The precision of several &ldqqo;standard” probabalistic process generators, as provided to the GPSS user in various textbooks and language manuals, are investigated in this paper. In particular, the unique requirements in GPSS for integral increments in its simulation clock, as well as its background integer and truncation features, are seen to have an inherent impact on methodologies for the generation of probabalistic processes. Thus, GPSS requires that we invoke a discrete approximation to any underlying continuous probabalistic process. The GPSS programmer should be keenly aware of these aspects; failure to do so could lead to a model lacking the desired verisimilitude to the object system under study.
A methodology is then developed in this paper which utilizes a least squares approach to yield process generators for the exponential, Gaussian and other continuous distributions. This approach is seen to overcome several of the difficulties associated with the conventional GPSS process generators, and will be useful in numerous instances.
[1]
Geoffrey Gordon.
Book Reviews : The Application of GPSS V to Discrete System Simulation
,
1975
.
[2]
A. W. Swan.
Handbook of Statistical Tables
,
1964
.
[3]
George S. Fishman,et al.
Solution of Large Networks by Matrix Methods
,
1976,
IEEE Transactions on Systems, Man, and Cybernetics.
[4]
Herbert Maisel,et al.
Simulation of discrete stochastic systems
,
1972
.
[5]
H. Stone.
Approximation of curves by line segments
,
1961
.
[6]
William Feller,et al.
An Introduction to Probability Theory and Its Applications
,
1951
.
[7]
Thomas J. Schriber,et al.
Simulation Using GPSS
,
1974
.
[8]
D. G. Beech,et al.
Handbook of Statistical Tables.
,
1962
.