$(max,+)$ 代数中随机矩阵乘积的失忆性质
概率论
2007-05-23 v3 最优化与控制
摘要
代数中的随机矩阵乘积被用作一类离散事件动力系统的模型。J. Mairesse 证明了这样的系统当且仅当具有失忆性质时,在有限时间内耦合到一个唯一的平稳机制。我们证明失忆性质是通有的,意义如下:若它不被满足,则测度的支撑包含于有限个仿射超平面的并中,并且在离散情形下测度的原子呈线性关系。
关键词
引用
@article{arxiv.math/0405452,
title = {Memory loss property for products of random matrices in the $(\max,+)$ algebra},
author = {Glenn Merlet},
journal= {arXiv preprint arXiv:math/0405452},
year = {2007}
}
备注
The article has been completely rewritten, in order to state more explicit results and allow the matrices' entries to be infinite. Moreover the results are illustrated on a simple production system