← 返回数学题库
490概率困难derivationlong

Martingale Construction for Hitting Time on a Non-Symmetric Chain

题目

A Markov chain on {0,1,2,3}\{0, 1, 2, 3\} has transition probabilities: p(1,0)=13,p(1,2)=23,p(2,1)=12,p(2,3)=12.p(1, 0) = \tfrac{1}{3}, \quad p(1, 2) = \tfrac{2}{3}, \quad p(2, 1) = \tfrac{1}{2}, \quad p(2, 3) = \tfrac{1}{2}. State 00 is absorbing and state 33 is reflecting: p(3,2)=1p(3, 2) = 1.

Let T=inf{n0:Xn=0}T = \inf\{n \ge 0 : X_n = 0\}.

(a) Find a function f:{0,1,2,3}Rf : \{0,1,2,3\} \to \mathbb{R} and a constant c>0c > 0 such that Mn=f(XnT)(nT)cM_n = f(X_{n \wedge T}) - (n \wedge T) \cdot c is a martingale. Use this to compute E[TX0=2]E[T \mid X_0 = 2].

(b) Find a function g:{0,1,2,3}Rg : \{0,1,2,3\} \to \mathbb{R} such that Nn=g(XnT)(nT)dN_n = g(X_{n \wedge T}) - (n \wedge T) \cdot d is a martingale for an appropriate constant dd, and use the optional stopping theorem to compute E[T2X0=2]E[T^2 \mid X_0 = 2]. Hence find Var(TX0=2)\text{Var}(T \mid X_0 = 2).

解题计时

0:00

提交作答时记录,用于后续平均用时统计。

你的答案

E[T | X0=2]

Var(T | X0=2)