#### DMCA

## Clones: CLOsed queueing Networks Exact Sampling

Citations: | 1 - 1 self |

### Citations

543 | Exact sampling with coupled Markov chains and applications to statistical mechanics.
- Propp, Wilson
- 1996
(Show Context)
Citation Context ... for the non-reversible Markov chain case. In the 1990’s, Propp and Wilson introduced a method for sampling a random variable according to the stationary distribution of a finite ergodic Markov chain =-=[9]-=-: the coupling from the past (CFTP) algorithm. The CFTP algorithm automatically detects and stops when the sample has the correct distribution. The main drawback of the original CFTP is that it consid... |

381 |
Markov Chains and Mixing Times.
- Levin, Peres, et al.
- 2008
(Show Context)
Citation Context ...mations [2] or simulation. The main difficulty of the Markov chain Monte Carlo approach is the stopping criterion. Usual tools are the asymptotic variance in the Central Limit Theorem or mixing times =-=[1, 7]-=-, unfortunately there is no generic technique for the non-reversible Markov chain case. In the 1990’s, Propp and Wilson introduced a method for sampling a random variable according to the stationary d... |

140 | Computational algorithms for closed queueing networks with exponential servers. - Buzen - 1973 |

68 |
Closed queueing systems with exponential servers,
- Gordon, Newell
- 1967
(Show Context)
Citation Context ...e past 1. INTRODUCTION Closed queueing networks are largely used in various application domains due to their modeling simplicity and productform stationary distribution in the unlimited capacity case =-=[5]-=-. This structure is no longer guaranteed when the queues have a finite capacity, due to the effect of blocking. There are different types of blocking, see [8] for a detailed discussion. According to t... |

39 |
Survey of closed queueing networks with blocking ,
- Onvural
- 1990
(Show Context)
Citation Context ...stribution in the unlimited capacity case [5]. This structure is no longer guaranteed when the queues have a finite capacity, due to the effect of blocking. There are different types of blocking, see =-=[8]-=- for a detailed discussion. According to the terminology introduced in that survey, our blocking model is RS-RD (repetitive service – random destination). Under this blocking policy, the stationary di... |

23 |
Stochastic simulation: algorithms and analysis, volume 57 of Stochastic Modelling and Applied Probability.
- Asmussen, Glynn
- 2007
(Show Context)
Citation Context ...mations [2] or simulation. The main difficulty of the Markov chain Monte Carlo approach is the stopping criterion. Usual tools are the asymptotic variance in the Central Limit Theorem or mixing times =-=[1, 7]-=-, unfortunately there is no generic technique for the non-reversible Markov chain case. In the 1990’s, Propp and Wilson introduced a method for sampling a random variable according to the stationary d... |

5 |
Randomized approximation scheme and perfect sampler for closed jackson networks with multiple servers
- Kijima, Matsui
(Show Context)
Citation Context ...e case of closed queueing networks the cardinality of the state space is exponential in the number of queues, which is untractable. When the network has a product-form distribution, Kijima and Matsui =-=[6]-=- proposed a perfect sampling algorithm with overall complexity O(K3 ln(KM)). However, their method strongly relies on the product form representation of the stationary distribution and it cannot be ap... |

2 |
Queueing networks with blocking: Analysis, solution algorithms and properties
- Balsamo
- 2011
(Show Context)
Citation Context ...oduct form only if the routing is reversible. When the stationary distribution does not have a product-form, the exact analysis may not be computationally tractable, and we may turn to approximations =-=[2]-=- or simulation. The main difficulty of the Markov chain Monte Carlo approach is the stopping criterion. Usual tools are the asymptotic variance in the Central Limit Theorem or mixing times [1, 7], unf... |

2 |
Perfect sampling for closed queueing networks
- Bouillard, Bušić, et al.
(Show Context)
Citation Context ...d strongly relies on the product form representation of the stationary distribution and it cannot be applied to the general case of queues with finite capacity. In a recent paper by Bouillard and al. =-=[3]-=-, a new representation of the sets of states has been proposed. This representation is used to derive a bounding chain for the CFTP algorithm for closed queueing networks, that enables exact sampling ... |