Apr 11/24 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, Nathan Heinrich, Muhammad Junaid, Alicia Postuma, Nacer Hamdi, Ali Usman, Vijay Kumar CSULA - Konrad Aniol York - Stephen Kay UVa - Richard Trotta JLab - Dave Gaskell CUA - Tanja Horn FIU - Pete Markowitz Richard ------- Resolving various issues in KaonLT LT-separation - RF time cut now removed - -t distribution offset between data and SIMC - found that proton mass was used for detected particle instead of K+ - SIMC recon script did not use the updated kinematics to recalculate the weight wt'= wt*(sig/sigcm_sim) - weights were not properly recalculated when doing an iteration - ROOT was always using the first iteration weights - an over-writing issue in ROOT script now fixed - with these corrections, there is a significant improvement in the phi-distributions at high and low epsilon for the kinematic setting shown, Q2=3.0, W=3.14 Outstanding issues - some mismatches remain between SIMC and data distributions - PMISS shape is off - Dave: this could be due to the SIMC model not yet being optimized, wait for more iterations to be completed before checking this again - pion subtraction needs improvement - "clean" pion sample used for Q2=3.0, W=3.14 leakage subtraction clearly shows some K+Lambda events - Garth: need to be very careful with the pion subtraction, as it has a different epsilon-dependence than the true K+ events, and you don't want to induce an artificial epsilon-dependence to the data - also very important that both K+ and pi+ samples populate the same focal plane limits, i.e. can't have an HGC hole cut on one but not the other Vijay ----- Low Q2 pi+ LT-sep update - unseparated cross sections shown for 3 epsilons - some weird issues remain with phi-distribution of lowest -t bin, although not as bad as last time this was shown - preliminary separated cross sections for L,T - no ratios or kinematic comparison plots shown, so it's impossible to provide helpful suggestion on what to look at - it's *VERY IMPORTANT* to also show diagnostic plots Ali --- piDelta BSA analysis - last week, showed some preliminary BSA results at Q2=3.0, W=3.14 - since then, the t-bin limits have changed - the issue is that the background in the Delta region grows immensely with -t - HGC cuts are applied to remove K+Lambda contamination in piDelta region - *IMPORTANT* any SIMC plots need to have the same cut - MM plots shown for new -t bins, 0-0.37, 0.37-0.60, 0.60-1.0 and total 0-1.0 - will exclude the 0.60-1.0 region since there are too few piDelta events there for a reliable fit - in process of doing MM fits for new t-bins - an issue with 0.37-0.60 region is that some phi bins (center setting) have very strong background, so getting a good piDelta fit is tricky - the SIDIS background grows much faster with -t than the piDelta does Alicia ------ pi+n BSA Q2-scan - delta(FIT) error will be treated separately from delta(CUTS) error, since they're not of comparable size - doing RMS of both together results in under-estimate of total uncertainties - LT'/sig0 data shown vs. Regge model curves from B-G Yu - much larger x-dependence than Q2-dependence - LT' at fixed t vs Q2 is mostly flat, but different for each x-bin - overlay of LT' at xB=0.25, 0.40 - xB=0.25 data sets at Q2=3.0,W=3.14 and Q2=2.1,W=2.95 overlap very well - xB=0.40 overlay has more scatter, but still small Q2 dependence - Dave: how different are the -t_min for the two xB=0.25 data sets? - the difference is quite small, about 0.01 - Dave: since the Q2-dep is small, how about overlaying all Q2 and look at the x-dependence - Alicia will make a plot for next week Nacer ----- Low Q2 K+ PID studies - using runs=6921-6978, Q2=0.5, TH_shms=9.79(left) for comparison - Main cuts: - H.cal.eTotTrackNorm>0.5 - H.cer.npeSum>2.0 - 0.15>RFtime(corrected)>0.95 - RF time correction - applied an adhoc correction to remove xptar dependence - Signal/Bkd study via a Gaussian fit + polynomial bkd - does fit vs RF time cut - apply both Low and High RF cuts, vary each one independently - optimum values: Low=0.15, High=0.95 - throw away events between these limits, correponding to pions - Best K+Lambda yield appears to be with only RF cut, RF+HMS cuts is similar - applying SHMS cuts greatly reduce the K+Lambda yield due to various inefficiencies, the extra clean-up of the MM spectrum is not worth it - Ali: there appears to be a small amount of piDelta contamination underneath the K+Sigma events - suggests to do a pi+ leakage subtraction, similar to what Richard is dong - Garth: since the RFtime is so effective, can use a tight RF cut to easily generate a clean pi+ spectrum for subtraction - Next steps - RF cut efficiency, Lumi checks Nathan ------ PionLT lumi scans - figured out a reasonable BCM offset for the runs in this scan - offset=-0.17uA - make scaler(ELREAL) normalized yield flat vs rate - RelYield-NoTrack has a rate dependence (anti-boiling) indicating a LiveTime issue - TotalLiveTime(TLT) is too low at high rate - Dave: as an intermediate step, try using CPULT instead of TLT, the Electronic DeadTime should be very small - Richard: can use the CPULT to fine tune your cuts, then re-introduce EDTM - Nathan will add EDTM timing plots to investigate cuts - Dave Mack sent a new BCM calib, but had difficulty w/replay, investigating - has a meeting with DM next Monday re. learning how to do BCM calib - Dave: do we know how complicated the BCM calib is? - the UNSER is very noisy, DM throws away UNSER points that are outliers, requires some judgement - Nathan: expecting to generate plots and have DM identify outliers to be removed, will still take less time than DM having to do all parts from scratch Junaid ------ - arriving at JLab on Tuesday night - Dave: beam studies on Wednesday, so a Hall C walk through is possible PionLT Heep analysis - Ali had pointed out that BPM calibs were wrong - slope values were right, but offsets wrong. All updated now - raster config files were also wrong - there was an issue for two run ranges where only one pair of raster coils were operational, so the other had to be run at double current - this is also fixed - Heep Effective Charge, corrected for TrkEff, but not Calor or Cer Eff - wrong BCM still used - for setting shown, the data normalized yield is ~2x SIMC yield, which is surprising - for a different run, the yield ratio is much closer to 1 - Richard: you need to use the kinematics recon script on SIMC data - unfortunately, this means the PM, etc. plots have to be redone again - Dave: make sure you have the correct_eloss flag turned off in the SIMC input file - by default, SIMC both does energy loss, and corrects for it in the event reconstruction, but hcana does not correct for eloss, so you need to turn it off Next Meeting ------------ - Thur April 18 @ 15:00 Eastern/13:00 Regina - PionLT will go first