Approximation and Optimization for Stochastic Networks

摘要: We describe a computational experiment directed at the problem of improving stochastic network such as those found in logistics planning. Standard methods simulation optimization can be very slow, especially for large networks. suggest two-phase approach using approximations place most runs required conventional approach. present simple example balancing an airlift network, which this successfully solves much less time than would have required. also discuss further work currently progress to refine and extend