高等概率论 Chapter 6 Construction of a Probability Measure

Chapter 6 Construction of a Probability Measure

南京审计大学统计学研究生第一学期课程，《高等概率论》。

欢迎大家来我的github下载源码呀，https://github.com/Berry-Wen/statistics-note-system

概率测度的构造 adveprobab 2

对于 $\mathcal{B}$ 是 { ω } ω ∈ Ω \{\omega\}_{\omega \in \Omega} {ω}ω∈Ω​ ，即 $\mathcal{B}$ 是单点集组成的事件类

如果概率 $P$ 建立在 $\mathcal{B}$ 上，这样构造的概率比较好计算，现在的想法是：想通过有限集合或者可数集合，来进行拓展，将概率测度扩展开

即将 $P$ 扩展到 $\mathcal{A}=2^{\Omega }$

核心：

$$finite\Rightarrow uncountable$$

---

• Assume given $\Omega$ (countable or uncountable) and a $\sigma$-algebra $\mathcal{A}\in 2^{\Omega }$

• $(\Omega ,\mathcal{A})$ is called a measurable space. 构造$P$

• Want to construct probability measure on $\mathcal{A}$

  • When $\Omega$ is finite or countable, we have already seen this is simple to do.

  • When $\Omega$ is uncountable, the same technique does not work. 失效

    • Indeed, a “typical” probability $P$ will have P ( { ω } ) = 0 P(\{\omega\})=0 P({ω})=0 for all $\omega$, and thus the family of all numbers P ( { ω } ) P(\{\omega\}) P({ω}) for $\omega \in \Omega$ does not characterize probability $P$ in general.

    • 怎么理解上一点？对于不可数集合。。。

• In many “concrete” situations — that it is often relatively simple to construct a “probability” on an algebra which generates the $\sigma$-algebra $\mathcal{A}$ , and the problem at hand is then to extend this probability to the $\sigma$-algebra itself.

  在由 $\mathcal{A}$ 生成的 sigma 代数里面构建一个代数上的概率是相对比较容易的，现在要将其拓展到 sigma 代数里面

• Suppose: 在一个代数上建立概率 $P$

  • ${\mathcal{A}}_0$ is an algebra and, $\mathcal{A}=\sigma ({\mathcal{A}}_0)$.

  • Given a probability $P$ on the algebra ${\mathcal{A}}_0$ : that is, a set function $P:{\mathcal{A}}_0\to [0,1]$ satisfying

    • $P(\Omega )=1$
    • (Countable Additivity) for any sequence { A n } n ≥ 1 ⊂ A 0 \{A_n\}_{n\ge 1} \subset \mathcal{A}_0 {An​}n≥1​⊂A0​, pairwise disjoint, and such that ${\cup }_{n\ge 1}{A}_n\in {\mathcal{A}}_0$, we have

$$P\left({\cup }_{n\ge 1}{A}_n\right)=\sum _{n=1}^{\infty }P(A_n)$$

It might seem natural to use for $\mathcal{A}$, the set of all subsets of $\Omega$, as we did in the case where $\Omega$ was countable.

We do not do so for the following reason, illustrated by an example: 不可数的不满足

• Suppose $\Omega =[0,1]$, and define a set function $P$ on intervals of the form $P((a,b])=b-a$, where $0\le a\le b\le 1$

• It is a natural “probability measure” that assigns the usual length of an interval as its probability.

  概率表示为长度

• Suppose we want to extend $P$ in a unique way to $2^{\Omega }=2^{[0,1]}=$ all subsets of $[0,1]$ such that

  • $P(\Omega )=1$
  • $P({\cup }_{n=1}^{\infty }{A}_n)={\sum }_{n=1}^{\infty }P(A_n)$ for any sequence of subsets { A n } n ≥ 1 \{A_n\}_{n\ge 1} {An​}n≥1​ with $A_n\cap A_m=\varnothing$ for $n\ne m$

• One can prove that no such $P$ exists! 怎么证明其不存在性？
• The collection of sets $2^{[0,1]}$ is simply too big for this to work.

Borel realized that we can however do this on a smaller collection of sets, namely the smallest $\sigma$-algebra containing intervals of the form $(a,b]$.

---

Borel set :

• the sigma-algebra generated by the open sets \qquad 所有开集生成的 sigma 代数

• the sigma-algebra generated by the open intervals \quad 所有开区间生成的 sigma 代数

• the sigma-algebra generated by the $(-\infty ,a],a\in \mathbb{Q}$ \quad 所有 $(-\infty ,a]$ 生成的 sigma 代数

  这三个定义构造的 Borel set 越来越小，事件类越小，构建概率 $P$ 越简单

---

由定理2.1：在$\mathbb{R}$上的博雷尔集是由$(-\infty ,a]$这种形式的区间所生成的sigma代数

Proof.

Let $\mathcal{C}$ denote all open intervals. Since every open set in $\mathcal{R}$ is the countable union of open intervals, we have $\sigma (\mathcal{C})=$ the Borel $\sigma -algebra$ of $\mathbb{R}$

Let $\mathcal{D}$ denote all intervals of the form $(-\infty ,a]$, where $a\in \mathbb{Q}$.

Let $(a,b)\in \mathcal{C}$

Let $(a_n{)}_{n\ge 1}$ be a sequence of rationals decreasing to a

Let $(b_n{)}_{n{\ge }_1}$ be a sequence of rationals increasing strictly to b.

Then

$$\begin{array}{rl}(a,b)& ={\cup }_{n=1}^{\infty }(a_n,b_n]\\ & ={\cup }_{n=1}^{\infty }\left((-\infty ,b_n]\cap (-\infty ,a_n{]}^c\right)\end{array}$$

Therefore , $\mathcal{C}\subset \sigma (\mathcal{D})$, where $\sigma (\mathcal{C})\subset \sigma (\mathcal{D})$

However, since each element of $\mathcal{D}$ is a closed set, it is also a Borel set, and therefore $\mathcal{D}$ is contained in the Borel sets $\mathcal{B}$, Thus we have
$$\mathcal{B}=\sigma (\mathcal{C})\subset \sigma (\mathcal{D})\subset \mathcal{B}$$

and hence $\sigma (\mathcal{D})=\mathcal{B}$

---

这个定理给出了更容易验证 $Borel\sigma -algebra$ 的一种方法，给出了验证博雷尔集的一个充要条件

这个定理的证明核心在于 $(a,b) =\cup_{n=1}^{ \infty} \left( (- \infty,b_n] \cap (- \infty,a_n]^c \right) \$

需要证明左边的开区间$(a,b)$所构成的 sigma 代数$\mathcal{B}$和右边的 $(-\infty ,a]$这种形式构成的sigma代数$\sigma (\mathcal{D})$是相等的，则需要证明两个包含关系

Theorem 6.1 概率延拓定理（唯一延拓）

Each probability $P$ defined on the algebra ${\mathcal{A}}_0$ has a unique extension (also call $P$) on $\mathcal{A}$.

每一个定义在代数 ${\mathcal{A}}_0$ 上的概率，在 $\mathcal{A}$ 上都有唯一的拓展

We will show only the uniqueness. For the existence one can consult any standard text on measure theory.

Definition 6.1

• A class $\mathcal{C}$ of subsets of $\Omega$ is closed under finite intersections $\mathcal{C}$

  对有限并封闭 （代数满足对有限并封闭）

  • If for when $A_1,A_2,...,A_n\in \mathcal{C}$ ,
    then $A_1\cap A_2\cap \cdots \cap A_n\in \mathcal{C}$ as well

    (n arbitrary but finite)

• A class $\mathcal{C}$ is closed under increasing limits $\mathcal{C}$

  对单増并封闭 （ $\sigma$-代数满足，且 $\sigma$-代数更小）

  • If wherever $A_1\subset A_2\subset \cdots \subset A_n\subset \cdots$ is a sequence of events in $\mathcal{C}$, then ${\cup }_{n=1}^{\infty }{A}_n\in \mathcal{C}$ as well.

• A class $\mathcal{C}$ is closed under differences $\mathcal{C}$

  对差封闭 （代数满足对差封闭，但满足对差封闭的不一定是代数）

  • If whenever $A,B\in \mathcal{C}$ with $A\subset B$, then $B-A\in \mathcal{C}$.

$\star \star \star \star \star$Monotone Class Theorem 单调类定理

Let $\mathcal{C}$ be a class of subsets of $\Omega$, closed under finite intersections and containing $\Omega$.

$\mathcal{C}$ 满足两个条件：包含 $\Omega$、对有限交封闭

Let $\mathcal{B}$ be the smallest class containing $\mathcal{C}$ which is closed under increasing limits and by difference.

$\mathcal{B}$ 满足两个条件：包含$\mathcal{C}$ 的最小类、对单増并、差封闭

Then $\mathcal{B}=\sigma (C)$

---

Proof.

Note that

• The intersection of classed of sets closed under increasing limits and differences is again a class of that type.

一列事件，对单増并封闭，则把之交起来也对单増并封闭

Proof.

题设：

• 有若干个事件类 ${\mathcal{A}}_{\alpha },\alpha \in \mathcal{I}$
• 对每个 $\alpha \in \mathcal{I},{\mathcal{A}}_{\alpha }$ 对单増并封闭

推导：

若 $A_1\subset A_2\subset \cdots$ 且 $A_i\in {\cap }_{\alpha \in I}{\mathcal{A}}_{\alpha }$

对任意固定的 $\alpha \in \mathcal{I},A_i\in {\mathcal{A}}_{\alpha },,i=1,2,...$

则 ${\bigcup }_{i=1}^{\infty }{A}_i\in {\mathcal{A}}_{\alpha }$ ，于是 ${\bigcup }_{i=1}^{\infty }{A}_i\in {\bigcap }_{\alpha \in I}{\mathcal{A}}_{\alpha }$ （每个都属于 ${\mathcal{A}}_{\alpha }$ ，则属于 ${\mathcal{A}}_{\alpha }$ 的交）

一列事件，对差封闭，则把之交起来也对差封闭

自己证明一下

So, by taking the intersection of all such classed,

• there always exists a smallest class containing $\mathcal{C}$ which is closed under increasing limits and by differences.

  总存在一个最小的类，满足：包含 $\mathcal{C}$ ，且对单増并封闭、差封闭

  所有满足（对单増并封闭、差封闭、包含 $\mathcal{C}$ ）的都交起来，最小

For each set $B$, denote ${\mathcal{B}}_B$ to be the collection of sets $A$ such that $A\in \mathcal{B}$ and $A\cap B$, $i.e.$

B B = { A : A ∈ B , A ∩ B ∈ B } \mathcal{B}_{B} = \left\{ A: A \in \mathcal{B}, A \cap B \in \mathcal{B} \right\} BB​={A:A∈B,A∩B∈B}

看上去 ${\mathcal{B}}_B$ 跟 $B$ 的选择有关，但实际上无关

Given the properties of $\mathcal{B}$, one easily checks that ${\mathcal{B}}_B$ is closed under increasing limits and by differences.

证明对单调并封闭，即证明： 若$A_1\subset A_2\subset ...$ 且 $A_i\in {\mathcal{B}}_B$ ，证明 ${\cup }_{i=1}^{\infty }{A}_i\in {\mathcal{B}}_B$

则证明其满足两条：

• ${\cup }_{i=1}^{\infty }{A}_i\in \mathcal{B}$
• $({\cup }_{i=1}^{\infty }{A}_i)B\in \mathcal{B}$

证明第一条：${\cup }_{i=1}^{\infty }{A}_i\in \mathcal{B}$

$\because \mathcal{B}$ 对单増并封闭（由定义）

$\therefore {\cup }_{i=1}^{\infty }{A}_i\in \mathcal{B}$

证明第二条：$({\cup }_{i=1}^{\infty }{A}_i)B\in \mathcal{B}$

$({\cup }_{i=1}^{\infty }{A}_i)B={\cup }_{i=1}^{\infty }(A_{i}B)$ 且 $A_{1}B\subset A_{2}B\subset ...$

则 ${\cup }_{i=1}^{\infty }(A_{i}B)\subset \mathcal{B}$ （定义）

则说明 ${\cup }_{i=1}^{\infty }{A}_i\in {\mathcal{B}}_B$ ，即 ${\mathcal{B}}_B$ 对单増并封闭

仿照这个证明，对差封闭的自己证明一下

Let $B\in \mathcal{C}$; 给定 $B\in \mathcal{C}$;

For each $C\in \mathcal{C}$ one has $B\cap C\subset \mathcal{C}\subset \mathcal{B}$ and $C\in \mathcal{B}$, thus $C\in {\mathcal{B}}_B$

Hence $\mathcal{C}\subset {\mathcal{B}}_B\subset \mathcal{B}$. 后一个是定义

$\because {\mathcal{B}}_B$ 是包含了 $\mathcal{C}$ 的对有限并、差封闭的

$\therefore \mathcal{B}={\mathcal{B}}_B$ (by the properties of $\mathcal{B}$ and of ${\mathcal{B}}_B$ )

说明在$B\in \mathcal{C}$时是可以证明$\mathcal{B}={\mathcal{B}}_B$的

下面将其拓展到 $\mathcal{B}$ 上

Now let $B\in \mathcal{B}$.

For each $C\in \mathcal{C}\subset \mathcal{B}$

由上一条可知，在$B\in \mathcal{C}$时是可以证明$\mathcal{B}={\mathcal{B}}_B$的，则在$C\in \mathcal{C}$时是可以证明$\mathcal{B}={\mathcal{B}}_C$的，则 $B\in \mathcal{B}={\mathcal{B}}_C$

we have $B\in {\mathcal{B}}_C$, and because of the preceding, $B\cap C\in \mathcal{B}$, hence $C\in {\mathcal{B}}_B$, whence $\mathcal{C}\subset {\mathcal{B}}_B\subset \mathcal{B}$, hence $\mathcal{B}={\mathcal{B}}_B$

$$\begin{array}{lllcc}\begin{array}{l}B\in {\mathcal{B}}_C\\ \Downarrow \text{定义}\\ 1.B\in \mathcal{B}\\ 2.BC\in \mathcal{B}\end{array}& \Rightarrow & \begin{array}{l}C\in \mathcal{B}\\ CB\in \mathcal{B}\\ \Downarrow \text{定义}\\ C\in {\mathcal{B}}_B\end{array}& \Downarrow & \\ \mathcal{C}\subset {\mathcal{B}}_B\subset \mathcal{B}& & & & \\ \mathcal{B}={\mathcal{B}}_B& & & & \end{array}$$
即${\mathcal{B}}_B$ 跟 $B$ 无关，但是$B$ 必须在 $\mathcal{B}$ 里面

Since $\mathcal{B}={\mathcal{B}}_B$ for all $B\in \mathcal{B}$, we conclude $\mathcal{B}$ is closed by finite intersections.

$\because {\mathcal{B}}_B$ 对有限并封闭

$\forall A,B\in \mathcal{B}$

$\because B\in \mathcal{B}A\in \mathcal{B}={\mathcal{B}}_B$

$\Rightarrow AB\in \mathcal{B}$

即 $\mathcal{B}$ 对有限并封闭

可以由定义推导

Furthermore $\Omega \in \mathcal{B}$, and $\mathcal{B}$ is closed by difference, hence also under complementation.

Since $\mathcal{B}$ is closed by increasing limits as well, we conclude $\mathcal{B}$ is a $\sigma$-algebra, and it is clearly the smallest such containing $\mathcal{C}$.

证明：$\mathcal{B}$ 是 $\sigma$-代数

缺一个可列并封闭：

证明：

若 $A_i\in \mathcal{B}$

令 $B_n={\cup }_{i=1}^n{A}_i$，则 ${\cup }_{i=1}^n{A}_i={\cup }_{i=1}^n{B}_i$ ，且 $B_1\subset B_2\subset ...$

$B_n^c={\cap }_{i=1}^n{A}_i^c\in \mathcal{B}$

• $A_i\in \mathcal{B}$
• $A_i^c\in \mathcal{B}$
• $A_i^c$ 有限并 $\in \mathcal{B}$

$\Rightarrow B_n\in \mathcal{B}$

则 $\mathcal{B}$ 对可列并封闭

$\mathcal{B}$ 是 $\sigma$-代数，$\mathcal{B}\supset \mathcal{C}\Rightarrow \mathcal{B}\supset \sigma (\mathcal{C})$

由 $\sigma (\mathcal{C})\supset \mathcal{C}$ 且 $\sigma (\mathcal{C})$ 对单増并、差封闭

由 $\mathcal{B}$ 的最小性，可得 $\sigma (\mathcal{C})\supset \mathcal{B}$

则 $\mathcal{B}=\sigma (\mathcal{C})$

The proof of the uniqueness in Theorem 6.1 is an immediate consequence of the following Corollary 6.1, itself a consequence of the Monotone Class Theorem

Corollary 6.1 推论：概率延拓定理

Let $P$ and $Q$ be two probabilities defined on $\mathcal{A}$

Suppose $P$ and $Q$ agree on a class $\mathcal{C}\subset \mathcal{A}$ which is closed under finite intersections.

If $\sigma (\mathcal{C})=\mathcal{A}$, we have $P=Q$

这个定理说明：两个定义在对交封闭的 $\mathcal{C}$ 上相等的概率测度 $P,Q$ ， 可以将其延拓到 $\sigma (\mathcal{C})$ 上

Proof. We can assume $w.l.o.g.$ that $\Omega \in \mathcal{C}$ , since

• $\Omega \in \mathcal{A}$, because $\mathcal{A}$ is a $\sigma$-algebra
• $P(\Omega )=Q(\Omega )=1$, because they are both probabilities.

Let

B = { A ∈ A : P ( A ) = Q ( A ) } \mathcal{B} = \left\{ A \in \mathcal{A}:P(A)=Q(A) \right\} B={A∈A:P(A)=Q(A)}
By the definition of a Probability measure and Theorem 2.3, $\mathcal{B}$ is closed by difference and by increasing limits.

Also $\mathcal{B}$ contains $\mathcal{C}$ by hypothesis.

Therefore since $\sigma (\mathcal{C})=\mathcal{A}$, we have $\mathcal{B}=\mathcal{A}$ by the Monotone Class Theorem.

不妨设 $\Omega \in \mathcal{C}$ （ 若不在里面，定义 C ′ = C ∪ { Ω } \mathcal{C}'=\mathcal{C} \cup \left\{ \Omega \right\} C′=C∪{Ω} ）

B = { A ∈ A : P ( A ) = Q ( A ) } \mathcal{B}= \left\{ A \in \mathcal{A}:P(A)=Q(A) \right\} B={A∈A:P(A)=Q(A)}

证明： $\mathcal{B}=\mathcal{A}$ ，则在 $\mathcal{A}$ 上 $P(A)=Q(A)$

$\because$ $\mathcal{C}$ 上 $P(A)=Q(A)$ $\therefore \mathcal{C}\subset \mathcal{B}$

若 $A_1\subset A_2\subset ...A_i\in \mathcal{B}$

则 ${\cup }_{i=1}^{\infty }{A}_i$

$$P\left({\cup }_{i=1}^{\infty }{A}_i\right)=\underset{n\to \infty }{\lim }P(A_n)=\underset{n\to \infty }{\lim }Q(A_n)=Q\left({\cup }_{i=1}^{\infty }{A}_i\right)$$

$\therefore {\cup }_{i=1}^{\infty }{A}_i\in \mathcal{B}$

对差也一样

又 $\mathcal{B}\supset \mathcal{C}$ 且 $\mathcal{B}$ 对单増并、差封闭

$\therefore \sigma (\mathcal{C})=\mathcal{A}\subset \mathcal{B}\subset \mathcal{A}$

$\therefore \mathcal{A}=\mathcal{B}$

Definition 6.2

Let $P$ be a probability on $\mathcal{A}$.

A null set (or negligible set) for $P$ is a subset $A$ of $\Omega$ such that there exists a $B\in \mathcal{A}$ satisfying $A\subset B$ and $P(B)=0$

可忽略的集合

$\mathcal{A}$ 是一个概率测度

因为 概率一定满足规范性和可列可加性

$A$ 不一定在 $\mathcal{A}$ 上

Remark

• We say that a property holds almost surely ($a.s.$ in short) if it holds outside a negligible set.

• This notion clearly depends on the probability, so we say sometimes P-almost surely , or P-a.s.

  几乎处处收敛

  $$\begin{array}{rl}{\xi }_n{\to }_{a.s.}\xi & \iff \exists A\subset B\\ & s.t.B\subset \mathcal{A}\text{ and }P(B)=0\\ & li{m}_{n\to \infty }{\xi }_n(\omega )=\xi \end{array}$$

  $r$ 收敛

  $${\xi }_n{\to }_r\xi \iff E|{\xi }_n-\xi {|}^r\to 0n\to \infty$$

Remark

• The negligible set are not necessarily in $\mathcal{A}$
• Nevertheless it is natural to say that they have probability zero.
• In the following theorem, we extend the probability to the $\sigma$-algebra which is generated by $\mathcal{A}$ and all $P$-negligible sets.

目标：将概率 $P$ 延拓到 A ∪ { all P -negligible } \mathcal{A} \cup \left\{ \text{all} P \text{-negligible} \right\} A∪{allP-negligible}

Theorem 6.4

Let $P$ be a probability on $\mathcal{A}$ and let $\mathcal{N}$ be the class of all $P$-negligible sets.

Then

A ′ = { A ∪ N : A ∈ A , N ∈ N } \mathcal{A}' = \left\{ A \cup N : A \in \mathcal{A}, N \in \mathcal{N} \right\} A′={A∪N:A∈A,N∈N}

is a $\sigma$-algebra, called the $P$-completion of $\mathcal{A}$

• ${\mathcal{A}}^{\prime }$ is the smallest $\sigma$-algebra containing $\mathcal{A}$ and $\mathcal{N}$

• $P$ extends uniquely as a probability (still denoted by $P$) on ${\mathcal{A}}^{\prime }$,

  by setting $P(A\cup N)=P(A)$ for $A\in \mathcal{A}$ and $N\in \mathcal{N}$

Proof.

由 Corollary 6.1 可知，存在 $\mathcal{A}\cup \mathcal{N}$

$\forall A\in \mathcal{A}\cup \mathcal{N}\text{ 即 }A\in \mathcal{A}orA\in \mathcal{N}$

下面证明 $\mathcal{A}\cup \mathcal{N}$ 对有限交封闭

1. 若 $A\in \mathcal{A},B\in \mathcal{A}\Rightarrow AB\in \mathcal{A}\subset \mathcal{A}\cup \mathcal{N}$
2. 若 $A\in \mathcal{A},B\in \mathcal{N}AB\in \mathcal{N}\subset \mathcal{A}\cup \mathcal{N}$

$$\begin{array}{rl}\because & B\in \mathcal{N}\to \exists C\in \mathcal{A}s.t.B\subset CandP(C)=0\\ \therefore & AB\subset B\subset C\\ \Rightarrow & AB\in \mathcal{N}\end{array}$$

3. 若 $A\in \mathcal{N},B\in \mathcal{N}\Rightarrow AB\in \mathcal{N}\subset \mathcal{A}\cup \mathcal{N}$

由 123 说明 $\mathcal{A}\cup \mathcal{N}$ 对有限交封闭

证明 $\mathcal{A}\cup \mathcal{N}\subset {\mathcal{A}}^{\prime }$

$\forall A\in \mathcal{A}\cup \mathcal{N}$ ， 即 $A\in \mathcal{A}orA\in \mathcal{N}$

$$\begin{array}{l}\because \varnothing \in \mathcal{A}and\varnothing \in \mathcal{N}\\ \forall B\in \mathcal{A},B=B\cup \varnothing \in \mathcal{A}\Rightarrow \mathcal{A}\subset {\mathcal{A}}^{\prime }\\ \forall N\in \mathcal{N},N=N\cup \varnothing \in {\mathcal{A}}^{\prime }\Rightarrow \mathcal{N}\subset {\mathcal{A}}^{\prime }\end{array}\}\Rightarrow \mathcal{A}\cup \mathcal{N}\subset {\mathcal{A}}^{\prime }$$

证明${\mathcal{A}}^{\prime }=\sigma (\mathcal{A}\cup \mathcal{N})$

Denote $\sigma (\mathcal{A}\cup \mathcal{N})={\mathcal{A}}^{\prime \prime }$

$$\begin{array}{c}\begin{array}{lll}\begin{array}{l}1.\\ \mathcal{A}\subset {\mathcal{A}}^{\prime }and\mathcal{N}\subset {\mathcal{A}}^{\prime }\text{已证}\\ \Updownarrow \\ \mathcal{A}\cup \mathcal{N}\subset {\mathcal{A}}^{\prime }\\ \Rightarrow {\mathcal{A}}^{\prime \prime }\subset {\mathcal{A}}^{\prime }\end{array}& & \begin{array}{l}2.\\ \forall B\in {\mathcal{A}}^{\prime }B=A\cup N,A\in \mathcal{A},N\in \mathcal{N}\\ \because {\mathcal{A}}^{\prime \prime }\text{ is σ-algebra}\\ \text{ and }A\in \mathcal{A}\subset {\mathcal{A}}^{\prime \prime },N\in \mathcal{A}\subset {\mathcal{A}}^{\prime \prime }\\ B=A\cup N\in {\mathcal{A}}^{\prime \prime }\Rightarrow {\mathcal{A}}^{\prime }\subset {\mathcal{A}}^{\prime \prime }\end{array}\\ & & \end{array}\\ \Downarrow \\ {\mathcal{A}}^{\prime }={\mathcal{A}}^{\prime \prime }=\sigma (\mathcal{A}\cup \mathcal{N})\end{array}$$

证明 ${\mathcal{A}}^{\prime }$ 是一个 $\sigma$-algebra

1. $\Omega \in {\mathcal{A}}^{\prime }\Omega =\Omega \cup \varnothing ,\Omega \in \mathcal{A},\varnothing \in \mathcal{N}$

2. 若 $B\in {\mathcal{A}}^{\prime }$ ，则 $\exists A\in \mathcal{A},N\in \mathcal{N}$

$$\begin{array}{rl}s.t.& \\ B& =A\cup N\\ B^C& =A^C\cap N^C=(A^C\cap C^C)\cup (A^C\cap N^C\cap C)\\ & (A^C\cap C^C)\in \mathcal{A}\\ & (A^C\cap N^C\cap C)\subset C\in \mathcal{N}\\ \Downarrow & \\ B^C& \in {\mathcal{A}}^{\prime }\end{array}$$

3. 可列可加

   $B_n\in {\mathcal{A}}^{\prime },n=1,2,...$

   $N_n\subset C_n\in \mathcal{A},P(C_n)=0$

   $P({\cup }_{n=1}^{\infty }{C}_n)\le {\sum }_{n=1}^{\infty }P(C_n)=0$

$$\begin{array}{rl}{B}_n& =A_n\cup N_n{A}_n\in \mathcal{A},N_n\in \mathcal{N}\\ {\cup }_{n=1}^{\infty }{B}_n& =\left({\cup }_{n=1}^{\infty }{A}_n\right)\cup \left({\cup }_{n=1}^{\infty }{N}_n\right)\in {\mathcal{A}}^{\prime }\\ & {\cup }_{n=1}^{\infty }{A}_n\in \mathcal{A}\\ & {\cup }_{n=1}^{\infty }{N}_n\in \mathcal{N}\end{array}$$

则 ${\mathcal{A}}^{\prime }$ 是一个 $\sigma$-algebra ， ${\mathcal{A}}^{\prime }=\sigma (\mathcal{A}\cup \mathcal{N})$

由 corollary 6.1 知, The uniqueness of the extension is straightforward.

现在想知道，若 $\mathcal{A}$ 上有不一样的分解，其概率是否会于分解有关？

Suppose now that $A_1\cup N_1=A_2\cup N_2$ with $A_i\in \mathcal{A}$ and $N_i\in \mathcal{N}$

The symmetrical difference $A_1△A_2=(A_1\cap A_2^c)\cup (A_1^c\cap A_2)$ is contained in $N_1\cup N_2$

Proof.

$$\begin{array}{rl}{A}_1{A}_2^c& \subset (A_1\cup N_1)A_2^c=(A_2\cup N_2)A_2^c=A_2{A}_2^c\cup N_2{A}_2^c=N_2{A}_2^c\subset N_2\\ A_1^c{A}_2& \subset A_1^c(A_2\cup N_2)=A_1^c(A_1\cup N_1)=A_1^c{A}_1\cup A_1^c{N}_1=A_1^c{N}_1\subset N_1\\ \therefore & A_1{A}_2^c\cup A_1^c{A}_2\subset N_1\cup N_2\end{array}$$

$\exists C_i\in \mathcal{A}P(C_i)=0,N_i\subset C_i,i=1,2,...A_i\in \mathcal{A}$

$A_1△A_2\subset N_1\cup N_2\subset C_1\cup C_2\Rightarrow A_1△A_2\in \mathcal{N},P(C_1\cup C_2)\le P(C_1)+P(C_2)=0$

$$\begin{array}{rl}P(A_1)& =P(A_1{A}_2+A_1{A}_2^c)\\ & =P(A_1{A}_2)+P(A_1{A}_2^c)\\ & =P(A_1{A}_2)\\ & A_1{A}_2^c\subset A_1△A_2\subset N_1\cup N_2\subset C_1\cup C_2\Rightarrow P(A_1{A}_2^c)\le P(C_1\cup C_2)=0\\ P(A_1)& =P(A_1{A}_2)\\ P(A_2)& =P(A_1{A}_2)\\ & \Rightarrow P(A_1)=P(A_2)\\ & \Rightarrow Q(A_1\cup N_1)=Q(A_2\cup N_2)\end{array}$$

则不管怎么分解，延拓的概率都相等

现在证明 $Q$ 是概率，概率需要验证两条：

1. $Q(\Omega )=Q(\Omega \cup \varnothing )=P(\Omega )=1$
2. 假设 $B_n\in {\mathcal{A}}^{\prime }$ 且 $B_n$ 两两互斥 $n=1,2,...$

Q ( ∪ i = 1 ∞ B n ) = Q { ∪ i = 1 ∞ ( A n ∪ N n } = Q { ∪ i = 1 ∞ A n ∪ ∪ i = 1 ∞ N n } = P { ∪ i = 1 ∞ A n }   A n  是 B n  的一部分，则  B n  两两互斥得到  A n  两两互斥 = ∑ n = 1 ∞ P { A n } Q ( B n ) = Q ( A n ∪ N n ) = P ( A n ) ∴   Q ( ∪ i = 1 ∞ B n ) = ∑ n = 1 ∞ Q ( B n ) \begin{aligned} Q( \cup_{i=1}^{ \infty} B_n) &= Q \left\{ \cup_{i=1}^{ \infty} (A_n \cup N_n \right\} \\ &= Q \left\{ \cup_{i=1}^{ \infty} A_n \cup \cup_{i=1}^{ \infty} N_n \right\} \\ &= P \left\{ \cup_{i=1}^{ \infty} A_n \right\} \\ & \qquad \text{ $A_n$ 是$B_n$ 的一部分，则 $B_n$ 两两互斥得到 $A_n$ 两两互斥}\\ &= \sum_{n=1}^{\infty} P \left\{ A_n \right\} \\ \\ Q(B_n) &= Q(A_n \cup N_n) = P(A_n) \\ \therefore & \ Q( \cup_{i=1}^{ \infty} B_n ) = \sum_{n=1}^{\infty} Q(B_n) \\ \end{aligned} Q(∪i=1∞​Bn​)Q(Bn​)∴​=Q{∪i=1∞​(An​∪Nn​}=Q{∪i=1∞​An​∪∪i=1∞​Nn​}=P{∪i=1∞​An​} An​ 是Bn​ 的一部分，则 Bn​ 两两互斥得到 An​ 两两互斥=n=1∑∞​P{An​}=Q(An​∪Nn​)=P(An​) Q(∪i=1∞​Bn​)=n=1∑∞​Q(Bn​)​

则 $Q$ 是 ${\mathcal{A}}^{\prime }$ 上的概率测度，仍然记 $Q$ 为 $P$

定理6.4证明完毕

回顾一下定理6.1以及其证明过程：

定理6.1： 概率 $P$ 是 $\mathcal{A}$ 上的概率， $\mathcal{N}$ 是所有的可忽略集合类

定义 A ′ = { A ∪ N : A ∈ A , N ∈ N } \mathcal{A}'= \left\{ A \cup N: A \in \mathcal{A}, N \in \mathcal{N} \right\} A′={A∪N:A∈A,N∈N}

则可以得到：

• ${\mathcal{A}}^{\prime }$ 是一个 $\sigma$-代数
• ${\mathcal{A}}^{\prime }=\sigma (\mathcal{A}\cup \mathcal{N})$
• 概率 $P$ （原来定义在 $\mathcal{A}$ 上）可在 ${\mathcal{A}}^{\prime }$ 上唯一延拓

定理6.1把概率又往前延拓，现在是可以延拓到一个 $\sigma$-代数和一个可忽略集的并集构成的 $\sigma$-代数上了。

定理的证明需要证明几步：

1. ${\mathcal{A}}^{\prime }$ 是一个 $\sigma$-代数 $\{\begin{array}{l}\varnothing \Omega \text{在里面}\\ A\text{在}{\mathcal{A}}^{\prime }\text{里面，则 }{A}^c\text{也在}{\mathcal{A}}^{\prime }\text{里面}\\ 对可列并封闭\end{array}$

2. 证明 ${\mathcal{A}}^{\prime }=\sigma (\mathcal{A}\cup \mathcal{N})$ ，由两个包含关系可以证明

   其中， 证明 $\mathcal{A}\in \mathcal{N}\subset {\mathcal{A}}^{\prime }$ 就是为了证明这一步

3. 证明概率 $P$ 延拓到 ${\mathcal{A}}^{\prime }$ 上之后（记为 $Q$ )仍然是一个概率（由概率的定义来证明）

4. 证明唯一性，延拓之后的概率是唯一的，则仍然可以记为 $P$.

   唯一性的证明由性质6.1可以得到，需要证明

   • $\mathcal{A}\cup \mathcal{N}$ 对有限交封闭

   • ${\mathcal{A}}^{\prime }=\sigma (\mathcal{A}\cup \mathcal{N})$

   • 概率与分解无关

     性质6.1： 概率 $P$ 和 概率 $Q$ 是定义在$\mathcal{A}$ 上的两个概率，若 $P,Q$ 在 $\mathcal{C}\subset \mathcal{A}$ 上相等，并且对有限交封闭，如果 $\sigma (\mathcal{C})=\mathcal{A}$ ，则 $P=Q$