# Notes on Probability Essentials - 1 - Axioms of Probability

403人阅读 评论(0)

“他说 / 你任何为人称道的美丽 / 不及他第一次遇见你。”
——《南山南》

Abstract：这里想谈一下直觉性的理解，为什么概率测度要与σ$\sigma$-algebra扯上关系呢？就概率事件而言，空集和全集肯定都得在A$\mathcal{A}$里头，这刚好是下文的性质1；任何一个事件必然有它的反面存在，这就是性质2——任何一个事件的补集也必须在A$\mathcal{A}$里头；性质4也是满足了概率的研究条件——因为概率测度(probability measure)的建立需要countable additivity, 那么显然这性质4（countable union and intersection）就是需要有的。所以σ$\sigma$-algebra就像是为概率测度量身打造的一种代数结构。

Let Ω$\Omega$ be an abstract space. Let 2Ω$2^\Omega$ denote all subsets of Ω$\Omega$, including the empty set denoted by $\emptyset$. With A$\mathcal{A}$ being a subset of 2Ω$2^\Omega$, we consider the following properties:

1. A$\emptyset\in\mathcal{A}$ and ΩA$\Omega\in\mathcal{A}$;
2. If AA$A\in\mathcal{A}$ then AcA$A^c\in\mathcal{A}$, where Ac$A^c$ denotes the complement of A;
3. A$\mathcal{A}$ is closed under finite unions and finite intersections: that is, if A1,...,An$A_1,...,A_n$ are all in A$\mathcal{A}$, then ni=1Ai$\bigcup_{i=1}^n{A_i}$ and ni=1Ai$\bigcap_{i=1}^n{A_i}$ are in A$\mathcal{A}$ as well;
4. A$\mathcal{A}$ is closed under countable unions and intersections: that is, if A1,A2,A3,...$A_1,A_2,A_3,...$ is a countable sequence of events in A$\mathcal{A}$, then ni=1Ai$\bigcup_{i=1}^n{A_i}$ and ni=1Ai$\bigcap_{i=1}^n{A_i}$ are both also in A$\mathcal{A}$.

Definition 1 A$\mathcal{A}$ is an algebra if it satisfies (1), (2) and (3) above. It is a σ$\sigma$-algebra (or a σ$\sigma$-field) if it satisfies (1),(2), and (4) above.
Note: (1)+(4) implies (3), hence any σ$\sigma$-algebra is an algebra (Exercise: Suppose that Ω$\Omega$ is an infinite set, countable or not, and let A$\mathcal{A}$ be the family of all subsets which are either finite or have a finite complement. Show that A$\mathcal{A}$ is an algebra, but not a σ$\sigma$-algebra).

Definition 2 If C2Ω$\mathcal{C}\subset 2^\Omega$, the σ$\sigma$-algebra generated by C$\mathcal{C}$, and written σ(C)$\sigma(\mathcal{C})$, is the smallest σ$\sigma$-algebra containing C$\mathcal{C}$.

Example:
1. A={,Ω}$\mathcal{A}=\{\emptyset,\Omega\}$ (the trivial σalgebra$\sigma-algebra$)
2. AΩ$A\subset\Omega$, then σ(A)={,A,Ac,Ω}$\sigma(A)=\{\emptyset,A,A^c,\Omega\}$
3. If Ω=R$\Omega=\mathbb{R}$, the Borel σalgebra$\sigma-algebra$ is the σalgebra$\sigma-algebra$ generated by the open sets.

Theorem 1 The Borel σ$\sigma$-algebra of R$\mathbb{R}$ is generated by intervals of the form (,a]$(-\infty,a]$, where aQ$a\in\mathbb{Q}$.

Proof: Let C$\mathcal{C}$ denote all open intervals. Since every open set in R$\mathbb{R}$ is the countable union of open intervals, we have σ(C)=$\sigma(\mathcal{C})=$ the Borel σ$\sigma$-algebra of R$\mathbb{R}$.
Let D$\mathcal{D}$ denote all intervals of the form (,a]$(-\infty,a]$, where aQ$a\in\mathbb{Q}$. Let (a,b)C$(a,b)\in\mathcal{C}$, and let (an)n1$(a_n)_{n\ge 1}$ be a sequence of rationals decreasing to a$a$ and (bn)n1$(b_n)_{n\ge 1}$ be a sequence of rationals increasing strictly to b$b$. Then

(a,b)=n=1(an,bn]=n=1((,bn](,an]c)

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

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

Definition 3 A probability measure defined on a σ$\sigma$-algebra A$\mathcal{A}$ of Ω$\Omega$ is a function P:A[0,1]$P:\mathcal{A}\rightarrow[0,1]$ that satisfies:
1. P(Ω)=1$P(\Omega)=1$
2. For every countable sequence (An)n1$(A_n)_{n\ge 1}$ of elements of A$\mathcal{A}$, pairwise disjoint(that is, AnAm=$A_n\cap A_m=\emptyset$ whenever nm$n\ne m$), one has

P(n=1An)=n=1P(An)

Note: Axiom (2) is called countable additivity, the number P(A)$P(A)$ is called the probability of the event A$A$.

Theorem 2 If P$P$ is a probability measure on (Ω,A)$(\Omega,\mathcal{A})$, then:
1. P()=0$P(\emptyset)=0$
2. P is finite additive.

Theorem 3 Let A$\mathcal{A}$ be a σ$\sigma$-algebra. Suppose that P:A[0,1]$P:\mathcal{A}\rightarrow [0,1]$ satisfies (1) and is additive. Then the following are equivalent:
1. Axiom(2) of Definition 3 (Countable Additivity)
2. If AnA$A_n\in\mathcal{A}$ and An$A_n\downarrow\emptyset$, then P(An)0$P(A_n)\downarrow 0$.
3. If AnA$A_n\in\mathcal{A}$ and AnA$A_n\downarrow A$, then P(An)P(A)$P(A_n)\downarrow P(A)$.
4. If AnA$A_n\in\mathcal{A}$ and AnΩ$A_n\uparrow\Omega$, then P(An)1$P(A_n)\uparrow 1$.
5. If AnA$A_n\in\mathcal{A}$ and AnA$A_n\uparrow A$, then P(An)P(A)$P(A_n)\uparrow P(A)$.

Note：这里的Notation AnA$A_n\uparrow A$是指AnAn+1$A_n\subset A_{n+1}$ and n=1An=A$\cup_{n=1}^\infty A_n=A$; AnA$A_n\downarrow A$是指An+1An$A_{n+1}\subset A_{n}$ and n=1An=A$\cap_{n=1}^\infty A_n=A$

Theorem 4 Let P$P$ be a probability measure, and let An$A_n$ be a sequence of events in A$\mathcal{A}$ which converges to A$A$. Then AA$A\in\mathcal{A}$ and limnP(An)=P(A)$\lim_{n\rightarrow\infty}P(A_n)=P(A)$.

Proof: Let us define

lim supnAn=n=1mnAmlim infnAn=n=1mnAm

Since A$\mathcal{A}$ is a σ$\sigma$-algebra, we have lim supnAnA$\limsup_{n\rightarrow\infty}A_n\in\mathcal{A}$ and lim infnAnA$\liminf_{n\rightarrow\infty}A_n\in\mathcal{A}$.
By hypothesis An$A_n$ converges to A$A$, which means limn1An=1A$\lim_{n\rightarrow\infty}1_{A_n}=1_A$, all ω$\omega$. This is equivalent to saying that A=lim supnAn=lim infnAn$A=\limsup_{n\rightarrow\infty}A_n=\liminf_{n\rightarrow\infty}A_n$. Therefore AA$A\in\mathcal{A}$.
Now let Bn=mnAm$B_n=\cap_{m\ge n}A_m$ and Cn=mnAm$C_n=\cup_{m\ge n}A_m$. Then limnP(Bn)=limnP(Cn)=P(A)$\lim_{n\rightarrow\infty}P(B_n)=\lim_{n\rightarrow\infty}P(C_n)=P(A)$, by Theorem 3. However BnAnCn$B_n\subset A_n\subset C_n$, therefore P(Bn)P(An)P(Cn)$P(B_n)\le P(A_n)\le P(C_n)$, so limnP(An)=P(A)$\lim_{n\rightarrow\infty}P(A_n)=P(A)$ as well.

Note: 这里通过引入lim sup$\limsup$lim inf$\liminf$，将An$A_n$转化成了一个递增与一个递减的序列。而后就可以直接利用Theorem 3证明结论了。

0
0

* 以上用户言论只代表其个人观点，不代表CSDN网站的观点或立场
个人资料
• 访问：22518次
• 积分：462
• 等级：
• 排名：千里之外
• 原创：20篇
• 转载：2篇
• 译文：0篇
• 评论：6条
阅读排行