Sep 26/25 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, Nermin Sadoun, Nathan Heinrich, Alicia Postuma,
   Nacer Hamdi, Muhammad Junaid, Jing Sin Hu
JLab - Dave Gaskell
York - Stephen Kay
CUA - BC Manoj, Sameer Jain, Tanja Horn, Chi Kin Tam
FIU - Pete Markowitz
MSU - Wenliang Li
Virginia - Richard Trotta
JMU - Ioana Niculescu
CSULA - Konrad Aniol

Alicia
------
- please have your comments on the BSA paper and response to referees by
  mid-next week, if possible

u-channel initial results
- improved MC fitting algorithm
  - using Pythia, rho, omega, eta, eta', pi0, as well as pi+, K+ leakthrough
Q2=3, W=3.14, SHMS center
- no t-phi binning, but a cut to exclude high -u where there are few omega
  counts
  - after subtracting backgrounds, the omega SIMC and Data have fairly good
    agreement over the 0.7<MM<1 GeV^2 region
- divide into 8-phi bins, fits for each work reasonably well
  - BSA for all phi-bins except one are non-zero, A=-0.2 +/- 0.2, no clear
    sin(phi) shape
  - Bkd Asymmetry is essentially zero
  - this is one of the higher statistics settings, BSA analysis does not look
    very promising

- Looked briefly at omega/eta, omega/eta' ratios
  - eta, eta' have different u-coverages, which makes forming ratios at common
    u difficult
  - maybe could do omega/eta, omega/eta' vs Q2, each one integrated over full
    u-range, so the integrated region would not be the same for different
    mesons
  - the uncertainties for this are likely high, not the best use of Alicia's
    time, but it would make a good summer student project

- Looking at options for L/T-separation
  - compared statistics w/Bill's thesis.  Looks to be a bit higher after
    diamond cut is added
    - lowest stats Q2=5.5, W=3.02 setting would be comparable to Bill's worst
      stats

To Do:
- run all SIMC models for all settings
- replay low epsilon data, push code to GitHub

Bill: the Pythia fit is really nice, it looks like you would be able to extract
  u-channel rho cross sections for the first time, since the MM distribution is
  clearly different from Pythia
  - he is also interested in seeing if this will allow rho to be extracted from
    his thesis data, since people have been asking about it

Nacer
-----
KaonLT Q2=0.5 LT-sep Iteration=3
- keeping LT,TT=0 in model but not in Rosenbluth fit
- both L,T use the same functions, but the exp(-p*(t-tmin)) parts have
  significantly different values
  - plots of sigL,T vs t show that sigT_model doesn't have a good fit compared
    to sigT_exp
  - sigL function goes crazy at low -t<0.08
- data/MC ratio ~1 for all t-bins except lowest one
  - some phi-oscillation for 3 highest t-bins, a bit flatter at low epsilon
    which makes sense due to the epsilon-dependence

- trying to keep sigL exponent from exploding at low -t, setting p5 in exp term
  - ad-hoc set a plateau for -t=0.08, for -t<0.08 sigL is flat at the -t=0.08
    value
    - then Data/MC=1 for lowest t-bin, but ratios in other bins are a bit
      higher
  - after looking closely at how the ratios behave, decide to fix p5=100 in the
    sigL model
    - sigL still rises, but only to ~8 at low -t, instead of ~80
    - get Data/MC ratios ~1 for all bins, some phi-oscillations

- asks whether it is better for him to next try a polynomial for sigL,T, or is
  it better instead to accept fixing p5 and move to LT,TT?
  - Dave: the behavior of sigL,T at low -t is suspicious, is there a slight
    shift in between the t-distributions at high,low epsilon that could cause
    this?
    - *NB* suggests to look at MM peak compared to SIMC, are these MM offsets
      much different between high, low epsilon?
    - it would also be interesting to see if there's a t-dependence to the
      offset?
   - *NB* Nacer will also recheck the pion-subtraction at low -t, as earlier
     suggested by Richard
   - Nathan: mentions the bug in Heepcheck that Gabi mentioned at the quarterly
     analysis meeting
     - *NB* Garth needs to investigate

Richard
-------
Working on Q2=4.4, 5.5, will present results next week

Junaid
------
PionLT Q2=3.85 changes to functional form
- LT,TT both have a p1/(p2+|t|^2) form
  - LT goes crazy at low -t, while TT seems better behaved
- try LT with p1/(|t|^p2) form, still sees big phi-oscillations in Exp/MC
  ratios, need to try something different for LT, happy with TT form
  - *NB* Richard: suggests to remove 1/Q^2 term in LT,TT
  - look at Fpi-2 parameteriation, LT function was complicated and TT simpler
    - *NB* Garth suggests to try TT form for LT
    - Junaid can also try Bosted LT form

Nathan
------
PionLT Coin Lumi update after Quarterly analysis meeting
- iFarm is back up, will have updated plots by end of day
- *NB* need to double check Tracked Boiling Study, the yield calc had a bug,
  was fitting instead of integrating the plot, hoping this will result in
  better Tracked Boiling results
- regarding the CoinTime Blocking Corr, at the Quarterly meeting Mark had
  suggested to place tighter timing cuts on individual hodoscope planes
  - *NB* planning to look at this, might need to ask Mark for a suggestion on
    how best to do this
    - the hope is that this would result in less scatter in the Normalized
      Yields used in the Coin Lumi vs Rate, used to determine the AdHoc
      correction
    - if the scatter is less, so will be the systematic uncertainty, which
      would be quite valuable since the AdHoc correction is at the moment one
      of our larger systematic uncertainties
    
Chi Kin
-------
KaonLT Q2=3.0 W=3.14 LT-sep
- finished iteration scripts
  - tried doing own t-phi binning as an exercise
    - Garth: it's important that the phi bins be of equal size
    - *NB* Nacer: plot high and low epsilon t-dependences on top of each other,
      so you can try to optimize the statistics between them, both epsilons
      require the same t-binning
- finding that the Data/MC ratios are diverging after more iterations
  - some suggestions on looking at all of the intermediate plots to better
    diagnose what is going on
    - both Richard and Nacer are available for consultations

Next Meeting
------------
- Fri Oct 3 @ 11:00 Eastern/9:00 Regina
  - PionLT will go first

- as the meetings are getting quite long, and we don't want to miss important
  issues in discussions, we should discuss various options, such as maybe 2
  shorter meetings per week