Graph Generation with several coupling constants

[mnklann]

I noticed that when generating graphs for models with more than three coupling constants, I get the following error. I encountered this while computing some beta functions and was wondering whether this limit is intentional, and whether lifting it could cause issues.

Symbols g1,g2,g3,g4;

Model FOUR;
  Particle q1, +1;
  Particle q2, +1;
  Particle q3, +1;
  Particle q4, +1;
  Vertex q1, q1, q1: g1;
  Vertex q2, q2, q2: g2;
  Vertex q3, q3, q3: g3;
  Vertex q4, q4, q4: g4;
EndModel;

gives the error message

Error : too many coupling constants in the model (GRCC_MAXNCPLG)

Is this limit intentional? If not, it might be worth considering whether it could be increased, since there are applications involving several couplings. At the moment, a possible workaround seems to be using a single coupling and assigning the couplings manually.
I’d be interested to know whether there is a technical reason for the current limit, or whether it could safely be lifted.

[tueda]

I'm not sure why this limit is set so low, other than to save memory, but the following line looks like it has a typo:

form/sources/grcc.cc

Lines 1989 to 1991 in 7a8f756

	if (ncouple >= GRCC_MAXNCPLG) {
	erEnd("too many coupling constants in the model (GRCC_MAXNCPLG)");
	}

compared with:

form/sources/diawrap.cc

Lines 76 to 80 in 7a8f756

	if ( m->ncouplings > GRCC_MAXNCPLG ) {
	MesPrint("Too many coupling constants in model. Current limit is %d.",(WORD)GRCC_MAXNCPLG);
	MesPrint("Suggestion: recompile Form with a larger value for GRCC_MAXNCPLG");
	return(-1);
	}

At least one of these conditions must be wrong.

[jodavies]

All over the place, grcc uses >= to compare with GRCC_MAXNCPLG and the other hardcoded limits, but then accesses arrays of those sizes in a 0-indexed way. So I suppose changing all of the grcc comparisons to > is the correct fix. But that has to be done carefully...

What is a better default value for GRCC_MAXNCPLG? 10?

[mnklann]

I found the problem while looking into the unbroken-phase gauge beta function, so I’d suggest setting it to at least 7. However, 10 gives some room to spare and would probably be a better default.