Objective:To study several types of ergodicity of the queue length of M/M/c queue with synchronous vacation. Methods: A matrix analytical method is applied to deal with it. Result: It is shown that {L ( t ), J (t) } i...Objective:To study several types of ergodicity of the queue length of M/M/c queue with synchronous vacation. Methods: A matrix analytical method is applied to deal with it. Result: It is shown that {L ( t ), J (t) } is geometrically ergodic if and only if it is ergodic. Conclusion:The criteria for the other types of ergodicity are obtained.展开更多
Abstract Let P be a transition matrix which is symmetric with respect to a measure π. The spectral gap of P in L2（π）-space, denoted by gap（P）, is defined as the distance between 1 and the rest of the spectrum of...Abstract Let P be a transition matrix which is symmetric with respect to a measure π. The spectral gap of P in L2（π）-space, denoted by gap（P）, is defined as the distance between 1 and the rest of the spectrum of P. In this paper, we study the relationship between gap（P） and the convergence rate of P^n. When P is transient, the convergence rate of pn is equal to 1 - gap（P）. When P is ergodic, we give the explicit upper and lower bounds for the convergence rate of pn in terms of gap（P）. These results are extended to L^∞ （π）-space.展开更多
This paper investigates the explicit convergence rates to the stationary distribution π of the embedded M/G/1 queue; specifically, for suitable rate functions r（n） which may be polynomial with r（n） = n^l, l 〉 0 ...This paper investigates the explicit convergence rates to the stationary distribution π of the embedded M/G/1 queue; specifically, for suitable rate functions r（n） which may be polynomial with r（n） = n^l, l 〉 0 or geometric with r（n） = α^n, a 〉 1 and ＂moments＂ f ≥ 1, we find the conditions under which Σ∞n=0 r（n）｜｜P^n（i,·） - π（·）｜｜f ≤ M（i） for all i ∈ E. For the polynomial case, the explicit bounds on M（i） are given in terms of both ＂drift functions＂ and behavior of the first hitting time on the state O; and for the geometric case, the largest geometric convergence rate α＊ is obtained.展开更多
We prove some transportation inequalities for hidden Markov chains, generalize the results proved by Kontorovich and Ramanan in two directions and give some applications to log-likelihood functions and hypothesis test...We prove some transportation inequalities for hidden Markov chains, generalize the results proved by Kontorovich and Ramanan in two directions and give some applications to log-likelihood functions and hypothesis testing.展开更多
This paper considers identification of the nonlinear autoregression with exogenous inputs(NARX system).The growth rate of the nonlinear function is required be not faster than linear withslope less than one.The value ...This paper considers identification of the nonlinear autoregression with exogenous inputs(NARX system).The growth rate of the nonlinear function is required be not faster than linear withslope less than one.The value of f(·) at any fixed point is recursively estimated by the stochasticapproximation (SA) algorithm with the help of kernel functions.Strong consistency of the estimatesis established under reasonable conditions,which,in particular,imply stability of the system.Thenumerical simulation is consistent with the theoretical analysis.展开更多
基金This work was partially supported by NNSF of China(No.10171009)"211 Project"+1 种基金"985 Project" Research Fund for Ph. D Programs of MOE of China(No.20010533001)
文摘Objective:To study several types of ergodicity of the queue length of M/M/c queue with synchronous vacation. Methods: A matrix analytical method is applied to deal with it. Result: It is shown that {L ( t ), J (t) } is geometrically ergodic if and only if it is ergodic. Conclusion:The criteria for the other types of ergodicity are obtained.
基金Supported in part by 985 Project,973 Project(Grant No.2011CB808000)National Natural Science Foundation of China(Grant No.11131003)+1 种基金Specialized Research Fund for the Doctoral Program of Higher Education(Grant No.20100003110005)the Fundamental Research Funds for the Central Universities
文摘Abstract Let P be a transition matrix which is symmetric with respect to a measure π. The spectral gap of P in L2（π）-space, denoted by gap（P）, is defined as the distance between 1 and the rest of the spectrum of P. In this paper, we study the relationship between gap（P） and the convergence rate of P^n. When P is transient, the convergence rate of pn is equal to 1 - gap（P）. When P is ergodic, we give the explicit upper and lower bounds for the convergence rate of pn in terms of gap（P）. These results are extended to L^∞ （π）-space.
基金Supported by National Natural Science Foundation of China(No.10171009)
文摘This paper investigates the explicit convergence rates to the stationary distribution π of the embedded M/G/1 queue; specifically, for suitable rate functions r（n） which may be polynomial with r（n） = n^l, l 〉 0 or geometric with r（n） = α^n, a 〉 1 and ＂moments＂ f ≥ 1, we find the conditions under which Σ∞n=0 r（n）｜｜P^n（i,·） - π（·）｜｜f ≤ M（i） for all i ∈ E. For the polynomial case, the explicit bounds on M（i） are given in terms of both ＂drift functions＂ and behavior of the first hitting time on the state O; and for the geometric case, the largest geometric convergence rate α＊ is obtained.
基金supported by the Youth Innovation Foundation of Zhongnan University of Economics and Law from the Fundamental Research Funds for the Central Universities of China (Grant No. 2009004/31540911202)
文摘We prove some transportation inequalities for hidden Markov chains, generalize the results proved by Kontorovich and Ramanan in two directions and give some applications to log-likelihood functions and hypothesis testing.
基金supported by the National Natural Science Foundation of China under Grant Nos. 60821091and 60874001Grant from the National Laboratory of Space Intelligent ControlGuozhi Xu Posdoctoral Research Foundation
文摘This paper considers identification of the nonlinear autoregression with exogenous inputs(NARX system).The growth rate of the nonlinear function is required be not faster than linear withslope less than one.The value of f(·) at any fixed point is recursively estimated by the stochasticapproximation (SA) algorithm with the help of kernel functions.Strong consistency of the estimatesis established under reasonable conditions,which,in particular,imply stability of the system.Thenumerical simulation is consistent with the theoretical analysis.
