Emulsion Simulation Y. Caffari (IPN Lyon) F. Juget (IPH Neuchatel) A. Marotta (CNRS-IN2P3 Lyon) CERN...

Emulsion SimulationY. Caffari (IPN Lyon)

F. Juget (IPH Neuchatel)A. Marotta (CNRS-IN2P3 Lyon)

CERN 3-5/05/2006

Outline ORFEO development program Event simulation for algorithm

comparison Event simulation for data/MC comparison Electron data/MC comparison A new effect of the reconstruction seen

in the data/MC comparison that must be simulated

Outlook

ORFEO development program

The next version of ORFEO will work under the new general simulation framework (OpRelease);

The ORFEO version actually on the web (working with OpRoot) has been validated (micro-tracks, smearing…) update to the simulation procedure and new packages (to add

new features and reproduce effects not related to physics or to emulsions) will be tested under this version till the new one will be fully operative;

The OpRelease framework has 3 working possibility: full detector+rock, full detector, single brick;

The ORFEO smearing features are implemented in OpRelease;

A new simulation of the brick structure (CS doublet, packing… ) is under development;

A package to read the new output (in the official OpRelease format) and process events with the emulsion reconstruction program will come soon;

A I/O package to reproduce the old output will be provided too for back compatibility.

Event simulation for algorithm comparison

The algorithm comparison is performed by using the simulated data stored @ http://ntslab01.na.infn.it/opera_sim/ORFEO_production/ ;

The default parameters are chosen in such a way to be in “reasonable agreement” with the OPERA test beam data BUT they are not supposed to be used for data/MC comparison; in

fact, in such a case a “good” data/MC agreement is not expected; The efficiencies are parameterized by using the reference brick

Actually, this parameterization does not fit with the efficiencies parameterized in the other tests beam made, so in data/MC comparison a parameterization obtained from data should be used;

The micro-tracks smearing, in position and angle, are obtained by using a 2 points simulation of the micro-track (see next slide) x=0.25 micron, y=0.25 micron, z=2.5 micron; These values should be tuned in the case of data/MC comparison;

2 points simulation of a micro-track

10. GeV muons y2

x= x1 +x2 - y= y1 +y2 - z= z1 +z2

)tan()cos()tan()cos(

)tan()cos()tan()cos(

0

0

z

z

Sxzz

zzz

ySy

Sxzz

zzz

xSx

y1

z1

x2

z2

x1

z

0

00

2220

2220

)tan1(

)cos()sin(

)sin()cos(

LT

L

zLL

T

L

SSS

SS

SySxS

SySxS

By taking into account that dx and dy are statistical errors gaussian distributed and dz is a maximal error with a flat distribution, these assumptions automatically give the expected dependences on :

“Since the base-tracks are constructed by using the 2 points at base, the same results are obtained for free for base-tracks too”

Transversal and longitudinal resolution

for simulated 10.GeV pions at =0.7 mrad

SlopLong1=0.037

SlopLong2=0.037

SlopTr1=0.008

SlopTr1=0.008

Smearing parameters are: Sx=0.25 micron, Sy=0.25 micron, Sz=2.5 micron.

Emulsion simulation for data/MC comparison: Event generation

Event generation in the ORFEO official version: Actually is used a modified version of OpRoot/Geant3

(the modified files are provided with the ORFEO tar file); The cut on the micro-tracks minimum energy is

removed; A smallest propagation step (as respect to the official

OpRoot) is used to reproduce the correct Moliere radius (as suggested in the Geant3 manual);

The compton production is on; To simulate a test beam: the beam momentum,

particles density and slopes can be chosen by using the standard OpRoot Config.C file.

Emulsion simulation for data/MC comparison: micro-tracks

digitization The micro-tracks digitization takes into account:

Position and angle micro-tracks smearing performed with the 2 point micro-tracks simulation.

The smearing parameters must be tuned to reproduce test beam data! The micro-tracks efficiencies (that must be parameterized from data) The micro-track number of clusters is assigned on a statistical basis.

A parameterization of the number of clusters as seen in data is needed;

Affine transformations between plates can be simulated To take into account the up-down link efficiency, the same

up-down linking algorithm used for real data is used to select the couples of micro-tracks that will survive after the linking.

Data vs MC: Pions 4 GeV The events were generated with

OpRoot The smearing parameters are:

Sx=0.25 micron, Sy=0.25 micron, Sz=2.5 micron.

these value are the same used for all the “official” productions

NO parameters optimization was made for this set of data.

TRIGGER

MC Tx=0.0055

MC Ty=0.0029

data Tx=0.0053

data Tx=0.0022

Slopes resolution(Tmicro-Tbase)

Tx-Tx1=0.01

Ty-Ty2=0.01

Ty-Ty1=0.01

Tx-Tx2=0.01

Tx-Tx1=0.009

Tx-Tx2=0.01

Ty-Ty2=0.009

Ty-Ty2=0.009

MC data

eCHI2P vs PHThis is the results of the micro slopes resolution simulation (eCHI2P) and of the

micro pulse height parametrization (bt PH)

bg rejectedMC Data

Base Track angle resolutionBin(i-1)=T(i)-T(1), i>1

MC Data

TxMean=11.5RMS=5.3

TxMean=11.7RMS=5.3

TxMean=11.2RMS=5.3

TyMean=11.4RMS=5.2

Base Pulse and eCHI2P

MC PHMean=26.6RMS=2.9

data PHMean=26.6RMS=3.0

MC eCHI2PMean=0.7RMS=0.5

data eCHI2PMean=0.8RMS=0.8

Number of segments followed without propagation (strongly related to micro track efficiencies)

Data

MC+eff.Assuming eff.=84.%

(84.% at theta<0.1mrad measured data)

MC: no correlation betweenMicro tracks momentum and pulse height by construction

Pulse height vs plate number

and energy loss

Data

MC

The events were generated with OpRoot The smearing parameters are:

Sx=0.4 micron, Sy=0.25 micron, Sz=2.5 micron.

(these value were obtained with a tuning on the data) Eff. Measured in the empty brick by using

cosmic muons Pulse Height parametrized by using the data

in the empty brick;

Data vs MC:Electrons 6 GeV

TRIGGER

MC Ty=0.0025

data Tx=0.0037

data Tx=0.0022

MC Tx=0.0043

Slopes resolution(Tmicro-Tbase)

Tx-Tx1=0.013

Ty-Ty2=0.007

Ty-Ty1=0.008

Tx-Tx2=0.014

Tx-Tx1=0.014

Tx-Tx2=0.014

Ty-Ty2=0.008

Ty-Ty2=0.008

MC data

eCHI2P vs PHThis is the results of the micro track slopes resolution simulation

(eCHI2P) and of the micro track pulse height parametrization (bt PH)

bg rejectedMC Data

MC vs data comparison

Selected tracks characteristics: The track starts in the 1st plate; Number of segments [3,15] ; The track is in a box with a

surface of 1.8x1.8cm2; The first segment of the track

is in a cone around the beam direction with open angle defined by the beam width;

YesNO

Base Track angle resolutionBin(i-1)=T(i)-T(1), i>1

MC Data

Base Pulse and eCHI2P

MC PHMean=26.4RMS=3.0

data PHMean=26.3RMS=3.0

MC eCHI2PMean=1.2RMS=0.9

data eCHI2PMean=1.1RMS=1.0

MC: no correlation between Micro tracks momentum and pulse height by construction

Pulse height vs plate number and

energy loss

Data

MC

about efficiencies… Brick not exposed

to the e beamReference brick eff.

Eff lead MC pionsEff no lead MC pions

Number of segments followed without propagation (strongly related to micro track efficiencies)

Data

MC without Efficiencies micro track

rejection

Eff. As measured inEmpty brick

(only cosmics exposition)

MC+eff.

A new effect of the reconstruction seen in the data/MC

comparison that must be simulated The effect as it was seen in the data and

in the MC; The source of the effect (as suggested by

Cristiano Bozza); Looking for a threshold effect in the

data…; Our suggestion for an upgrade of the

simulation procedure; A set of test that we need/suggest;

The effect seen in the data/MC comparison

Frederic

Yvan

The effect seen in the data/MC comparison

Blu real dataRed simulation

The effect seen in the data/MC comparison

Blu Geant4Red Geant3Green real data

The effect seen in the data/MC comparison

Blu real dataRed simulation

The source of the effect

“ When SmartTracker7 finds that two micro-tracks have one grain in common, only the one with more grains survives, and the other is discarded… …Background can “kill” good tracks (e.g. a Compton electron might “cross” some good track, “eating” its grains), and this is not obtained by just adding background to MC data at a micro-track level.”

(From an email of C. Bozza)

Looking for a threshold in the data…

the plot of the minimum distance between micro-tracks in a view will give us the real dimension of the “problem”. Coming soon…

Our suggestion for an upgrade of the simulation procedure.

A transversal dimension must be added to the micro-track

this will introduce a threshold effect… related to the cluster size, function of the energy…

If the distance between 2 micro-tracks at fixed z is less then the threshold, a function f(nc1,nc2) will help to decide if 1 micro must be rejected

In such a way the effect is additive in fact by adding the bg micro-tracks the overlap

probability will increase!!!

Tests program All the tests should be done on both oil and dry

objective and for different configuration parameters (grey level…)

From any test beam: Measure of the threshold (minimum distance between

micro-tracks couples in a view); …

From e test beam: Cluster size dimension as a function of the energy

(number of plate…); …

From test beam: Number of times that the first base-track of a secondary

track is seen as a function of the distance of the primary vertex from the lead surface;

…

Conclusions The emulsion simulation package (ORFEO) is

going to be fully implemented in the OpRelease framework;

2 different simulation procedure for algorithms comparison and for data/MC comparison defined;

A new tracking effect was seen in data/MC comparison. A procedure to reproduce it in the simulation is

under development; Some tests on data are needed to improve the

simulation properly.