| commit | 50b83d153e53173d2eb55d13fed63c0363494288 | |
| author | Carlos R. Mafra <crmafra@gmail.com> | |
| Fri, 30 Sep 2016 22:59:22 +0000 (30 23:59 +0100) | ||
| committer | Carlos R. Mafra <crmafra@gmail.com> | |
| Sun, 2 Oct 2016 01:29:29 +0000 (2 02:29 +0100) | ||
| tree | 62824b8215e985d5282c088e5f3915217bed9e65 | treesnapshot (tar.gz zip) |
| parent | 35846b3bcd2458fc5bfeddb8dfdf8a8da3a9dca5 | commitdiff |
BGexpand: More performance improvements
Sorting intermediate results when loading the regular integrals
avoids a huge pile-up of terms that need to be sorted later.
This improves performance a lot. One example case is
L tmp = Zint(1,...,9,[p],1,...,9);
#call BGexpand(6)
which goes from 110 to 75 secs, a ~30% improvement.
Sorting intermediate results when loading the regular integrals
avoids a huge pile-up of terms that need to be sorted later.
This improves performance a lot. One example case is
L tmp = Zint(1,...,9,[p],1,...,9);
#call BGexpand(6)
which goes from 110 to 75 secs, a ~30% improvement.
| BGap.h | diffblobblamehistory |