Oct 10/24 PionLT/KaonLT Analysis Meeting Notes ---------------------------------------------- (Notes by GH) Today: KaonLT will be discussed first Present ------- Regina - Garth Huber, Alicia Postuma, Nathan Heinrich, Muhammad Junaid, Vijay Kumar, Nacer Hamdi, Argha Das CUA - Casey Morean, Tanja Horn JLab - Dave Gaskell Virginia - Richard Trotta Ohio - Julie Roche FIU - Pete Markowitz Vijay ----- Finalized PhaseSpace Volume Cut Systematic Study - varied delta_HMS cut by -10%, as discussed last week - calculate R=Y_DATA/Y_MC for each (t,phi) bin, then do error-weighted average of R of 16 phi-bins, for each of 7 t-bins - R changes 1-5% from low to high -t bin - quadrature sum of delta(-), xptar(+-), yptar(+-) gives 1.2% to 6.5% from low to high -t bin - Tanja: have you done any checks yet on whether the new replay is the same as this or different? It would be good to not waste time by having to repeat the study if the new replay is different? - files directly from hcana replay were lost, files after python analysis were correctly archived - replay of Q2=0.42 should happen next week, first have to modify replay scripts for ALMA9 then submit - will replay Q2=0.38 at same time, still have to check that results are unchanged - Garth: it will be important to archive the code used to generate the results, including the replay scripts Alicia ------ Response to Referees on BSA PRL draft - key authors promised to give feedback on the response letter, as well as the manuscript, no later than Monday Nacer ----- KaonLT Heep singles analysis - ran singles replay after lots of help from Junaid - still debugging results Nathan ------ PionLT Luminosity scans (COIN SIDIS and exclusive pi+) - initial results shown, but still have to refine cuts and implement random subtraction - remove goodscinhit==0 cut - remove hgc>1.5npe cut, as protons giving knock-on electron Cherenkov signal will give a rate dependence - remove beta-cut, as we discussed a few weeks ago - issues seen with CPULT - Exclusive: number OK but errors weird - SIDIS: CPULT up to 300%, so wrong variable somehow used in calculation - SIDIS#1 COIN yields, LH2 12200-12216 - anti-boiling LH2 seen for scaler, no-track and track analyses, but should look better after randoms subtracted - SIDIS#2, LH2 anti-boiling is less - Excl#1,3 LH2 14601-14606, 15816-15846 - no-track is flat, will go down after random subtraction - track has 20% boiling, which is larger than SIDIS (~10%) - Excl#4,5 16220-16242 - similar to Excl#1,3 Junaid ------ Hodo-3/4 efficiencies - Mark Jones made significant changes to hcana code which seem to have fixed the issue, now getting ~99% efficiency for both Heep and Physics data - recompiled hcana for both PionLT and KaonLT, so everyone has access to the changes now - still investigating exactly what Mark changed, but as a summary: - modified module for event selection used to calc Hodo-3/4 Eff - need to add parameters to "phodo_cut.param" file - essentially, events w/HMS singles need to be removed when calculating SHMS efficiencies - added module to check missing track positions, fTrack_Miss_Xpos, Ypos - added goodstarttime cut - with the new efficiencies, Heep ratios are 0.95-1.013, excluding one outlier setting with R=0.85 - Richard's boiling correction is applied, awaiting result of Nathan's analysis New HMS matrix elements - Christine will release 5.6 GeV/C ME in ~1 week - 5.9 GeV/c is half done so far, should also not be long - 4.9 GeV/c is planned to be analyzed, apparently some evidence of saturation already there - 6.8 GeV/c, unsure of ME fitting status PionLT pi+ PID studies in progress - looking at 3 settings, one of which has Theta_SHMS=5.50deg - iterating over RF, CoinTime, Aerogel, HGC cuts to understand particle distributions and efficiencies Richard ------- KaonLT L/T-separations - varying exponent "n" in 1/(W^2-mp^2)^n factor in cross section parameterizations - varies with increasing W, from n=3.40 at W=2.32 to n=2.25 at W=3.14 - also trying two different TT-parameterizations sigTT=(p13*Q^2 * exp(-Q^2))*ft*sin^2(theta) sigTT=(-p13*|t|+p14)*|t|^(Q^2/p15)-p16*Q^2 - more complicated expression generally does better, but there are few t-bins so it is difficult to constrain so many parameters - Garth: suggests to figure out which parts of the complicated expression are more important, and simplify the expression to just that part - also can play the strategy of fixing some parameters and releasing others in the fit, and then fixing them and releasing the others. The downside is that it takes many more iterations to converge, but it helps avoid the issue of too many degrees of freedom in the fit - Ratios after 5 iterations are generally near 1 now - Q2=4.4: Ratio drops linearly with -t - sig_UNS is consistently higher at high epsilon than at low epsilon now, which is very good news - will arrange a meeting with Garth and Tanja next week to discuss the ratios in more detail Next Meeting ------------- - Thur Oct 17 @ 16:30 Eastern/14:30 Regina - PionLT will go first