Geant3/Geant4 Comparisons - status David Ward University of Cambridge Where we were in November. Looking into the guts of the hadronic packages. Effect of a bug in Gheisha. Sensitivity of results to tracking cutoffs. A few comments on Fluka implementation in Geant3. D.R. Ward CALICE meeting, Amsterdam; 31 March 23 1
Procedure Comparing physics content of Geant4 with Geant3, especially for hadronic showers. Using Mokka+Geant4. Write out calorimeter geometry (Tesla TDR geometry Si-W ECAL; scintillator tile HCAL), and use in Brahms+Geant3. At Cambridge, performing similar studies for MINOS. MINOS is a fine-grained iron-scintillator calorimeter (4 cm 1 cm strips; 2.54 cm thick steel). Minos has a test module in a beam at CERN now. relevant info for CALICE. Comparisons made at the hits level just using deposited in sensitive detectors. Energies crudely converted to MIPs using muons to calibrate. D.R. Ward CALICE meeting, Amsterdam; 31 March 23 2
Pions 2 GeV π + G3 5 GeV π + G3 6 5 4 3 1 5 1 4 1 8 6 1 5 1 4 2 1 1 3 4 2 1 3 2 4 6 8 1 1 2 2 4 6 8 1 1 2 18 16 14 12 1 8 6 4 2.25.5.75 1 1 3 1 2 2 4 6 5 4 3 2 1.25.5.75 1 1 4 1 3 1 2 2 4 Hadronic showers for mesons seem to be reasonably similar between Geant3 and Geant4. (Also electrons, muons were OK.) D.R. Ward CALICE meeting, Amsterdam; 31 March 23 3
Protons 2 GeV proton G3 5 GeV proton G3 5 4 3 2 1 2 4 6 8 1 1 5 1 4 1 3 1 2 1 2 1 8 6 4 2 2 4 6 8 1 1 5 1 4 1 3 1 2 14 12 1 8 6 4 1 3 1 2 3 25 2 1 1 4 1 3 2.25.5.75 1 2 4 5.25.5.75 1 1 2 2 4 Big discrepancy seen for protons (neutrons similar). This effect also seen in MINOS, where Geant3 seems disfavoured by data. D.R. Ward CALICE meeting, Amsterdam; 31 March 23 4
Summarise hadron energy response Geant4 predicts approximately equal energy response for and protons, at the same kinetic energy, while Geant3 predicts much higher for protons. Also seen in Minos. Minos data indicate and p approximately equal. suspicious about Geant3. Response Mip / unit energy Response Mip / unit energy 11 1 9 8 7 6 5 4 3 4 35 3 25 2 1 ECAL 1 1 1 1 proton proton G3 π - π - G3 π + π + G3 HCAL KE /GeV KE /GeV D.R. Ward CALICE meeting, Amsterdam; 31 March 23 5
What s wrong? Trying to investigate details of hadronic showers. After several false trails, focussed on first interaction of incoming hadrons. Including Geant3-Fluka in this comparison. Cross-sections seem OK. Big difference in energy of daughters for interactions of low energy baryons. Geant3-Gheisha appears to violate energy conservation. Geant4 and Fluka are different, but at least have some features in common. suspect bug in G3-Gheisha. 1 8 6 4 2 25 2 1 5 n.5 GeV G3-Fluka G3-Gheisha 25 5 75 1 125 175 2 225 Depth of first interaction /mm.2.4.6.8 1 1.2 1.4 1.6 1.8 2 KE fraction (daughters/parent) D.R. Ward CALICE meeting, Amsterdam; 31 March 23 6
Bug in Gheisha! Have identified a bug in G3-Gheisha. Seems to compute target mass by counting produced baryons, including projectile! Simple one line fix (in twoclu.f and genxpt.f) seems to fix up the kinematic limit problem. 1 8 6 4 2 5 4 3 2 1 n.5 GeV G3-bug fix G3-Gheisha 25 5 75 1 125 175 2 225 Depth of first interaction /mm.2.4.6.8 1 1.2 1.4 1.6 1.8 2 KE fraction (daughters/parent) D.R. Ward CALICE meeting, Amsterdam; 31 March 23 7
Bug fix continued... Significant differences in detail between models remain. Subsequently discovered that the same problem was identified a couple of years by D.Bailey (ATLAS) see url below. He made the same diagnosis and cure. Hence fixed in Geant4 (but not Geant3). n.5 GeV G3-bug fix G3-Gheisha 1 2 1 2 1 1 1 1.25.5.75 1.5 1 1.5 2 KE fraction (protons/parent) KE fraction (neutrons/parent) 1 3 1 3 1 2 1 2 1 1 1.2.4 1.25.5.75 1 KE fraction (nuclei/parent) KE fraction (mesons/parent) www.physics.utoronto.ca/ dbailey/atlas_geant4_6nov2/ D.R. Ward CALICE meeting, Amsterdam; 31 March 23 8
Effect of bug fix? Implement Gheisha bug fix in BRAHMS and rerun. π + 2 GeV G3-fixed G3 π + 5 GeV G3-fixed G3 6 5 4 3 2 1 2 4 6 8 1 1 5 1 4 1 3 1 2 1 2 1 8 6 4 2 2 4 6 8 1 1 5 1 4 1 3 1 2 225 2 175 125 1 75 5 25.25.5.75 1 1 3 1 2 2 4 9 8 7 6 5 4 3 2 1.25.5.75 1 1 4 1 3 1 2 2 4 Bug fix has little effect. As expected no effect on original interaction; only on the nucleon daughters secondary interactions. D.R. Ward CALICE meeting, Amsterdam; 31 March 23 9
Effect on protons? p 2 GeV G3-fixed G3 p 5 GeV G3-fixed G3 7 6 1 5 1 1 5 5 4 3 2 1 2 4 6 8 1 1 4 1 3 1 2 1 2 8 6 4 2 2 4 6 8 1 1 4 1 3 1 2 225 2 175 125 1 75 5 25.25.5.75 1 1 3 1 2 2 4 45 4 35 3 25 2 1 5.25.5.75 1 1 4 1 3 1 2 2 4 Bug fix has excellent effect G3 and now in quite good agreement. D.R. Ward CALICE meeting, Amsterdam; 31 March 23 1
Summarise hadron energy response Response Mip / unit energy 11 1 9 8 ECAL proton proton G3 proton G3 fixed Response Mip / unit energy 4 35 3 π + π + G3 π + G3 fixed proton HCAL 7 π + π + G3 25 proton G3 proton G3 fixed 6 π + G3 fixed 2 5 4 3 1 1 KE /GeV 1 1 1 KE /GeV G3-Gheisha bug fix clearly fixes the gross discrepancy between G3 and. D.R. Ward CALICE meeting, Amsterdam; 31 March 23 11
Hadron energy response ratio Look at ratios between G3 and. After G3-Gheisha bug fix; agreement at 1% level or better in most cases. But small differences remain. next check tracking cutoffs. Geant3/Geant4 Geant3/Geant4 2 1.8 1.6 1.4 1.2 1.8.6.4.2 2 1.8 1.6 1.4 1.2 1.8.6.4.2 π + G3 fixed/ π + G3/ π + G3 fixed/ π + G3/ 1 1 π + G3 fixed/ π + G3/ π + G3 fixed/ π + G3/ 1 1 ECAL HCAL KE /GeV KE /GeV D.R. Ward CALICE meeting, Amsterdam; 31 March 23 12
Cut dependence Geant4 In Geant4, set a cut on step length (by default) to control tracking cutoff. 2 GeV Results, especially energy, seem rather insensitive to this cutoff. Mokka uses 5 m by default. ECAL energy /MIPs ECAL Nhits 2 195 19 185 18 175 17 165 16 5 55 54 53 52 51 5 49 48 47 46 45 44 42 4 38 36 34 32 1 1 2 1 3 3 Cutoff /µm 1 1 2 1 3 Cutoff /µm HCAL energy /MIPs HCAL Nhits 2 19 18 17 16 14 13 12 11 1 1 1 2 1 3 Cutoff /µm 1 1 2 1 3 Cutoff /µm D.R. Ward CALICE meeting, Amsterdam; 31 March 23 13
Cut dependence Geant3 2 GeV ECAL energy /MIPs 2 195 19 185 18 175 17 165 16 5 1 1 2 1 3 EM cutoff /kev HCAL energy /MIPs 4 38 36 34 32 3 28 26 1 1 2 1 3 EM cutoff /kev ECAL energy /MIPs 2 195 19 185 18 175 17 165 16 5 1-1 1 1 Had. cutoff /MeV HCAL energy /MIPs 42 4 38 36 34 32 3 28 1-1 1 1 Had. cutoff /MeV ECAL Nhits 55 54 53 52 51 5 49 48 47 46 45 1 1 2 1 3 EM cutoff /kev HCAL Nhits 2 19 18 17 16 14 13 12 11 1 1 1 2 1 3 EM cutoff /kev ECAL Nhits 55 54 53 52 51 5 49 48 47 46 45 1-1 1 1 Had. cutoff /MeV HCAL Nhits 22 21 2 19 18 17 16 14 13 12 1-1 1 1 Had. cutoff /MeV In contrast, Geant3 uses energy cuts, and results are quite sensitive. Geant3 defaults are 1 MeV for and 1 MeV for hadrons. Need to reduce? D.R. Ward CALICE meeting, Amsterdam; 31 March 23 14
Hadron energy response I was previously using cuts of 1 kev for and 1 MeV for hadrons in Geant3. Looks like 1 kev for and 1 MeV for Geant3/Geant4 2 1.8 1.6 1.4 1.2 1.8.6.4.2 π + G3 fixed/ π + G3/ proton G3 fixed/ proton G3/ 1 1 ECAL KE /GeV hadrons should be a better choice. Inconclusive some things better; some worse. Geant3/Geant4 2 1.8 1.6 1.4 1.2 1.8.6.4.2 π + G3 fixed/ π + G3/ proton G3 fixed/ proton G3/ 1 1 HCAL KE /GeV D.R. Ward CALICE meeting, Amsterdam; 31 March 23
Fluka in Geant3 An implementation of Fluka is available in Geant3; easy to use. But quite an old version; now deprecated by Fluka authors. π + 2 GeV G3-Fluka G3-fixed π + 5 GeV G3-Fluka G3-fixed 6 5 4 3 2 1 2 4 6 8 1 1 5 1 4 1 3 1 2 1 2 12 1 8 6 4 2 2 4 6 8 1 1 5 1 4 1 3 1 2 14 12 1 8 6 4 2.25.5.75 1 1 3 1 2 2 4 4 35 3 25 2 1 5.25.5.75 1 1 4 1 3 1 2 2 4 D.R. Ward CALICE meeting, Amsterdam; 31 March 23 16
Fluka in Geant3 45 4 35 3 25 2 1 5 4 35 3 25 2 1 5 p 2 GeV 2 4 6 8 1.25.5.75 1 G3-Fluka 1 5 1 4 1 3 1 2 1 3 1 2 G3-fixed 1 2 2 4 1 8 6 4 2 35 3 25 2 1 5 p 5 GeV 2 4 6 8 1.25.5.75 1 G3-Fluka G3-fixed 1 5 1 4 1 3 1 2 1 4 1 3 2 4 Fluka is significantly different showers deeper; more energy in HCAL; broader in HCAL. Could be useful for checks. [Minos data seem to disfavour Fluka compared to Geant4/Geant3-Gheisha.] D.R. Ward CALICE meeting, Amsterdam; 31 March 23 17
Summary Helpful symbiosis between work on CALICE and MINOS. MINOS has data now. MINOS studies (and data) pointed at problems with Gheisha implementation in Geant3. A bug leading to one gross problem has been identified, and hopefully fixed. (Fixed long ago in Geant4). Calice simulation in Geant3 now looks healthier. Old version of Fluka in Geant3 may be useful if one wants a significantly different model. Work proceeds on implementing the current version of Fluka (N.Watson). More to be learned from study of Minos data before Calice beam test. D.R. Ward CALICE meeting, Amsterdam; 31 March 23 18