We would like to make a request for a new MC production with the SMASH [1] as an input event generator. The production will be used for physics studies within PWG1 (Global observables) according to our RFBR grant project. It could be of some use by other PWGs as well. It also will be the first MC production in MPD with the SMASH generator as an input.
Collisions: BiBi @ 9.46 GeV, min. bias.
Number of events: 2 mln (the final number will be smaller due to empty events in SMASH, see below).
Configs and macros
The code that is necessary to work with SMASH was added to dev branch of mpdroot several days ago, and updated today (commit 91ad0946). Here are the necessary files:
Output from reconstruction to MiniDst is switched On.
In reco.C the time resolution for TOF is set to 80 ps, tofHit->SetTimeResolution(0.080), according to suggestions from experts.
Details of the SMASH configuration (config.yaml):
In order to get min. bias collisions, impact parameter range is set to a range (0, 20) fm.
As an output, SMASH will produce ROOT-files only (Oscar1999 output is Off).
The current setting is 500 events per run.
The flag Only_Final is set to IfNotEmpty, which means that events with no interaction are not written. => Number of events in output root-files from SMASH will be smaller than 500. With the current configuration, a fraction of empty events is expected to be ≈45%. Please launch it “as is”.
Random seed is set to “-1” - it means that the seed will be randomly generated by the operating system.
Details of the request
We would like to keep generator-level events (i.e. root-files from SMASH) in order to be able to analyse the output from SMASH directly.
Files produced after the runMC.C execution are considered as temporary ones to be used for reconstruction - we do not require to save them. After reconstruction, we need both standard mpddst and miniDst.
The output files will have to be copied and stored at NICA cluster, if possible.
To validate the production configuration, we need ~100,000 events reconstructed for QA.
It would be good to tag the mpdroot version used for this production.
[1] SMASH (Simulating Many Accelerated Strongly-interacting Hadrons) is a relativistic hadronic transport approach for the dynamical description of heavy ion reactions - https://smash-transport.github.io/.
There was no notification about a new topic on the forum. I just read request. It takes some time to deploy smash on govorun and tier clusters. I can’t do it myself, just ask the admins. Hopefully tomorrow I’ll be ready to produce 100,000 QA events.
brief QA showed that the produced events (gen-level, sim, Dst-rec, miniDst-rec) seem to be Ok,
except one thing: since SMASH omits empty events in its output, the number of events in root-files from SMASH is less than 500 (with the current configuration, ~55% of events are non-empty).
At the same time, the runMC.C and reco.C were parameterized to process exactly 500 events per job, which caused a repetition of the last entry of the smash sim-tree (by ~220 times).
An obvious workaround is to check if the next event is the same as the previous one, for instance, by looking at a value of the Impact Parameter, if yes - skip event (continue;).
It is agreed with Andrey that this problem will be solved by setting nEvents=0 in runMC.C and reco.C, and the large-scale production (2 mln) should start in coming days.
2 000 000 events on nica cluster in
MiniDst: /eos/nica/mpd/sim/data/MiniDST/dst-BiBi-09.5GeV-mp07-20-pwg1-500ev/BiBi/09.5GeV-mb/Smash/BiBi-09.5GeV-mp07-20-pwg1-500ev-1
dst: /eos/nica/mpd/sim/data/exp/dst-BiBi-09.5GeV-mp07-20-pwg1-500ev/BiBi/09.5GeV-mb/Smash/BiBi-09.5GeV-mp07-20-pwg1-500ev-1
Everything is fine with eos. Copying to the NICA cluster in production scripts did not work. I had to start copying to the NICA cluster separately today. In 5-6 hours all files will be in place.
Dear Andrey,
thanks a lot again for this SMASH production! We are analyzing the output, everything seems to be rather Ok.
In order to get better statistical significance for some observables, we would like to ask you to produce 8 mln extra events with the same configuration of the gen/sim/rec.
Thanks a lot in advance!
Cheers,
Igor