Sep 12/25 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, Nathan Heinrich, Nacer Hamdi, Muhammad Junaid, Alicia Postuma, Nermin Sadoun FIU - Pete Markowitz CUA - Tanja Horn, Chi Kin Tam, BC Munoz, Sameer Jain York - Stephen Kay Ohio - Julie Roche JMU - Ioana Niculescu JLab - Dave Gaskell Nacer ----- KaonLT Q2=0.5 LT-separations - replicated sig_factorized_2007 model in physics_iterate.f - this model uses different cross-section units than the usual d2sigma/dt-dphi so a different Jacobian was needed - will use sig_factorized_2007 model for first iterations, which are still in progress Alicia ------ pi+n BSA paper modifications - completed radiative correction systematic requested by referee - Asymmetries for radcor=ON/OFF are nearly identical - max discrepancy is 0.002 (or 0.35% of Asymmetry) for Q2=3.0, W=3.14 - still running other settings, max discrepancy so far is 0.003 or 0.6% of Asymmetry - received reconstructed VR Regge model from CKY, now all models in Fig 6 use exact kinematics as data - this gives much better agreement with data at low -t - VR model is now also included in Fig 7 - VR model gives good agreement with data - now both Regge-based models agree better with data than GPD model, which also addresses a question by referees - will try to smooth the theory curves in Fig 6, probably kinks are due to varying kinematics of data, the kinks are most noticeable in GPD model - planning to circulate both modified paper and referee comments early next week Ioana ----- KaonLT Lambda/Sigma/Lambda* ratios - looking at KaonLT Heepcheck analysis - having difficulty locating the offsets Richard used, found some discrepancies among GitHub files - *NB* both Alicia and Chi Kin will send what they're using Chi Kin ------- KaonLT high Q2 analysis - lots of broken files in cache, replayed all 10 settings in Q2=3.0 W=3.14 and Q2=2.1, W=2.95 settings - for runs which were good before, the new file size is somewhat larger than Richard's (e.g. Run 5013: new=847MB old=713MB) - Ioana: also noticed some missing files in cache - Stephen: some farm jobs had failed and output was not properly closed - MM is identical - *NB* Garth: please also check #events between old and new versions - Nathan, Junaid: believe the correct GitHub to use is MASTER, not LTSep_Analysis_2022, which does not include most recent updates *NB* please check with Richard on this - Diamond cuts rechecked, low epsilon nicely centered in big epsilon diamond - *NB* Garth: please also compare the low epsilon Data and SIMC diamonds, if there is a shift between Data and MC, it's important to take the SMALLER of the low epsilon diamonds - MM-shift fits - fitting Lambda peak w/ Gaussian, same shift applied also to pion leakthrough data - see a small pi+n MM shape difference between pion leakthrough and pure pion samples - Ioana: leakthrough and clean pi+n samples have slightly different delta distributions, which accounts for the shape difference - SIMC negative weights - using what believes to be Richard's latest parameters - Left SHMS settings have negative weight, which is surprising, still checking with Richard on whether using the correct parameters - found that the weights are negative in a specific region of t-phi - for now, just reduce LT function parameter causing this, to give positive weight - also found that TT gives negative weight at high -t, outside expt acceptance - Garth: in this region the TT model is unphysical, but since it's outside the expt acceptance, it will have no impact upon the results Junaid ------ PionLT Q2=3.85 LT-separations - making changes to LT-sep functions due to problems identified last week with T,TT - changed T=(p1/Q2)*exp(p2*Q2)*exp(p3/|t|) - TT also changed, first part of function is (p14/Q2)*exp(p15*|t|)+polynomial - with initial parameters (0th iteration) high epsilon Data/MC ratios are low - 1st iteration: high epsilon Data/MC ratios near 1 but with significant phi-dependence similar to low epsilon - 2nd iteration: high epsilon ratio osc got stronger, ratios still centered near 1 - some parameters in L,LT,Tt functions are held fixed during iterations - *NB* Garth: with the fixed params, LT,TT have no freedom in fitting Q2=dep, suggest to freeze one of the t-dep params and unfreeze the Q2-dep param - Nacer: suggests to fix exponential param to a small value and allow the polynomial term to float in fit - *NB* Garth: also please extend the model curves to end of t-range of data, and add curve of previous model to plots, so you can more easily see how the model varies between iterations Nathan ------ PionLT Coin Blocking Studies - plot integration problems from last week fixed, calculating OUTSIDE-CUTS/FULL-RANGE ratio per setting, using CUT=30-130ns RawCoinTime range - Plots of correction vs Rate go from near 1 at low rate to ~0.95 at high rate, lots of variation between settings at same rate - SHMS Rate has the tightest correlation between settings - fitting with an exponential to get the time constant - get 53-83ns vs SHMS rate, depending on setting - Dave: this time constant is an effective dead time - implemented the correction into CoinLumi yields - *NB* something is wrong, curves vs. Current are steeper instead of shallower, need to recheck - changed RawCoinTime cut to 30-100ns from 30-130ns based on narrower distribution seen for 1 run - this gives a larger correction, up to 10%, still working on it - Garth: it's important to confirm that the correction is ~1 at very low rate, where there would be no blocking - Dave: what were the HMS+SHMS gate widths? - Dave finds https://logbooks.jlab.org/entry/3892668 which has HMS=30ns and SHMS=50ns widths - *NB* Nathan needs to look more carefully at logbook and check for other entries - Garth: the obtained time constant and gate widths should be approximately consitent with each other, as they're effectively measuring the same thing - also investigating effect of including GoodTrack and GoodStartTDC cuts in this analysis, Mark Jones had suggested using both cuts - if the GoodTrack cut is removed, the correction is up to 20% - Dave: there will be additional inefficiencies caused by these cuts - as long as they are treated separately, so that this study is uncorrelated from them, it should be okay to keep the cuts Next Meeting ------------- - Fri Sept 19 @ 11:00 Eastern/9:00 Regina - PionLT will go first ** NOTE THE NEW TIME ** - this time will be in effect until USA daylight savings time change, after which it will switch to 10:30 Eastern/9:30 Regina