Convergence of Transition Probability Matrix in CLVMarkov Models

2nd International Conference on Mathematics, Science, Education and Technology

数学;自然科学;教育

Permana, D.^1 ; Pasaribu, U.S.^2 ; Indratno, S.W.^2 ; Suprayogi, S.^3

Study Programme in Statistics, Faculty of Mathematics and Natural Sciences, Universitas Negeri Padang, Indonesia^1

Statistics Research Division, Faculty of Mathematics and Natural Sciences, Institut Teknologi Bandung, Indonesia^2

Industrial System and Techno-Economics Research Group, Faculty of Industrial Engineering, Institut Teknologi Bandung, Indonesia^3

关键词: Diagonalizations;    Limiting method;    Markov chain models;    Matrix of transition probabilities;    Step transitions;    Transition matrices;    Transition probabilities;    Transition probability matrix;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/335/1/012046/pdf
DOI  :  10.1088/1757-899X/335/1/012046

来源: IOP

【 摘 要 】

A transition probability matrix is an arrangement of transition probability from one states to another in a Markov chain model (MCM). One of interesting study on the MCM is its behavior for a long time in the future. The behavior is derived from one property of transition probabilty matrix for n steps. This term is called the convergence of the n-step transition matrix for n move to infinity. Mathematically, the convergence of the transition probability matrix is finding the limit of the transition matrix which is powered by n where n moves to infinity. The convergence form of the transition probability matrix is very interesting as it will bring the matrix to its stationary form. This form is useful for predicting the probability of transitions between states in the future. The method usually used to find the convergence of transition probability matrix is through the process of limiting the distribution. In this paper, the convergence of the transition probability matrix is searched using a simple concept of linear algebra that is by diagonalizing the matrix.This method has a higher level of complexity because it has to perform the process of diagonalization in its matrix. But this way has the advantage of obtaining a common form of power n of the transition probability matrix. This form is useful to see transition matrix before stationary. For example cases are taken from CLV model using MCM called Model of CLV-Markov. There are several models taken by its transition probability matrix to find its convergence form. The result is that the convergence of the matrix of transition probability through diagonalization has similarity with convergence with commonly used distribution of probability limiting method.