MathGroup Archive 2002

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: maximal entropy method

  • To: mathgroup at smc.vnet.net
  • Subject: [mg32864] Re: [mg32854] maximal entropy method
  • From: Sseziwa Mukasa <mukasa at jeol.com>
  • Date: Sat, 16 Feb 2002 04:35:21 -0500 (EST)
  • Organization: JEOL (USA) Ltd.
  • References: <200202150750.CAA09943@smc.vnet.net>
  • Sender: owner-wri-mathgroup at wolfram.com

Borut L wrote:

> Hello,
>
> I am modeling a problem with it. Now, this method is nicely implemented in
> C's Numerical Recipes. Is something alike done in Mathematica somewhere and
> by somebody?
>
> Thank You for your help,
>
> Borut

I assume you are referring to the regularization method developed by Skilling
and Gull, not Berg's method.  There is an incomplete implementation of
Skilling's Cambridge method in Applied Mathematica:Getting Started, Getting It
Done by William T. Shaw and Jason Tigg (Addison-Wesley, 1994).  I say it's
incomplete because it does not do the line search to constrain the step size
in the search subspace.  Some of the code from the book is here:
http://www.mathsource.com/Content22/Publications/BookSupplements/ShawTigg-1993/0205-366,
but as far as I can tell the Maximum Entropy code is not.

Regards,

Sseziwa



  • Prev by Date: Re: A Learning TableForm Problem
  • Next by Date: Tracing Evaluation
  • Previous by thread: maximal entropy method
  • Next by thread: Re: Re: maximal entropy method