Neutron Background Simulation: Infrastructure and Validation Vadim Khotilovich, Rick Wilkinson, with help from Piet Verwilligen, Alexei Safonov, Tim Cox.

Neutron Background Simulation: Infrastructure and Validation

Vadim Khotilovich, Rick Wilkinson, with help from

Piet Verwilligen, Alexei Safonov, Tim Cox

Simulation MeetingAugust 30, 2010

Outline• Intro• Neutron simulation infrastructure• Basic SimHit level results for muon systems

– From high-stat samples produced in 3_6_3 QGSP_BERT_HP physics list

– Comparison to older results– Plans

• Validation of QGSP_BERT_HP_EML and QGSP_BERT_EMLSN physics lists– low stat samples in 3_9_0_pre2

2

Intro

• Long-lived neutrons created, diffuse around collision hall

• They get captured by nuclei, emitting a photon

• Compton scattering or photoelectric effect makes MeV electrons, which cause hits in muon chambers

From Rick’s previous presentation:http://indico.cern.ch/conferenceDisplay.py?confId=55712

Why neutrons are hard to simulate?

• Because neutrons can live up to a second before making a signal

• They can’t be treated like ordinary minimum-bias pileup, because millions of collisions in the past can contribute

• log10TOF vs log10Ekin– Plot made in 2001

From Rick’s previous presentation:http://indico.cern.ch/conferenceDisplay.py?confId=55712

5

New Neutron Simulation Infrastructuregenerator

“Pythia6GeneratorFilter”generatorNeutrons

“Pythia6GeneratorFilter”

g4SimHits“OscarProducer”

g4SimHitsNeutrons“OscarProducer”

cscNeutronWriter“CSCNeutronWriter”

rpcNeutronWriter“RPCNeutronWriter”

dtNeutronWriter“DTNeutronWriter”

g4SimHits“NeutronHitsCollector”

generator“EmptyHepMCProducer”

mix“MixingModule”

generatorAny desired generator

g4SimHits“OscarProducer”

signal

MinBiasPileup

Sample

NeutronPileup

Sampleto digitizers

NPU

f*NPU

f = fraction of filled BXs in LHC orbit

MinBias simulated with:- special physics list- tracking time up to 1s- no pt and neutron cuts

Transform TOFs of >250ns SimHits into25ns wide time intervalwhile preservinghits’ time ordering

Special utility modulesto standardize collectionsfor MixingModule S

imM

uo

n/N

eu

tron

Neutron SIM Samples• CMSSW_3_6_3• QGSP_BERT_HP physics list• MinBias generated @ 14 TeV• ~105K events• Geometries

– Current detector geometry– Most of results: TDR RPC & YE4 shielding

• Along with transformed neutron SimHit collections keep the original ones produced by GEANT

Muon SimHits TOF vs. Ekin

• TDR geometry with YE4

• CSC&DT have lots of low energy hits– RPCs are

not so sensitive

CSC: hits vs. clusters of hits• Simple clustering of CSC hits in a

chamber layer:1. Sort by closes wire group, strip, TOF2. Take first hit3. Cluster other hits around it requiring

WG ≤ 1 & S ≤ 3 & TOF < 2 ns4. Repeat recursively…

• Very low energy hits are from secondary interactions

• Fraction of events with neutron simhits– Can be used for probabilistic

interpretations

• Comparing two geometries:– Current– TDR with YE4

• ME4/2 gets ~4.4 times more neutron hits without YE4 shielding– There is smaller effect in

other ring=2 stations– Should expect similar

magnitude effect in RPC’s RE4/2 & RE4/3

9

Fractions of evt. with n-hits: YE4 effect

Geometry:currentTDR with YE4

• SimHits from regular minbias vs. SimHits from neutrons

• TDR geometry with YE4

• CSC:– Most stations get hits from

neutrons more often then from MinBias (except ME1/1)

– Even with shielding ME4/2 is dominated by neutron hits

• RPCf:– MinBias contribution is higher in

most of stations (except RE 4/3, 4/2, 3/3)

• b/c neutron hits energies are lower

10

Fractions of evt. with hits: MB vs. neutrons

SimHits from:MinBiasNeutrons

SimHits from:MinBiasNeutrons

•T

DR

geo

met

ry w

ith Y

E4

11

Fractions of events with neutron hits

• TDR geometry with YE4

• NPU=25

• Fraction of filled BXs = 0.795

• Only neutron bars have text markers

• DT&CSC n-cluster fluxes ~2 times lower then n-hits fluxes 12

Neutron Hits Fluxes at Nominal LSimHits from:MinBiasNeutrons

SimHits from:MinBiasNeutrons

SimHits from:MinBiasNeutrons

SimHits from:MinBiasNeutrons

• CMS Trigger TDR results:– Full circles = Neutron hits– Solid lines + open circles =

MinBias

• Current estimates for neutron rates are– ~2-6 times lower in CSC– ~4-30 times lower in DT– Reason not clear

• Current estimates for MinBias rates are– CSC

• Somewhat higher, especially for closer to beamline stations

– DT• Reasonably similar• Slightly higher in outer stations

• Plan to produce somewhat similar plot for easier comparison 13

Neutron Hits Fluxes: Trigger TDR

• Neutron simulation infrastructure:– Extending the configuration for heavy ion simulation

• With help from Catherine Silvestre– Consolidation of all relevant modules in SimMuon/Neutron– Maybe adding an option to cmsDriver– Adopting new physics lists– Possibly: employ endcaps’ rotational and +-z symmetry to effectively increase

neutron samples statistics x34 for trigger rates studies• Plans for studies:

– Comparison to data• Special runs with 1-layer CSC pretrigger are planned to be taken soon• Will need 7 TeV neutron sample simulated

– DIGI level• Preliminary: fraction of events with CSC digis from neutrons is ~5 times lower than

fraction of events with neutron simhits– Trigger primitives level

• Preliminary: – Without YE4, ME4/2 gets ~40% more of LCTs b/c of neutrons – Much lower effect in other chamber types

– GMT• Combining CSC and RPCf in high pileup environment

14

Plans

Physics Lists Validation

15

• Standard one for use in CMS simulation:– QGSP_BERT_EML

• Too crude for neutron modelling

• Special “high precision” lists for modeling of long-lived neutrons:– QGSP_BERT_HP

• Older detailed multiple scattering model– QGSP_BERT_HP_EML

• New simplified/faster multiple scattering model

• Special parameterized neutron cross section table:– QGSP_BERT_EMLSN

• The newest parameterization

16

Physics Lists

• For each physics list:– 2K events sample– Produced with CMSSW_3_9_0_pre2 in the same conditions– Standard geometry– GEANT settings:

g4SimHitsNeutrons.Physics.type = 'SimG4Core/Physics/QGSP_BERT_***‘g4SimHitsNeutrons.Physics.FlagBERT = Trueg4SimHitsNeutrons.StackingAction.NeutronThreshold = 0.g4SimHitsNeutrons.StackingAction.MaxTrackTime = 1e9g4SimHitsNeutrons.SteppingAction.MaxTrackTime = 1e9# the following two enable simulation in the Quad regiong4SimHitsNeutrons.StackingAction.MaxTrackTimes[2] = 1e9g4SimHitsNeutrons.SteppingAction.MaxTrackTimes[2] = 1e9# no cuts on generator-level particlesg4SimHitsNeutrons.Generator.ApplyPCuts = Falseg4SimHitsNeutrons.Generator.ApplyEtaCuts = False

17

Validation Samples

QGSP_BERT_HP 17.2QGSP_BERT_HP_EML 16.2QGSP_BERT_EMLSN 1.4QGSP_BERT_EML 1.2

• For HP lists Most of time is spent in transport of neutrons from quad regions

• EMLSN timing: too good to be true?...

18

Timing (min/evt)

• ~x4 lower for EMLSN

• HP and HP_EML: within stat errors– Except ME4/2 where

quad region simulationis important

• DT & RPC:– Small statistics– x2-5 lower for EMLSN

in DT

19

Fraction of Events with n-hits in CSC

• _

20

Validation: TOF vs. Ekin in CSCQGSP_BERT_HP QGSP_BERT_HP_EML

QGSP_BERT_EMLSN QGSP_BERT_EML

Both exhibitTOF cut-offat 104 ns

• P

21

Validation: Ekin & TOF in CSC

Conclusions• QGSP_BERT_HP and QGSP_BERT_HP_EML show close

results, except for ME4/2 station• QGSP_BERT_EMLSN

– Neutron hit yield lower by x4– It doesn’t track particles longer then 104 ns

2222

BACKUP

23

• Neutrons leave hits in 1-2 layers most of time

• By chamber type (no YE4 schielding):

24

CSC # of layers with hitsMinBias

Neutrons

25

r-z Neutron hits Heatmap

Endcap in currentgeometry

TDR geometry with YE4

26

x-y Neutron hits Heatmap in Barrel