Abstract
In this paper, we propose the first fully polynomial-time randomized approximation scheme (FPRAS) for closed Jackson networks with single servers. Our algorithm is based on the Markov chain Monte Carlo (MCMC) method, and our scheme returns an approximate solution, for which the size of error satisfies a given error rate. We propose two Markov chains: one is for approximate sampling, and the other is for perfect sampling based on the monotone coupling from the past algorithm.
| Original language | English |
|---|---|
| Pages (from-to) | 1484-1503 |
| Number of pages | 20 |
| Journal | SIAM Journal on Computing |
| Volume | 38 |
| Issue number | 4 |
| DOIs | |
| Publication status | Published - Nov 7 2008 |
| Externally published | Yes |
All Science Journal Classification (ASJC) codes
- Computer Science(all)
- Mathematics(all)