Tradeoff between Accuracy and Efficiency in the Time-Parallel Simulation of Monotone Systems

We present a new version of the time-parallel simulation with fix-up computations for monotone systems. We use the concept of monotony of a model related to the initial state of the simulation to derive upper and lower bounds of the sample-paths. For a finite state space with some structural constraints, we prove that the algorithm provides bounds at the first step. These bounds are improved at every fix-up computation steps leading to a natural trade-off between accuracy of the simulation results and efficiency of the parallel computations. We also show that many queueing networks models satisfy these constraints and show the links with the monotone version of the Coupling From The Past technique.

[1]  Ana Busic,et al.  Perfect simulation and non-monotone Markovian systems , 2008, VALUETOOLS.

[2]  Mohamed Nassim Seghir,et al.  A Lightweight Approach for Loop Summarization , 2011, ATVA.

[3]  Tobias Kiesling.  Using Approximation with Time-Parallel Simulation , 2005, Simul..

[4]  Jean-Michel Fourneau,et al.  Time Parallel Simulation and hv-Monotonicity , 2011, ISCIS.

[5]  Albert G. Greenberg,et al.  Algorithms for unboundedly parallel simulations , 1991, TOCS.

[6]  Jean-Marc Vincent,et al.  Perfect simulation of index based routing queueing networks , 2006, PERV.

[7]  Richard M. Fujimoto,et al.  Parallel and Distribution Simulation Systems , 1999 .

[8]  Jean-Michel Fourneau,et al.  Monotonicity and Efficient Computation of Bounds with Time Parallel Simulation , 2011, EPEW.

[9]  R.M. Fujimoto,et al.  Parallel and distributed simulation systems , 2001, Proceeding of the 2001 Winter Simulation Conference (Cat. No.01CH37304).

[10]  Jean-Michel Fourneau,et al.  Monotone Queuing Networks and Time Parallel Simulation , 2011, ASMTA.

[11]  P. Rousseeuw,et al.  Wiley Series in Probability and Mathematical Statistics , 2005 .

[12]  Jean-Marc Vincent,et al.  Bounds for the Coupling Time in Queueing Networks Perfect Simulation , 2006 .

[13]  Nihal Pekergin,et al.  Statistical Model Checking Using Perfect Simulation , 2009, ATVA.

[14]  Jean-Michel Fourneau,et al.  An Algorithmic Approach to Stochastic Bounds , 2002, Performance.

[15]  Jean-Michel Fourneau,et al.  Improving Time Parallel Simulation for Monotone Systems , 2009, 2009 13th IEEE/ACM International Symposium on Distributed Simulation and Real Time Applications.

[16]  Ana Busic,et al.  Perfect Sampling of Networks with Finite and Infinite Capacity Queues , 2012, ASMTA.

[17]  P. Glasserman,et al.  Monotone Structure in Discrete-Event Systems , 1994 .

[18]  Albert G. Greenberg,et al.  ICASE MASSIVELY PARALLEL ALGORITHMS FOR TRACE-DRIVEN CACHE SIMULATIONS , 1991 .

[19]  Hagit Attiya,et al.  Wiley Series on Parallel and Distributed Computing , 2004, SCADA Security: Machine Learning Concepts for Intrusion Detection and Prevention.

[20]  J. Propp,et al.  Exact sampling with coupled Markov chains and applications to statistical mechanics , 1996 .

[21]  Allan Clark,et al.  State-Aware Performance Analysis with eXtended Stochastic Probes , 2008, EPEW.

[22]  J. Fourneau,et al.  Perfect simulation and monotone stochastic bounds , 2007, Valuetools 2007.

[23]  Sigrún Andradóttir,et al.  Parallel simulation of transfer lines by time segmentation , 2004, Eur. J. Oper. Res..

[24]  Yi-Bing Lin,et al.  A time-division algorithm for parallel simulation , 1991, TOMC.