Apr 24/25 PionLT/KaonLT Analysis Meeting Notes ---------------------------------------------- (Notes by GH) Today: KaonLT will be discussed first Please remember to post your slides at: https://redmine.jlab.org/projects/kltexp/wiki/Kaon_LT_Meetings Present ------- Regina - Garth Huber, Ivan Zhenchuk, Ali Usman, Alicia Postuma, Nathan Heinrich, Vijay Kumar, Nacer Hamdi JLab - Dave Gaskell FIU - Pete Markowitz Virginia - Richard Trotta Ohio - Julie Roche Richard ------- KaonLT both Q2=3.0 settings update - W=2.32 setting - modified fitting algorithm, had some constraints on the fit range of some parameters, the ranges for some parameters were widened - it=10 shown - Data/MC ratios generaly better, particularly for lower 2 t-bins - higher 2 t-bins have weird Data/MC ratios near zero for phi~0 at high epsilon, where there's good data, Richard is investigating - L/T/LT/TT t-dependences look fairly reasonable - L>T everywhere - LT is still large, similar magnitude to L - TT~15, which is still large, but quite a bit smaller than last week - believes the differences are due to change in fitting parameters - *NB* Garth asks to next time please show "money plots", i.e. the plots of unseparated cross sections vs phi at both epsilon used to determine the separated cross sections - *NB* some discussion about the fitting algorithm - Richard says he checks if the chi-square is stable after some iterations - Garth stresses the importance of also checking if the output parameters are close to the edge of a fitting window. If they are, the chi-square will be stable, but the fit results not optimized - W=3.14 setting - Data/MC ratios nicely between 0.75-1.25 for lowest t-bin - *NB* SIMC might not yet have Ali's resolution correction, needs to confirm this Next steps: - finish W=2.32 setting - reparameterize W=3.14 with new form - parameterize Q2=2.115 and do first tests, should have this by next week - planning full replay of Data and SIMC at end of May - Dave asks if a full replay is really necessary at this time - Richard points out it would be helpful to do before starting on systematics, as otherwise they would have to be redone later - Garth suggests to try replaying the "worst" setting and see if anything changes before deciding whether to replay everything else Vijay ----- PionLT Q2=0.425 MM offset study - low epsilon, ~7 MeV shift between Data and MC, differences between peak positions are clearly evident before offset is applied - mid epsilon, ~2 MeV shift - high epsilon, ~4 MeV shift but not all SHMS settings done yet. Some SIMC settings appear to have a normalization error Next steps: - will try calculating pion missed triggers correction using Alicia's G4 code Ali --- piDelta BSA analysis, new binning of 3 of 5 kinematic settings - starting with Q2=3.0, W=2.32 "The worst setting" - hoping for 2 t-bins instead of 1 - low -t (0-0.6) center SHMS setting - fit of MCs to Data MM distribution generally looks good, except for a systematic discrepancy on left side of Delta peak - in comparison to the shown MM Data region (>1.0), the brown pi+n MC looks too low - however, the pi+n MC is actually fit to the top of the pi+n peak that is outside of this MM range, there is not sufficient freedom to raise the MC normalization to fully reproduce the data in the 1.0-1.1 MM region - the issue is that the resolution correction doesn't fully describe the pi+n tail - fortunately, the MM integration range for piDelta BSA excludes this poor fit region - Alicia: how will the systematic uncertainty be calculated, will it take into account this discrepancy? - one systematic will be determined from the difference between data and MC shapes - another systematic will be determined by varying the background normalization within a "reasonable range", for the pi+n this should be based on the statistical variation at the top of the pi+n peak, and so will still not come all the way up to the data in the 1.0-1.1 MM region - low -t right SHMS setting - good statistics and fits in all 8 phi-bins - high -t left SHMS setting - 4 of 8 phi bins have good enough statistics and MM shapes to be fit - high -t right SHMS setting - the other 4 phi bins have good data, so the sum of left and right gives good phi coverage - this setting has more statistical variation than the left - high -t center SHMS setting still in progress - the issue is that at high -t, there is no pi+n tail between 1.0-1.1 GeV in some phi bins and Ali's MC fiting code breaks - shows plots at end of meeting, where Ali removed the pi+n MC from the fit - phi bins 4,5 do not have Delta MM shape and will not be included despite the fact that the code successfully gave a fit (i.e. fit is too uncertain to extract). bins 1,8 are empty Alicia ------ BSA paper: No update yet from PLB KaonLT u-channel study - replayed all 10.6 GeV data with proton selected and beam helicity enabled - interesting discovery re. background underneath omega peak - Henry Klest showed a PYTHIA simulation in the talk for his exclusive phi production proposal - gets a MM distribution similar to Alicia's data - Alicia got Henry's code working and will try passing PYTHIA events through SIMC - Ali has a bit of experience with Python for his MSc work with CMS - if that turns out to not work very well, the other option would be to modify one of the existing SIDIS generators in SIMC to handle proton fragmentation instead of pion fragmentation - Garth spoke to Bill about this. They are planning to make such a generator, GH told that we would be very interested if they could make one soon rather than later - RF PID studies - Q2=3.0, W=3.14 center SHMS setting - the RF works well for this setting, but it is not available for the right SHMS setting - RF vs Aerogel shows nice delineation between particle types - Q2=3.0, W=2.32 - RF is not good, a wide distribution - similarly RF not helpful for Q2=2.1 and 4.4 settings - Q2=5.5 - RF is helpful and there for all SHMS settings - MM plots - Q2=3.0, W=3.14 helicity=+1 - pi+n peak is heavily suppressed (~1000x) with PID cuts - Q2=3.0, W=2.32 - MM resolution is poorer, omega overlaps with pi+n peak so it might be difficult to extract u-channel results for this setting - Q2=2.1, W=2.95 - MM resolution is good, will have to apply pi/K sample subtraction since no RF info available - Q2=4.4 - same - Q2=5.5 - should have nice results despite lower statistics. Can even separate eta-prime after pi+n suppressed with RF cuts - Cherenkov plots - Q2=3.0, W=3.14 - HGC cut <2npe, Aerogel cut <3npe probably will change to <2npe - will evaluate a cut efficiency for the Aerogel, even if it is not strictly needed for BSA, the pi+n PRL referee asked for it so we should be prepared - Q2=3.0, W=2.32 - plots shown of aerogel with HGC cut and vice versa - Planned analysis order of settings 1) Q2=3.0, W=3.14: RF=Good Stats=High 2) Q2=5.5, W=3.02: RF=Good Stats=Low 3) Q2=2.1, W=2.95: RF=No Stats=High 4) Q2=4.4, W=2.74: RF=No Stats=Low 5) Q2=3.0, W=2.32: RF=No Stats=Med Nacer ----- KaonLT Q2=0.5 LTsep setup - now has Data and SIMC Yields per t-phi bin - wants to cross compare with Richard's results as a check - working on scripts for Data-MC comparison plots and Ratios - then will look at Richard's iteration scripts, with aim to simplify where possible Junaid ------ PionLT Q2=3.85, W=2.62 LTsep setup - presents full equations for Normalized Yields and Error Calculations - very nice! We appreciate the explicit documentation - shows spreadsheet of Normalized Yield per t-phi bin - Dave: the errors seem too small - Note added later: JM found an error in his script, the errors were normalized twice by mistake - *NB* it would also be good to add columns of #Counts/bin and %error/bin - working next on SIMC yields per t-phi bin Nathan ------ PionLT Coincidence Blocking studies - met with Dave last Tuesday - plots made of CTime.CoinTime_RAW_ROC1,2 - ROC2 version has main CT distribution sitting on a very wide pedestal, from -600 to +600 - ROC1 version has more structure, same pedestal to right of main CT distribution to +600, but a sloping decline to left of CT distribution extending to -400 and some sharp peaks at -400 - Dave is mystified, assuming a 50ns coincidence time window, the full CT distribution should only be about 100ns wide, so what is causing the distribution to be ~1000 wide? - Dave: what timing window cuts are being applied? - Nathan says the timing cuts are 2000 bins winde, but isn't sure offhand what is the conversion to ns - NOTE: Nathan is using ROC1 version in the analysis - Dave suggests that the shoulder on the left of the ROC1 plot indicates we should use ROC2 - plotting ROC1 vs ROC2 should indicate that it's running against a trigger window limit - Dave will look at the Timing Cuts used on the files Nathan sent him, to see if he can figure out what's going on - Nathan will try changing the TCOIN.param TDC cuts and see if it makes any difference - this would mean switching between offline and online versions - not expecting the deadtime effect in the Lumi study to go away, but even it changing would tell us a lot about the origin of the extra deadtime Ivan ---- - NSERC summer student starting with us next Thursday May 1 - we will have him look at BSA data for low Q2 KaonLT and PionLT settings - Alicia has kindly offered to replay these data with helicity enabled Next Meeting ------------- - Thur May 1 @ 15:30 Eastern/13:30 Regina - PionLT will go first ** THIS TIME WILL BE USED AT LEAST UNTIL THE END OF AUGUST **