### 统计代写|随机过程作业代写stochastic process代考|Special Chains and Foster Type Theorems

statistics-lab™ 为您的留学生涯保驾护航 在代写 随机过程stochastic process方面已经树立了自己的口碑, 保证靠谱, 高质且原创的统计Statistics代写服务。我们的专家在代写随机过程stochastic process方面经验极为丰富，各种代写 随机过程stochastic process相关的作业也就用不着说。

• Statistical Inference 统计推断
• Statistical Computing 统计计算
• Advanced Probability Theory 高等楖率论
• Advanced Mathematical Statistics 高等数理统计学
• (Generalized) Linear Models 广义线性模型
• Statistical Machine Learning 统计机器学习
• Longitudinal Data Analysis 纵向数据分析
• Foundations of Data Science 数据科学基础

## 统计代写|随机过程作业代写stochastic process代考|Special chains

If the Markov Chain is infinite, the number of equations given by $\pi(P-I)=0$ will be infinite involving an infinite number of unknowns. In some particular cases we can solve these equations. The following examples will illustrate this point.
Example $2.5 \quad$ Birth and-Death Chain (Non-Homogeneous Random Walk) Consider a birth and death chain on ${0,1,2, \ldots, d}$ or a set of non-negative integers i.e. where $d=\infty$. Assume that the chain is irreducible i.e. $p_{j}>0$ and $q_{j}>0$ in case $0 \leq j \leq d$ (i.e. when $d$ is finite) $p_{j}>0$ for $0 \leq j<\infty$ and $q_{j}>0$ for $0<j<\infty$ if $d$ is infinite. Consider the transition matrix

$$X=\left(x_{0}, x_{1}, x_{2}, \ldots\right)=\left(x_{0}, x_{1}, x_{2}, \ldots\right)\left(\begin{array}{ccccc} 0 & 1 & 0 & 0 & \cdots \ q_{1} & 0 & p_{1} & 0 & \cdots \ 0 & q_{2} & 0 & p_{2} & \ldots \ & \cdots & \cdots & & \end{array}\right)$$
or $X=X P$. Let $x_{0} \neq 0$. Then
\begin{aligned} &x_{0}=x_{1} q_{1}, \ &x_{1}=x_{0}+x_{2} q_{2}, \ &x_{3}=x_{2} p_{2}+x_{4} q_{4}, \ &x_{4}=\ldots \ &\ldots \ &y_{i}=\frac{x_{i}}{x_{0}}, y_{0}=1, i=1,2,3, \ldots \end{aligned}
Define
Then
\begin{aligned} &y_{1}=1 / q_{1}, y_{1}=1+y_{2} q_{2} \text { or } y_{2}=\frac{y_{1}-1}{q_{2}}=\frac{1-q_{1}}{q_{1} q_{2}}=\frac{p_{1}}{q_{1} q_{2}} \ &y_{3}=\frac{p_{1} p_{2}}{q_{1} q_{2} q_{3}}, \ldots, y_{n}=\frac{p_{1} p_{2} \ldots p_{n-1}}{q_{1} q_{2} \ldots q_{n}}>0 \quad \text { for all } n=1,2, \ldots \end{aligned}
(by assumption that all $p, q$ ‘s are $>0$ ).
By Theorems $2.9$ and $2.11$, the non-homogeneous random walk is positive recurrent if $0<\sum_{0}^{\infty} x_{i}<\infty$ i.e. $\sum_{1}^{\infty} y_{i}<\infty$ i.e. iff $$\sum_{n=1}^{\infty} \frac{p_{1} p_{2} \cdots p_{n-1}}{q_{1} q_{2} \cdots q_{n}}<\infty$$ Note that $\sum_{0}^{\infty} y_{i}=\sum_{0}^{\infty} x_{i} / x_{0}=1 / x_{0}$, since $\sum_{0}^{\infty} y_{i}=1$. Therefore $x_{n}=y_{r} x_{0}=y_{n} / \sum_{0}^{\infty} y_{i}$ gives the stationary distribution provided $\sum_{0}^{\infty} y_{\mathrm{i}}<\infty \cdot x_{0}$ still has to be determined. Now $$1=x_{0}+\sum_{i=1}^{\infty} x_{i}=x_{0}+\sum_{i=1}^{\infty} x_{0} y_{i} \Rightarrow x_{0}=\frac{1}{1+\sum_{i=1}^{\infty} y_{i}}$$ and so $x_{0}>0$, iff $\sum_{i=1}^{\infty} y_{i}<\infty$ i.e. iff $\sum_{i=1}^{\infty} y_{i}<\infty\left(\right.$ since $\left.y_{0}=1\right)$.
In fact if $\sum_{i} y_{i}=\infty$, the solution to $(2.18)$ is either identically zero or has infinite sum $\left(\sum_{i} x_{i}=\infty\right)$ and hence has no stationary distribution.

## 统计代写|随机过程作业代写stochastic process代考|Foster type theorems

The following theorems, associated with Foster, give criteria for transient and recurrent chains in terms of solution of certain equations. Assume that the M.C. is irreducible.

Theorem 2.11 (Foster, 1953) Let the Markov chain be irreducible. Assume that there exists $x_{k}, k \in S$ such that $x_{k}=\sum_{k \in S} x_{i} p_{i k}$ and $0<\sum_{k \in S}\left|x_{k}\right|<\infty$. Then the Markov Chain is positive recurrent (this is a sort of converse of Theorem $2.9$ ). Proof Since $y_{k}=\frac{1}{\sum_{k \in S}\left|x_{k}\right|}>0, \sum_{k \in S} y_{k}=1$.

Without loss of generality $\left{x_{k}, k \in S\right}$ is a stationary distribution of a M.C. Then

$$x_{k}=\sum_{k \in S} x_{i} p_{i k}^{(n)} \text { for all } n=1,2, \ldots$$
Suppose that there is no positive state.
Since the M.C. is irreducible, then all the states are either transient or null. In that case $p_{i k}^{(n)} \rightarrow 0$ as $n \rightarrow \infty$ for all $i, k \in S$. By Lebesgue Dominated Convergence Theorem, taking $n \rightarrow \infty$ in (2.19)
$$x_{k}=\sum_{i \in S}\left(x_{i}\right) .0=0 \text { for all } k \in S$$
But $0<\sum_{k \in S} x_{k}<\infty$ is a contradiction to $(2.20)$.
Hence, there is at least one positive recurrent state. Since M.C. is irreducible, by Solidarity Theorem the M.C. must be positive recurrent. Conclusion An ireducible aperiodic M.C. has a stationary distribution iff all states are positive recurrent.

Theorem 2.11(a) If the M.C. is positive recurrent the system of equations $x_{i}=\sum_{j=0}^{\infty} x_{j} p_{j i}$ has a solution such that $0<\sum_{j=0}^{\infty} x_{j}<\infty$.
(Proof may be found in Karlin and Taylor’s book.)
Theorem 2.12 The M.C. is transient iff $x_{i}=\sum_{j=0}^{\infty} p_{i j} x_{j}$ has a solution for $i \neq 0$, which is bounded and non-constant i.e. all $x_{i}$ ‘s are not equal.

Theorem $2.13$ The M.C. is positive recurrent if $x_{i} \geq \sum_{j=0}^{\infty} p_{i j} x_{j}$ has a solution such that $x_{i} \rightarrow \infty$ as $i \rightarrow \infty$ (see Chung’s book on Markov Chains with Stationary Transition Probabilities).

## 统计代写|随机过程作业代写stochastic process代考|Theorems Regarding Finite Markov Chain

Theorem 2(a). In a M.C. with a finite number of states, there is no null state and not all states can be transient.

Proof Suppose the chain has $N<\infty$ states. If all states are transient, then letting $n \rightarrow \infty$ in the relation $\sum_{j=0}^{N} p_{i j}^{(n)}=1$ we get $0=1$ (since by Theorem $2.8$, $\lim {n \rightarrow \infty} p{i j}^{(n)}=0$ for each $j$, which is absured and hence not all states in a finite M.C. are transient. Consider the subchain $C_{1}$ formed by a closed set of null recurrent states. Then $\sum_{j \in C_{1}} p_{i j}^{(n)}=\alpha$ (say) $>0$. Letting $n \rightarrow \infty, 0=\alpha>0$ which is also absurd. So there cannot be any null recurrent state in a finite M.C.
Theorem 2(b). An irreducible M.C. having a finite number of states is positive recurrent.

Proof By previous theorem, there is no null recurrent state and not all states are

transient. Suppose there is one transient state. Then all states are transient by Solidarity Theorem. Hence, all states are positive recurrent.

Exercise $2.6$ If a finite M.C. is irreducible, aperiodic and has doubly stochastic transition matrix, then show that $\lim {n \rightarrow \infty} p{i j}^{(n)}=1 / k$, where $k$ is the number of states in the chain.

Solution If $j$ is a positive recurrent state in an aperiodic irreducible chain then $p_{i j}^{(n)} \rightarrow \pi_{j}>0($ by Theorem $2.9)$
Hence $1=\sum_{i=1}^{k} p_{i j}^{(n)}$ for all $j$ and $n \geq 1$,
Therefore $k \pi_{j}=1 \Rightarrow \pi_{j}=\frac{1}{k}$.

## 统计代写|随机过程作业代写stochastic process代考|Special chains

X0=X1q1, X1=X0+X2q2, X3=X2p2+X4q4, X4=… … 是一世=X一世X0,是0=1,一世=1,2,3,…

（假设所有p,q是>0）。

## 统计代写|随机过程作业代写stochastic process代考|Foster type theorems

Xķ=∑一世∈小号(X一世).0=0 对全部 ķ∈小号

（证明可以在 Karlin 和 Taylor 的书中找到。）

## 广义线性模型代考

statistics-lab作为专业的留学生服务机构，多年来已为美国、英国、加拿大、澳洲等留学热门地的学生提供专业的学术服务，包括但不限于Essay代写，Assignment代写，Dissertation代写，Report代写，小组作业代写，Proposal代写，Paper代写，Presentation代写，计算机作业代写，论文修改和润色，网课代做，exam代考等等。写作范围涵盖高中，本科，研究生等海外留学全阶段，辐射金融，经济学，会计学，审计学，管理学等全球99%专业科目。写作团队既有专业英语母语作者，也有海外名校硕博留学生，每位写作老师都拥有过硬的语言能力，专业的学科背景和学术写作经验。我们承诺100%原创，100%专业，100%准时，100%满意。

## MATLAB代写

MATLAB 是一种用于技术计算的高性能语言。它将计算、可视化和编程集成在一个易于使用的环境中，其中问题和解决方案以熟悉的数学符号表示。典型用途包括：数学和计算算法开发建模、仿真和原型制作数据分析、探索和可视化科学和工程图形应用程序开发，包括图形用户界面构建MATLAB 是一个交互式系统，其基本数据元素是一个不需要维度的数组。这使您可以解决许多技术计算问题，尤其是那些具有矩阵和向量公式的问题，而只需用 C 或 Fortran 等标量非交互式语言编写程序所需的时间的一小部分。MATLAB 名称代表矩阵实验室。MATLAB 最初的编写目的是提供对由 LINPACK 和 EISPACK 项目开发的矩阵软件的轻松访问，这两个项目共同代表了矩阵计算软件的最新技术。MATLAB 经过多年的发展，得到了许多用户的投入。在大学环境中，它是数学、工程和科学入门和高级课程的标准教学工具。在工业领域，MATLAB 是高效研究、开发和分析的首选工具。MATLAB 具有一系列称为工具箱的特定于应用程序的解决方案。对于大多数 MATLAB 用户来说非常重要，工具箱允许您学习应用专业技术。工具箱是 MATLAB 函数（M 文件）的综合集合，可扩展 MATLAB 环境以解决特定类别的问题。可用工具箱的领域包括信号处理、控制系统、神经网络、模糊逻辑、小波、仿真等。