열렬히.뛰기

과제2

수학 & 통계 > 확률과정론 > 과제2

2018111373 최인렬

문제

{X_t} 는 포아송 과정.

  1. E(X_{10}~|~X_{11})
  2. Var(X_{11}|X_{10})
  3. Cov(X_1,~X_{10}|X_9)

1번

P(X_{10}=x_{10}~|~X_{11}=x_{11}) = {P(X_{10}=x_{10},~X_{11}=x_{11}) \over P(X_{11} = x_{11})} \\[15pt] ={P(X_{10}=x_{10,}~ X_{11}-X_{10}=x_{11}-x_{10}) \over P(X_{11} = x_{11})} \\[15pt] = {P(X_{10}=x_{10})~ P(X_{11}-X_{10}=x_{11}-x_{10}) \over P(X_{11} = x_{11})}
={\dfrac{e^{-10\lambda}~{(10\lambda)}^{x_{10}}}{(x_{10})!} \cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{11}-x_{10}}} {(x_{11}-x_{10})!} \over \dfrac{e^{-11\lambda}~ {(11\lambda)}^{x_{11}}} {(x_{11})!} }
= {x_{11}! \over x_{10}! ~(x_{11}-x_{10})!} ~\cdot~ {{(10\lambda)}^{x_{10}}~~ \lambda^{x_{11}-x_{10}} \over {(11\lambda)}^{x_{10}}~~ {(11\lambda)}^{x_{11}-x_{10}}} \\[10pt] = {x_{11}! \over x_{10}! ~(x_{11}-x_{10})!} ~\cdot~ \bigg(\dfrac{10}{11}\bigg)^{x_{10}} ~\bigg(\dfrac{1}{11}\bigg) ^{x_{11} - x_{10}}
\therefore~X_{10}~|~X_{11} = x_{11} \sim binominal \bigg(x_{11},~{10\over 11}\bigg) \\[10pt] E(X_{10}~|~X_{11}) = {10\over 11} \cdot x_{11}

2번

P(X_{11}~|~X_{10}) = {P(X_{10}=x_{10},~X_{11}=x_{11}) \over P(X_{10} = x_{10})} \\[15pt] ={P(X_{10}=x_{10,}~ X_{11}-X_{10}=x_{11}-x_{10}) \over P(X_{10} = x_{10})} \\[15pt] = {P(X_{10}=x_{10})~ P(X_{11}-X_{10}=x_{11}-x_{10}) \over P(X_{10} = x_{10})}
= P(X_{11}-X_{10}=x_{11}-x_{10}) = P(X_{1} = x_{11}-x_{10}) \\[15pt] \\[15pt] \therefore~ P(X_{11}~|~X_{10}) = \dfrac{e^{-\lambda}~ {\lambda}^{x_{11}-x_{10}}} {(x_{11}-x_{10})!}

3번

Cov(X,Y~|~Z) = E(XY| Z) - E(X | Z)~E(Y|Z) 를 이용한다.

Cov(X_{1},~X_{10}|X_{9}) = E(X_{1}X_{10}|X_{9}) - E(X_{1}|X_{9})~ E(X_{10}|X_{9})

E(X_{1}|X_{9})를 구해보자.

P(X_{1}=x_{1}~|~X_{9}=x_{9}) = {P(X_{1}=x_{1},~X_{9}=x_{9}) \over P(X_{9} = x_{9})} \\[15pt] ={P(X_{1}=x_{1}, ~X_{9}-X_{1}=x_{9}-x_{1}) \over P(X_{9} = x_{9})} \\[15pt] ={P(X_{1}=x_{1})~ P(X_{9}-X_{1}=x_{9}-x_{1}) \over P(X_{9} = x_{9})} \\[15pt] ={P(X_{1}=x_{1})~ P(X_{8}=x_{9}-x_{1}) \over P(X_{9} = x_{9})}
={\dfrac{e^{-\lambda}~{\lambda}^{x_{1}}}{(x_{1})!} \cdot \dfrac{e^{-8\lambda}~ {8\lambda}^{x_{9}-x_{1}}} {(x_{9}-x_{1})!} \over \dfrac{e^{-9\lambda}~ {(9\lambda)}^{x_{11}}} {(x_{9})!} }
= {x_{9}! \over x_{1}! ~(x_{9}-x_{1})!} ~\cdot~ {{(\lambda)}^{x_{1}}~~ {(8\lambda)}^{x_{9}-x_{1}} \over {(9\lambda)}^{x_{1}}~~ {(9\lambda)}^{x_{9}-x_{1}}} \\[10pt] = {x_{9}! \over x_{1}! ~(x_{9}-x_{1})!} ~\cdot~ \bigg(\dfrac{1}{9}\bigg)^{x_{1}} ~\bigg(\dfrac{8}{9}\bigg) ^{x_{9} - x_{1}}
\therefore~X_{1}~|~X_{9} = x_{9} \sim binominal \bigg(x_{9},~{1 \over 9}\bigg) \\[10pt] E(X_{1}~|~X_{9}) = {1 \over 9} \cdot x_{9}

이번에는 E(X_{10}|X_{9})를 구해보자.

P(X_{10}~|~X_{9}=x_{9}) ~=~ {P(X_{9}=x_{9},~X_{10}=x_{10}) \over P(X_{9} = x_{9})} \\[15pt] ={P(X_{9}=x_{9}, ~X_{10}-X_{9}=x_{10}-x_{9}) \over P(X_{9} = x_{9})} \\[15pt] ={P(X_{9}=x_{9})~ P(X_{10}-X_{9}=x_{10}-x_{9}) \over P(X_{9} = x_{9})}
= P(X_{1}=x_{10}-x_{9}) =~\dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!} \\[15pt] \therefore~X_{10}~|~X_{9}=x_9 \sim poisson(\lambda). \kern{10pt} E(X_{10}|X_{9}) = \lambda

마지막으로 E(X_{1}X_{10}|X_{9})를 구해보자.

\begin{align*} &P(X_1 = x_{1,}~ X_{10} = x_{10}~|~X_9 = x_9) \\[10pt] &= P(X_1 = x_{1,}~ X_{10} = x_{10,}~X_9 = x_9|~X_9 = x_9) \\[10pt] &= {P(X_{1}=x_{1}, ~X_{8}=x_{9}-x_{1}, ~X_{1}=x_{10}-x_{9}) \over P(X_{9} = x_{9})} \\[10pt] &={P(X_{1}=x_{1}) ~P(X_{8}=x_{9}-x_{1}) ~P(X_{1}=x_{10}-x_{9}) \over P(X_{9} = x_{9})} \\[20pt] &=~~{ \dfrac{e^{-\lambda} ~\lambda^{x_1}} {x_1!} \cdot \dfrac{e^{-8\lambda}~ {(8\lambda)}^{x_{9}-x_{1}}} {(x_{9}-x_{1})!} \cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!} \over \dfrac{e^{-9\lambda}~ {(9\lambda)}^{x_{9}}} {(x_{9})!} } \\[30pt] &= { \dfrac{e^{-\lambda} ~\lambda^{x_1}} {x_1!} \cdot \dfrac{e^{-8\lambda}~ {(8\lambda)}^{x_{9}-x_{1}}} {(x_{9}-x_{1})!} \over \dfrac{e^{-9\lambda}~ {(9\lambda)}^{x_{9}}} {(x_{9})!} } \cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!} \\[30pt] &= \dfrac{x_{9}!} {x_1!~(x_{9}-x_{1})!}~ \bigg({1\over9}\bigg)^{x_1} \bigg({8\over9}\bigg)^{x_9-x_1} \cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!} \end{align*}
\begin{align*} &E(X_1X_{10}|X_9) = \sum_{X_1=0}^{x_9}~ \sum_{X_{10}=x_{9}}^{\infty}~ x_1\cdot x_{10} \cdot P(X_1 = x_{1,}~ X_{10} = x_{10}~|~X_9 = x_9) \\[20pt] &= \sum_{X_1=0}^{x_9}~x_1 \cdot \dfrac{x_{9}!} {x_1!~(x_{9}-x_{1})!}~ \bigg({1\over9}\bigg)^{x_1} \bigg({8\over9}\bigg)^{x_9-x_1}~ \Bigg[ \sum_{X_{10}=x_{9}}^{\infty}~x_{10}\cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!} \Bigg] \\[30pt] &= E(X_1|X_9) \times \Bigg[ \sum_{X_{10}=x_{9}}^{\infty}~ (x_{10}-x_{9}+x_{9}) \cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!} \Bigg] \\[30pt] &= E(X_1|X_9) \times \Bigg[ \sum_{X_{10}=x_{9}}^{\infty} (x_{10}-x_9)\cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!} + \sum_{X_{10}=x_{9}}^{\infty}(x_{9})\cdot \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9})!}\Bigg] \\[30pt] &= {x_9 \over 9} \cdot \Bigg[ \sum_{X_{10}=x_{9}}^{\infty} \dfrac{e^{-\lambda}~ {\lambda}^{x_{10}-x_{9}}} {(x_{10}-x_{9}-1)!} + x_9\Bigg] \\[30pt] &= {x_9 \over 9} \cdot \Bigg[ \sum_{k=0}^{\infty} \dfrac{e^{-\lambda}~ {\lambda}^{k}} {(k-1)!} + x_9\Bigg] ~=~ {x_9 \over 9} \cdot \Bigg[ \sum_{k=1}^{\infty} \dfrac{e^{-\lambda}~ {\lambda}^{k}} {(k-1)!} + x_9\Bigg] \\[30pt] &= {x_9 \over 9} \cdot \big(\lambda + x_9 \big) \end{align*}

따라서 공분산을 구해보면 다음과 같다.

Cov(X_{1}X_{10}|X_{9}) = E(X_{1}X_{10}|X_{9}) - E(X_{1}|X_{9})~E(X_{10}|X_{9}) \\[15pt] = {8 \over 9} \cdot (x_{9}-x_{1}) - {1 \over 9} \cdot x_{9} \cdot \lambda