Comments on: A new pseudo-random number generator for Haskell http://www.serpentine.com/blog/2009/09/19/a-new-pseudo-random-number-generator-for-haskell/ Bryan O'Sullivan's blog Wed, 08 Feb 2012 06:41:38 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Michael Lesniak http://www.serpentine.com/blog/2009/09/19/a-new-pseudo-random-number-generator-for-haskell/comment-page-1/#comment-254087 Michael Lesniak Fri, 29 Jan 2010 11:13:48 +0000 http://www.serpentine.com/blog/?p=434#comment-254087 Just wanted to thank you for this. Really nice! Just wanted to thank you for this. Really nice!

]]> By: Bryan O'Sullivan http://www.serpentine.com/blog/2009/09/19/a-new-pseudo-random-number-generator-for-haskell/comment-page-1/#comment-246902 Bryan O'Sullivan Fri, 02 Oct 2009 06:41:02 +0000 http://www.serpentine.com/blog/?p=434#comment-246902 Bart, the problem is that the ziggurat method as originally described and usually implemented is broken (see Doornik's "An Improved Ziggurat Method to Generate Normal Random Samples" for an explanation). A less-than-stellar uniform PRNG merely makes the brokenness more obvious. My implementation uses a fast-and-good underlying PRNG and a better, but slower, ziggurat (Doornik's), so it should equal or beat ISAAC-plus-broken-ziggurat on both performance and quality. Bart, the problem is that the ziggurat method as originally described and usually implemented is broken (see Doornik’s “An Improved Ziggurat Method to Generate Normal Random Samples” for an explanation). A less-than-stellar uniform PRNG merely makes the brokenness more obvious.

My implementation uses a fast-and-good underlying PRNG and a better, but slower, ziggurat (Doornik’s), so it should equal or beat ISAAC-plus-broken-ziggurat on both performance and quality.

]]> By: PO8 http://www.serpentine.com/blog/2009/09/19/a-new-pseudo-random-number-generator-for-haskell/comment-page-1/#comment-246901 PO8 Fri, 02 Oct 2009 05:59:54 +0000 http://www.serpentine.com/blog/?p=434#comment-246901 I've used ISAAC as the underlying PRNG for generating normally-distributed variates via the Ziggurat method, since I haven't found anything else robust enough to get a good distribution from this sensitive filter. I'll give the PRNG you suggest a try and see what I find out. Sure would like to have a pure-Haskell ISAAC implementation, but that's a hard one and I don't think I'm up to it. I’ve used ISAAC as the underlying PRNG for generating normally-distributed variates via the Ziggurat method, since I haven’t found anything else robust enough to get a good distribution from this sensitive filter. I’ll give the PRNG you suggest a try and see what I find out. Sure would like to have a pure-Haskell ISAAC implementation, but that’s a hard one and I don’t think I’m up to it.

]]> By: Vermue http://www.serpentine.com/blog/2009/09/19/a-new-pseudo-random-number-generator-for-haskell/comment-page-1/#comment-246324 Vermue Tue, 22 Sep 2009 10:02:12 +0000 http://www.serpentine.com/blog/?p=434#comment-246324 Please try this prng. on http://www.number.com.pt/index.html Can you tell me the comparative results on it. Thank you Vermue Please try this prng.

on

http://www.number.com.pt/index.html

Can you tell me the comparative results on it.

Thank you

Vermue

]]> By: brian http://www.serpentine.com/blog/2009/09/19/a-new-pseudo-random-number-generator-for-haskell/comment-page-1/#comment-246210 brian Sun, 20 Sep 2009 02:37:22 +0000 http://www.serpentine.com/blog/?p=434#comment-246210 That's really cool! Thanks. That’s really cool! Thanks.

]]>