Dec 18-19/25 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, Nermin Sadoun, Muhammad Junaid CUA - Sameer Jain, Chi Kin Tam, Tanja Horn Virginia - Richard Trotta Ohio - Julie Roche FIU - Pete Markowitz JLab - Dave Gaskell Richard ------- KaonLT Q2=3.0 W=2.32 LT-sep - using same paramterization and parameters as Q2=4.4 so far, no iterations yet - looking at outliers in Exp/MC Ratios bin-by-bin, either high or low - low epsilon, -t=0.46, phi=30deg - some possible undersubtraction of background underneath Lambda region - low epsilon, -t=0.546, phi=150deg - oversubtraction of background, also some issues with propagation of uncertainties when no backgrounds remain after subtraction - other bins shown: similar issues - new Exp/MC ratios (no parameter changes) after fixes - still a few outliers remain, but others now more in line, some with substantially larger (but correct) error bars - higher -t bins show a hump in the Exp/MC ratio near phi=0, which would be an LT issue, lower -t bins are flatter Chi Kin ------- KaonLT Q2=3.0 W=3.14 LT-sep - looking at background subtractions as well - background subtraction was not properly applied for some low epsilon bins - changed to: SIMC(Lambda)+exp(Cheybshev)*(x-xmin)*(xmax-x) - this results in an oversubtraction - one issue is due to SIMC underprediction of Lambda peak width - Dave: this could be caused by SHMS saturation effect - setting has: P_HMS=4.2 GeV/c, P_SHMS=6.8 GeV/c - *NB* suggests to plot MM vs delta_SHMS - *NB* also suggests to double-check that low momentum HMS matrix elements are being used, rather than high momentum ones - DG and Tennessee students recently stumbled upon an unforeseen SHMS optics effect - W resolution was surprisingly bad - discovered a step function in W-reconstruction, different values for yptar<0 and yptar>0 like a step function - possibly a HB effect not properly taken into account in the optics reconstruction? - *NB* suggests to plot MM vs yptar_SHMS - investigated changes to the resmult factor that Ali had tuned for his data - Lambda peak width consistently wider at high epsilon than low epsilon - *NB* Dave: before trying to tune the resmult factor to your data, it's better to look first at correlations in the data and try to understand the optics first - whether things change with setting, depend on the origin of the effect e.g. in the 6 GeV era, there was an observed rate-dependence to the tracking resolution that traced as due to bad hits in the DC at high rate - the SIMC Lambda resolution mismatch causes some background oversubtraction (i.e. the SIMC Lambda is fit and subtracted from the data to yield a flatter distribution that is then fit for the background subtraction, if this peak subtraction has problems, the background that is fit is too large) - temporary solution: add a Gaussian smearing function to SIMC when doing the background fit - hasn't checked Exp/MC ratios yet - the uncertainty in the background subtraction is calculated as: - fit the polynomial bkd and normalized SIMC to data - use MINUIT params: vol, cov sampled from multivariable normal dist - discard regions where the uncertainty explodes - Richard: we eventually will want a discussion on the best way to do the background subtraction, since RT and CKT have different methods - Garth: if they're both reasonably valid, we should take the difference between them as a systematic - a full replay of KaonLT data will be done over the break, including new HMS 0th order matrix offsets that Nacer identified, and CoinBlocking correction Sameer ------ KaonLT CoinBlocking correction for 10.6 GeV data - shows plot of correction vs CoinRate - 3 plateaus observed, different correction values for <1kHz, 1.0-2.5kHz, >2.5kHz - obtained corrections of ~0.94 for >2.5kHz CoinRate, SHMS-Left setting - Dave: are these rates corrected for the beam off periods? - SJ checks the code, and indeed the rate is not corrected for this - *NB* need to apply a beam threshold when incrementing the clock, to get a more accurate rate - lumi studies typically apply a higher threshold than physics analysis, in order to get a higher resolution rate number - overall, the correction values themselves look reasonable, but we need a more accurate rate dependence in order to compare to Nathan's results at similar rate - shows some CoinTimeRaw plots and applied cuts, apparently there are some differences setting-by-setting - Nathan: provided the electronic setup is unchanged, find a set of cuts that work with your widest set of data and use that for all runs in that setting Friday ------ Regina - Garth Huber, Nathan Heinrich, Alicia Postuma, Vijay Kumar York - Stephen Kay CUA - Sameer Jain, Tanja Horn, Chi Kin Tam Virginia - Richard Trotta JMU - Gabriel Niculesu, Ioana Niculescu Alicia ------ KaonLT u-channel Q2=3.0 W=2.32 Pythia model - suppressing rho and rho-diffractive processes did not fix the MM double bump issue, can make the bump bigger, but not go away - turning off charged rho production makes the bump worse, as it suppresses featureless background from pi+pi0 production - is there someone more familiar with Pythia that she should contact? - *NB* Gabriel suggests Harut Avakian from Hall B, either he will know, or can put in contact with a grad student who is familiar - other options if this doesn't work - could try running SIMC rho MC and subtract it from Pythia when doing background fit - could try fitting a higher W-setting Pythia to these data - the model cross section will have the wrong angular distribution, but since the background is fit separately for each t-phi bin, the normalization error would not matter Vijay ----- Low Q2 PionLT LT-sep - started working on systematic uncertainty studies - PID cut is determined with CoinTime and RFTime, not HGC Aerogel - CoinTime cut systematics: will vary cut width by +/-0.1ns (10% of cut width) and evaluate the difference in sig_uns at low, med, high epsilon - draft PRL manuscript - made progress on experiment section Nathan ------ PionLT Q2=3.85 W=2.62 LT-sep - comparing his results with Junaid's - separate data replay but what should be the same scripts - found that the Exp/MC ratio is a bit different Nathan: 0.973 +/- 0.060 Junaid: 0.952 +/- 0.059 - apparently the SIMC iterations are not the same, which is easily fixed - also finds small differences in the MM peak distribution - Gabriel: this is a dual-purpose exercise 1) cross-check of steps in analysis 2) if alternate analses are equivalent, we have to treat it as a systematic - Nathan: would like to know what the source of the difference is first before deciding how to handle it - strongly suspects cut difference in LT-sep Python package - expects to have a conclusion at the next meeting in January - TrackingEff systematic studies - Dave had suggested to plot Normalized Yield vs S1X rate instead of 3/4 rate - indeed finds the spread of points to be smaller against S1X, so Dave is right, S1X is the better rate to use - will make the Mean of Deviations from the trend as a scale systematic - and the Standard Deviation as a random systematic Gabriel ------- KaonLT Sigma0 Lambda* analysis - follow-up about using HallC:p as the beam energy - found the deviation in HallC:p at the time of the Arc Energy Measurement - about 1.5 MeV at 10.6 GeV, more for lower beam energies - has incorporated this - has a question about the beam energy spread - SIMC has a flat beam energy spread about the peak energy, unchanged from 6 GeV era - CEBAF Accelerator paper says the Hall C energy spread is 5E-4 RMS at 10.6 GeV - *NB* should switch to a Gaussian beam energy distribution, with width corresponding to the RMS values from the Accelerator paper - can have a flag in the input file to use either the current or new beam energy spead code - Gabriel will follow-up Next Meetings ------------------ - we will continue with the current meeting schedule for now, can adjust once people better know their 2026 timetables - Thurs: Jan 8 @ 16:00 Eastern/15:00 Regina - PionLT will go first - Fri: Jan 9 @ 11:00 Eastern/10:00 Regina - we will continue where we left off