Decay Chain Hangs

[Zach]

Hi Olivier,

I'm attempting a decay chain of particles, here "add indirect_detection n2 n2, ( n2 > z n1, n1 > e+ e- ve)". The process hangs during the decay process generation. Not the numerical evaluation, and after the individual decay diagram generation, but in some intermediate step. There is no error output (because it never stops), and the corresponding standard model thing where I decay leptons works just fine.

It is technically possible that it is working just fine and is just technically complex, however I believe normally this portion of the process takes ~< 10 min or so, and I have so far run the above process for about 22 hr's after which I stopped it.

Further, I can generate/fully compute any part of the decay chain just fine by itself, it is only the combination of all three steps that hangs. I would appreciate even an estimate for a max time this should take, or any output that can help me understand the issue.

The last few lines of output:

INFO: Combine n2 > z n1 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 @0 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 with decays n1 > e+ e- ve SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 @0 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0
INFO: Combine ~psi ~psibar > n2 n2 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 @1995 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 with decays n2 > z n1 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 @0 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0
  Decay: n1 > e+ e- ve SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 @0 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 , n2 > z n1 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 @0 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0
  Decay: n1 > e+ e- ve SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0 @0 SDEFFF=0 SDEFFV=0 SIEFFF=0 SIEFFS=0 SIEFFV=0

Best,
Zach

[Olivier Mattelaer (olivier-mattelaer)]

Hi,

Like this I have simply no clue.
If you do a Ctrl-C, it should creates a debug file that can help me to know in which part of the code it is taking a lot of time.

For the maximum time, this is a very complicated question that I do not even know for standard Madgraph. This is something very complex to evaluate.
This depend of the matrix-element complexity (yours is a 10 particles amplitude which is quite complex) but also of the number of channel of integration (which depends of the number of Feynman Diagram) (here I guess it is not too much) and also of the number of resonance conflict (which I would bet you have quite a lot). Finally it depends also of the initial energy, cuts.
I was planning to try to have an euristic to get a [very large] over-estimate to help cluster schedulling but this is not top priority.

Cheers,

Olivier