Closed jstupak closed 2 years ago
Will be hard. Adding to the Pythia cards (as per the CMS tune, for instance):
PDF:pSet=LHAPDF6:NNPDF31_nnlo_as_0118
Tune:preferLHAPDF=2
i.e., trying to use the LHAPDF from within the MadGraph + Pythia combo crashes the whole thing with:
PYTHIA Info from SLHAinterface::initSLHA: No MODSEL found, keeping internal SUSY switched off
PYTHIA Warning in SLHAinterface::initSLHA: ignoring MASS entries for id = {1,2,3,4,5,11,12,13,14,15,16,21,22,23,24} (m0 < SLHA:minMassSM)
PYTHIA Warning in SLHAinterface::initSLHA: ignoring DECAY tables for id = {23,24,1,2,3,4,5,11,12,13,14,15,16,21,22} (m0 < SLHA:minMassSM)
PYTHIA Warning in SLHAinterface::initSLHA: ignoring empty DECAY tables for id = {6,25} (total width provided but no Branching Ratios)
terminate called after throwing an instance of 'LHAPDF::ReadError'
what(): Could not find a pdfsets.index file
So I guess using LHAPDF is going to be hard... however, it seems that a simplified version of that PDF is already available in Pythia, e.g. see here: https://pythia.org/latest-manual/PDFSelection.html
option 20 : NNPDF3.1 QCD+LUXQED NNLO alpha_s(M_Z) = 0.118.
More investigation soon.
I tried compiling LHAPDF standalone and it seems to be working: https://github.com/Snowmass21-software/MCProd/commit/b4a29f228abc8ef16924e742f75be1ce3913a309
along with d4d3057dff4099a71af4a38e3f5233a6fdf32abd.
These on both on the gridpacks
branch.
I think the commits I just pushed to gridpacks
should be good to go now
I don't think so... when we do "install lhapdf6" within MadGraph, it install another version of LHAPDF (6.3.0) instead of the one we installed previously (as per install.sh
). Are we sure we are picking the right version?
I think maybe I should have removed echo "install lhapdf6" | python ./MG5_aMC_v2_9_7/bin/mg5_aMC
from install.sh
. But i'll check. I had some trouble using LHAPDF installed via madgraph, so I installed it standalone instead
Yea, I tested and everything runs fine without (additionally) installing lhapdf through madgraph:
INFO: Using LHAPDF v6.4.0 interface for PDFs
I can reproduce that it works now, but we just have to be careful that the test case (test.mg
) doesn't have hardcoded paths in. And in general, for the gridpack generation we want to avoid hardcoded paths as much as possible! If you agree we can close this issue!
Hello Thiago, You are right, and I don't think the hard coded paths are necessary. I was trying to debug something and wanted to rule out an issue with the relative paths, but we can remove this. I'll create a separate issue so as not to forget
Switch to more modern pdf set?