Alberto Oliva INFN/University of Perugia Tracker meeting 24/10/2006 Beam test 2003Beam test 2003 Goal: charge discrimination algorithm with high efficiency

Alberto Oliva Alberto Oliva INFN/University of PerugiaINFN/University of Perugia

Tracker meetingTracker meeting24/10/200624/10/2006

• Beam test 2003Beam test 2003• Goal: charge discrimination algorithm with high efficiency and low Goal: charge discrimination algorithm with high efficiency and low

contaminationcontamination• Signal characterization for charges from Beryllium to NeonSignal characterization for charges from Beryllium to Neon• Probability PProbability PZZ that a cluster is generated by a particle of charge Z that a cluster is generated by a particle of charge Z• Implementation of a probability test for the estimation of the Implementation of a probability test for the estimation of the

charge associated to a track charge associated to a track • Status of the workStatus of the work

Beam Test Oct 2003: Beam Test Oct 2003: Ion Charge AnalysisIon Charge Analysis



Beam test 2003Beam test 2003

Ions of average energy Ions of average energy »» 10 GeV/n 10 GeV/n extracted from an Indium beam of 158 GeV/n extracted from an Indium beam of 158 GeV/n on a Beryllium target, selected by A/Z ratio:on a Beryllium target, selected by A/Z ratio:- A/Z = 1.00, 6% of tot. ev., mainly - A/Z = 1.00, 6% of tot. ev., mainly protonsprotons- A/Z = 2.00, 77 % of tot. ev., - A/Z = 2.00, 77 % of tot. ev., HeHe component is dominant (Be is suppressed) component is dominant (Be is suppressed)- A/Z = 2.25, 16 % of tot. ev., - A/Z = 2.25, 16 % of tot. ev., 99BeBe component is dominant component is dominant - A/Z = 2.35, 1 % of tot. ev., - A/Z = 2.35, 1 % of tot. ev., 77LiLi component is dominant (He is suppressed) component is dominant (He is suppressed)



Sample selectionSample selectionCalibrationCalibration ● pedped »» 2.5(3) for p(n)-side 2.5(3) for p(n)-side

Clusterizzation Clusterizzation ● Seed with Signal/Noise Seed with Signal/Noise (SN) > 5.(SN) > 5.● Neightboring strips with Neightboring strips with SN > 2.SN > 2.

Identification of an event of Z>2Identification of an event of Z>2● One and only one cluster One and only one cluster for each ladder/side (only 12 clusters)for each ladder/side (only 12 clusters)● Cluster charge amplitude > helium total charge Cluster charge amplitude > helium total charge ● A good probability for A good probability for 22 of the of the linear fitlinear fit between the 6 points of p(n) side between the 6 points of p(n) side



Cluster amplitude: Cluster amplitude: total charge of a cluster, all strips (total charge of a cluster, all strips (ssii))Impact Point (IP): Impact Point (IP): obtained with a 5 point fit obtained with a 5 point fit

impact point VS cluster amplitude: impact point VS cluster amplitude: n-siden-side

NeNe

FFOO

NNCC

BBBeBe

SaturationSaturation



Cluster amplitude: Cluster amplitude: total charge of a cluster, all strips (total charge of a cluster, all strips (ssii))Impact Point (IP): Impact Point (IP): obtained with a 5 point fit obtained with a 5 point fit

impact point VS cluster amplitude: impact point VS cluster amplitude: p-sidep-side

SuperpositionSuperposition

NeNe

FFOONNCCBBBeBe

The charge separation will be done with the n-side knowledge...The charge separation will be done with the n-side knowledge...



n-side: n-side: distribution distribution

Three “Three “ regions” regions”● Readout regionReadout region: : ·· 0.15 or 0.15 or > 0.85 > 0.85● Interstrip regionInterstrip region: 0.15 < : 0.15 < ·· 0.35 or 0.65 < 0.35 or 0.65 < ·· 0.85 0.85 ● Not read strip regionNot read strip region: 0.35 < : 0.35 < ·· 0.65 0.65

RR RRNN104 104 mm 104 104 mm

: : charge center of gravity between the two higher strips charge center of gravity between the two higher strips

Charge samplesCharge samples● 3 different charge specta3 different charge specta● Charge peak selection on the sixth ladder Charge peak selection on the sixth ladder



n-side: signal characterization n-side: signal characterization ● Sixth ladder amplitude selection in a restricted window around the energy loss peaks Sixth ladder amplitude selection in a restricted window around the energy loss peaks

(first ladder peak selection to study the sixth ladder)(first ladder peak selection to study the sixth ladder)● Study the charge distributions on the other laddersStudy the charge distributions on the other ladders● Signal characterization with a Signal characterization with a Landau Landau Gauss + exponential tail Gauss + exponential tail● Low efficiency and great purity samplesLow efficiency and great purity samples

7 parameters7 parameters● 1 for the position (1 for the position (MPVMPV))● 4 shape parameters (slope, 4 shape parameters (slope, ,...),...)● 2 normalizations2 normalizations



n-side: purity of the samplesn-side: purity of the samples● fit the all charges spectrum of the sixth ladder: the only free parameters are the normalizations of the fit the all charges spectrum of the sixth ladder: the only free parameters are the normalizations of the

charge distributions charge distributions ● integration of the charge contributions around the peaksintegration of the charge contributions around the peaks

● sample puritysample purity : : »» 98% 98% with a Z – 1 contamination of with a Z – 1 contamination of »» 1% 1%

● low efficiencylow efficiency

B C NB C N

contamination of B contamination of B In the C sampleIn the C sample

Z – 1Z – 1

improve the efficiency using a likelihood based testimprove the efficiency using a likelihood based test



n-side: shape parametersn-side: shape parameters● Shape parametersShape parameters

– : width of the Gaussian distribution: width of the Gaussian distribution– : width of the Landau distribution: width of the Landau distribution– : slope of the exponential function: slope of the exponential function– wwexpexp: weight of the exponential function (Area: weight of the exponential function (Areaexpexp/Area/Areatottot) )

● Given Z and Given Z and these parameters are these parameters are similarsimilar for the different ladders (the maximum for the different ladders (the maximum variation is of the order of variation is of the order of »»10%) 10%)

● MPV values differ from ladder to ladderMPV values differ from ladder to ladder

wwexpexp



n-side: relative gainn-side: relative gain● Ladder to ladder differences are parametrized by a traslation coefficient:Ladder to ladder differences are parametrized by a traslation coefficient:

● The parameter GThe parameter Gll is independent from Z and is independent from Z and relative gainrelative gain● A unique charge parametrization can be used for all the laddersA unique charge parametrization can be used for all the ladders

GG22 »» 2% 2%

GGll function of Z for different ladders ( function of Z for different ladders ( fixed) fixed) GGll function of Z for different function of Z for different (ladder fixed) (ladder fixed)



n-side: probability and likelihoodn-side: probability and likelihood

Probability distribution on the single ladderProbability distribution on the single ladder● The probabilityThe probability PPZZ that a cluster corresponds to a Z ion is defined as: that a cluster corresponds to a Z ion is defined as:

Some Oxygen in the Some Oxygen in the Fluorine sampleFluorine sample

Likelihood on n laddersLikelihood on n ladders

● we want 0 we want 0 ·· L L ·· 1, then: 1, then:– if L(Z) < 10if L(Z) < 10-50-50 L(Z) = 10 L(Z) = 10-50-50

– L(Z) = 1 + L(Z)/(50L(Z) = 1 + L(Z)/(50¢¢n)n)L(Z = 8)L(Z = 8)



n-side: likelihood ratio test (6 ladders)n-side: likelihood ratio test (6 ladders)

InefficiencyInefficiency

(Nitrogen impurity)(Nitrogen impurity) contaminationcontaminationcontaminationcontamination

(Oxygen impurity)(Oxygen impurity)inefficiencyinefficiency

● which test for the charge estimation?which test for the charge estimation?– Maximum likelihoodMaximum likelihood: has the maximum efficiency : has the maximum efficiency – Recursive likelihood ratioRecursive likelihood ratio: a parametrized contamination : a parametrized contamination

Separation between Z and Z – 1 Separation between Z and Z – 1 Separation between Z and Z + 1 Separation between Z and Z + 1



n-side: inefficiency and contamination (6 ladders)n-side: inefficiency and contamination (6 ladders)

● cc11 and c and c22 are nearly charge independent are nearly charge independent● for for cc11 = 0.53 e c = 0.53 e c22 = 0.40 = 0.40 efficiency efficiency »»98% and contamination 98% and contamination »»1%1%

● tests have been applied to clean samples (tests have been applied to clean samples (»» 98%) tuning c 98%) tuning c11 and c and c22● inefficiencyinefficiency: percentage of not recognized Z events: percentage of not recognized Z events● contaminationcontamination: percentage of not–Z events recognized as Z: percentage of not–Z events recognized as Z



n-side: inefficiency and contamination (6 ladders)n-side: inefficiency and contamination (6 ladders)

● Inefficiency and contamination can be an Inefficiency and contamination can be an overestimatedoverestimated because the sample are not because the sample are not clean (there is clean (there is »»1% of Z – 1 events in each Z sample) 1% of Z – 1 events in each Z sample)



n-side: charge with only one laddern-side: charge with only one ladderC

harg

e re

cons

truc

ted

with

the

C

harg

e re

cons

truc

ted

with

the

six

thsi

xth

ladd

er la

dder

Charge reconstructed with the likelihood ratio of Charge reconstructed with the likelihood ratio of 5 ladder5 ladder

1010 0.000.0066

0.990.9933

99 0.010.0133

0.940.9455

0.040.0411

88 0.010.0144

0.950.9533

0.020.0299

0.000.0011

77 0.010.0111

0.950.9588

0.020.0277

0.000.0011

66 0.010.01 0.960.9611

0.020.0266

55 0.000.0099

0.960.9622

0.020.0266

0.000.0011

44 0.940.9455

0.040.0422

0.000.0033

ZZ 44 55 66 77 88 99 1010

● ““real charge” estimated with 5 ladders real charge” estimated with 5 ladders ● the charge on the last ladder is reconstructed with the maximum likelihood methodthe charge on the last ladder is reconstructed with the maximum likelihood method● the right charge is reconstructed at the right charge is reconstructed at »»95%95% while at while at »»3% is reconstructed as Z – 13% is reconstructed as Z – 1

iteration of the charge selection procedure ...iteration of the charge selection procedure ...



p-side: problemsp-side: problems● superposition of the total charge profile function of I.P. for the different ions superposition of the total charge profile function of I.P. for the different ions ● not linear relation between amplitude and energy deposited [ not linear relation between amplitude and energy deposited [ B. Alpat et al., NIM A B. Alpat et al., NIM A

540 (2005) 121–130540 (2005) 121–130 ] ]● a deformed a deformed distribution (the spatial resolution has a strong systematic component) distribution (the spatial resolution has a strong systematic component)

NeNe

FFOONNCCBBBeBe



p-side: problemsp-side: problems

p-sidep-side

● superposition of the total charge profile function of I.P. for the different ions superposition of the total charge profile function of I.P. for the different ions ● not linear relation between amplitude and energy deposited [ not linear relation between amplitude and energy deposited [ B. Alpat et al., NIM A B. Alpat et al., NIM A




p-side: problemsp-side: problems

HeliumHelium CarbonCarbon

MM RRRR

● superposition of the total charge profile function of I.P. for the different ions superposition of the total charge profile function of I.P. for the different ions ● not linear relation between amplitude and energy deposited [ not linear relation between amplitude and energy deposited [ B. Alpat et al., NIM A B. Alpat et al., NIM A




p-side: strip correctionp-side: strip correction

ProofProof● A n-cluster is associated to each p-clusterA n-cluster is associated to each p-cluster● On p-side only the seed strip is considered On p-side only the seed strip is considered ● Interstrip region: 0.45<Interstrip region: 0.45<pp<0.55<0.55● On n-side the cluster amplitude is On n-side the cluster amplitude is

consideredconsidered● Combine the information of all the ladder Combine the information of all the ladder

using the relative gainusing the relative gain

Proposed correctionProposed correction● a polynomial fit of the seed VS n-side amplitudea polynomial fit of the seed VS n-side amplitude● Linearization Linearization (hard, soft)(hard, soft)● Correction applied to all the p-cluster strips Correction applied to all the p-cluster strips (the same for all the ladders)(the same for all the ladders)

no correctionno correctionsoft correctionsoft correctionhard correctionhard correction

HypothesisHypothesis● The The deformation could arise from a not linear behaviour in the single strip signal deformation could arise from a not linear behaviour in the single strip signal



p-side: p-side: distribution distribution

● moremore uniform uniform● Implantation structureImplantation structure



p-side: spatial resolutionp-side: spatial resolution

● ladders allignement ladders allignement ● fit of a 5 point track fit of a 5 point track ● analisys of the analisys of the residual residual x distributionx distribution● evaluation of the spatial resolution (evaluation of the spatial resolution (»» x x - - fitfit))

residual



p-side: charge characterizationp-side: charge characterization● Charge samples are obtained with the n-side likelihood testCharge samples are obtained with the n-side likelihood test● Clean separation in Clean separation in region is now possible also on p-side region is now possible also on p-side● The ladders differences can be parametrized with a single parameter The ladders differences can be parametrized with a single parameter

(relative gain)(relative gain)● A set of p.d.f. have been calculated for the different ions A set of p.d.f. have been calculated for the different ions ● Costruction of a p-side likelihood testCostruction of a p-side likelihood test



p-side: test efficiencyp-side: test efficiency

99 0.000.0055

0.050.05 0.940.9433

88 0.090.0955

0.900.9033

0.000.0022

77 0.100.1099

0.880.8855

0.000.0022

0.000.0011

66 0.000.0022

0.020.0255

0.980.9844

0.000.0055

0.000.0011

55 0.010.0122

0.960.9611

0.020.0266

44 0.940.9499

0.050.05

ZZ 44 55 66 77 88 99

Charge reconstructed with 6 Charge reconstructed with 6 n-siden-side ladders ladders

Cha

rge

reco

nstr

ucte

d w

ith 6

C

harg

e re

cons

truc

ted

with

6 p

-sid

ep

-sid

e la

dder

s la

dder

s

● the p-side charge is reconstructed with the maximum likelihood methodthe p-side charge is reconstructed with the maximum likelihood method● the right charge is recognized in the right charge is recognized in »» 90 90% % of the eventsof the events



Status of the workStatus of the work

Implementation of a Implementation of a charge discrimination algorithm with high charge discrimination algorithm with high efficiency and low contamination. An AMS-note is in preparation:efficiency and low contamination. An AMS-note is in preparation:

Use the algorithm to find fragmentation events and to study their Use the algorithm to find fragmentation events and to study their topologytopology

Compare these result with a FLUKA Monte Carlo simulation Compare these result with a FLUKA Monte Carlo simulation

Documents

Alberto Oliva INFN/University of Perugia Tracker meeting 24/10/2006 Beam test 2003Beam test 2003 Goal: charge discrimination algorithm with high efficiency