Hi all,
I was seeing very large variations by doing muon PID systematics.
I went back to the log files and I noticed that the seed seems to set
differently for all jobs.
An example for muon ID
redo muon ID killing
< Running under process ID 6620
< Setting gRandom seed: 6620
< Opening
redo muon ID killing with eff +3%
> Running under process ID 28140
> Setting gRandom seed: 28140
> Opening
I see that an extra flag is needed when running anaQA (-r 1234) that in
runQA is commented out.
##my $mcFlags = "-mc -pk -r 1234 $opt_o ";
my $mcFlags = "-mc $opt_o ";
that was changed in the version 1.2 of runQA by Urs last July.
I tried to check what Riccardo did for the inclusive production but I
don't find the log files for that (Riccardo did you delete them?).
I hope I am wrong but it seems that we have to rerun everything for
systematics.
Urs, could you confirm that? Could you also help us in checking that the
version of runQA in the HEAD is making sense? This caused many problems in
this production. It is a pain now to rerun everything.
Daniele
|