PAPER: Numerical Replication of Computer Simulations: Some Pitfalls and How To Avoid Them

[Theodore C. Belding]

[This is a new, expanded draft. -Ted]

The following paper is available; comments are welcome!

Title: Numerical Replication of Computer Simulations: Some Pitfalls 
and How To Avoid Them
Author: Theodore C. Belding, University of Michigan Center for the Study
of Complex Systems <address@hidden>
Date: January 25, 2000

Abstract: 
A computer simulation, such as a genetic algorithm, that uses IEEE
standard floating-point arithmetic may not produce exactly the same
results in two different runs, even if it is rerun on the same
computer with the same input and random number seeds. Researchers
should not simply assume that the results from one run replicate those
from another but should verify this by actually comparing the
data. However, researchers who are aware of this pitfall can reliably
replicate simulations, in practice. This paper discusses the problem
and suggests solutions.

Submitted to GECCO 2000

Available at:

HTML:
http://www-personal.umich.edu/~streak/papers/numerical-replication/
http://www-personal.umich.edu/~streak/papers/numerical-replication-html.tar.gz

PostScript:
http://www-personal.umich.edu/~streak/papers/numerical-replication.ps.gz

PDF: 
http://www-personal.umich.edu/~streak/papers/numerical-replication.pdf

Also submitted to http://arXiv.org/ (available January 27, 2000)

--
Ted Belding                              address@hidden 
University of Michigan Center for the Study of Complex Systems
Homepage: http://www-personal.umich.edu/~streak/
PGP key:  http://www-personal.umich.edu/~streak/pgp-key.html



                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.