Re: 4/6-dimensionl numerical integration? [message #32493 is a reply to message #32459] |
Fri, 11 October 2002 11:48   |
Craig Markwardt
Messages: 1869 Registered: November 1996
|
Senior Member |
|
|
Arne Gennerich <agenner@gwdg.de> writes:
> Craig Markwardt wrote:
>
>> Arne Gennerich <agenner@gwdg.de> writes:
>>> Hi Folks!
>>>
>>> Who can provide me IDL-functions which can calculate 4- and
>>> 6-dimensional numerical integrals?
>>
>> Probably you will have to provide it for yourself. I have wanted to
>> make an extension to my QUADPACK integrator to support
>> multi-dimensional integrals, but haven't gotten around to it yet. If
>> your integration hypervolume is rectangular then it is pretty trivial.
>> However, when you get to high enough dimensions, you probably want to
>> use Monte Carlo, no?
>>
>> Craig
>>
>
> Thanks Craig for your answer! Probably the Monte Carlo Integration is a good
> suggestion. Is the random number generator RANDOMN usefull for this approach?
> Any experience?
I am satisfied with RANDOMN and RANDOMU, but I have not subjected it
to the torture tests that statisticians would. There have been
problems with non-randomness in those routines in the past.
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|