Feb 19-20/26 PionLT/KaonLT Analysis Meeting Notes ------------------------------------------------- (Notes by GH) Today: PionLT will be discussed first Thursday: Present ----------------- Regina - Garth Huber, Nacer Hamdi, Muhammad Junaid, Vijay Kumar, Alicia Postuma, Nermin Sadoun JLab - Dave Gaskell CSULA - Konrad Aniol CUA - Tanja Horn, Chi Kin Tam, Sameer Jain FIU - Pete Markowitz Junaid ------ PionLT Q2=3.85 analysis - started working on PhD thesis and on systematic studies in parallel - will meet w/Nathan on systematic studies already completed Vijay ----- Low Q2 PionLT analysis - presented new model systematics studies - followed instructions in Blok et al. paper: - varied sigL, sigT +/-10% then 2 iterations - effect is very small overall, with 1st t-bin is more sensitive: sigL only 0.004 ub/GeV2 variation - apparently this is not what Tanja did, not sure why the text in the paper is written this way. A change in the parameters should result in negligible change after 2 iterations as the parameters return to their optimal values - Tanja suggests instead to change the functional form in some way and see what that change is - Garth: Vijay already has a slightly different parameterization (from Nov/25), so all he needs to do is compare the latest paramterization result with that one - the difference is that the old one has no Q2-dep in the functional form, while the new one does Alicia ------ u-channel analysis - comparing data and MC kinematics plots for Q2=3.0, W=3.14, high eps, center - apply MC normalization params from MM shape study to other kinematic variables - two types of plots presented: sum of MC compared to data, subtract bkd MCs from data and compare difference with omega MC - encouraging level of agreement, but more checks needed - ssypfp, ssxpfp distributions have a lot of zero events, need to check if 1-2 runs for the setting were replayed improperly - removed the zero events for now, and compared shapes with MC - need to recheck MC normalizations to data - working in parallel on Phi-007 MM reconstruction resolution for Expt Readiness Review w/ Henry Klest Chi Kin ------- KaonLT high Q2 analysis - looking into discrepancy between his results and Richard's - narrowed down that difference is not coming from cuts - next possibility is that it might be the SIMC weights - meeting tomorrow w/ Richard and will report again tomorrow - started drafting paper on Overleaf Sameer ------ PionLT luminosity analysis - working with Nathan's lumi scripts to learn how to use them and apply to his own analysis Nacer ----- Low Q2 KaonLT analysis - further work on MC model optimization - prior best model: - functional form only for sigT, sigL,LT,TT=0 functional form - ratios were fairly good, but some wiggles - fit of function to sigL,T results vs t not very good - new changes: - added another t-bin, since sigL shows a large jump between bins 1,2 - New limits (8 bins): 0.070, 0.084, 0.090, 0.096, 0.102, 0.108, 0.116, 0.128, 0152 Old limits (7 bins): 0.070, 0.086, 0.094, 0.100, 0.106, 0.114, 0.126, 0.152 - has ~10k events/bin - then used only sigT model for new 8-bins, Data/MC ratios changed more than expected but still near 1 - change to simple functional forms for L,LT,TT same as Chi Kin sigL = p2*exp(-|p3*t|) sigLT= p4*exp(-|p5*t|)*sin(theta*) sigTT= p6*exp(-|p7*t|)*sin^2(theta*) - Data/MC ratios look very good, close to 1 and much flatter than before, even the highest -t bins look good - fit of function to sigT,LT,TT vs t also looks pretty good - new 2nd t-bin has sigL slightly higher than zero by ~1.5 sigma, bins 3-8 remain statistically consistent with zero - kinematic comparison plots (Data vs MC): - HMS xptar looks better after adjusting HMS ME offset - Dave: a few nagging discrepancies but not bad agreement overall - Dave: still concerned that rise in sigL at low -t is anti-correlated with drop in sigT for same bin, need to demonstrate that this is real, and not an analysis artifact - are the MM offsets different at low and high epsilon? - the difference in offsets is ~2.5 MeV - The MM offset is applied only to MM, and is not propagated to the other kinematic variables. The concern is that MM is correlated with t, so significantly different MM offsets at high and low epsilon could imply the t-distributions are shifted by different amounts at high and low epsilon and this could give rise to anti-correlated behavior between sigT,L *NB* Garth will work with Nacer to estimate the implied shift in t by a 2.5 MeV MM shift Friday: Present --------------- Regina - Garth Huber, Ailcia Postuma, Nacer Hamdi, Nermin Sadoun, Muhammad Junaid, Vijay Kumar York - Stephen Kay JMU - Gabriel Niculescu CUA - Tanja Horn, Chi Kin Tam Virginia - Richard Trotta FIU - Pete Markowitz Glasgow - Rachel Montgomery Nermin ------ PionLT LD2 analysis - met w/ Kathleen Ramage yesterday on analysis issues - RF times offset correction for LD- Q2=1.6, W=3.08, high eps=9.16GeV - before: RF time distribution had 2 peaks separated by 2ns, both consistent with pi-p missing mass - determined that some runs needed the RF time to be offset by 2ns compared to the others - the issue, as we learned during the experiment, is that sometimes Hall C receives the even beam bunches, and sometimes the odd beam bunches, and we have to correct the odd ones by 2ns to bring them into time with the even ones - the shifted runs were: 12043,44 which were in the middle of the setting - after: RF time has 1 nice peak - shows RF time plot w/o and w/ CT, Hcer, Hcal, Acceptance Cuts: background is much reduced with these cuts applied - NGC plots have most events near zero Mean=0.134NPE #Entries=646774 - apply Pcal>0.85 cut to select electrons in SHMS Mean=0.153NPE #Entries=305096 - ~50% of SHMS events are electrons - Nermin looks up LD analysis report by Argha Das - NGC mean is ~4 NPE for a different setting, much higher than here - Garth: suggests to look at low epsilon data for same setting to see if the issue is just these runs, or overall - Nermin will look into the NGC calibration scripts Rachel ------ PionLT LD2 analysis - Kathleen looking at a different LD- setting than Nermin, sees a very large #randoms, hard to tell where the prompt peak is - Nacer: suggests to look at CoinTime with MM cut applied, it should make the prompt peak stand out more - Garth: the LD- settings have high random rates due to electrons in the SHMS, it is easier to set up the analysis first on LD+ data for same kinematics Richard ------- KaonLT high Q2 analysis - looking into discrepancy between his results and Chi Kin's - compared results bin-by-bin - Kin's yields ~50% higher - found 2 differences in cuts: - CKT has optimized the diamond cut - HGC hole cut also slightly different, Richard needs to confirm that x,y axes are applied correctly (so the hole cut is where it is intended) - SIMC weights were consistent, so that's not the issue - believes the issue is caused by the order of cuts being applied - in principle, cuts are supposed to be commutative (i.e. the order doesn't matter), but possibly there is a NumPy array issue which screws this up - Gabriel: is there a milestone plan on who checks what? - Tanja: it's important to try to keep some things constant (such as identical cuts between RT and CKT) so the cause of the difference can be identified more readily Next Meetings ------------------ - Thurs: Feb 26 @ 16:00 Eastern/15:00 Regina - KaonLT will go first - Fri: Feb 27 @ 11:00 Eastern/10:00 Regina - we will continue where we left off