Management of the Monte Carlo simulation process in

  • Slides: 18
Download presentation
Management of the Monte Carlo simulation process in the MPD experiment on the NICA

Management of the Monte Carlo simulation process in the MPD experiment on the NICA cluster Łukasz Sawicki Faculty of Physics Warsaw University of Technology Supervisor Prof. Adam Kisiel

 Assigned Outline tasks Used tools How to run simulations? Testing of generated data

Assigned Outline tasks Used tools How to run simulations? Testing of generated data Summary Łukasz Sawicki Faculty of Physics WUT 2

Assigned tasks Mass events generation Correctness check of generated events Łukasz Sawicki Faculty of

Assigned tasks Mass events generation Correctness check of generated events Łukasz Sawicki Faculty of Physics WUT Automatisation of simulation process to create events Comparison of simulation models 3

Used tools Simulation Ur. QMD models 3. 4 v. HLLE Mpd. ROOT NICA framework

Used tools Simulation Ur. QMD models 3. 4 v. HLLE Mpd. ROOT NICA framework cluster Łukasz Sawicki Faculty of Physics WUT 4

How to run simulations Required Install Set package avaliable on Git. Hub Mpd. ROOT

How to run simulations Required Install Set package avaliable on Git. Hub Mpd. ROOT Ur. QMD/v. HLLE your paths Necessary access to NICA cluster and data folder https: //github. com/Lukaszz 99/work Łukasz Sawicki Faculty of Physics WUT 5

How does it work? Reconstrution • Ur. QMD • v. HLLE Simulation Łukasz Sawicki

How does it work? Reconstrution • Ur. QMD • v. HLLE Simulation Łukasz Sawicki Faculty of Physics WUT • run. MC. C – detector response • reco. C – collision reconstruction • mpddst. root – file with reconstructed collision Ready to analysis 6

 All generated data is reconstructed Stored in /eos/nica/mpd/data folder Separate folders for each

All generated data is reconstructed Stored in /eos/nica/mpd/data folder Separate folders for each energy and simulation model Ur. QMD v. HLLE 4, 7, 9, 11 Ge. V 7. 7 Ge. V Łukasz Sawicki Faculty of Physics WUT 7

 Data folder 7 PB No of space quota per user Used as a

Data folder 7 PB No of space quota per user Used as a normal disc of a computer Łukasz Sawicki Faculty of Physics WUT 8

data template Urqmd. 4 Gev Urqmd. 11 Ge. V v. HLLE. 7. 7 Ge.

data template Urqmd. 4 Gev Urqmd. 11 Ge. V v. HLLE. 7. 7 Ge. V JOB_ID Macros for simulation gen Ur. QMD simulation files Łukasz Sawicki Faculty of Physics WUT Test. Ev. Urqmd Detector v. HLLE response, simulation Particle tracks, files Reconstructed collision gen output Same in Test. Ev. Urqmd 9

inputfile v. HLLE. C Łukasz Sawicki Faculty of Physics WUT 10

inputfile v. HLLE. C Łukasz Sawicki Faculty of Physics WUT 10

Testing generated data Included Works in Git. Hub package for Urqmd and v. HLLE

Testing generated data Included Works in Git. Hub package for Urqmd and v. HLLE simulations Information about type of errors, list of correct mpddst. root files Łukasz Sawicki Faculty of Physics WUT 11

Scheme of testing create list of JOB_ID folders Do Test. Ev. Urqmd and gen

Scheme of testing create list of JOB_ID folders Do Test. Ev. Urqmd and gen folders exist? Does evetest. root exist? Does mpddst. root exist? big enough? Łukasz Sawicki Faculty of Physics WUT Ur. QMD files? n. Events from inputfile add mpddst. root path to list with correct files 12

Program output errors and error_category lists failed jobs list correct mpddst. root files list

Program output errors and error_category lists failed jobs list correct mpddst. root files list Łukasz Sawicki Faculty of Physics WUT 13

Program output List of errors Commands to run reco. C or run. MC. C

Program output List of errors Commands to run reco. C or run. MC. C for failed jobs Łukasz Sawicki Faculty of Physics WUT 14

Summary 4 Ge. V 203600 Ur. QMD 7 Ge. V 9 Ge. V 11

Summary 4 Ge. V 203600 Ur. QMD 7 Ge. V 9 Ge. V 11 Ge. V reconstructed events 448346 593851 721652 v. HLLE 7. 7 Ge. V 151393 2 120 000 reconstructed events – 60 TB Numbers from 24. 08. 2019 Łukasz Sawicki Faculty of Physics WUT 15

 Event generation and reconstruction for 500 events – 22 h 2 120 000

Event generation and reconstruction for 500 events – 22 h 2 120 000 events – 93 280 h – 10. 6 years on personal computer We did it in 4 weeks Łukasz Sawicki Faculty of Physics WUT 16

Thank you for attention Łukasz Sawicki Faculty of Physics WUT

Thank you for attention Łukasz Sawicki Faculty of Physics WUT

Contact info l. saw 99@gmail. com https: //github. com/Lukaszz 99/work Łukasz Sawicki Faculty of

Contact info l. saw 99@gmail. com https: //github. com/Lukaszz 99/work Łukasz Sawicki Faculty of Physics WUT 18