Mihael MakekWeizmann Institute of Science

HBD Meeting, 2/6/2010

Clustering algorithm („Weizmann clusterizer“)

• loop through all fired pads (“CellList”):• find pads with 3 pe < charge < 100 pe seed pads • build clusters around seed pads by summing the first neighbour pads with charge > 1 pe

• loop through the clusters („BlobList“) and merge clusters if they have overlaping pads

Matching of the CA tracks to HBD:• loop through the clusters again and find the one that is the closest to the track projection point

Minimum bias data 200 GeV Au+Au

nCentral = 278, nCells = 1655

Subtraction of pedestals• Method 1, define the average charge per pad:

• Method 2, define the average charge per fired pad:

where a[pad] normalizes pad area • Look at the <ch>pp as a function of centrality (nCentral) • Derive this dependence module by module• Subtract charge from each pad according to nCentral

pads of no. Total

]charge[pad a[pad] ch pp

pads fired of No.

]charge[pad a[pad] ch fp

Subtraction of pedestals 1• Correction functions:

example: WN2all modulesPossible causes of variations:• zero suppression + gain difference• reverse bias voltage

After subtraction of pedestals 1

nCentral = 278, nCells = 670

After subtraction of pedestals 1 • Run WIS clusterizer on subtracted data

• Electron tracks

• Matching distribu-tions for different nCentral bins

• 3 cuts on hbdd and hbddz (mom. corrected)

• Select only clusters with size > 1 pad

• Subtracted random background gene-rated by projecting tracks to different module

Subtraction of pedestals 2• Correction functions:

example: WN2

After subtraction of pedestals 2

nCentral = 278, nCells = 530

Comparison of the subtraction methods

Comparison of the subtraction methods

Comparison of the subtraction methods

Comparison of the subtraction methods

Comparison of the subtraction methods

Comparison of the subtraction methods

Summary • After subtraction of pedestals we do see electron signal in the most central events!

• Even after subtraction we are still picking additional charge (need to subtract more?!)

• Needs to optimized:• S/B ratio is dropping for the most central events• Electron „efficiency“ decreasing for the most central events

Outlook• Ideas to proceed:

• Optimize subtraction empirically monitoring the electron efficiency and validating results with the Accumulator

• Optimize lower threshold of the seed pad (e.g. 35 pe) expected effect: low charge background reduction

• Optimize upper threshold of the seed pad (e.g. 10050 pe): expected effect: reduction of the high charge clusters

• Optimize lower threshold for the neighbouring pads according to centrality (e.g. 1 0.06 * nCentral). Expected effects:

• reduction of the cluster size according to centrality• reduction of the cluster charge to centrality

BACKUP slide: electron ID cuts

Using central arms:• quality = 31, 51, 63• n0 > 2• abs(emcdz+1.0) < 10.0, abs(emcdphi-0.00023) < 0.030• ecore/mom > 0.6• chi2/npe0 < 10• disp < 5• prob > 0.01• ecore > 0.15

Using HBD:• abs(phbdz) < 27.0• remove tracks projecting to EN2• 3 sigma matching in hbddz and hbddphi• HBD clustersize > 1