Posted by Stephen Earl on May 08, 1998 at 13:31:42:
In Reply to: Re: deconvolution posted by Scott Betts on October 08, 1997 at 09:42:23:
: : Hi,
: : I'm interested in numerical deconvolution:
: : [knowing y(k) and g(k) fing x(k) such that:
: : y(k)=Sum(g(k+m) x(m),for(m,1 to N))]
: : and I would like to know the existing algorithm
: : Can I find them on the Web? or somewhere?
: : Thanks
: The math guru here isn't sure if there's anything
: appropriate here at Netlib, but suggests that if you can't
: find it here, you may want to look at the NAG manuals,
: which sometimes point to public domain versions of their code.
: Hope this helps!
As a first off, trying using the Gerchburg/Gerchburg-Saxton routine, this is a very simple algorithm to program and use, but can produce good results. If you can find a copy have a look at a book entitled Deconvolution by P Jansson. The best method (if there is one!) is the maximum entropy method, there are free downloads of software using this method available somewhere on the web. However, I have yet to find one that is suitable for easy insertion into an exsiting program. Perhaps if you manage to find one, you can send me a copy as I am currently looking for this particular form of nirvana.