1 * ACM Computing Surveys (CSUR)
2 Volume 39 , Issue 4 (2007)
4 Gaussian random number generators
5 David B. Thomas, Wayne Luk, Philip H.W. Leong, John D. Villasenor
9 * add Erlang dist back in?
13 Note that we need to get rid of mu when it is not the mean.
15 From: Brian Gough <bjg@network-theory.co.uk>
16 To: briggsk@info.bt.co.uk
17 Cc: gsl-discuss@sourceware.cygnus.com
18 Subject: Re: Pareto Distribution
19 Date: Sun, 9 Jul 2000 20:05:03 +0100 (BST)
21 Yes, we should adopt the conventions from a standard reference book --
22 the existing functions are drawn from a variety of sources, mostly
23 Devroye's book on Random Variates (which is public domain, but not
24 available electronically unfortunately). Maybe the three volumes of
25 Johnson & Kotz on Univariate Distributions would do, for
26 example. Patches are welcome from anyone who wants sort this out.
29 > Another thing to think about: some of the other distributions
30 > have a argument `mu' to the C function which is a parameter
31 > which is not the mean. This is non-standard and confusing.
32 > (Also, in the Pareto function, `a' is normally called beta,
33 > `b' is normally called alpha.)
37 > +-------------------------------------------------------------------+
38 > | Dr. Keith M. Briggs, Complexity Research Group, BT Research Labs. |
39 > | Adastral Park admin2 pp5, Martlesham Heath, IP5 3RE, Suffolk, UK |
40 > | Tel. 01473 641 911 Fax. 01473 647 410. Home tel: 01473 625 972 |
41 > | www.bt.com | personal homepage: www.labs.bt.com/people/briggsk2/ |
42 > +-------------------------------------------------------------------+
45 * The exponential power distribution method could be speeded up by
46 using a rational function approximation for the rejection scaling
49 * Do something about the possibility of the user providing invalid
50 parameters (e.g. negative variance etc). Not sure what to do though,
51 since returning an error code is not possible. Maybe just return zero.
52 We should return NAN in this case, and for the CDFs.
54 * Add the triangular distribution.
56 * Look at Marsaglia & Tsang, "The Monte Python Method for generating
57 random variables", ACM TOMS Vol 24, No 3, p341
59 and the paper on the Ziggurat Method: Journal of Statistical Software,
60 Volume 05 Issue 08. George Marsaglia and Wai Wan Tsang. "The ziggurat
61 method for generating random variables"
63 * Should 0 be included in distributions such as the exponential
64 distribution? If we want a consistent behaviour, is it included in
65 others? Note that 1-gsl_rng_uniform() can have a slight loss of
66 precision when the random float is small.