1 / 4

Advanced Multiprocessing Techniques in MARS15 for Resistive Magnet Simulations

This document outlines the implementation of advanced multiprocessing techniques in the MARS15 simulation code, specifically for the study of resistive magnets within the proposed IDS120j geometry. It details two simulation codes, mars1510 and mars1512, highlighting their performance metrics and efficiencies. Notably, the new module shows improvements in result stability despite longer computation times per history due to expanded physics models. A summary of geometry configurations, neutron energy cutoffs, and simulation settings is provided, emphasizing the capability to run multiple jobs for extensive event analysis.

demitrius
Télécharger la présentation

Advanced Multiprocessing Techniques in MARS15 for Resistive Magnet Simulations

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Multiprocessing/MARS15(2012)/Princeton Cluster(IDS120j w/t Resistive Magnets: New Hg Module) X. Ding UCLA (4/11/2013)

  2. Multiprocessing (Motivation) Key Points in Nicholas Souchlas’s talk (https://pubweb.bnl.gov/~kirk/Target_Studies/2013_03_28/IDS120j_NEW_Hg_MODULE_m1510_VS_m1512.pdf) 1. IDS120j GEOMETRY, NO RESISTIVE MAGNETS: WITH 20 cm GAPS BETWEEN CRYOSTATS. 2. SIMULATIONS CODE: mars1510 (Desktop)/ mars1512 (Princeton Cluster). mars1510: 1E5 ~ 7.72 hrs, [5E5 ~ 38.33 hrs] mars1512: 1E5 ~ 111.06 hrs, [5E5 ~ 555.3 hrs ( 23.6 d ) !!! ( ESTIM)] Explanation (http://physics.princeton.edu/mumu/target/Ding/ding_020713.pdf): The newest code is slower per one history because of substantially extended physics models and more sophisticated algorithms that now result in lesser fluctuations and give better smoother results with a lesser number of histories. (Comment from N. V. Mokhov) (about 10 times slower of m1512 vs. m1510) 2 IDS120j GEOMETRY, NO RESISTIVE COILS: WITH 20 cm GAPS *********************************************************** # SIMULATIONS USING LOWEST GRADE W BEADS IN SHIELDING ( OF 15.8 g/cc ) # BP#2 SEGMENTATION STUDIES WITHIN THE FIRST CRYOSTAT AND RIGHT FLANGE OF Hg POOL INNER VESSEL. *********************************************************** >SIMULATIONS CODE:mars1512 ( USING MCNP CROSS SECTION LIBRARIES ) >NEUTRON ENERGY CUTOFF:10-11 MeV >SHIELDING: 60% W + 40% He( WITH STST VESSELS) >PROTON BEAM POWER: 4 MW >PROTON ENERGY: E = 8 GeV >PROTON BEAM PROFILE: GAUSSIAN, σx = σy = 0.12 cm

  3. Multiprocessing (Cont’d) 3. High TDPD (mW/g) is given by mars1512 (1E5) then m1510 (5E5). mars1510: ~ 0.07 mW / g AT ( r, z, phi) = (125 cm, 35 cm, 225 deg) IN SC#1 [ 4 x 5E5 EVENTS ] mars1512: ~ 0.19 mW / g AT ( r, z, phi) = (125 cm, 115 cm, 255 deg) IN SC#2 [ 1E5 EVENTS ] A multiprocessing option has been developed and implemented into MARS15 (2012) by N.V. Mokhov. Multiprocessing is job-based, i.e. the processes, replicating the same geometry of the setup studied, run independently with different initial seeds. So for requirement of high number of incidents (primary events), we can run multiple MARS15 (2012) jobs.

  4. Multiprocessing (Setting & Simulation) Setting & Running with 100 jobs (50,000 events for each) Setting at /mnt/dayabay/dayabay/nsouchlas/INRNHMWDSSS (_100) (100 jobs) or /mnt/dayabay/dayabay/ding/mars1512/ED (100 jobs) All running of 100 jobs are finished in INRNHMWDSSS directory (each job took almost 100 CPU hrs). Nicholas will analysis the results.

More Related