 유한 교집합 · 유한 합집합에 대하여 닫혀 있으며, <math>\varnothing\in\mathcal A</math>이다. 이를 만족시키는 집합족을 '''집합체'''({{llang|en|field of sets}})라고 한다. * (λ′) <math>X\in\mathcal A</math>이며, 임의의 <math>A_0,A_1,\dots\in\mathcal A</math>에 대하여, <math>A_0\subseteq A_1\subseteq\cdots</math>라면, <math>\textstyle\bigcup_{i=0}^\infty A_i\in\mathcal A</math>이며, 임의의 <math>A_0,A_1,\dots\in\mathcal A</math>에 대하여, <math>A_0\supseteq\mathcal A_1\supseteq\cdots</math>라면, <math>\textstyle\bigcap_{i=0}^\infty A_i\in\mathcal A</math>이다. 이를 만족시키는 집합족을 '''단조류'''({{llang|en|monotone class}})라고 한다. 그렇다면, 다음이 성립한다. 집합 <math>X</math> 속의 집합족 <math>\mathcal A\subseteq\mathcal P(X)</math>에 대하여, 다음 세 조건이 서로 [[동치]]이다. * [[시그마 대수]]이다. * (π) 조건과 (λ) 조건을 만족시킨다. * (π′) 조건과 (λ′) 조건을 만족시킨다. 집합 <math>X</math> 속의 집합족 <math>\mathcal A\subseteq\mathcal P(X)</math>에 대하여, 다음이 성립한다. * ('''딘킨 π-λ 정리''' {{llang|en|Dynkin π–λ theorem}}) <math>\mathcal A</math>가 (π) 조건을 만족시킨다면, <math>\mathcal A</math>로부터 생성되는, (λ) 조건을 만족시키는 최소의 집합족은 [[시그마 대수]]이다.<ref name="Billingsley">{{서적 인용 |성=Billingsley |이름=Patrick |제목=Probability and Measure |url=https://archive.org/details/probabilitymeasu0000bill |언어=en |판=3 |총서=Wiley Series in Probability and Mathematical Statistics |출판사=Wiley-Interscience |위치=New York, N.Y. |날짜=1995 |isbn=978-0-471-00710-4 }}</ref>{{rp|42, §1.3, Theorem 3.3}} * ('''단조류 정리''' {{llang|en|monotone class theorem}}) <math>\mathcal A</math>가 (π′) 조건을 만족시킨다면, <math>\mathcal A</math>로부터 생성되는, (λ′) 조건을 만족시키는 최소의 집합족은 [[시그마 대수]]이다.<ref name="Billingsley" />{{rp|43, §1.3, Theorem 3.4}} {{증명|부제=딘킨 π-λ 정리}} <math>\mathcal A</math>로부터 생성되는, (λ) 조건을 만족시키는 최소의 집합족을 <math>\lambda(\mathcal A)</math>라고 하자. 그렇다면, <math>\lambda(\mathcal A)</math>가 (π) 조건을 만족시킴을 보이면 된다. :<math>\mathcal L_1=\{B\in\lambda(\mathcal A)|\forall A\in\mathcal A\colon A\cap B\in\lambda(\mathcal A)\}</math> 라고 하자. 그렇다면, <math>\mathcal A</math>가 (π) 조건을 만족시키므로 <math>\mathcal A\subseteq\mathcal L_1</math>이다. 또한 <math>\mathcal L_1</math>는 (λ) 조건을 만족시킨다. 따라서, <math>\mathcal L_1=\lambda(\mathcal A)</math>이다. :<math>\mathcal L_2=\{A\in\lambda(\mathcal A)|\forall B\in\lambda(\mathcal A)\colon A\cap B\in\lambda(\mathcal A)\}</math> 라고 하자. <math>\mathcal L_1=\lambda(\mathcal A)</math>이므로 <math>\mathcal A\subseteq\mathcal L_2</math>이며, <math>\mathcal L_2</math>는 (λ) 조건을 만족시키므로, <math>\mathcal L_2=\lambda(\mathcal A)</math>이다. 즉, <math>\lambda(\mathcal A)</math>는 (π) 조건을 만족시킨다. {{증명 끝}} {{증명|부제=단조류 정리}} 위 증명을 (π′) 조건과 (λ′) 조건에 적용하면 단조류 정리의 증명을 얻는다. {{증명 끝}} === 범주론적 성질 === 가측 집합과 [[가측 함수]]의 범주 <math>\operatorname{Measble}</math>은 [[구체적 범주]] :<math>F\colon\operatorname{Measble}\to\operatorname{Set}</math> :<math>F\colon(X,\Sigma)\mapsto X</math> 를 이루며, <math>F</math>는 [[위상 함자]]이다. 따라서, <math>\operatorname{Measble}</math>은 [[완비 범주]]이며 [[쌍대 완비 범주]]이다. <math>\operatorname{Measble}</math>의 [[시작 대상]]은 (유일한 가측 공간 구조를 갖춘) [[공집합]]이며, <math>\operatorname{Measble}</math>의 [[끝 대상]]은 (유일한 가측 공간 구조를 갖춘) [[한원소 집합]]이다. == 예 == <math>X</math>를 임의의 집합이라고 할 때, 다음은 모두 <math>X</math> 위의 가측 공간 구조이다. * 양의 정수가 아닌 임의의 [[기수 (수학)|기수]] <math>\kappa</math>에 대하여, <math>\left\{S\subseteq X|\min\{|S|,|X\setminus S|\}\le\kappa\right\}</math>는 [[시그마 대수]]를 이룬다. 즉, 이는 <math>X</math>의 부분집합 가운데 크기가 <math>\kappa</math> 이하이거나 그 여집합의 크기가 <math>\kappa</math> 이하인 집합들의 족이다. ** 만약 <math>\kappa=0</math>이라면, 이는 <math>\{\varnothing, X\}</math>이다. 이 집합족은 '''자명 가측 공간''' 구조이며, <math>X</math> 위의 가장 엉성한 가측 공간 구조이다. ** 만약 <math>\kappa\ge|X|</math>라면, 이는 <math>X</math>의 [[멱집합]] <math>\mathcal P(X)</math>이다. 이를 '''이산 가측 공간'''이라고 하며, 이는 <math>X</math> 위의 가장 섬세한 가측 공간 구조이다. * <math>X</math>의 분할 <math>\textstyle X=\bigsqcup_{A\in\mathcal P}A</math>이 주어진다면, <math>\{\textstyle\bigcup\mathcal Q|\mathcal Q\subseteq\mathcal P\}</math>는 <math>X</math> 위의 가측 공간 구조를 이룬다. === 유한 집합 위의 시그마 대수 === [[유한 집합]] <math>X</math> 위의 모든 가측 공간 구조는 <math>X</math>의 [[집합의 분할|분할]]로서 정의된다. 즉, 유한 집합 위의 가측 공간 구조들은 그 [[집합의 분할|분할]]과 [[일대일 대응]]하며, 크기가 <math>n=0,1,2,\dots</math>인 유한 집합 위의 가측 공간 구조의 수는 [[벨 수]] :1, 1, 2, 5, 15, 52, 203, 877, 4140, 21147, 115975, … {{OEIS|A000110}} 이다. 예를 들어, 크기가 3인 유한 집합 <math>\{a,b,c\}</math> 위의 분할과 가측 공간 구조들은 다음과 같다. {| class="wikitable" |- ! 분할 !! 가측 집합 !! 비고 |- | {a, b, c} || {}, {a, b, c} || 자명 가측 공간 |- | {a}, {b, c} || {}, {a}, {b, c}, {a, b, c} || |- | {a, b}, {c} || {}, {a, b}, {c}, {a, b, c} || |- | {a, c}, {b} || {}, {a, c}, {b}, {a, b, c} || |- | {a}, {b}, {c} || {}, {a}, {b}, {c}, {a, b}, {b, c}, {a, c}, {a, b, c} || 이산 가측 공간 |} === 보렐 시그마 대수 === {{본문|보렐 집합}} [[위상 공간 (수학)|위상 공간]] 위의 [[보렐 집합]]들의 집합은 [[시그마 대수]]를 이루며, 이를 '''[[보렐 시그마 대수]]''' <math>\operatorname{Borel}(X)</math>라고 한다. 위상 공간 <math>X</math>에 대하여 다음 두 조건이 서로 [[동치]]이다. * <math>(X,\operatorname{Borel}(X))</math>는 분리 가측 공간이다. * <math>X</math>는 [[콜모고로프 공간]]이다. [[유한 집합]] 위의 모든 가측 공간 구조는 적절한 위상의 [[보렐 시그마 대수]]로 나타낼 수 있다. 그러나 [[무한 집합]]의 경우 임의의 위상의 보렐 시그마 대상으로 나타낼 수 없는 가측 공간 구조 또한 존재한다.<ref>{{저널 인용|제목=On the problem of generating sigma-algebras by topologies|저널=Statistics & Decisions|권=2|호=3–4|날짜=1984-04|쪽=377-388|이름=Albert|성=Ascherl|doi=10.1524/strm.1984.2.34.377|issn=0721-2631|언어=en}}</ref><ref>{{저널 인용|제목=A simple example of a non-Borel σ-field|저널=Statistics & Decisions|권=4|호=1|날짜=1986-01|쪽=97–98|이름=Robert|성=Lang|doi=10.1524/strm.1986.4.1.97|issn=0721-2631|언어=en}}</ref> 구체적으로, <math>S</math>가 크기 <math>2^{\aleph_0}</math>의 [[비가산 집합]]이라고 하자. 크기 2의 [[이산 공간]] <math>\{0,1\}</math>의 비가산 [[곱공간]] <math>\{0,1\}^S</math>을 생각하자. 이 [[곱위상]]은 [[기저 (위상수학)|기저]] :<math>\mathcal B=\left\{ \prod_{s\in S}U_s \colon U_s\subseteq\{0,1\}\forall s\in S,\; |\{s\in S\colon U_r\ne\{0,1\}\}|<\aleph_0\right\}</math> 로 생성된다. 그렇다면, <math>\mathcal B</math>로 생성되는 가측 공간 구조는 <math>\{0,1\}^S</math> 위의 임의의 위상의 보렐 시그마 대수로 나타낼 수 없다. <div class="mw-collapsible mw-collapsed toccolours"> '''증명:''' <div class="mw-collapsible-content"> 우선, <math>|\mathcal B|=|S|=2^{\aleph_0}</math>이므로, :<math>\sigma(\mathcal B)\le|S|^{\aleph_0}=2^{\aleph_0}</math> 이다. [[귀류법]]을 사용하여 <math>\sigma(\mathcal B)=\operatorname{Borel}(\mathcal U)</math>가 되는 위상 <math>\mathcal U</math>가 존재한다고 하자. 그렇다면 <math>\sigma(\mathcal B)</math>가 분리 시그마 대수이므로 <math>\mathcal U</math>는 [[콜모고로프 공간|콜모고로프 위상]]이다. 그러므로 :<math>\{0,1\}^S\to\sigma(\mathcal B)</math> :<math>x\mapsto\operatorname{cl}_{\mathcal U}(\{x\})</math> ([[한원소 집합]]의 <math>\mathcal U</math>-[[폐포 (위상수학)|폐포]])는 [[단사 함수]]이며, :<math>2^{\aleph_0}=|\sigma(\mathcal B)|\ge|\{0,1\}^S|=2^{2^{\aleph_0}}</math> 인데, 이는 모순이다. </div></div> == 각주 == {{각주}} == 외부 링크 == * {{eom|title=Algebra of sets}} * {{eom|title=Measurable set}} * {{eom|title=Measurable space}} * {{매스월드|id=Sigma-Algebra|title=Sigma-algebra}} * {{매스월드|id=MeasurableSpace|title=Measurable space}} * {{매스월드|id=MeasurableSet|title=Measurable set}} * {{nlab|id=sigma-algebra|title=Sigma-algebra}} * {{nlab|id=measurable space|title=Measurable space}} * {{nlab|id=measurable subset|title=Measurable subset}} * {{웹 인용|url=https://terrytao.wordpress.com/2010/09/25/245a-notes-3-integration-on-abstract-measure-spaces-and-the-convergence-theorems/|제목=245A, Notes 3: Integration on abstract measure spaces, and the convergence theorems|이름=Terry|성=Tao|저자링크=테런스 타오|웹사이트=What’s New|언어=en}} * {{웹 인용|url=http://mathoverflow.net/questions/87838/is-every-sigma-algebra-the-borel-algebra-of-a-topology|제목=Is every sigma-algebra the Borel algebra of a topology?|출판사=Math Overflow|언어=en}} [[분류:측도론]] [[분류:집합족]] [[분류:공간 (수학)]]
이 문서에서 사용한 틀:
틀:Eom
(
원본 보기
)
틀:Llang
(
원본 보기
)
틀:Nlab
(
원본 보기
)
틀:OEIS
(
원본 보기
)
틀:Rp
(
원본 보기
)
틀:각주
(
원본 보기
)
틀:매스월드
(
원본 보기
)
틀:본문
(
원본 보기
)
틀:서적 인용
(
원본 보기
)
틀:웹 인용
(
원본 보기
)
틀:위키데이터 속성 추적
(
원본 보기
)
틀:저널 인용
(
원본 보기
)
틀:증명
(
원본 보기
)
틀:증명 끝
(
원본 보기
)
가측 공간
문서로 돌아갑니다.
둘러보기 메뉴
개인 도구
로그인
이름공간
문서
토론
한국어
보기
읽기
원본 보기
역사 보기
더 보기
검색
둘러보기
대문
최근 바뀜
임의의 문서로
미디어위키 도움말
특수 문서 목록
도구
여기를 가리키는 문서
가리키는 글의 최근 바뀜
문서 정보