the geant3 tale ➢ motivation... ➢ why aliroot+g3 are not reproducing any more the measurements ?...

8
The Geant3 tale Motivation ... Why AliROOT+G3 are not reproducing any more the measurements ? Why stand alone code do reproduce measurements ? ource: http://www-linux.gsi.de/~abercuci/Contributions/Talks/Software

Upload: lewis-wright

Post on 17-Jan-2018

214 views

Category:

Documents


0 download

DESCRIPTION

Software Meeting GSI May 23, Too much memory economy ? ● location G3STXIN ● location G3STTAB EMAX=LOG(DCUTE*1.E9) EMAX= Calculate d2N/dEdx Calculate dN/dx ● What are CERN people saying ? “This is not a bug but a feature. [...] If you decide to produce explicitly d-rays above a certain threshold, then you should adjust the BB so that these are not counted twice. The "correction" you propose would lead to a double counting of the energy lost by the particle, so it is definitely wrong.” *) see G3 Manual pg. 264

TRANSCRIPT

Page 1: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

The Geant3 tale

➢ Motivation ...➢ Why AliROOT+G3 are

not reproducing any more the measurements ?

➢ Why stand alone code do reproduce measurements ?

talk source: http://www-linux.gsi.de/~abercuci/Contributions/Talks/Software

Page 2: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

Software Meeting GSIMay 23, 2006 2

G3 tracking

Initialise material constants for all

the physics mechanisms used by GEANT Initialization of tables

for energy lossstraggling in thin

gas layersInitialization of the values needed in computation of

dN^2/dE/dx as in Grishin,Ermilova,Kotelnikov,

NIM A307(1991),273

dN^2/dEdx table for a gamma factor

Controls tracking of currentparticle up to end of track

for sequential tracking mode, or through

current volume for parallel tracking mode.Electron type track.

Computes step size and propagates particle

through step.Subroutine to decide

which method is used to simulate the straggling around the

mean energy loss.

Simulation of energy loss straggling in thin layers. Sampling is done from the tables which are prepared in

GSTINI.

Generates Delta raysCharged hadron type track.

Computes step size and propagates particle

through step.

Page 3: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

Software Meeting GSIMay 23, 2006 3

Too much memory economy ?

● location G3STXIN

● location G3STTAB

EMAX=LOG(DCUTE*1.E9)EMAX=29.9335

Calculate d2N/dEdxCalculate dN/dx

● What are CERN people saying ?

“This is not a bug but a feature. [...] If you decide to produce explicitly d-rays

above a certain threshold, then you should adjust the BB so that these are not counted twice.

The "correction" you propose would lead to a double counting of the energy lost by the

particle, so it is definitely wrong.”

*) see G3 Manual pg. 264

Page 4: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

Software Meeting GSIMay 23, 2006 4

Who's afraid of 0 ?

● What are CERN people saying ?

AMU = DAN*STEPCALL G3POISS(AMU,N,1)NICOLL = MAX(N,1)NICOLL = N

● location G3STREN

The reason for this protection is to have at least one collision,

otherwise there is no fluctuation and no straggling and thedistribution is too narrow.

Page 5: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

Software Meeting GSIMay 23, 2006 5

Redundancy ?

● location G3TELECIF((IPART.EQ.2).OR.((IPART.EQ.3).AND.(GEKIN.GT.2.*DCUTE))) CALL G3DRAY

● Moeller theory for e-● Bethe theory for e+

4 GeV/c electronsG3 11.73 % delta per primary track 124.1 keV mean energy

Stand alone code based on PAI22.2 % delta per primary track 167.8 keV mean energy

“PAI should not be used. If it is (used), it should be switched off and all the

tests redone with the standard energy fluctuation. PAI is for very

specific studies, and not for general use. There is no point in discussing

it for a general case. [...]This is not the standard GEANT.”

Page 6: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

Software Meeting GSIMay 23, 2006 6

What about AliROOT ...

● AliTRD::CreateMaterials()

● AliTRDv1::FixStep()

wXeCO2[3] = { 0.85 , 0.0375 , 0.1125};wXeCO2[3] = { 8.5 , 1.5 , 3. };.....................................................AliMixture(10,"Gas mixture",aXeCO2, zXeCO2, dgm, 3, wXeCO2);AliMixture(10,"Gas mixture", aXeCO2, zXeCO2, dgm, -3, wXeCO2);

if (( gMC->TrackCharge() ) && (!gMC->IsTrackStop() ) && (!gMC->IsTrackDisappeared())) {

*) Thanks to Andreas Morsch for pointing it out.

Page 7: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

Software Meeting GSIMay 23, 2006 7

Results and answers ...Why AliROOT + G3 are not reproducing any more the measurements ?

Most probable because of:a) DCUTEb) material definition(s)c) FixStepManager()

Why stand alone code do reproduce measurements ?

Because of luck !Why using the PAI model for TRD ?

a) Because the number of inelastic collisions is small !2 GeV/c pi+ = 64.5/15end is even more reduced by the magnetic field !b) Because we need the number of inelastic collisions.c) Alternatives ... ?! more results can (or will) be found @:

http://www-linux.gsi.de/~abercuci/MyWork/ALICE-TRD/dEdx/FixStep/Results/New

Page 8: The Geant3 tale ➢ Motivation... ➢ Why AliROOT+G3 are not reproducing any more the measurements ? ➢ Why stand alone code do reproduce measurements ? talk

Software Meeting GSIMay 23, 2006 8

The future ...

● Can we use G3 in AliROOT (TRD) ?● What do we learn from the G3 lesson ?● Efficiency ?● What about the G3 community ?

Disclaimer:I have used for the current analysis AlirROOT v4-04-02 and ROOT v511-01-alice some results may not be in agreement with AlirROOT v4-04-09 and ROOT v511-02 or the G3 distribution which is accompanying them.