Re: [Help-gsl] numerical integration with strong discontinuities

[eknecronzontas]

Hmm....more details would be helpful, but since you mention particle physics, I 
thought maybe I should point out that I am indeed using GSL (in combination 
with CERNLIB) to do integrations over propagators which look very similar to 
what you described. In my case, though, I'm doing Cauchy principal value 
integration...I'm not sure why you'd want Monte Carlo for this...

----- Original Message ----
From: Nicolau Werneck <address@hidden>
To: address@hidden
Sent: Wednesday, August 8, 2007 10:27:03 PM
Subject: [Help-gsl] numerical integration with strong discontinuities

Hello. I'm trying to numerically integrate a function that has some
ugly discontinuties like

Int{x/(x-y)}  x=0:1  y=0:1

It seems that most standard algorithms don't like things like that.
The one I tried out in Mathematica  needs to have a previous
estimation of where to look for discontinuities, but this is only OK
for one dimension, I have a more complex "locus"  of discontinuity...

What is the "killer" algo for this kind of integration?

I've read somewhere that VEGAS bases his sampling on the absolute
value of the function (|f|).  Isn't that bad for  functions with 1/x
discontinuities?... Afterall, it's just a place where the function is
odd.

There must be an adaptative algorithm that handles this kind of
discontinuities naturally... Any ideas, suggestions?...

It's for a particle physics problems, BTW... I'm helping out someone.
Does anybody here has experience on the kind of integrals the particle
folks calculate?

Thanks...

-- 
Nicolau Werneck <address@hidden>         9F99 25AB E47E 8724 2F71
http://cefala.org/~nwerneck                   EA40 DC23 42CE 6B76 B07F


_______________________________________________
Help-gsl mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/help-gsl