May 1/25 PionLT/KaonLT Analysis Meeting Notes --------------------------------------------- (Notes by GH) Today: PionLT 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, Ivan Zhenchuk, Alicia Postuma, Vijay Kumar, Nacer Hamdi, Muhammad Junaid, Ali Usman JLab - Dave Gaskell CUA - Tanja Horn Virginia - Richard Trotta Ohio - Julie Roche Nathan ------ PionLT "Extra" Deadtime Investigation - question being investigated this week: Are the Coincidence Time Reference Cuts applied correctly? - changed the reference time cuts, as was suggested last week. Found that the intended cuts (see https://redmine.jlab.org/attachments/2244) were not applied correctly for some reason. They are now. - plots shown of ROC1,2 CoinTime with old and new Ref Cuts - the new plot of CTime.CoinTime_RAW_ROC2 shows a tower with an antenna sitting on a plateau that is an order of magnitude smaller than the version with incorrect cuts. The plateau is a bit smaller than before, particularly on the left, but still about 900 channels wide. - Reran Lumi analysis with the new cuts, and did not see any difference - *NB* plots only compared so far, it would be best if Nathan could directly check the numbers to make sure they are slightly different. Identical numbers to before would indicate a problem with the new analysis. - Dave: the new cuts have improved the size of the background in the wings of the distribution, the plateau has also gotten slightly narrower, it seems odd that it's still nearly 1000 channels wide - in the version with the wrong cuts sent last week, Dave applied a cut to the time in the flat plateau region and saw good pi+n MM events there, so good events were being excluded - *NB* it's VERY important for Nathan to check if the number of pi+n events in the plateau is reduced now compared to the wrong cuts - Nathan refreshes us on how the Reference Time cuts work: - if no cuts are applied, the first found time in the buffer is used - with cuts, the the first time in the good region is used; if no good time is found, then the first time outside the good region is taken - this means the total number of events is unchanged. Times migrate from "bad" to "good" regions depending on which time is taken - *NB* it would be very interesting to see if the real to randoms ratio in the "good" region has changed. One would expect it to. - the user can change the behavior of what to do if no time within the good region is found - *NB* Garth: it might be worthwhile to try excluding events with no good time and see if this alters the shape of the CTime.CoinTime_RAW_ROC2 distribution - Vijay points out that the "antenna" peak at the very top might be slightly wider than before. - *NB* possibly this is a ROOT plotting issue. Nathan will check - due to the difference in shape near the top of the CTime.CoinTime_RAW_ROC2 distribution, Nathan agrees that the yields should have changed - *NB* Nathan will carefully check things over, there is one step in the Lumi analysis where numbers have to be manually copied from a CSV file and maybe the wrong numbers were copied by mistake (this is why it's important to check the numerical values as mentioned above) - *NB* Dave suggests to look at one run in detail by hand Junaid ------ PionLT LTsep preparations - found error in Physics Yield error calculation - some t-phi bins still have weird 100% errors despite having ~200 counts - Nacer: it looks like you're using relative error instead of absolute error in the dummy subtraction - Junaid will talk with Richard about this after the meeting - based on the need to have SIMC t-phi errors substantially less than the experimental errors, SIMC has been rerun with 900k events - errors per t-phi bin now ~1% for 10k events, which is good Next steps: - working on plotting script for Data/MC comparison Vijay ----- Low Q2 PionLT analysis - working on pion absorption correction using Alicia's G4 code - to switch from Proton to Pion, only need to change an input flag - to remove NGC from the code, use NGC=0 flag, don't comment NGC out as that will cause the code to crash (ill defined parameter) Alicia ------ - working on PID and shape study for u-channel, hopefully some results to show next week Nacer ----- Low Q2 KaonLT analysis - found an optimal way to run jobs in ifarm - has a table of yields per t-phi bin, will show next week - already has a plotting script for Data/MC comparison Richard ------- High Q2 KaonLT analysis - looking at more detail at Q2=3.0, W=3.14,2.32 settings to understand why the fit LT are so big - money plots show that phi=100-250deg range has negative cross section, even though there are lots of events there (which actually indicates a large positive cross section) - this can only happen if the model cross section is negative there, clearly the model needs to be constrained better to avoid this - in the Rosenbluth equation procedure, Bill had commented out (for u-channel analysis) Step 6: fix L,T fit LT,TT, Richard had kept this commented out, but will restore it - it's not clear that will be sufficient, Richard will have to carefully keep track of what's going on here Ali --- piDelta BSA finer t-binning - slowly progressing on shape studies for high statistics settings, plans to show some results next week - has part of a code written to calculate fitting systematic uncertainty - spent a lot of time preparing slides for defense later in May Ivan ---- Started summer workterm today, will start by learning ROOT and reading up on many topics Next Meeting ------------- - Thur May 8 @ 15:30 Eastern/13:30 Regina - KaonLT will go first ** THIS TIME WILL BE USED AT LEAST UNTIL THE END OF AUGUST **