29/4 /2010 – II Panda Russia Meeting – Maria Pia Bussa

29/4/2010 – II Panda Russia Meeting – Maria Pia BussaRecent results from Muon simulation and software in

Torino

Report on software activities with PandaRoot in Torino

Design of the Muon Tracker

(George Serbanut)

MDT Pattern Recognition

(Stefano Spataro)

Muon vs Pion discrimination in PID

(Marco Destefanis)

Design of Muon Tracker (MDT)

highest efficiency in

muon identification

benchmark channel

Drell-Yan

signal-to-noise 10-6

Xpp

strong hadronic background

collaborationJINR – Dubna (RU)

Gieen – Germany (D)

29/4/2010 –II Panda Russia Meeting – Maria Pia BussaRecent results from Muon simulation and software in

Torino

Barrel

Endcap

Muon Filter

Forward CPU Time

Simple

3 sec

Real 3 min

MDT Design

Room for improvement?

Magnet not aligned with MDT(dec 2009)

10 @3GeV/c

Geometry Implementations

Waiting for new design


Torino

Muon System in PandaRoot

TS_BarrelTS_Endcap

MF

FRS

Credits to Valery

Rodionov

All available in

PandaRoot and tested

Barrel Encap Muon Filter

PndMdt *Muo = new PndMdt("MDT",kTRUE):

Muo->SetBarrel("torino");

Muo->SetEndcap("torino");

Muo->SetMuonFilter("torino");

fRun->AddModule(Muo);

Simplified Geometry(George Serbanut)

simplified geometry

ArCo2 planes

2,5 cm thickness


Torino

Magnet Design

PndMdt *Muo = new PndMdt("MDT",kTRUE);Muo->SetBarrel…Muo->SetMdtMagnet(kTRUE);Muo->SetMdtMFIron(kTRUE);fRun->AddModule(Muo);

FairModule *Magnet= new PndMagnet("MAGNET");Magnet->SetGeometryFileName

("FullSolenoid_V842.root");fRun->AddModule(Magnet);

FairModule *Magnet= new PndMagnet("MAGNET");Magnet->SetGeometryFileName

("FullSuperconductingSolenoid_V831.root");fRun->AddModule(Magnet);

Full CAD conversion (Tobias Stockmanns)

Coils CAD conversion (Tobias Stockmanns)

MDT Design - TDR (George Serbanut)

MDTPattern

Recognition

MdtHit Energy Loss > 0MdtHit Position Smearing 0.3 cm -> 1 cm bar


Torino

Stefano Spataro

Pattern Recognition

MdtHit from inner layer

one tracklet PndMdtTrk

closest hit in next layerin a search cone

and so on…

Endcap and Muon Filterthreated as single

module

and so on…


Torino

Track Propagation to MDT layers

centraltracking

GEANEextrapolatio

n

MDT hit(layer 0)


Torino

@ 1 GeV/c

@ 3 GeV/c @ 3 GeV/c

Barrel Propagation


Torino

Residuals for track propagated to first MDT

layer

Test Simulation Data

5000 events PID , P 1, 3 GeV/c [5°, 90°] [0°, 360°]

@ 1 GeV/c

Momentum Loss

Vertex – MDT Layer 0

EC

BARREL

@ 1 GeV/c

EC

BARRELDISC is missing

Muon Detection @barrelMDT layer multiplicity

vs pMUON (barrel) PION (barrel)

Single track reconstructed in 0-4 GeV

range

@ 3 GeV/c @ 3 GeV/c

Fired Layers – 3 GeV/c

Fired Layers – 1 GeV/c

@ 1 GeV/c @ 1 GeV/c

muon global tracking with the Muon System - pattern recognition based on GEANE track follower - implementation of MDT inside the Kalman filter - Multivariate algorithm for muon identification

rejection of the secondary muon background

acceptance study for dimuon in J/y formation

just started search for discriminating variables for PID

29/4/2010 – II Panda Russia Meeting – Maria Pia BussaRecent results from Muon simulation and software in

Torino

Work in PandaRoot @Torino

Requirements for Particle Identification

MVD, TPC/STT, Cherenkov, EMC, MDT…

different detectors for PID

covering

different momentum/angle ranges

handling of different PID signals (dE/dx, C, EMC shower,

…)

combining several PID detectors to improve identification

if one detector does not contribute to PID, it should not

decrease the identification perfomances

8/3/2010 – XXXII Panda Meeting, GSI - Stefano Spataro

Bayesian Approach for Particle Identification in Panda(Root)

Requirements for Particle Identification (II)

PID procedure should be as much as possible automatic

PID depends also on analysis

we need to separate Detector response (i.e. resolution)

Event/track selection (analysis)

handling of different PID signals (dE/dx, C, EMC shower,

…)

combining several PID detectors to improve identification

if one detector does not contribute to PID, it should not

decrease the identification perfomances



The Bayes Theorem

If many detectors/algorythms constributing to PID

k = MVD dE/dx, DRC C…

kk hxphxL )|(|

Global Likelihood

Probability that a given track with given params x

corresponds to particle type h

pKeh

hPhxL

hPhxLhxP

,,,,

)()|(

)()|(|




Torino

M. DestefanisM. Destefanis

Search for Search for parameters parameters

discriminating discriminating vs vs

100000 events PID -, -

P 0.5, 4.5 GeV/c [0°, 90°] [0°, 360°] no Forward RS100% efficiency

angle

angle

- -

Barrel Barrel+EC+MF EndcapC+MF

Total nuber of layer hit vs momentum@vertex

8/3/2010 – XXXII Panda Meeting, GSI – Marco Destefanis

Muon subgroup meeting

muon

pion

- -

Barrel Barrel+EC+MF EndCap+MF

Last layer vs total number of layer hit



muon

pion

- -

Barrel Barrel&EC&MF EC&MF

muon

pion

Iron crossed by the particles



Sailing through pions

Counts

Counts

cm

cm

Interaction and decay of Interaction and decay of --

decay

interaction

Detector description ready in PandaRoot

Realistic simulation needed to evaluate layout and number of channel required

Test on prototype essential to check MC expectations

Muon System included in pattern recognition, track reconstruction and PID

lack of human resources for software work and beam test


Torino

Final remarks

29/4 /2010 – II Panda Russia Meeting – Maria Pia Bussa

Documents

Transcript of 29/4 /2010 – II Panda Russia Meeting – Maria Pia Bussa