When, and When Not to Use Warm-up Periods in Discrete Event Simulation

Winfried K. Grassmann

2nd International ICST Conference on Simulation Tools and Techniques

Research Article

When, and When Not to Use Warm-up Periods in Discrete Event Simulation

Download905 downloads

Cite: BibTeX Plain Text

@INPROCEEDINGS{10.4108/ICST.SIMUTOOLS2009.5603,
    author={Winfried K. Grassmann},
    title={When, and When Not to Use Warm-up Periods in Discrete Event Simulation},
    proceedings={2nd International ICST Conference on Simulation Tools and Techniques},
    publisher={ICST},
    proceedings_a={SIMUTOOLS},
    year={2010},
    month={5},
    keywords={Discrete Event Simulation Time Averages Initialization Bias Queues Agent-based Systems},
    doi={10.4108/ICST.SIMUTOOLS2009.5603}
}

Winfried K. Grassmann
Year: 2010
When, and When Not to Use Warm-up Periods in Discrete Event Simulation
SIMUTOOLS
ICST
DOI: 10.4108/ICST.SIMUTOOLS2009.5603

Winfried K. Grassmann¹^,*

1: Department of Computer Science, University of Saskatchewan, 110 Science Court, Saskatoon, SK S7N 5C9

*Contact email: grassman@cs.usask.ca

Abstract

This paper demonstrates, both theoretically and by using numerical examples, that if one has good starting states, one should not use warmup periods in discrete event simulation when estimating equilibrium expectations by means of time averages. The numerical methods used are deterministic, and they are based on randomization or uniformization. We also show that if estimating expectations of sums by simulation, good starting states are sometimes diffcult to fnd and are often inconvenient, which justifes warmup periods.

Keywords: Discrete Event Simulation Time Averages Initialization Bias Queues Agent-based Systems

Published: 2010-05-16
Publisher: ICST
Modified: 2010-05-16

: http://dx.doi.org/10.4108/ICST.SIMUTOOLS2009.5603