GEM-DHCal Performance and Energy Flow Algorithm Studies

1/9/2004 GEM DHCALJ. Yu

1

GEM-DHCal Performance and Energy Flow Algorithm Studies

*On behalf of the HEP group at UTA.

• Single Pion Performance Study• Study of Pythia events• Energy Flow Algorithm

– Single pion Track – cluster matching studies • Conclusions

jetsttee 6

ALCW 2004, SLACJae Yu

University of Texas at Arlington


2

Introduction• DHCAL a solution for keeping the cost manageable for EFA• Finer cell sizes are needed for efficient calorimeter cluster

association with tracks and subsequent energy subtraction • UTA focused on DHCAL using GEM for

– Flexible geometrical design, using printed circuit pads– Cell sizes can be as fine a readout as GEM tracking chamber!!– High gains, above 103~4,with spark probabilities per incident less

than 10-10

– Fast response• 40ns drift time for 3mm gap with ArCO2

– Relatively low HV• A few 100V per each GEM gap

– Possibility for reasonable cost• Foils are basically copper-clad kapton• 3M produces foils in large quantities (12”x500ft rolls)


3

UTA GEM Simulation • Use Mokka as the primary tool

– Kept the same detector dimensions as TESLA TDR– Replaced the HCAL scintillation counters with GEM

(18mm SS + 6.5mm GEM, 1cmx1cm cells) • Single Pions used for initial studies

– 3 – 100 GeV single pions– Analyzed them using ROOT

• Compared the results to TDR analog as the benchmark– GEM Analog and Digital (w/o threshold)

• ECal is always analog


4

UTA Double GEM Geometry

3.4 mm ArCO2

GEM3.1 mm

Simple GEM

Detailed GEM

0.

00

51

.

0

Cu

Kapton

ArCO2

G10

0.

00

5

6.5mm


5

Performance Comparisons of Detailed and Simple GEM Geometries

Detailed GEM75GeV

• 25.2sec/event for Simple GEM v/s 43.7 sec/event for Detailed GEM• Responses look similar for detailed and simple GEM geometry• Simple GEM sufficient

<E>=0.80 0.007MeV <E>=0.81 0.008MeV

Simple GEM75GeV


6

GEM-Digital: Elive vs # of hits for π-


7

EM-HCAL Weighting Factor• ELive=EEM+ W GEHCAL• For analog:

– L + G is used to determine the mean values as a function of incident pion energy for EM and HAD

– Define the range for single Gaussian fit using the mean– Take the mean of the Gaussian fit as central value– Choose the difference between G and L+G fit means as the systematic uncertainty

• For digital:– Gaussian for entire energy range is used to determine the mean– Fit in the range that corresponds to 15% of the peak– Choose the 15% G fit mean as the central value– Difference between the two G as the systematic uncertainty

• Obtained the relative weight W using these mean values for EM only v/s HCAL only events

• Perform linear fit to Mean values as a function of incident pion energy• Extract ratio of the slopes Weight factor W• E = C* ELive


8

Response - Comparison


9

GEM Analog & Digital Converted: 15 and 50 GeV π-

15GeV Analog 50GeV Analog

15GeV Digital 50GeV Digital


10

Resolution - Comparison


11

• GEM digital and analog responses comparable– Large remaining Landau fluctuation in analog mode

observed– Digital method removes large fluctuation Become

more Gaussian• GEM Energy resolutions

– Digital comparable to TDR – Analog resolution worse than GEM digital or TDR

• GEM is as good a detector as others for DHCAL

GEM Performance Study Summary


12

Does more Gaussian Behavior of GEM digital make sense?

• Gas detectors have small number of primary ionization electrons Very Landau like distributions

• Large amplification only stretches out the Landau distribution– Amplification does not increase the number of primary electrons– It only worsens the fluctuation

• The cells with large energy due to the fluctuation get saturated – Suppressing the large energy tail– While preserving low energy distributions


13

10 GeV

• Saturation occurs at every energy• Is this a good thing? I think so, as long as the

linear region in each energy bin is sufficiently wide

100 GeV


14

• Energy distribution of final state particles in jets• Choose a ΔR = 0.5 cone around a quark to define a jet• Determine energy fraction of jets carried by EM, Neutral

and Hadrons• Determine the relative distances between all pairs of

charged, neutral particles in the cone• Use two pions to study effective charged hadron energy

subtraction• Study of centroid finding algorithm

Analysis of jetsttee 6


15

Energy distribution in a jet


16

Fraction Energy of Particles in Jets

Electromagnetic

Neutral Hadrons

Charged Hadrons


17

R Between All Particles in Jets


18

• Pions E π- = 7.5 GeV chosen for study• Studied the energy distribution of pions in jet events

• Find the centroid of the shower ( HCAL ) using– Energy weighted method– Hits weighted method– Density weighted method

• Matched the extrapolated centroid with TPC last layer hit to get Δ and Δφ distribution

Energy Flow Studies for π-

TeVsjetsttee 0.16


19

• Identify layers with hits (at least 3 hits required)• Fit a line through all layers (at least 2 layers with

3 or more hits required) • Extrapolate the line to TPC last layer• Compare tpc with hcal and tpc with hcal

Determination of Calorimeter Centroid


20

Methods for determination of centroid

n

jij

n

jijij

i

E

E

1

1

n

jij

n

jijij

i

E

E

1

1

n

ijj iji Rd

,1

1

n

jij

n

jijij

i

d

d

1

1

n

jij

n

jijij

i

d

d

1

1

n

n

jij

i

1

n

n

jij

i

1

Hits Weighted Method

Energy Weighted Method

401:

:

1:

i

ilayerinhitsofNumbern

nj

Density Weighted MethodFor all three methods:


21

Density Weighted Method



- 7.5 GeV π-


22

- 7.5 GeV π-

Density Weighted Method




23

• GEM Analog and digital performance studies completed– GEM Analog resolution a nit worse than TDR and other studies due to large

Landau like fluctuation– GEM Digital resolution comparable with TDR and other studies– Threshold dependence in progress– Translate these resolutions into jet energy resolution

• A energy flow algorithm study using single pion events began R of single particles in typical jets and using 3 different methods– Compared the three methods

• Durham jet algorithm is being implemented• Resolving 2 pions as function of R using Mokka• Kaushik is working on his thesis that will contain both the studies• A visiting professor and an undergraduate student will join this

semester for EFA studies

Conclusions

Documents

GEM-DHCal Performance and Energy Flow Algorithm Studies