Mar 7/24 PionLT/KaonLT Analysis Meeting Notes ---------------------------------------------- (Notes by GH) Today: PionLT 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, Nathan Heinrich, Muhammad Junaid, Nacer Hamdi, Alicia Postuma, Ali Usman, Vijay Kumar FIU - Pete Markowitz CSULA - Konrad Aniol CUA - Richard Trotta, Casey Morean, Tanja Horn JLab - Dave Gaskell Nathan ------ PID plots for SHMS NIM paper - HGC efficiency was found to be low, ~80% - removed zero bin entries as Vijay had found in his PID study, then ~99% outside of hole region - indicates this is still a problem during PionLT data taking - DG: are you sure these aren't otherwise bad events? - GH: Vijay had shown that bad HGC events corresponded to good MMK, not due to other expt backgrouns - believed to be due to either a bad reference time or bad pedestal caused by high rates per PMT due to low HGC granularity (only 4 channels covering focal plane) - will eventually need to use fADC mode-10 data to determine an apporpriate correction - *IMPORTANT NOTE* Richard uses HGC NPE<1.5 cut, so maybe a correction is not needed for his data, but he will have to confirm, for example if a good HGC time is required to calculate the NPE, then a correction would still be needed - shows efficiency plots vs X,Y and vs delta for Aerogel, NGC, HGC - shows NPE correlations between detectors for PID Next steps: - will look through Lumi scripts and try to understand recent changes Junaid ------ DC efficiency plots for SHMS NIM paper - removed outliers - finalizing calorimeter plots for paper Next steps: - setting up scripts for Heep COIN data analysis Richard ------- Fixes to xsect uncertainty calcs - met w/GH on Monday afternoon and found the problem - errors were not propagated correctly - uncertainties now stable vs iteration Pion contamination subtraction - Ali provided clean pi+ samples for Q2=2.115 and both Q2=3.0 settings - a sample subtraction is shown - pi+ subtraction is red curve, stops at MM=1.24 for some reason - DeltaPi contribution seems surprisingly small - some rechecks should be done Xsect comparo w/ and w/o pion subtraction - Q2=3.0, W=3.14 shows ~20-25% reduction in sigL at low -t, no MM cut applied - makes sense that low -t region is most affected, as pi+ cross section should have much steeper t-depenence than K+ - comparison with 1.10epsHi yields - sigL works out negative, sigT unrealistically small - Q2=2.115, sigL consistent w/zero for lowest -t bin - compare to Marco's K+ data for Q2=3.0, W=2.32 - his epsLO value is close to our epsHI value - having difficulty getting things to converge sensibly for different iterations - Vijay: what do your Ratio plots look like? - RT shows Q2=3.0, W=2.32 ratios vs phi - sees strong epsilon-dependence to Ratios - Ratios at epsLO larger than Ratios at epsHI - GH: it is *very important* to look at the Ratios plots for each iteration - the goal is to get the Ratios having similar values for both epsLO and epsHI, otherwise any separated cross sections are not reliable - look more closely at the parameterization - GH: suggests to start with only P1 for sigL and P5 for sigT, setting others to zero initially Ali --- Quick updates - iFarm is very slow, but finished replays for two higher Q2 settings - will get these pi+ spectra to RT soon - working on hodoscope efficiency plots for SHMS NIM paper SIMC resolution study - no HMS delta-corr needed for 3 of 5 kinematics, so can finalize SIMC resolution tuning - Q2=5.5, W=3.02 comparison of MM vs t-bin between new and Feb analysis - Feb had overall good agreement, but poor agreement vs t-bin, SIMC was wider than data at low -t - now doing tuning for all t-bins, to make sure SIMC is not wider than data anywhere - DC resolution factor is now 4x instead of 9x - Q2=3.0, W=3.14 center: now 5 t-bins instead of 4 - Q2=4.4, W=2.74 center - reasonable agreement of MM shape w/data - now running MC for all DeltaPi BSA kinematics w/new resolution correction - will start MM shape study for 3 of 5 settings - Dave: still don't understand why the correction factor is so large, but at least it's no longer 9x - Vijay: suggests to look at xptar, yptar comparison, to make sure this is not adversely affected by the MC resolution tuning - definitely a *GOOD IDEA* to check this Vijay ----- Summer 2019 LT-sep analysis, Q2=0.38 - shows Yield Ratios vs phi-bin at epsLO w/o iterations - sees some unusual phi-dependence in the Ratios that doesn't make sense - GH: suggest to turn of LT, TT in initial parameter fit - Dave: notices a weird Left-Right asymmetry in the Ratio error bars, one side has dramatically smaller errors than the others, if there is reasonably uniform phi-coverage these errors should be more uniform - Tanja: suggests to look at Ratios separately for each SHMS setting to see if things make sense, will be helpful diagnostic for debugging Alicia ------ BSA update: looking at the 3 Q2-settings that don't need HMS delta-corr - MM cut study - compare MM from data and SIMC, more zoomed in than Ali's plots - Q2=3.0, W=3.14, center Data is wider than MC - Data-SIMC agreement is best at Q2=5.5, but still wider than SIMC - what MM cut values to use? - Dave: prefers no lower limit cut at all, unless the lack of a cut introduces a problem (e.g. extra background) - Dave: upper cut 0.98 is too tight, likes 1.0 better - suggest to look at Data/MC vs MM and look where the Ratio plateaus for deciding your cut values - Tanja: the Ratio plateau will occur somewhere around 5-10 sigma from pi+n peak - CoinTime windows for Random subtraction - CT spectra are now flat after earlier error fixed - can use 6 peaks Left and Right of main peak without issues - PID cuts are applied to the plot, but no HGC, and loose MM cut applied - Dave: suggests to look at yield of each bucket and stay away from that region where yield/bucket starts to decrease - GH: the roll off in the yeild will in principle occur at different CT for different settings, as the coincidence signal timing is TOF dependent - Nacer: where is there a region where peaks are merged together next to the first random peak? - Dave: it's just due to finite timing resolution and the fact that the prompt peak is so large Nacer ----- Resuming KaonLT Q2=0.50 analysis after calibrations - H.gtr.beta plot is now centered at 1.0 after new calib - SHMS P.gtr.beta vs P.gtr.dp shows a wiggle - Nathan: this is due to a minor fault in the Hodo calib but you can ignore it - P.aero.npeSum vs P.hger.npeSum for K+ identification - HGC<3, Aero>3 NPE cuts used so far - will have to tune the cuts according to the data distributions - CTime.eKcointime_ROC1 vs P.kin.secondary.MMK - can see small K+Lambda and K+Sigma dots amid large pi+ background - GH: can apply a MMK>1.05 cut to remove the big pi+n CoinTime peak and make it easier to select K+ CT cuts - Ali: for these kinematics, you will find the RF cut to be very effective - Vijay: agrees, looked at these data earlier, with RF cut, you can get the pi+ CT peak to be smaller than K+ peak - Nathan: eK cointime is not quite centered at 1 - Stephen may have not set the CT offset quite correctly, you can tune it - testing some Real/Random cuts to apply - MMK after random subtraction shows negative counts in DeltaPi region - needs more optimization as part of PID studies Next Meeting ------------ - Thur Mar 14 @ 15:00 Eastern/13:00 Regina - KaonLT will go first - note the change in time for Regina people, because of USA daylight savings switch