Open ssiccha opened 3 years ago
Even finding the name is reasonably fast. Next thing to test would be to see how slow/fast calling FindHomMethodsProjective.ClassicalNatural
is (which calls RECOG.FindStdGens_SL
among other things). And then of course one could run the profiler to see where else we spend/waste time.
I have looked into this.
The function 'FindHomMethodsProjective.ClassicalNatural' was pretty fast. I summarise the lines which used a lot of time: 250ms: rifac := RecogniseGeneric( Group(List(GeneratorsOfGroup(H), x->ImageElm(Homom(ri),x))), methodsforfactor(ri), depth, forfactor(ri) );
200ms: Setpregensfac( ri, CalcNiceGens(rifac,ri!.gensHmem) );
1850ms:
succ := CallFuncList(methgensN.method,
Concatenation([ri],methgensN.args));
These three lines are basically the bottlenecks for the call 'RecognizeGroup(SL(40,3));'
@fingolfin and @danielrademacher mentioned this during a jitsi call: The call
RecognizeGroup(SL(40,3));
takes pretty long. Once one knows the name of the group, doing the constructive recognition byRECOG.FindStdGens_SL(SL(40,3),GF(3));
is quick.