Bug in random number generator - (nf)
utzoo!decvax!microsof!uw-beave!cornell!vax135!ariel!orion!lime!houti!hogpc!houxz!ihnp4!ixn5c!inuxc!pur-ee!uiucdcs!hall
utzoo!decvax!microsof!uw-beave!cornell!vax135!ariel!orion!lime!houti!hogpc!houxz!ihnp4!ixn5c!inuxc!pur-ee!uiucdcs!hall
Thu Mar 24 22:26:15 AEST 1983
#R:ritcv:-19100:uiucdcs:8200012:000:529
uiucdcs!hall Mar 24 21:00:00 1983
I suggest replacing the random number generator with one based
on the linear congruential method described in Knuth, "The Art
Of Computer Programming, Vol 2: Seminumerical Algorithms".
There is an EXCELLENT article in the Bell System Technical
Journal, Vol. 61, No. 8, October, 1982 by C.S. Roberts that
describes just what you want. The appropriate constants are
listed for a generator with period 2**48; test data is also
tabulated to check your implementation. I recommend this
article highly.
--John
(...pur-ee!uiucdcs!hall)
More information about the Comp.bugs.4bsd.ucb-fixes
mailing list