Phase-2 Muon Simulations

Alexei SafonovTexas A&M University

Background Assumptions• GE-1/1 will save the day for

muon trigger between LS-2 and LS-3– GE-1/1 covers the

region 1.6<|h|<2.2 planned to be installed in LS2 as part of “early” Phase-2 upgrades

– Post-LS-2 is the worst time ever: no track trigger yet• Critical post LS-3 concerns– Avert loss of triggering in 2.2<|h|<2.5 (region beyond GE-1/1)– Take advantage of a terrific opportunity to expand physics reach

by extending offline muon coverage to 2.4<|h|<4.0

Bending Angle

• An increased lever arm of the combined CSC+GEM system allows accurate measurement of the bending angle– Excellent discrimination power to

distinguish soft muons from hard– Larger lever arm for “far” chambers

provides even better separation

View from the top of the CMS down

Trigger Rate

• Illustration of the achievable trigger rate reduction in the region covered by GEM station GE-1/1 using bending angle measured using GEM and CSC stations– Each Level-1 muon track of a given moment is required to have its measured bending angle be less than

the working point Df cut value defined for the momentum range, to which the track in question belongs

• In this simplified scenario, the tracks are required to satisfy a requirement of having hits in at least two stations (out of four possible)

• Results are compared with that for the standard CMS configuration used in 2012 (red line in the rate vs pT curve)

GE-1/1: Bending Angle Cut

• Illustration of the achievable trigger rate reduction in the region covered by GEM station GE-1/1 using bending angle measured using GEM and CSC stations– Each Level-1 muon track of a given moment is required to have its measured bending angle be less than

the working point Df cut value defined for the momentum range, to which the track in question belongs

• In this simplified scenario, the tracks are required to satisfy a requirement of having hits in three or more stations (out of four possible)

• Results are compared with that for the standard CMS configuration used in 2012 (red line in the rate vs pT curve)

MUON TRIGGER: POST LS-3

7

Trigger Concerns Past LS-3• Muon Level-1 Trigger will rely on tracking trigger and

Muon matching• The “double problem” region is 2.15<|h|<

2.5– Either large efficiency losses or high fake rate in

L1 Track Trigger– The exact same region where muon trigger

rates shoot up

• Solvable if we can suppress muon trigger rate by about ~x5 Muon gun pT>5 GeV

Efficiency includes track finding only. No muon system

inefficiencies incorporated.

Reco’ed stubs pT>2Stubs from true particles w/ pT>2

Tracker

Muon System

Post-LS3 Trigger Scenarios• Motivation: prevent collapse of CMS muon trigger coverage from

the current |h|<2.4 down to |h|< 2.1 or less• Strategy: build “Maximum Scenario” and see what works best• Maximum configuration:

– Near tagger ME-0 at the back of present HE with trigger capabilities in 2.1<|h|<2.4 (can be long or short)

– GEM stations:• “Old ”GE-1/1

– Already there in LS2 • “New” GE-2/1

– iRPC stations:• “New” GE-3/1 and 4/1

• Goal: seek a factor of ~x5 in trigger rate reduction– Evaluate impact of each new

component on the trigger

9

Bending Angle and Distance to IP• Apply the same technique in Station 2:

– It works, but not as well as in station 1

• Muon bending reduces due to radial B-field turning muons back

• Multiple scattering smearing reduces discrimination

• Reducing trigger rate in 2.1<|h|~2.4 requires measuring bending angle in close to IP stations

Trigger Rate Reduction: Preliminary2.1< |h| < 2.4 Signal muons

pT=30 GeVRate

reduction

Near stations: bending angle & new redundancyRequire ME-1/1 and cut on GE-0 – ME-1/1a bending angle

~98% X3.2

Require ME-2/1 and GE-2/1-ME-2/1 bending angle

~98% x1.4

Near stations: combined ~96% X3.75Additionally utilize redundancy in far stations:Additionally require stubs in YE- 3, 4 (either CSC or GRPC)

~100% (?) x1.2

Combined all stations ~96% X4.5

• The “all of the above solution” provides a strong rate reduction– Phase-2 muon trigger design: define “lose track trigger tracks” to

match with standalone L1 muons, use combined tracking and muon information to control the rate

GE-1/1: 1.6<|h|<2.1

* A combined effect of new redundancy in all 4 stations exclusive of bending angle reductions is x1.5

Simulation used pitch of 1.9 mrad (~2 mm @ R=1m)• Precision important for bending

angle, not for redundancy

Precision Timing in YE-3 and 4• New physics with B-mesons (Bs/d)– Trigger: two soft forward muons (on the same side)

• High precision timing (100 ps range?) can be used to confirm that both muons come from the same vertex – Need to measure t1-t2, many systematics effect can potentially cancel

– Need simulation to evaluate if it can help or not

• Reduce neutron hits by utilizing timing windows– Lower background can potentially benefit the single muon

trigger (more reliable points means better momentum measurement and thus lower trigger rate)

– It appears that windows can’t be less than a few nsec • That gives the required level of precision for the detectors

OFFLINE MUON RECONSTRUCTION: EXTENSION TO ETA=4.0

13

Forward Region: Physics• The hard part is the

region of 2.1 <|h|>2.4:– Highest background rates

yet least redundancy, most vulnerable at high luminosity

– Challenging B-field topology• Radial field turning muons

back

• Awkwardly, if there is one place to make large physics acceptance gains, it is in the forward region– Also improve MET by tagging muons in the forward region

HZZ4m : ~50% acceptance increase if hmax=2.44.0

14

Phase-2 Near Tagger ME-0 Scenario

• Near tagger ME-0 at the back of present HE• Coverage: 2.1<|h|<4.0

• Upper portion of 2.1<|h|<2.5 has trigger capabilities• Lower portion is only used in the offline

– Muon reconstruction based on matching tracks reconstructed in forward muon extension with hits in the muon system

General strategy for performance estimates (same strategy will be used to provide extrapolation tools)

In absence of GEANT simulation of the extended muon detector use Fastsim with forward pixel geometry muon detector emulated as a flat surface at |z| = 560 cm, covering |η| = 2.4-4.0 → only 2D hits for now material effects can be studied using parametrization in SteppingHelixPropagator

Propagate the generated-track initial state and covariance matrix (null at IP) to a surface at z = 560 cm, using the SteppingHelixPropagator

after the propagation, the covariance matrix will include the uncertainty from material effects only (multiple scattering, energy-loss fluctuations, bremsstrahlung, etc.)

Use position error on the muon detector surface to smear the propagated position and “emulate” a sim-hit

gen track

“sim-hit”

IP

r = (x2

+ y2

)

½

z560 cm

reco pixeltrack

rec-hit

y

x

r · Δφ

Δr

longitudinal view transversal view

rec-hit

“sim-hit”

IP

High h offline extension

Contributions to the total muon resolution after propagation to ME0, from

detector effects (multiple scattering): from RMS b/w “GEN” and “SIM” hits

pixel detector resolution: from RMS b/w “RECO” and “GEN” hits

Total resolution can be used to determine “matching windows” for muon tagging in ME0

e.g. at pT = 5 GeV/c, we can chooseΔη × Δφ = 0.002 ⇒ 2 RMS = 95% matching prob.

With this window, and the average expected pile-up in phase-2 (N ~ 0.25 tracks/pp inter. or 50 tracks/BX) we can estimate the mis-tag rates for topologies with

known vertex (e.g. H → 4μ): < 0.004 unknown vertex (e.g. VH → γγμ): < 0.8

*** Performance studies are on hold and will be redone with new pixel geometries

rφ coordinate RMS

Muon gun with pT = 5 and 20 GeV/c

Example: performance of ME0 as a tagger

THINKING OF SIMULATIONS STRATEGY FOR TP

Simulation Strategy Options for TP• Need reliable results on a short time scale - “emulated”

simulation?– Trigger: Use simHit or digis information for CSC and

extrapolators to make fake “hits” in detectors being simulated• Faithful representation of magnetic field and material budget important

for multiple scattering• Smear to fake resolution effects• Combine with track trigger simulation in FullSim, develop algorithms

– Use extrapolations for the new ME-0• Muon hits: use gen level particles, extrapolate to muon chambers using

realist extrapolators to create muon “hits”– Will wrap the machinery in producers making CMSSW objects usable in full

sim or fast sim studies • Inner tracks: Use whatever available for tracking (currently using fastSim

but waiting for a new layout, can switch to FullSim easily – whatever tracking people do, we will do the same)

Trigger: “Emulated Simulation”• Use the same emulator as used for making estimates

for ME0 for eta>2.1 and for GE2/1.• As a reference, below is the comparison of the full-sim

GE1/1 bending angle and the emulated-sim version

• The emulated-sim performance is reproduced well

GE1/1 FullSimGE1/1 FastSim

Extended Offline Coverage: Options

• Emulated simulation (slides 14-15) is working well• However, given that calorimeter simulation goes into CMSSW,

we can also add sensitive volumes behind it to emulate ME-0– SimHits created by GEANT, use simple smearing to emulate

detector granularity, wrap things up to make CMSSW objects • Will work in fullSim or fasSim

• Doable on the time scales of the TP

Trigger Simulation: Options• Can do the same trick and add sensitive volumes for

exploring trigger scenarios:– Add detectors:

• GE-1/1 is already there in all detail• GE-2/1 (almost done and will be in CMSSW in ~0.5 week)• RE-3/1 and RE-4/1 (already available)

Caveat: Non-Prompt Backgrounds• Neutron backgrounds are a potential concern

– Details in the next talk • Need flux estimations:

– Close to where we can modify FLUKA geometries and calculate rates where we need them, including in the space behind the new calorimeter

– Currently can already use dual-readout calorimeter geometry (thanks to them for their help!)

• Relies on FLUKA and is decoupled from CMSSW developments• Can go on a parallel track

Strategy: Options

• Seems like we can factorize simulation work:– In-CMSSW: performance studies of the “detector

package” using a combination of true GEANT simulation and reliable extrapolation techniques • But assume neutron backgrounds will be taken care of (they

are not in simulation)

– Outside-CMSSW: Optimization of what’s inside the “detector package”• Fluka for background rates and working out shielding issues• Optimize the inner structure of the “detector package”

– How many layers? Do we need material between detector sensitive layers for background rejection (where and how much)?