Download presentation
Presentation is loading. Please wait.
Published byClaud Doyle Modified over 9 years ago
1
Mark Thomson Timing, Tungsten and High Energy Jets
2
Timing: Recap Distribution dominated by time of flight to HCAL Long tail from low energy neutrons out to ~1 s Recently looked at timing for 250 GeV jets in ILD (Steel-Scintillator) Correct for time of flight using hit position assuming propagates at speed of light 90 % of energy deposited in first few ns 2 CERN, 3 rd August 2010Mark Thomson
3
Steel HCAL HCAL 95 % of energy in 10 ns 99 % in 50 ns Corrected for ToF Suggests optimal timing window in range 5-10 ns 3 CERN, 3 rd August 2010Mark Thomson
4
Tungsten What about Tungsten? Iron (and lead) doubly magic nuclei, i.e. particularly stable Tungsten: both n and p far from closed shells naively would expect more nuclear interactions with Tungsten a priori not a problem (e.g. Uranium for compensation) but expect longer time profile (decays, secondary interactions) + not clear how well modeled in Geant 4 Study with CLIC_ILD model generated single K L s (QGSP_BERT) copied uds 91, 200, 500 from Grid (thanks Stephane) repeated previous studies… NOTE: all at reconstructed PFO level uses 0.3 MiP cut rejection of very isolated hits 4 CERN, 3 rd August 2010Mark Thomson
5
Tungsten vs Steel: 25 GeV K L Steel Tungsten Tungsten much “slower” only 80 % of energy in 25 ns only 90 % in 100 ns how much due to thermal n ? 5 CERN, 3 rd August 2010Mark Thomson
6
Tungsten: Time vs Energy Tungsten much “slower”, but not the only difference distribution of single energy depositions much harder significant number of single hits have energy depositions > few GeV nuclear fragments? Time/ns Hit Energy/GeV Previously, PandoraPFA reconstruction had (evil) maximum single hit energy of 1 GeV responsible for poor performance reported by J-J. B. last meeting now removed (.xml steering) 6 CERN, 3 rd August 2010Mark Thomson
7
Tungsten vs Steel: 25 GeV K L Study HCAL resolution vs time window for Tungsten vs Steel removed max hit energy cut and recalibrated for each cut Dependence much stronger for W HCAL reflects larger time spread For decent HCAL performance, i.e. need to integrate over 20 ns ! 7 CERN, 3 rd August 2010Mark Thomson
8
PFA Performance vs time cut: uds Look at PFA performance for CLIC_ILD For no time cut (1000 ns) peformance of CLIC_ILD v. good somewhat better than ILD (thicker HCAL, larger B) For high(ish) energy jets – strong dependence on time cut suggests time window of > 10 ns need something like 50 ns to get into “flat region” 8 CERN, 3 rd August 2010Mark Thomson
9
Tungsten: Summary Tungsten leads to a longer time distribution of hits activity on the timescale of a full CLIC bunch-train for “reasonable” performance need to integrate over 10s of ns Is Tungsten is reasonable choice for a CLIC HCAL absorber? Not clear at this stage – a number of questions how good is simulation? what about digital calorimetry with gaseous active material? although digital may bring problems of its own… how much can be recovered offline i.e. integrate over some part of bunch train in reconstruction and then tag BX for clusters My Conclusions: Tungsten NOT an obvious choice for the endcap HCAL where background is significant In barrel region HCAL occupancy sufficiently low: Tungsten probably OK. needs serious study 9 CERN, 3 rd August 2010Mark Thomson
10
CLIC_ILD and High Energy Jets Started to look at performance of CLIC_ILD for high energy jets Looked at uds events 91, 200, 500 GeV events generated at CERN using Cambridge stdhep files 1 TeV, 2 TeV, 3 TeV events generated at CERN using SLAC stdhep Jet Energy /sqrt{E} E /E 45 GeV24.0 %3.5 % 100 GeV27.4 %2.7 % 250 GeV43.7 %2.8 % CLIC_ILD (no timing cut) E /E 3.7 % 2.9 % 3.3 % ILD “Low energy” performance looks very good, better than ILD model: HCAL resolution better (no timing cuts) Thicker HCAL Higher B “High Energy” performance – much worse than expected ! ~110%/√E c.f. ~80%/√E for 500 GeV jets Events looked suspicious – track multiplicities too low, no thee jet events, jets very narrow… 10 CERN, 3 rd August 2010Mark Thomson
11
1 TeV uds (SLAC stdhep) 1 TeV uds (Cambridge stdhep): Pythia, gluon radiation on, OPAL tune Example 1 TeV Events 11 CERN, 3 rd August 2010Mark Thomson
12
Strongly suspect that the stdhep files used for 1, 2, and 3 TeV production have gluon radiation off… Currently Generating 1 TeV events in Cambridge Preliminary (only 1500 event) performance looks good… David Ward produced new 2 TeV and 3 TeV stdhep files Jet Energy /sqrt{E} E /E 45 GeV24.0 %3.5 % 100 GeV27.4 %2.7 % 250 GeV43.7 %2.8 % 500 GeV77 %3.4 % About as expected (from previous ILD studies) However, some obvious PFA issues… tail at very high energies due to split tracks CLIC version of LDCTracking helps somewhat, but doesn’t solve problem – needs further study tail at low energies (e.g. >100 GeV missing) bad track-cluster matches, e.g. 100 GeV track – 250 GeV cluster should be handled in reco – needs further study 12 CERN, 3 rd August 2010Mark Thomson
13
Conclusions Time structure is an issue Tungsten may not be viable for endcap region not where it is needed – so not a problem background in barrel region is less of an issue but needs full bunch train studies with background Tungsten: High-Energy Jets: No obvious problems with CLIC-ILD model work needed: Tracking optimisation PFA optimisation Study 1 TeV and 1.5 TeV jets General: Full simulation/full reconstruction studies essential to demonstrate: PFA with CLIC bunch time structure Viability of Tungsten calorimetry Viability of DHCAL at CLIC … 13 CERN, 3 rd August 2010Mark Thomson
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.