2
$\begingroup$

To do some calculation on QCD on Lattice, I needed a Monte Carlo multi integration. I wrote a a C++ program according to what I understood from the paper “ Lepage (1978) “ A New Algorith For Adaptive Multidimensional Integration” SLAC_Pub_1839 November 1976” The results I got, however, were not accurate, so I copied the written program from the books: a) Numeerical Recipies in C and b) Numeerical Recipies in C Examples. The program is running for 4 days on My PC with Intel Core 2 Due, @ 2.9 GHz. My question is from anyone who has experience on Vegas algorithm that this long running is normal or my program is in a non-ending loop. Thanks.

$\endgroup$

migrated from physics.stackexchange.com Dec 14 '11 at 20:01

This question came from our site for active researchers, academics and students of physics.

  • $\begingroup$ Have you tested your algorithm against functions with known anti-derivatives? If the algorithm is not the focus of your research, I would suggest using a pre-existing (read: tested) library. In particular, GNU Scientific Library and the CUBA Library both have VEGAS implemented. $\endgroup$ – dls Dec 14 '11 at 20:04
2
$\begingroup$

I used VEGAS some time ago to calculate the hadronic cross section using partonic cross section and parton distributions and none of my calculations lasted that long. Around ~10-20min at most. I would go with the precompiled CUBA libraries. There are four different algorithms Cuhre, Divonne, Suave and Vegas, consult with the manual/paper for more information and then choose the one fits best at your needs.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy