From landsberg@hep.brown.edu Fri Aug 1 16:08:25 2003 Date: Mon, 12 May 2003 17:27:30 -0400 From: Greg Landsberg To: Marco Verzocchi , burdin@fnal.gov, jonas@fnal.gov Cc: tamburel@fnal.gov, Eric Thomas Subject: RE: Skimming Gents, I have summarized the proposed common skims in a presentation to the NP group 2.5 weeks ago. Here it is for the reference: http://www-d0.fnal.gov/Run2Physics/np/d0_private/meetings/NP-04-24-03-GL 1.pdf (p.4) Looking back at my notes, I see the following things that Marco and Peter agreed to look at: 1) Single muon skim: make a 2D-distribution of pT(local) vs. max(pT(central),pT(global)) to see whether we can come up with a more optimal cut then max(pT(local),pT(central),pT(global)) > 10 GeV (i.e., lowering the cut on pT(central,global), while keeping intact (or raising) the cut on local-only muons. 2) What't the size of the single EM stream as a function of the ET(EM) cut (in the 17-18 GeV range)? 3) What's the size of the all-jet stream with Higgs group cuts, agreed to be the basis for the common cuts? 4) jets+MET skim: while we agreed to use the trigger definition as a temporary solution, we also talked about looking at the size of the skim for th following two sets of cut ("low" and "high"): (ET(j1),ET(j2),MET) > (15,15,15) and (20,20,20). Peter, could you send us numbers, please? Thanks, Greg From mverzocc@fnal.gov Fri Aug 1 16:08:50 2003 Date: Wed, 21 May 2003 20:15:12 -0400 From: Marco Verzocchi To: Greg L. Landsberg Subject: skimming Hi Greg, I've finally found the time to work on the skimming (p14.02 data are finally available, no excuse to delay this any more)... So far I've done testing with a single file (20k events) and these are my suggestions: single EM: require ID=10,+/-11, pT>17.5 GeV (or 17) approximately 10% of the events are selected (no other cuts, except the requirement of ID=10,+/-11) variation of the pT cut (sample of 20527 events) 15: 2361 / 20527 16: 2246 17: 2097 17.5: 2043 18: 1977 19: 1839 20: 1681 at 17.5 GeV of pT cut removing the ID=10,+/-11 requirement brings the size of the same from 2043 to 4146 (double.... 20%) maybe we could even afford 15 or 16 GeV with the ID=10,+/-11 requirement (this is in line with the trigger strategy) to be tested tomorrow: remove the ID=10,+/-11 for pT>2x GeV (keep total size below 15%) diEM: require ID=10,+/-11, pT> 6 GeV (this is in line with the trigger strategy) selects approximately 2.5% of the events variation of the pT cut (again 20527 events) 5: 983 6: 509 7: ... 8: 427 9: 367 10: 281 11: 213 12: 158 to be tested tomorrow: removing the ID=10,+/-11 for pT>10 GeV (keep total size below 5%) single m: require pT>6.5 GeV for muons which have a track matched to central tracker. pT is max(pT from global muon fit, pT from tracker). For muons which don't have a track matched to the central tracker require pT>8 GeV (this is kept fixed below) variation of the pT cut (again 20527 events) 5: 3269 6: 2229 6.5: 1908 7: 1839 8: 1670 9: 1336 10: 990 at 6.5 GeV of pT cut changing the cut on the local muons by 2 GeV (from 8 to 6 or 10) changes the number of selected events by +/-10% maybe we could even afford 6 GeV dimuon: require 2 muons (pT>1 GeV which all dimuons satisfy) avoding double counting, select also events with 1 muon and one calMTC (the calMTC has a central track match, with at least X GeV) select between 910 and 995 events (from 4.4 to 4.8%) when varying X between 5 and 10 GeV (the 1 m plus 1 calMTC events are not too many) I want to play a little bit to see if we can recover some more electrons with ID not 10,+/-11, by raising the pT cut for the single electrons and allowing in the diEM case the following: 2 em with ID=10,+/-11 low pT cut 1 em with ID=10,+/-11 low pT cut on the ID=10,+/-11 high pT cut on the other add also the viceversa 0 em with ID=10,+/-11 higher pT cut This will take me the morning.... I'll make a more extensive study overnite (larger statistics, there are 560k events available in SAM)... I think that the skimming could start early next week.... Marco From landsberg@hep.brown.edu Fri Aug 1 16:09:02 2003 Date: Thu, 22 May 2003 08:07:32 -0400 From: Greg Landsberg To: Marco Verzocchi , Eric Thomas Cc: Peter Tamburello , burdin@fnal.gov, jonas@fnal.gov Subject: RE: skimming Hi Marco, Thanks a lot for the detailed information! A few comments: 1) It's clear that p14.03 will be late for the summer conference sample, so we won't be able to start the common skimming in time for this sample. I hope that each group is aware of that and continues with its own skimming. In the NP group we are evolving our selections to approach the common skimming cuts. I suggest that we delayed the common skimming till the p14.03 or 04, which fixes the calorimeter problems, is installed on the farm. 2) The issue of |ID| = 10/11 is clearly important, and since we are delaying skimming, it might be worth our while to look in it. The tests you have done are quite revealing, but the two-folded increase in the rate is not something we could afford. I'd study straight ISO cut instead of the ID cut, which might suffice to bring things under control. Personally I am against "split" definition of a stream, when cuts are pT-dependent, as it generally always results in somebody misusing the data-set. If necessary, I'd rather have an additional stream with no ID cut but with low-ET electron cut and a track match to study low-ET electrons seed by tracks. 3) Could you please confirm Levan's numbers for jets+MET stream with the new p14.02? The stream definition is the OR of the jets+MHT trigger and the following cuts: HT > 25 GeV, MHT > 20 GeV. The rest are trivial comments on detailed stream definitions, which I interspersed with your original e-mail. I have included other streaming reps on the distribution list for this message, so that they can participate in fine-tuning. Many thanks for staying on top of this business! Greg > -----Original Message----- > From: Marco Verzocchi [mailto:mverzocc@fnal.gov] > > Hi Greg, > I've finally found the time to work on the skimming (p14.02 data > are finally available, no excuse to delay this any more)... > > So far I've done testing with a single file (20k events) and these > are my suggestions: > > single EM: require ID=10,+/-11, pT>17.5 GeV (or 17) approximately 10% of > the events are selected (no other cuts, except the requirement > of ID=10,+/-11) > > variation of the pT cut (sample of 20527 events) > > 15: 2361 / 20527 > 16: 2246 > 17: 2097 > 17.5: 2043 > 18: 1977 > 19: 1839 > 20: 1681 > > at 17.5 GeV of pT cut removing the ID=10,+/-11 requirement > brings the size of the same from 2043 to 4146 (double.... 20%) > > maybe we could even afford 15 or 16 GeV with the ID=10,+/-11 > requirement (this is in line with the trigger strategy) 17.5 GeV cut looks a bit funny to me - like we know things to such high precision. I'd go with 17 or 18 GeV instead. > to be tested tomorrow: remove the ID=10,+/-11 for pT>2x GeV > (keep total size below 15%) > > diEM: require ID=10,+/-11, pT> 6 GeV (this is in line with the > trigger strategy) selects approximately 2.5% of the events > > variation of the pT cut (again 20527 events) > > 5: 983 > 6: 509 > 7: ... > 8: 427 > 9: 367 > 10: 281 > 11: 213 > 12: 158 > > to be tested tomorrow: removing the ID=10,+/-11 for pT>10 GeV > (keep total size below 5%) Looks good! > single m: require pT>6.5 GeV for muons which have a track matched to > central tracker. pT is max(pT from global muon fit, pT from > tracker). For muons which don't have a track matched to > the central tracker require pT>8 GeV (this is kept fixed below) > > variation of the pT cut (again 20527 events) > > 5: 3269 > 6: 2229 > 6.5: 1908 > 7: 1839 > 8: 1670 > 9: 1336 > 10: 990 > > at 6.5 GeV of pT cut changing the cut on the local muons > by 2 GeV (from 8 to 6 or 10) changes the number of selected > events by +/-10% > > maybe we could even afford 6 GeV Looks good again; however, I think it's worth playing a bit with the pT(local) cut, rather than fix it at 8 GeV. The biplot of max(pT) vs. pT(local) we discussed a couple of times before would help here. > dimuon: require 2 muons (pT>1 GeV which all dimuons satisfy) > avoding double counting, select also events with 1 muon > and one calMTC (the calMTC has a central track match, with > at least X GeV) > > select between 910 and 995 events (from 4.4 to 4.8%) > when varying X between 5 and 10 GeV (the 1 m plus 1 calMTC > events are not too many) Sounds good, and we could easily afford to go as low as 5 GeV on the MTC-muon pT-cut. That's all! > I want to play a little bit to see if we can recover some more > electrons with ID not 10,+/-11, by raising the pT cut for the single > electrons and allowing in the diEM case the following: > > 2 em with ID=10,+/-11 low pT cut > 1 em with ID=10,+/-11 low pT cut on the ID=10,+/-11 high pT cut on the > other > add also the viceversa > 0 em with ID=10,+/-11 higher pT cut > > This will take me the morning.... I'll make a more extensive study > overnite (larger statistics, there are 560k events available in SAM)... > I think that the skimming could start early next week.... > > Marco From tamburel@fnal.gov Fri Aug 1 16:09:28 2003 Date: Thu, 22 May 2003 23:56:34 -0400 From: Peter Tamburello To: Greg Landsberg Cc: 'Marco Verzocchi' , 'Eric Thomas' , burdin@fnal.gov, jonas@fnal.gov Subject: RE: skimming I wonder if we are making the cuts too loose based on the current data. As the luminosity increases and the trigger list changes the size of the streams could grow as a percentage of the total. Here's what I find for the size of some streams [% of total] for runs with different prescale sets: stream 15e30 20e30 30e30 35e30 SingleMuon 6.081 6.105 6.588 7.012 SingleEM 10.279 13.258 17.056 17.165 EleTrk 3.940 5.071 6.644 6.963 DiMuon 1.754 2.074 2.606 2.882 DiEM 2.332 3.065 2.749 1.873 EMMuon 2.714 3.329 3.898 3.917 JetsMEt 5.327 6.386 8.184 10.496 MultiJet 5.677 7.147 9.781 12.339 SingleMuon stays pretty flat, but that is because the single muon trigger starts to get prescaled at 30e30. The stream definitions are from Greg's NP slide, except that MTC muons are not included in DiMuon and JetsMEt is from the last iteration with Levan. Peter On Thu, 22 May 2003, Greg Landsberg wrote: > Hi Marco, > > Thanks a lot for the detailed information! A few comments: > > 1) It's clear that p14.03 will be late for the summer conference sample, > so we won't be able to start the common skimming in time for this > sample. I hope that each group is aware of that and continues with its > own skimming. In the NP group we are evolving our selections to approach > the common skimming cuts. I suggest that we delayed the common skimming > till the p14.03 or 04, which fixes the calorimeter problems, is > installed on the farm. > > 2) The issue of |ID| = 10/11 is clearly important, and since we are > delaying skimming, it might be worth our while to look in it. The tests > you have done are quite revealing, but the two-folded increase in the > rate is not something we could afford. I'd study straight ISO cut > instead of the ID cut, which might suffice to bring things under > control. Personally I am against "split" definition of a stream, when > cuts are pT-dependent, as it generally always results in somebody > misusing the data-set. If necessary, I'd rather have an additional > stream with no ID cut but with low-ET electron cut and a track match to > study low-ET electrons seed by tracks. > > 3) Could you please confirm Levan's numbers for jets+MET stream with the > new p14.02? The stream definition is the OR of the jets+MHT trigger and > the following cuts: HT > 25 GeV, MHT > 20 GeV. > > The rest are trivial comments on detailed stream definitions, which I > interspersed with your original e-mail. I have included other streaming > reps on the distribution list for this message, so that they can > participate in fine-tuning. > > Many thanks for staying on top of this business! > > Greg > > > -----Original Message----- > > From: Marco Verzocchi [mailto:mverzocc@fnal.gov] > > > > Hi Greg, > > I've finally found the time to work on the skimming (p14.02 data > > are finally available, no excuse to delay this any more)... > > > > So far I've done testing with a single file (20k events) and these > > are my suggestions: > > > > single EM: require ID=10,+/-11, pT>17.5 GeV (or 17) approximately 10% > of > > the events are selected (no other cuts, except the > requirement > > of ID=10,+/-11) > > > > variation of the pT cut (sample of 20527 events) > > > > 15: 2361 / 20527 > > 16: 2246 > > 17: 2097 > > 17.5: 2043 > > 18: 1977 > > 19: 1839 > > 20: 1681 > > > > at 17.5 GeV of pT cut removing the ID=10,+/-11 requirement > > brings the size of the same from 2043 to 4146 (double.... > 20%) > > > > maybe we could even afford 15 or 16 GeV with the > ID=10,+/-11 > > requirement (this is in line with the trigger strategy) > > 17.5 GeV cut looks a bit funny to me - like we know things to such high > precision. I'd go with 17 or 18 GeV instead. > > > > to be tested tomorrow: remove the ID=10,+/-11 for pT>2x GeV > > (keep total size below 15%) > > > > diEM: require ID=10,+/-11, pT> 6 GeV (this is in line with the > > trigger strategy) selects approximately 2.5% of the events > > > > variation of the pT cut (again 20527 events) > > > > 5: 983 > > 6: 509 > > 7: ... > > 8: 427 > > 9: 367 > > 10: 281 > > 11: 213 > > 12: 158 > > > > to be tested tomorrow: removing the ID=10,+/-11 for pT>10 > GeV > > (keep total size below 5%) > > Looks good! > > > single m: require pT>6.5 GeV for muons which have a track matched to > > central tracker. pT is max(pT from global muon fit, pT from > > tracker). For muons which don't have a track matched to > > the central tracker require pT>8 GeV (this is kept fixed > below) > > > > variation of the pT cut (again 20527 events) > > > > 5: 3269 > > 6: 2229 > > 6.5: 1908 > > 7: 1839 > > 8: 1670 > > 9: 1336 > > 10: 990 > > > > at 6.5 GeV of pT cut changing the cut on the local muons > > by 2 GeV (from 8 to 6 or 10) changes the number of selected > > events by +/-10% > > > > maybe we could even afford 6 GeV > > Looks good again; however, I think it's worth playing a bit with the > pT(local) cut, rather than fix it at 8 GeV. The biplot of max(pT) vs. > pT(local) we discussed a couple of times before would help here. > > > dimuon: require 2 muons (pT>1 GeV which all dimuons satisfy) > > avoding double counting, select also events with 1 muon > > and one calMTC (the calMTC has a central track match, with > > at least X GeV) > > > > select between 910 and 995 events (from 4.4 to 4.8%) > > when varying X between 5 and 10 GeV (the 1 m plus 1 calMTC > > events are not too many) > > Sounds good, and we could easily afford to go as low as 5 GeV on the > MTC-muon pT-cut. > > That's all! > > > I want to play a little bit to see if we can recover some more > > electrons with ID not 10,+/-11, by raising the pT cut for the single > > electrons and allowing in the diEM case the following: > > > > 2 em with ID=10,+/-11 low pT cut > > 1 em with ID=10,+/-11 low pT cut on the ID=10,+/-11 high pT cut on the > > other > > add also the viceversa > > 0 em with ID=10,+/-11 higher pT cut > > > > This will take me the morning.... I'll make a more extensive study > > overnite (larger statistics, there are 560k events available in > SAM)... > > I think that the skimming could start early next week.... > > > > Marco > > From landsberg@hep.brown.edu Fri Aug 1 16:09:42 2003 Date: Fri, 23 May 2003 04:57:14 -0400 From: Greg Landsberg To: Peter Tamburello Cc: Marco Verzocchi , Eric Thomas , burdin@fnal.gov, jonas@fnal.gov Subject: RE: skimming Hi Peter, Thanks for the cross-checks. The numbers are interesting, but they do not surprise me as our suite of triggers is designed to get higher fraction of interesting physics at high luminosity; hence the relative fraction of physics-motivated streams do increase. I think that we should design the set of cuts which is good for average luminosity around 30E30, which should be good up to instantaneous luminosity of about 50E30 at the beginning of the store. (Right now the average is about 25E30.) For higher luminosities we will be using completely different trigger set 12.xx, so the numbers you showed are likely to change. Looking at your list, I see that all the triggers except for the single EM are within the budget. What cut did you use for single EM - 17 or 18 GeV? Marco, could you please look into this and see how to curb the rate? As far as the jets+MET and multijet skims are concerned, they getting pretty close to the limit. Let's try the latest Levan's cut to see what the rate is going to be. It was HT > 25-30 GeV (uncorrected jets) + MHT > 20 GeV. Could something similar be applied to the multijet stream? I.e. an HT + multiplicity cut rather than a cut on individual jet ETs? Greg > -----Original Message----- > From: Peter Tamburello [mailto:tamburel@fnal.gov] > Sent: Thursday, May 22, 2003 11:57 PM > To: Greg Landsberg > Cc: 'Marco Verzocchi'; 'Eric Thomas'; burdin@fnal.gov; jonas@fnal.gov > Subject: RE: skimming > > I wonder if we are making the cuts too loose based on the current data. As > the luminosity increases and the trigger list changes the size of the > streams could grow as a percentage of the total. Here's what I find for > the size of some streams [% of total] for runs with different prescale > sets: > > stream 15e30 20e30 30e30 35e30 > > SingleMuon 6.081 6.105 6.588 7.012 > SingleEM 10.279 13.258 17.056 17.165 > EleTrk 3.940 5.071 6.644 6.963 > DiMuon 1.754 2.074 2.606 2.882 > DiEM 2.332 3.065 2.749 1.873 > EMMuon 2.714 3.329 3.898 3.917 > JetsMEt 5.327 6.386 8.184 10.496 > MultiJet 5.677 7.147 9.781 12.339 > > SingleMuon stays pretty flat, but that is because the single muon trigger > starts to get prescaled at 30e30. The stream definitions are from Greg's > NP slide, except that MTC muons are not included in DiMuon and JetsMEt is > from the last iteration with Levan. > > Peter > > > > On Thu, 22 May 2003, Greg Landsberg wrote: > > > Hi Marco, > > > > Thanks a lot for the detailed information! A few comments: > > > > 1) It's clear that p14.03 will be late for the summer conference sample, > > so we won't be able to start the common skimming in time for this > > sample. I hope that each group is aware of that and continues with its > > own skimming. In the NP group we are evolving our selections to approach > > the common skimming cuts. I suggest that we delayed the common skimming > > till the p14.03 or 04, which fixes the calorimeter problems, is > > installed on the farm. > > > > 2) The issue of |ID| = 10/11 is clearly important, and since we are > > delaying skimming, it might be worth our while to look in it. The tests > > you have done are quite revealing, but the two-folded increase in the > > rate is not something we could afford. I'd study straight ISO cut > > instead of the ID cut, which might suffice to bring things under > > control. Personally I am against "split" definition of a stream, when > > cuts are pT-dependent, as it generally always results in somebody > > misusing the data-set. If necessary, I'd rather have an additional > > stream with no ID cut but with low-ET electron cut and a track match to > > study low-ET electrons seed by tracks. > > > > 3) Could you please confirm Levan's numbers for jets+MET stream with the > > new p14.02? The stream definition is the OR of the jets+MHT trigger and > > the following cuts: HT > 25 GeV, MHT > 20 GeV. > > > > The rest are trivial comments on detailed stream definitions, which I > > interspersed with your original e-mail. I have included other streaming > > reps on the distribution list for this message, so that they can > > participate in fine-tuning. > > > > Many thanks for staying on top of this business! > > > > Greg > > > > > -----Original Message----- > > > From: Marco Verzocchi [mailto:mverzocc@fnal.gov] > > > > > > Hi Greg, > > > I've finally found the time to work on the skimming (p14.02 data > > > are finally available, no excuse to delay this any more)... > > > > > > So far I've done testing with a single file (20k events) and these > > > are my suggestions: > > > > > > single EM: require ID=10,+/-11, pT>17.5 GeV (or 17) approximately 10% > > of > > > the events are selected (no other cuts, except the > > requirement > > > of ID=10,+/-11) > > > > > > variation of the pT cut (sample of 20527 events) > > > > > > 15: 2361 / 20527 > > > 16: 2246 > > > 17: 2097 > > > 17.5: 2043 > > > 18: 1977 > > > 19: 1839 > > > 20: 1681 > > > > > > at 17.5 GeV of pT cut removing the ID=10,+/-11 requirement > > > brings the size of the same from 2043 to 4146 (double.... > > 20%) > > > > > > maybe we could even afford 15 or 16 GeV with the > > ID=10,+/-11 > > > requirement (this is in line with the trigger strategy) > > > > 17.5 GeV cut looks a bit funny to me - like we know things to such high > > precision. I'd go with 17 or 18 GeV instead. > > > > > > > to be tested tomorrow: remove the ID=10,+/-11 for pT>2x GeV > > > (keep total size below 15%) > > > > > > diEM: require ID=10,+/-11, pT> 6 GeV (this is in line with the > > > trigger strategy) selects approximately 2.5% of the events > > > > > > variation of the pT cut (again 20527 events) > > > > > > 5: 983 > > > 6: 509 > > > 7: ... > > > 8: 427 > > > 9: 367 > > > 10: 281 > > > 11: 213 > > > 12: 158 > > > > > > to be tested tomorrow: removing the ID=10,+/-11 for pT>10 > > GeV > > > (keep total size below 5%) > > > > Looks good! > > > > > single m: require pT>6.5 GeV for muons which have a track matched to > > > central tracker. pT is max(pT from global muon fit, pT from > > > tracker). For muons which don't have a track matched to > > > the central tracker require pT>8 GeV (this is kept fixed > > below) > > > > > > variation of the pT cut (again 20527 events) > > > > > > 5: 3269 > > > 6: 2229 > > > 6.5: 1908 > > > 7: 1839 > > > 8: 1670 > > > 9: 1336 > > > 10: 990 > > > > > > at 6.5 GeV of pT cut changing the cut on the local muons > > > by 2 GeV (from 8 to 6 or 10) changes the number of selected > > > events by +/-10% > > > > > > maybe we could even afford 6 GeV > > > > Looks good again; however, I think it's worth playing a bit with the > > pT(local) cut, rather than fix it at 8 GeV. The biplot of max(pT) vs. > > pT(local) we discussed a couple of times before would help here. > > > > > dimuon: require 2 muons (pT>1 GeV which all dimuons satisfy) > > > avoding double counting, select also events with 1 muon > > > and one calMTC (the calMTC has a central track match, with > > > at least X GeV) > > > > > > select between 910 and 995 events (from 4.4 to 4.8%) > > > when varying X between 5 and 10 GeV (the 1 m plus 1 calMTC > > > events are not too many) > > > > Sounds good, and we could easily afford to go as low as 5 GeV on the > > MTC-muon pT-cut. > > > > That's all! > > > > > I want to play a little bit to see if we can recover some more > > > electrons with ID not 10,+/-11, by raising the pT cut for the single > > > electrons and allowing in the diEM case the following: > > > > > > 2 em with ID=10,+/-11 low pT cut > > > 1 em with ID=10,+/-11 low pT cut on the ID=10,+/-11 high pT cut on the > > > other > > > add also the viceversa > > > 0 em with ID=10,+/-11 higher pT cut > > > > > > This will take me the morning.... I'll make a more extensive study > > > overnite (larger statistics, there are 560k events available in > > SAM)... > > > I think that the skimming could start early next week.... > > > > > > Marco > > > > From tamburel@fnal.gov Fri Aug 1 16:09:54 2003 Date: Thu, 22 May 2003 23:56:34 -0400 From: Peter Tamburello To: Greg Landsberg Cc: 'Marco Verzocchi' , 'Eric Thomas' , burdin@fnal.gov, jonas@fnal.gov Subject: RE: skimming I wonder if we are making the cuts too loose based on the current data. As the luminosity increases and the trigger list changes the size of the streams could grow as a percentage of the total. Here's what I find for the size of some streams [% of total] for runs with different prescale sets: stream 15e30 20e30 30e30 35e30 SingleMuon 6.081 6.105 6.588 7.012 SingleEM 10.279 13.258 17.056 17.165 EleTrk 3.940 5.071 6.644 6.963 DiMuon 1.754 2.074 2.606 2.882 DiEM 2.332 3.065 2.749 1.873 EMMuon 2.714 3.329 3.898 3.917 JetsMEt 5.327 6.386 8.184 10.496 MultiJet 5.677 7.147 9.781 12.339 SingleMuon stays pretty flat, but that is because the single muon trigger starts to get prescaled at 30e30. The stream definitions are from Greg's NP slide, except that MTC muons are not included in DiMuon and JetsMEt is from the last iteration with Levan. Peter On Thu, 22 May 2003, Greg Landsberg wrote: > Hi Marco, > > Thanks a lot for the detailed information! A few comments: > > 1) It's clear that p14.03 will be late for the summer conference sample, > so we won't be able to start the common skimming in time for this > sample. I hope that each group is aware of that and continues with its > own skimming. In the NP group we are evolving our selections to approach > the common skimming cuts. I suggest that we delayed the common skimming > till the p14.03 or 04, which fixes the calorimeter problems, is > installed on the farm. > > 2) The issue of |ID| = 10/11 is clearly important, and since we are > delaying skimming, it might be worth our while to look in it. The tests > you have done are quite revealing, but the two-folded increase in the > rate is not something we could afford. I'd study straight ISO cut > instead of the ID cut, which might suffice to bring things under > control. Personally I am against "split" definition of a stream, when > cuts are pT-dependent, as it generally always results in somebody > misusing the data-set. If necessary, I'd rather have an additional > stream with no ID cut but with low-ET electron cut and a track match to > study low-ET electrons seed by tracks. > > 3) Could you please confirm Levan's numbers for jets+MET stream with the > new p14.02? The stream definition is the OR of the jets+MHT trigger and > the following cuts: HT > 25 GeV, MHT > 20 GeV. > > The rest are trivial comments on detailed stream definitions, which I > interspersed with your original e-mail. I have included other streaming > reps on the distribution list for this message, so that they can > participate in fine-tuning. > > Many thanks for staying on top of this business! > > Greg > > > -----Original Message----- > > From: Marco Verzocchi [mailto:mverzocc@fnal.gov] > > > > Hi Greg, > > I've finally found the time to work on the skimming (p14.02 data > > are finally available, no excuse to delay this any more)... > > > > So far I've done testing with a single file (20k events) and these > > are my suggestions: > > > > single EM: require ID=10,+/-11, pT>17.5 GeV (or 17) approximately 10% > of > > the events are selected (no other cuts, except the > requirement > > of ID=10,+/-11) > > > > variation of the pT cut (sample of 20527 events) > > > > 15: 2361 / 20527 > > 16: 2246 > > 17: 2097 > > 17.5: 2043 > > 18: 1977 > > 19: 1839 > > 20: 1681 > > > > at 17.5 GeV of pT cut removing the ID=10,+/-11 requirement > > brings the size of the same from 2043 to 4146 (double.... > 20%) > > > > maybe we could even afford 15 or 16 GeV with the > ID=10,+/-11 > > requirement (this is in line with the trigger strategy) > > 17.5 GeV cut looks a bit funny to me - like we know things to such high > precision. I'd go with 17 or 18 GeV instead. > > > > to be tested tomorrow: remove the ID=10,+/-11 for pT>2x GeV > > (keep total size below 15%) > > > > diEM: require ID=10,+/-11, pT> 6 GeV (this is in line with the > > trigger strategy) selects approximately 2.5% of the events > > > > variation of the pT cut (again 20527 events) > > > > 5: 983 > > 6: 509 > > 7: ... > > 8: 427 > > 9: 367 > > 10: 281 > > 11: 213 > > 12: 158 > > > > to be tested tomorrow: removing the ID=10,+/-11 for pT>10 > GeV > > (keep total size below 5%) > > Looks good! > > > single m: require pT>6.5 GeV for muons which have a track matched to > > central tracker. pT is max(pT from global muon fit, pT from > > tracker). For muons which don't have a track matched to > > the central tracker require pT>8 GeV (this is kept fixed > below) > > > > variation of the pT cut (again 20527 events) > > > > 5: 3269 > > 6: 2229 > > 6.5: 1908 > > 7: 1839 > > 8: 1670 > > 9: 1336 > > 10: 990 > > > > at 6.5 GeV of pT cut changing the cut on the local muons > > by 2 GeV (from 8 to 6 or 10) changes the number of selected > > events by +/-10% > > > > maybe we could even afford 6 GeV > > Looks good again; however, I think it's worth playing a bit with the > pT(local) cut, rather than fix it at 8 GeV. The biplot of max(pT) vs. > pT(local) we discussed a couple of times before would help here. > > > dimuon: require 2 muons (pT>1 GeV which all dimuons satisfy) > > avoding double counting, select also events with 1 muon > > and one calMTC (the calMTC has a central track match, with > > at least X GeV) > > > > select between 910 and 995 events (from 4.4 to 4.8%) > > when varying X between 5 and 10 GeV (the 1 m plus 1 calMTC > > events are not too many) > > Sounds good, and we could easily afford to go as low as 5 GeV on the > MTC-muon pT-cut. > > That's all! > > > I want to play a little bit to see if we can recover some more > > electrons with ID not 10,+/-11, by raising the pT cut for the single > > electrons and allowing in the diEM case the following: > > > > 2 em with ID=10,+/-11 low pT cut > > 1 em with ID=10,+/-11 low pT cut on the ID=10,+/-11 high pT cut on the > > other > > add also the viceversa > > 0 em with ID=10,+/-11 higher pT cut > > > > This will take me the morning.... I'll make a more extensive study > > overnite (larger statistics, there are 560k events available in > SAM)... > > I think that the skimming could start early next week.... > > > > Marco > > From tamburel@fnal.gov Fri Aug 1 16:10:06 2003 Date: Wed, 14 May 2003 12:42:00 -0400 From: Peter Tamburello To: Greg Landsberg Cc: Marco Verzocchi , burdin@fnal.gov, jonas@fnal.gov, Peter Tamburello , Eric Thomas , blevan@fnal.gov, hobbs@fnal.gov, avto@fnal.gov Subject: Re: jets+met skim (fwd) Levan checked the size, as a percent of the total data, for the jets+MEt cuts (ET(j1),ET(j2),MET) (15,15,15) and (20,20,20). He got 13% and 5% respectively. The jets+MEt trigger isn't in yet, so these are lower bounds. Upper bounds would be his results plus the amount of the trigger (4%?). Peter ---------- Forwarded message ---------- Date: Tue, 13 May 2003 18:11:49 -0500 From: Levan Babukhadia To: Peter Tamburello Cc: John Hobbs , Avto Kharchilava , Levan Babukhadia Subject: Re: jets+met skim Hi Peter, Ok, so as a sidetrack of my looking at met+jets in data, I looked at a recent run and I see 13% and 5% for those two sets of cuts (I used uncorrected jet Et's and the so-called original met, meto). Hope this helps, Levan On Tue, 13 May 2003, Peter Tamburello wrote: > > It'll be part of the p14 skims, but they haven't started (p14.02 isn't > even frozen yet). Are you interested in this? Can you answer > what fraction of the data passes the low and high sets of cuts below? > > Peter > > > 4) jets+MET skim: while we agreed to use the trigger definition as a > > temporary solution, we also talked about looking at the size of the skim > > for th following two sets of cut ("low" and "high"): (ET(j1),ET(j2),MET) > > (15,15,15) and (20,20,20). > > > On Tue, 13 May 2003, Levan Babukhadia wrote: > > > Hi Peter, > > > > Have you by chance started on the jets+met skim? > > I understand from Avto/John that this has been in > > works just recently, but John had suggested I check > > with you in case you might have something already... > > > > Thanks, > > Levan ------------------------------------------------------------------- Levan Babukhadia http://www-d0.fnal.gov/~blevan Research Scientist blevan@fnal.gov SUNY at Stony Brook, NY 11794. +1-630-840-3105 (Work) MS 357, Fermilab P.O. Box 500, +1-630-848-1268 (Home) Batavia, IL 60510. +1-630-840-6650 (Fax) -------------------------------------------------------------------