From spagnolo@le.infn.it Thu Sep 11 19:41:44 2008 Date: Fri, 5 Sep 2008 14:56:15 +0200 (CEST) From: Stefania Spagnolo To: Darren Price Cc: Alastair Dewhurst , Jonatan Ginzburg , Vato Kartvelishvili , Lee Benjamin De Mora , Maria Smizanska , Thorsten Stahl , walkowiak@hep.physik.uni-siegen.de, Karl Jakobs , Samira Hassani Subject: RE: Low pt muon and di-muon rates in ATLAS / comments to the NOTE Dear Darren, thanks for your reply. Here is a quick message just to let you know that I'm just back after a a break and I'll read carefully your reply and come back to you soon. Best regards, Stefania On Tue, 26 Aug 2008, Darren Price wrote: > Date: Tue, 26 Aug 2008 17:07:12 +0200 > From: Darren Price > To: Stefania Spagnolo > Cc: Alastair Dewhurst , > Jonatan Ginzburg , > Vato Kartvelishvili , > Lee Benjamin De Mora , > Maria Smizanska , > Thorsten Stahl , > walkowiak@hep.physik.uni-siegen.de, > Karl Jakobs , > Samira Hassani > Subject: RE: Low pt muon and di-muon rates in ATLAS / comments to the NOTE > > > Dear Stefania, > > Thank you for your detailed comments to our note! All the changes you suggested are now implemented / I have listed my responses to some of your points below: > > Points 1--8: > ------------ > > Changed as requested > > Point 9: > ----------- > > a) You are correct that bbmu6X also includes contributions from bb->J/Psi(mu6mu4)X. I have added a small note in section 2.2 to this effect. > > b) Regarding Figure 1, the ccmu6mu4/mu4mu4 contribution is not included as anything other than a single number because this process is particularly inefficient in generation due to kinematics. It is extremely time-consuming to produce high statistics samples of this process without severely skewing the kinematic distributions, and at the time of the study the resources for producing such statistics was not available. As such, the channel is added as a cross-section for completeness, but is done so more as a comment and not as a main part of the study. > > Point 10: > ------------ > > a) Yes, the procedure for the selection of bb->mu6X events is as you wrote. They key point here is that Pythia however is not restricted in its selection: all the cuts on b quarks/muons and so on, are done at a later filtration stage (although the event is dumped if the b-quarks do not pass pT/eta requirements). This ensures that the distributions of the b quarks, muons are not skewed by the selections. This does have the effect of meaning that efficiency of generation is very low however, as Pythia may generate thousands of events before one passes the selection criteria in the tables. > > One cut is unavoidable though: the ckin(3) cut for the min pt of the hard interaction, which must be set to an appropriate level without which efficiency would be so low as to effectively allow no events through in any reasonable time frame. This does unavoidably skew low pT results somewhat, the effects are documented in Appendix B. > > b) The muon is not required to come from a B-hadron for good reason! For example, the ATLAS Level-1 trigger for muons of a particular pT threshold will see a particular rate irrespective of whether the muon comes from a B-hadron, even in a B-decay event, and as such if any additional muons are present in these events they are as legitimate as those from B-decays. > > c) In the generation b-quarks can be produced not just in f+f->f+f but also through gluon fragmentation. > In fact, isub processes 11, 12, 13, 28, 53 and 68 are all included. All generation for B-Physics uses the 'PythiaB' wrapper to Pythia in order to correctly simulate such events. > > You can find details of this procedure in the internal communication ATL-COM-PHYS-2003-038. > As this procedure is used universally for such processes I felt it best not to detail this in this note, but have added a reference to PythiaB in the introduction. Anyone reproducing these results with the jobOptions/parameters as defined will be using PythiaB as standard. > > d) I believe that there is some overlap between the bb->mu6X and cc->mu6X samples. I think a study was done on this some years back. Maybe Maria can comment on this? > > e) Regarding comments on Section A.4: there are two points here -- firstly, when selecting the cuts for ckin(3), pT cuts on the quarks and so on, these are tuned from many iterations of generator level tests using two main guiding (and often contrary) principles, that the cuts should be as loose as possible, and that the final tuning should be efficient enough so as allow generation of datasets in a 'reasonable' time in the central ATLAS production system. Having the cuts too loose means a loss of efficiency, so these cuts always represent some pragmatic 'happy-medium'. > > Having said that, in this particular case, only generation of a cc->mu4X dataset was performed, although cc->mu6X information was still requested at the time as being of interest, so in this case for cc->mu6X analysis, the dataset produced was with c-quark cuts of 4 GeV (for cc->mu4X) , explaining the seemingly low cut. This also explains why the muon pT cuts seem to be done in two steps: this is because they indeed are. For transparency, we note that in generation, only a mu4 cut was required, but in the analysis for the ccmu6X rate, an additional 6 GeV cut was imposed. > > f) Quite right: I have rephrased "This means that all muon sources are possible > porvided that at least one b quark is present in the hadron leading to > the accepted muon" to be "This means that all muon sources are possible > provided that at least one b quark is present in the event with an accepted muon" > > Point 12: > ---------- > > Changed to nb/GeV > > > Point 14: > ----------- > > Appendix B issues: Yes, this is problematic. In principle the 160 nb result should be closer to what we expect than the 110 nb, but this additional contribution does not represent an overall scaling, but dominates at low pTs, so it is difficult to correct for, as official simulation with lower cuts can not be done in the production system (this has been tried without success over the past few years), due to extremely low efficiency of generation. > > In that respect this is a general problem afflicting B-Physics production -- maybe Maria would like to comment about this too? > > This is the reason no conclusion has been made in Appendix B: this is effectively just a warning to people looking at this process near the pT threshold that the cross-section may be somewhat suppressed. > > In the same way, ccmu6mu4X should be affected, but was not studied in depth for this note, as mentioned above. > > I do not believe the problem is as pronounced for single muon rates, due to less stringent requirements on the pT of the final state. As I recall, the bbmu6mu4X process was particularly flagged for further study of the kinematic suppression precisely because the problem was known to be particularly pronounced there. > Production of Jpsi and Upsilon does not suffer -- quarkonia generation in Pythia is handled in a specialised way and are less sensitive to a cut on the hard interaction. > > The generator production for the mu4X thresholds was performed using jobOptions with a number of varying parameters. I have added an addendum to the note in Appendix A documenting the jobOptions that were used for each mu4X production. > > > I have uploaded the new version which replaces the old ATL-COM-PHYS-2007-089 on CDS. > If you have any further comments please let me know. > > Best regards, > Darren > > > > ________________________________ > > From: Stefania Spagnolo [mailto:spagnolo@le.infn.it] > Sent: Fri 7/11/2008 13:48 > To: Darren Price > Cc: Alastair Dewhurst; Jonatan Ginzburg; Vato Kartvelishvili; Lee Benjamin De Mora; Maria Smizanska; Thorsten Stahl; walkowiak@hep.physik.uni-siegen.de; Karl Jakobs; Samira Hassani > Subject: Low pt muon and di-muon rates in ATLAS / comments to the NOTE > > > > > > Dear all, > > let me, first of all, apologize for sending my comments after a so long > delay. > > I think the document is a nice and useful detailed documentation of the > procedures followed for obtaining theoretical estimates of cross sections > used atlas-wide in a special moment of the collaboration scientific > production (CSC notes). > > Here are the points I'd like to discuss with you. > > 1/ In the abstract, I would suggest the following rewording: > ... are presented as predicted by PYTHIA 6.403 for the 14 TeV run. Event > generation was performed usig a specific setting of the event generator > discussed in the paper; no consideration of ... > > 2/ Introduction, first sentence: > ... di-muon rates for various decays of B hadrons and direct ... > there's no mention of the charmed hadrons - I would change in something > like > ... di-muon rates from various decays of hadrons with beauty or charm and direct ... > > 3/ Introduction, second sentence: > Inclusive muon rates will provide a key signature for commissioning of the > detector .... > I would change in: > Inclusive muon triggers will provide a key data set for commissioning the > detector .... > > 4/ Introduction, 3th sentence: > I will completely rewrite as follows: > In addition, it will be possible to perform measurements of known > properties of B hadrons to be compared to results from previous > experiments (UA1, CDF, D0). > > 5/ Introduction, 2nd par., last sentence: > We also provide a breakdown ... > change in > For each process, we provide a breakdown of the muon rate into large > eta regions (barrel-barrel, barrel-endcap, and endcap-endcap) with an > accompanying .... > > 6/Introduction, 3rd par.... I would not mention here the appendices (D > was already mentioned, A/B/C should be referred later in the text of > section 2) > > 7/ Section 2: I would change the title in "Generation of data" > > 8/ Section 2, 1st par, 2nd sentence: > The appendix to this note ..... > change to > Appendix A contains further details of the generation: > generator-level parameters and cuts for each process and DC3 jobOptions > files used ... > also later on "The appendix contain ..." specify which appendix contains > what ... > > > 9/ In section 2.1 I would appreciate a few comments about the > existing relations between the processes listed in table 1; for example: > - bb->mu6X includes the contribution from b-> c-> mu (this is preatty > obvious from the discussion in 2.2.1); it includes also, along with the > 2-muon signature, the process bb->J/Psi(mu6mu4)X, am I right ? > > In fig. 1: why there's no line on the plot for the cc->mu6mu4X > contribution ? Actually, the only mention of the study of > cc->mu6mu4X is in table 1 where the cross section is given > (also for mu4mu4X) nut no discussion, no generation details > (in Appendix A) are given whatsoever ... or did I miss them ? > In the caption: with pT -> as a function of the pT ... > > 10/ In section 2.2.1. I find the first sentence criptic. Actually, I'd > like to see if I understood exactly what the procedure for the estimate of > the mu6X rate was. Here is my undertanding: > - you run pythia with msel 1 (hard qcd) and ckin(3)=6 GeV; > - you process the output by selecting the events with > -- a b-bbar pair with pt of both quarks >=7GeV and |eta|<=4.5 > -- a muon with pt >=6 GeV and |eta|<2.5 > ----- you do not require this muon to come from the decay of a B hadron! > the selected / total n. of events give you the scaling factor for the > cross section predicted by Pythia > Did I understand properly the procedure ? > Moreover I have a question: is the b-bbar pair required to be the primary > qqbar pair coming from the 2f->2f hard scattering or it can be produced, > for example, in the fragmentation of a gluon in ffbar->gg (isub 13) ? > In the latter case, is the bbar pair required to come from a single source > (or might the a b over the pt threshold come from one gluon, the other b > from the second gluon) ? > Sorry is these are maybe naive questions (and maybe related to very rare > processes), but, besides of my own curiosity, I think, it might be useful > to spend a few words to explain in the text the procedure up to this sort of > details with the purpose (which underlying the whole document) to provide > reproducible results. > > If I got the picture right, then I have a question: is there any overlap > between the bb->mu6X and cc->mu6X selected samples and estimated cross > sections ? > > Still on the same subject: in section A.4, why the cut on the pt of > the ccbar quarks is 4 GeV (it was 7 for the bbar) and why the cuts on the > muon pt is described as if it was done in two steps (generator level cut > 4 GeV, offline cut 6 GeV) ?? > > Finally, in 2.2.1: "This means that all muon sources are possible > porvided that at least one b quark is present in the hadron leading to > the accepted muon"... I have a dout here: since no parental link is > enforced between the muon and B hadrons (or C hadrons from B decays) could > the muon accidentally come from the underlying event ? > > > 11/ In table 5 (and in all other tables, for example table 9, table > (un-numbered) on page 17, etc): remove any reference to trigger > logics (LVL1, LVL2), since -as I understand- there's no trigger emulation > at all but just cuts applied on kinematic variables as given by the event > generator. > > 12/ In table 13: units should be > nb/GeV in the first coulumn and nb in the second ? > > 13/ in section 2.2.4, 1st par., 2nd sentence: ist is written that the b > quarks in the event are constrained to ... change to c > > 14/ the most critical point (as far as I can see): Appendix B. > My impression is that the tests discussed here show that the ckin(3)=10 GeV > cut was not adequate to generate bb->mu6mu4X and the results with > ckin(3)=6 should be used... or at least one should make a statement about > what number, 110nb or 160nb, should be thrusted more! > What about the cuts applied in the cc->mu6mu4X study ? > Do we, then, expect any underestimate in the result for bb->J/Phi(mu6mu4)X > where ckin(3) is required to be >=9 GeV ? > Then maybe the question of how the predictions on single muon rates depend > on this parameter might also be raised ? Are we safe with ckin(3)=6 GeV > for bb->mu6X or there's any chance of loosing some contribution at the > border of the kinematic region ? > > Finally, concerning the cross sections given in table 1 for the 4 GeV > thresholds (there's no explicit mention of the specific cuts used for > these estimates): have this numbers been obtained with a ckin(3) cut equal > to 4 GeV ? > > That's all from my side. Please be patient if I made naive questions due > to a not adequate background on the field. > > Best regards, > Stefania > > ------------------------------------------------------------------------- > Stefania Spagnolo > Dip. di Fisica, Univ. di Lecce / +39 0832 297458 (office) 297569 (lab) > INFN Lecce / +39 0832 325128 (fax) > ------------------------------------------------------------------------- > > > > > > ------------------------------------------------------------------------- Stefania Spagnolo Dip. di Fisica, Univ. di Lecce / +39 0832 297458 (office) 297569 (lab) INFN Lecce / +39 0832 325128 (fax) -------------------------------------------------------------------------