Long Term Average Control of a Local Time Process
Long Term Average Control of a Local Time Process
复制标题
DOI：
10.1007/978-1-4613-0265-0_27
复制
发表时间：
2002
期刊：
影响因子：
--
通讯作者：
M. Mendiondo;R. Stockbridge
中科院分区：
文献类型：
--
作者：
M. Mendiondo;R. Stockbridge
本文为长期平均随机控制问题提供了一种易于处理的数值方法，其中成本包括基于扩散的局部时间过程的项。控制问题被重新表述为过程不变分布集上的线性程序。特别是，长期平均本地时间成本用不变分布表示。利用马尔可夫链逼近将无限维线性规划简化为有限维线性规划，并给出了最优值收敛的条件。
This paper provides a tractable numerical method for long-term average stochastic control problems in which the cost includes a term based on the local time process of a diffusion. The control problem is reformulated as a linear program over the set of invariant distributions for the process. In particular, the long-term average local time cost is expressed in terms of the invariant distribution. Markov chain approximations are used to reduce the infinite-dimensional linear programs to finite-dimensional linear programs and conditions for the convergence of the optimal values are given.