Status of MC Simulation

Huaimin Liu

BESIII Collaboration meetingIHEP, Beijing, Jan. 12, 2006

What to cover?

• Retrospect

• Current status

• Performance

• Planned development

MC Project: BOOST on GEANT4

BESIII Object Oriented Simulation Tool

(proposal: August, 2002)

BES MC Software Evolution

BESI BESII BESIII

SOBER SIMBES BOOST

EGS GEANT3 GEANT4

1980s 1990s 2000s

Working group

• IHEP Deng Ziyan (TOF, framework)

He Miao (EMC)

Liu Huaimin (framework, project coordinator)

Yuan Ye (MDC)

• PKU Mao Yajun (MUC)

You Zhengyun (MUC, XML)

GeometryGeant4

Tracking

DetectorDigitization

Response

Hitobjects

EventGENBESGenerator

HepEvtformat

Raw dataMC truth

BOOSTarchitecture

BOOST: 3 phases

Phase 1: Framework prototype, simple geometry with hit

Finished in May 2003

Phase 2: Detailed Geometry with simple digitization

Finished in May 2004, MC data were produced to

tune the reconstruction program

Phase 3: Detailed digitization, physics validation

Need information from beam test and data.

We are here!

Generator

Simulation

Calibration

Reconstructin

Anaysis

Generator

Simulation

BOOST working in BOSS

BOOST based on GEANT4

BOSS based on GAUDI

MC data

Status

• Event generator

All BESII event generators work in BESIII

• MC-truth data (similar to MCMADE in BESII)

Add more information for each sub-detector

Useful for comparison between reconstruction and MC

also for track-match extrapolation

Still in Ascii format, binary format needed

Status

• Detector Description (new hardwire design)

MDC: new MDC geometry with larger-angle axial layers

TOF: new barrel TOF with double layer structure

EMC: both barrel and end-cap EMC with support structure

MUC: geometry with new alignment data

(MDC axial cells and EMC irregular crystals)

Insensitive regions:

Beam-pipe

super-conducting magnet

Status

• GDML (XML) Two options for detector description 1. Build the detector by Gean4 classes

2. Use XML file with GDML

can be shared by reconstruction and event display

Detector G4 objects

Geometry

XML file

GEANT4classes

Simulation

Geometry

Data file

GDMLconverter

DetectorRoot objects

ROOTconverter

ReconstructionEvent display

Status

• Digitization

MDC: wire resolution from beam test (two-Gaussian forms)

other effects like Q-t correction, etc

work with calibration group

TOF: revisited, all physics mechanism considered

EMC: photon statistics, noise, and electronic readout

MUC: noise added

Status

• Magnetic field Still 1Tesla uniform field used inside the magnet

Geant4 parameters (step size, intersection) optimized

sensitive to momentum resolution

3-D simulated field map in the whole detector will come

• Physics interactions 2 EM packages: standard (1 KeV) and low (100eV)

Almost the same for BESIII

Geant4 recommended hadronic packages (QGSP, …)

Problem: some have strange behavior

Performance

• Running status Installed on different IHEP linux-platforms

lxplus, besfarm, koala

Running stable for about 1M physics events

no crash, no dead-loop

Easy to use in BOOST and BOOS

user control card

Performance

• Simulation speed Depends on event type, cuts, machine, …

Using Tester, Bhabha, Dimuon, Hadronic,

inclusive J/Psi events

CPU time, roughly, 1 second for one event

on besfarm (PIII-933)

Still slow compared with SIMBES

Geant4 is 2 times slower than Geant3

Performance

• Memory consumption Roughly 80MB memory needed

About 4MB memory leak for 10K events

Memory leak depends on particle type,

mainly comes from G4 interaction packages

G4 is still in improvement

Performance

• Physics quantities Key physics quantities/parameters checked

momentum/time/energy resolutions,

reconstruction efficiencies, …

They are close to the Design Report!

(For detail, see other talks in this meeting)

A good starting point for physics study

Performance

Sub-detector Designed MC MDC σw =130 μm 110 (beam test)

σp/p = 0.5% 0.4%

σdE/dx = 6-7% 6%

TOF σt = 90 ps 85

EMC σE/E =2.5% 2.2%

σxy < 6mm 4.8

MUC ε(μID) =95% 96%

ε(π->μ)= 6% 6.2%

MDC tracking performance

Note: Fitted with single Gaussian in each bin.

μ- at pt = 1GeV/cMomentum resolutionσ = 0.4MeV

Momentum resolution Vs Pt (e-,μ-,π,p)

Effective speed Veff = 17.7 cm/ns

Effective attenuation length λ= 318cm

Intrinsic time resolution at z=0 : 85 ps

beam test result

EMC (barrel) Energy Resolutionsingle gamma

4E/E=[(p0/E) (p1/ E) p2]%

Planned development

• Move to new G4 and GDML – 02/2006 Gean4.7.1 Geant4.8.0 (12/2005)

GDML1.0 GDML2.3.0 (08/2005)

• Interaction packages – 12/2005 Study more hadronic packages

Important but not urgent

Planned development

• Background simulation – 04/2006 BESIII background much higher than BESII

Event mixing with beam related background

Bian Jianguo is working on it

• Trigger simulation – 12/2006 Implement trigger algorithm in offiline

Cao Guofu is working on it

Planned development

• Real detector simulation – 12/2006? Detector performance changes with running time

Rethink about BESII scheme

BESII: a separate procedure for realization

Using Calibration constants? Wire resolution/ efficiency, dead/hot channels, …

• Detector response – 01/2006… Long-term work, work with calibration group

Summary

• Large MC data samples can be produced for physics study

• MC performance is generally satisfactory

• Physics validation in progress

Thank you!