May 8/25 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, Ali Usman, Alicia Postuma, Ivan Zhenchuk,
   Muhammad Junaid, Nathan Heinrich, Nacer Hamdi
JLab - Dave Gaskell
CUA - Tanja Horn
Virginia - Richard Trotta
Ohio - Julie Roche
FIU - Pete Markowitz

Richard
-------
KaonLT LT-separations Q2=4.4, W=2.74
- fixing weird dip in unseparated cross sections near phi=180deg discussed last
  week
  - the problem was a ROOT issue, a cut was applied to eliminate from the
    Rosenbluth fit those phi bins with negative cros section, however, this cut
    seemed to not be used
  - needed to put this cut directly into the Fortran code used for the
    Rosenbluth fit
  - after a few iterations, the unseparated cross sections are now positive for
    all phi, and these bins are no longer excluded from the fit
  - L,T,LT,TT functional forms unchanged from before
  - added step 6 (discussed last week) in the Rosenbluth fitting algorithm,
    i.e. fit LT,TT while keeping L,T fixed
- new separated results look much more sensible
  - LT: now ~5, previously ~20
  - TT: now ~8, previously 10-35
  - L:  now 8-0, before ~18
  - T:  now 6-10, before 4 to -3
  - now T>L for all t-bins, which is more in line with theoretical expectations
  - TT has decreasing trend to smallest -t bin
  - LT is flatter, no clear decreasing trend with smaller -t

Future steps:
- given these large changes, will need to eventually demonstrate that both
  sig_uns and the separated cross sections are stable
- Data|MC comparison plots can be shown next week
- will present a detailed systematic errors plan next week

Ali
---
piDelta BSA finer t-binning
- shape study for lowest -t bins completed
- ran into a problem with higher -t bins, the issue is that the pi+n process
  doesn't populate some bins, which caused fits to fail
  - in the process of removing pi+n entirely from high -t fit, taking longer
    than expected
- plans to show final shape study, and possibly some early BSA results, next
  week

Alicia
------
u-channel Q2=3.0, W=3.14 missing mass studies
- SHMS center setting, covering all of -pi<phi<+pi
  - peak near MM~0 is fit better with DVCS MC than pi0 MC
    - somewhat surprising that DVCS would be so prominent in comparison to pi0,
      Alicia plans to also look at MM^2 as that would better indicate if it's
      really DVCS or not
    - in the end, it really doesn't matter what process it is, since we're not
      likely to extract results for the MM~0 peak, just as long as it's fit
      well to describe non-omega background in the histogram
  - high MM region: still need to figure out Pythia generator for this region
  - MM plots shown for 8 phi bins, did MC fits to each
    - eta' peak is surprisingly large, but there is no proper background
      underneath yet, so likely it's overestimated
  - some discussion about pi+|K+ leakthrough
    - applying HGC hole cut to reduce leakage, no K+|pi+ subtraction yet

- SHMS left setting
  - MM distribution cuts off earlier than for center setting, around 1.35 GeV
  - MM plots for 8 phi bins: 4 filled bins near phi~+/-pi, bins near phi~0
    mostly empty (as expected)
  - clear omega peaks are seen, also a more prominent K+Lambda peak than for
  - center setting, nearly as big as omega

- SHMS right setting
  - MM distribution extends up to ~1.5 GeV
  - MM plots for 8 phi bins: 4 filled bins near phi~0, so left+right settings
    give full phi coverage
  - some evidence of pi+n leathrough, smaller than K+Lambda, but these events
    are on shoulder of omega peak, so we will have to treat them carefully
  - all plots are for full t(u) range

Next steps:
- will present a plan on how the analysis should proceed next week
  - next step will be the Pythia generator for high MM region
    - if it down't work, Plan B would be to work with Hem on modifying the
      SIDIS generator for u-channel protons in SIMC
  - then will look at K+|pi+ leakthrough subtraction
  - too soon to know if any u-binning is possible, for this setting Ali has 3
    t-bins, omega is likely less statistics than piDelta

Nacer
-----
Q2=0.5 KaonLT analysis
- Data|MC comparison plots shown
  - P.dc.xp_fp shows some difference in width between Data and MC
  - a weird dip is seen near center of H.gtr.th
    - some discussion whether it's statistically significant or not
    - Dave suspects an instrumental issue, the dip is close to where the two
      mirrors overlap in the HMS Cherenkov, however Nacer is applying only a
      HCer>2npe cut, so that should't really cause this

- MM histo shows the Lambda peak at 1.12 GeV instead of expected 1.16
  - what is weird is that the Sigma peak is not similarly shifted, but appears
    at the expected 1.19 GeV
  - however, SIMC predicts correctly the observed spacing between these two
    peaks
    - a fixed offset of 3.68 MeV is applied to all MM data (both Lambda and
      Sigma) to get them to line up with the MCs

- working on yield scripts per t-phi bin, would like to have some preliminary
  cross sections to show at the CAP Congress in a month

Junaid
------
Q2=3.85, W=2.62 PionLT analysis
- showed t-binning for this setting last week
- found a small bug in error calculation script
  - only bins with ~100% errors are those with 1-2 events (SHMS left or right)
    so the errors appear to be sensible now
  - ran SIMC more so that SIMC yield errors are much smaller than data, about
    1%

- calculated Data/MC ratios, where MC uses default Bosted pion model
  - even with no iterations, the ratios already 0.9-1.0(!)

- Data|MC comparisons
  - generally the comparisons are very good, except for:
    - HMS-delta has a significant dip at +6% in low epsilon data
    - Dave: this looks like a mis-calibration of some segmented detector, such
      as calorimeter or hodoscope
      - suggests to look at calorimeter E/p vs delta, as well as beta vs delta
    - Dave: do these data use Nathan's corrected Reference Times?
      - while looking in more detail at Nathan's Lumi scan ROOT trees earlier
        today, noticed a block shift at H.gtr.dp~6 when using the updated time
        window cuts
      - the same Run 14616 with the old cuts have no problem
      - Nathan: it appears that somehow a bad calibration for the hodoscope
        bars slipped in when changing the reference cuts, will investigate

Next steps:
- working on LTsep iteration framework
  - GH giving a tutorial on this topic next Monday morning

Nathan
------
PionLT Coin Lumi deadtime problem
- last week, we discussed the effect of changing the reference time cuts, and
  that the yields mysteriously did not change, although it was clear from the
  CoinTime plots that they should have
- spent last few days looking at this
  - found that #Events passing cuts makes no sense, seems way too low
  - tried removing PID cuts and #Events went down instead of up!
  - clearly something is massively wrong in the script

Next steps:
- will briefly consult with Junaid and Richard on the analysis script
  - Nathan will give script debug info to Richard for him to look at
- if can't quickly find the problem, then will write a new script from scratch
- will also look at the calibration issue identified above by Dave

Ivan
----
- reading up on detectors, things starting to make a bit more sense now
Next Meeting
-------------
- Thur May 15 @ 15:30 Eastern/13:30 Regina
  - KaonLT will go first
  ** THIS TIME WILL BE USED AT LEAST UNTIL THE END OF AUGUST **