Apr 9-10/26 PionLT/KaonLT Analysis Meeting Notes ------------------------------------------------ (Notes by GH) Today: KaonLT will be discussed first Thursday: Present ----------------- Regina - Garth Huber, Alicia Postuma, Nathan Heinrich, Vijay Kumar, Nacer Hamdi, Muhammad Junaid, Nermin Sadoun Virginia - Richard Trotta CUA - Tanja Horn, Chi Kin Tam, Sameer Jain Ohio - Julie Roche FIU - Pete Markowitz Chi Kin ------- KaonLT high Q2 analysis - double checking differences between high and low epsilon data - improved HGC hole cut, based on a plot of HGC efficiency vs hgcer.xAtCer,yAtCer - does not resolve MM shift at high epsilon, but now the pion bkd underneath Lambda peak is less, should result in smaller systematic uncertainty in the bkd subtraction - double checking HMS matrix elements - 2018 ME instead of NPS used - still checking on correct value of hphi_offset to use for both spectrometers - redid t-binning - increased lower and upper limits of used t-range - new SIMC event processing script - did not like how long it took to reweight SIMC events after each iteration - changed to pre-process SIMC events as a table in Qi,ti bin indices and weight - all operations are vectorized, one-shot reweighting - reweighting is now nearly instantaneous, takes 2min instead of 2 hrs - takes some work to set up initially, but gain that investment back in speed - systematics studies - varied upper end of MMmax cut from 1.12 to 1.19 - cut below 1.14 has poor statistics, 1.14 is the nominal cut used - yields from 1.14 to 1.17 look vairly stable - *NB* Tanja: please look at how sigU=sigT+epsilon*sigL varies w/cut - *NB* Garth: the nominal MM cut should really be near the middle of the varied MM cut range, this indicates that your nominal MMmax is too low - also suggested the same last week, the high sensitivity to the radiation correction on/off is due to the MM cut being too tight Richard ------- KaonLT high Q2 replays - set up several sets of replay pass4: testing pass4a: the new April 2026 replay pass4b: will be done after the Matrix Elements question (above) is settled - planning to use hphi_offset=0 - Nacer: is the CoinTimeBlock Correction working properly in your new report files? In my new replay, the output variables are zero. Still trying to figure out what is wrong - Richard will check more carefully, and they will meet offline if needed - new skim_ROOTFiles - 7 branches: no cuts, 3 for RF cut, 3 for no RF cut - the 3 for RF cut can be removed, since RFcut is not used for this analysis - Everyone should note: collected batch submission scripts in farm template area on GitHub - added new diagnostic tools for failed Run Codes - SWIF2 jobs are much faster, data written to node, so many fewer communication bottlenecks that slowed down jobs - need to update efficiency scripts to include CoinBlocking Nacer ----- KaonLT low Q2 analysis - cleaning and organizing scripts used to compute yields - ran a replay including brances for CoinTimeBlocking studies - addition of blocking factor to script gave zeros, sill investigating Sameer ------ PionLT Q2=1.6 W=3.08 low epsilon (2022) THshsms=5.50deg PID studies - HMS cuts: Hcer>1.5 Hcal>0.7 - *NB* Tanja: before finalizing the cuts, need to find out the cut efficiency for these values - Hcal efficiency requires a special analysis. Junaid sent out an efficiency report explaining the method - SHMS cuts: - Junaid: the appropriate aerogel tray cuts for each n-index are automatically selected in the replay script, only need to confirm that they are applied correctly - Nathan: link to the "good" replay script which picks the def-cuts file for you based onAerogel tray https://github.com/heinricn/hallc_replay_lt/blob/master/SCRIPTS/COIN/PRODUCTION/PionLT_REPLAY/FullReplay_PionLT_Phys_Prod.C - PCal: after CoinTime prompt cut is applied, a noticeable peak at 1 remains, which is a bit surprising - *NB* Note added by GH after more thought: Given the small SHMS angle, and the low Q2 of this setting, it is conceivable that e+e- real coincidences from vector meson decay are being detected with small probability. It is the only way he can think of e+e- pairs surviving the CoinTime cut. If so, these events should be excluded when the Q2-W diamond cut is applied Please make a note to check this again later - Coin PID cuts: - *NB* suggestions to make plots of RFtime vs MM and PCal vs MM Friday: Present --------------- Regina - Garth huber, Vijay Kumar, Nacer Hamdi, Muhammad Junaid, Nermin Sadoun, Nathan Heinrich, Alicia Postuma Virginia - Richard Trotta JMU - Gabriel Niculescu CUA - Tanja Horn, Chi Kin Tam Glasgow - Kathleen Ramage FIU - Pete Markowitz Ohio - Julie Roche York - Stephen Kay Alicia ------ KaonLT u-channel shape study - data are binned in 3 u bins (0-0.06, 0.06-0.12, 0.12-0.40), 8 phi bins - first make plots for each u-bin (phi integrated) of: - bkd subtracted data vs omega MC - rho vs omega subtracted data (for lowest -u bin, where only rho MC contributes, no Xphasespace) - residuals (data-omega-bkd) vs MM - then separate into plots for each phi bin - middle u-bin: pi+n leakthrough has a notable phi-dependence - large u-bin: Xphasespace has significant contribution, residuals are worse for MM>1.0GeV, fit range goes up to 1.2GeV - added a print box to indicate the exclusion criterion triggered (when it happens) - some bins have an extended omega tail, still need to implement radiation criterion (if SIMC indicates most of omega peak comes from radiation, not from single photon exchange, the bin should be excluded) - plans to do a Pass4b (as Richard's) for proton kinematics - also needs to look at the SHMS proton collimator routine in SIMC, this might be needed for the proper modeling of the extended omega tail Vijay ----- PionLT low Q2 analysis - applied t-shift to Q2=0.375, then did 5 iterations - sigL: first 3 t-bins are now much flatter, possibly some evidence of turn-over at -t<0.02 - had to change sigL fit function to account for this, as previous function could not reproduce the turnover - sigT: changed to a linear t-dependence - Data/MC ratios: eps=0.25 have a noticeable phi-dependence in t-bin #4 - eps=0.63, 0.75 plots are much flatter for this t-bin - Gabriel: there is some evidence that the Rosenbluth plots are not left-right symmetric, which is also reflected in the Data/MC ratios - they see this in their data too. It indicates some small problem in the data that is not yet corrected, but have difficulty understanding where it's coming from - Garth: agrees. The only solution he can think of is to assign a larger sysematic to LT, TT to account for the asymmetry, perhaps in the Model Dependence - *NB* Vijay did not have his Model Dependence results immediately available, it's important that we have a chance to look them over and thing about them - analysis difficulties: - getting an unusual error when try to use jput -cache FATAL Class not found: jlab.scicomp.shell.io.InputOutputSubprocess, caused by jlab/scicomp/shell/io/InputOutputSubprocess - unfortunately, no one has any suggestions. Best to contact CC helpdesk Junaid ------ - no report, working on thesis - most recent LT-sep results were presented in March 27 meeting - requests Gabriel to post his kinematic offsets info, interested in seeing if that gives better results - after meedting: uploaded the 28-page kinematic offset study on Redmine https://redmine.jlab.org/issues/1238 - Also posted the presentation (from Feb. 2026) on the topic https://redmine.jlab.org/issues/1221 Nathan ------ PionLT LT-sep cross checks - finally getting same SIMC cross sections and yields as Junaid - my mistake Junaid was using 1 wrong file in the earlier comparison - Q2=5.0 W=2.95 data replay w/ all def-cut files, calibrations, etc. - this setting has 3 epsilons - shows plots from each epsilon, indicating the replays worked - the default PID cuts need a lot of optimization, lots of K+Lambda, p+omega events seen in pion MM histo Nermin ------ PionLT LD2 analysis, setting up for cryotarget boiling study - looking at Nathan's Lumi analysis scripts, tried replaying some Carbon runs - analysis difficulties: - frozen out of JLab account since Wednesday, contacted CC helpdesk Kathleen --------- PionLT LD2 analysis - working on poster for IOP meeting at Univ of Brighton next week - Stephen sent helpful comments, poster presentation is Monday - giving a talk at Early Career workshop on Thursday, will circulate slides - started on detector efficiency studies - analysis difficulties: - group area is full - Stephen: Do "du -hd1" in /group/hallc/c-kaonlt (or c-pionlt) and to get a rough idea of where lots of stuff is being stored - it appears that Nathan is the culprit, using 25 GB - cc.jlab.org/utils/disk-quota/ Next Meetings ------------- - Thurs: Apr 16 @ 16:00 Eastern/14:00 Regina - PionLT will go first - Fri: Apr 17 @ 11:00 Eastern/9:00 Regina - we will continue where we left off