1 MOEDAL SIMULATION UPDATE Matt King 09/06/2015
Overview 2 Plan for paper simulation Overall plan Details Issues MoEDAL collaboration meeting Plan for talk 09/06/2015 Matthew King, IFIC
Plan for paper simulation 3 Parameterize efficiency over several model points: First pass model points and production range: Points: m = 100GeV, 1000GeV, 3000GeV q = 0.5, 1.0, 2.0 Production: θ = 2.4 3.0 (MMT acceptance, wide coverage) φ = -2.7 -0.5 (MMT acceptance, wide coverage) KE = 0 1000GeV (feedback? ck?) Parameterize efficiency for each point in terms of: Ez Y projection of θ , φ (at approximate MMT z position) Run generation phase only for new models to get θ , φ , KE values Obtain efficiency from parameterized graphs 09/06/2015 Matthew King, IFIC
Plan for paper simulation 4 Parameterized efficiency tables: Take a long time to create. Do not take up much space. Enable fast calculation of efficiency from models Can generate several tables for different geometry variations May allow for systematic estimation of error from geometry May take of order a week to generate new set of tables from a new geometry Geneva batch system availability is variable Grid may be better, but have had problems trying to get it to work 09/06/2015 Matthew King, IFIC
Issues 5 Code for generation of efficiency tables is already in place Have initial testing tables (1M events) ready Will require a set of “final” geometries to actually produce results Currently writing new nTuple writer algorithm for Gauss Generation phase Original plan was to run, but immediately kill Simulation phase Slow, simulation still needs to be set up and started. Messy. Instead, run only the Generation phase MonopoleTupleAlg is part of the SimMonitor Gaudi chain, never gets called New tool (GenTupleAlg) sits in the GenMonitor chain and outputs generator results Cannot simply copy MonopoleTupleAlg code. “ StoreGate ” classes used are different between Sim and Gen. Looking into using Gauss built in Ntuple output. This is the official LHCb method, but could never make it work before. 09/06/2015 Matthew King, IFIC
MoEDAL collaboration talk 6 We were in very early stages for the talk last year https://indico.cern.ch/event/320391/contribution/5/material/slides/1.pdf A reminder of some advances for software framework alone: Proper infrastructure: svn, batch system, CernVM image Addressed problems arising from “Magnetic field drift” Fixed memory usage problems, allowing for longer simulations Removal of non-monopoles from Simulation phase, vastly speeding up process LHE based generation as well as improved particle gun Far fuller understanding of the Geant4 dE/dx calculation Implementation of MMT/NTD hit generation in Geant4 MMT efficiency calculation (Identified LHCb Random Number Generator issue) 09/06/2015 Matthew King, IFIC
MoEDAL collaboration talk 7 Basic idea: Outline developments made, want to demonstrate that real progress has been made. In more detail: G4 dE/dx equation research dE/dx eqn is the main thing people are conceptually interested in. It’s validity is essential, low beta model concerns may come up. MMT efficiency calculations and current results This is the main tool we will use for the paper, everyone must understand what the situation is. Plan for paper: Similar to previous slides in this document, but with more detail and explanation Estimation of Geometry errors: should this be in mine or Rafal’s talk? 09/06/2015 Matthew King, IFIC
Recommend
More recommend
