P. Gay Energy flow session 1

Analytic Energy FlowAnalytic Energy Flow

F. Chandez P. GayS. Monteil

CALICE Coll.

P. Gay Energy flow session 2

OutlineOutlineSample & frameworkSample & framework

Photon reconstructionPhoton reconstruction–Photon reconstructionPhoton reconstruction

–Fake rates Fake rates

–hh/photon separation/photon separation

–K° /photon separation K° /photon separation

–Neutron/photon separationNeutron/photon separation

Neutral hadron reconstructionNeutral hadron reconstruction–K° and neutron reconstruction K° and neutron reconstruction

–Neutral/charged hadron separation Neutral/charged hadron separation

First step in EFlow First step in EFlow

SparsificationSparsification

P. Gay Energy flow session 3

Framework Framework Charged w/ Tracking system Si/W e.m. Calorimeter Digital hadron calorimeter

CALIMERO

SampleSampless WW pair production at 800 GeV

And particules extracted from those events

Simulated w/ MOKKA

More info. about CALIMERO in simulation session

P. Gay Energy flow session 4

ProcedureProcedure

Identification and reconstruction of all eflow objects

Ejet = Σ Eem + Eneutral h + Echarged

P. Gay Energy flow session 5

Photon reconstructionPhoton reconstruction E M I L E

Beside the Standard approach ( e.g. REPLIC), new one is developped

Main directions•3D•Long range•No seed

Energy Measurement Intended for Low Energy em showers (1)

(1) Cf. Talk of ECFA-DESY Workshop in PadovaP. Gay, F. LeDiberder, S. Monteil, F. Yermia

P. Gay Energy flow session 6

Energy Measurement Intended for Low Em showers

A terminal pad defines an EMILE object •Every characteristic of the cluster is built through the dij weighing matrix •The information from a pad could be shared by many objects

•The dij terms are determined between every pair of pads in the event but pad j should be on a layer outer than the pad i ie follows the e.m. shower development•All pads are connected w/o any initiate pad (in constrast w/ max. energy pad rule)

P. Gay Energy flow session 7

Photon reconstruction & IDPhoton reconstruction & ID

Variables involved longitudinal and Transverse profiles pad multiplicity Centre-of-gravity position

photons

Charged hadrons≈65% @100 MeV ↙

↙

P. Gay Energy flow session 8

Photon energy resolutionPhoton energy resolution

Fake rate (h±) = 8%

13%/√E ± 1.2%

P. Gay Energy flow session 9

Photon separationPhoton separationMethod based on pattern recognition with mip identification and vertexing (if any) (2)

MOKKA GeometryFew examples w/ γ 1 GeV /π±

10 GeV

(2) Cf. Talk of ECFA-DESY Workshop in Padova

P. Gay Energy flow session 10

Fake rateFake rate Tested with charged and

neutral hadrons from WW

Impact of track information

P. Gay Energy flow session 11

Charged hadron/photon separationCharged hadron/photon separation Tested with pions, kaons and protons

Efficiency as a function of the distance

14.5%/√E ± 2%

P. Gay Energy flow session 12

Neutral hadron/photon separationNeutral hadron/photon separation

Tested with neutral kaons and neutrons

Efficiency as a function of the distance

13%/√E ± 2% neutrons

16%/√E ±.5% kaons

P. Gay Energy flow session 13

Neutral hadron reconstructionNeutral hadron reconstructionA crude approach based on pad multiplicity and em energy

deposited in ECAL

Dedicated work performed by Anatoli and Arthur, cf their talks

Pad multiplicity and em energy are linearly combined

EK≤3GeV↑

36%/√E ± 6%

Behaviour is preciably different w/ neutron

Pad multiplicity

P. Gay Energy flow session 14

Neutral / charged hadron separationNeutral / charged hadron separation

pattern recognition, mip id and vertexing

In development presently will be improved soon

Ereco/EMC

P. Gay Energy flow session 15

Angle reconstructionAngle reconstructionPhoton Neutral Hadron

Isolated particles

Barycentre thru EMILE Geometric Barycentre of fired pads

P. Gay Energy flow session 16

Preliminary resultsPreliminary results

Preliminary means •No tuning per isolated particle •No weighing of the three components•No tuning of the efficiency (wrt fake rate for instance)

P. Gay Energy flow session 17

Preliminary resultsPreliminary results Assuming an ideal sparsification for WW pair production including fake contribution from other particles

Very hard photon are lost, PHOTID bias? Can be easily solved↓

Photons Neutral hadrons

P. Gay Energy flow session 18

Preliminary resultsPreliminary results WW pair production events

(Erec-Emc)/Emc

P. Gay Energy flow session 19

Preliminary resultsPreliminary results

Previous result

Neutral hadron reconstructed replaced by the Truth MC contribution↙

↙

P. Gay Energy flow session 20

SparsificationSparsificationPrevious results assumed an ideal sparsification, not necessarily an advantage e.g. a charged/neutral combination is considered as charged one even if they are separated by 10 cm A complete event has to be divided in rather small sub-objects w/ a

majority of the pads linked to a given particle in only one sub-object.

Sparsification based on simple neighbouring rules on Virtual Tower Pads are grouped in a given tower according a [Θ,φ] direction

θ

φ

High granularity and similar geometry in both ECAL & HCAL

w/ a neighbouring rule of ~10X10, 88% of the Tracks are 88% included in the sub-object and 2 Tracks are connected to a sub-object

Neighbouring rule3X3

5X5

7X7 9X915X15

25X25

<#Track/#sub-object>0.8 1.3 1.6 1.8 2.5 3.5

Then the study based on couple of particles is realistic

P. Gay Energy flow session 21

ConclusionsConclusions No clustering (only sparsification to reduce the size of the problem) New (Stochastic) approach in photon reconstruction (pad can be linked to different particles) dedicated to low energy photons Thanks to High granularity both in ECAL and HCAL Pattern recognition is allowed High granularity and similar geometry in both E- &HCAL allow pseudo-projectivity Results presented are based on study of couple of particles from Physics event configurationMany features have to be tuned and be improved; only a crude

approach has been performed here An effort on neutral hadrons is essential

Anyway the preliminary results based on Pattern recognition (high granularity and similar geometry in both E- & HCAL ) are very promising

⁂ Sparsification w/ virtual tower opens the way for EFlow package