Aug 14/25 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, Alicia Postuma, Ivan Zhenchuk, Nacer Hamdi
Virginia - Richard Trotta
JLab - Dave Gaskell
JMU - Ioana Niculescu
CUA - Tanja Horn, Chi Kin Tam

Richard
-------
KaonLT Q2=4.4, W=2.74 LT-separations
- improvements made to background subtraction underneath Lambda peak
  - first do pion subtraction, clear proton contamination remains (small omega
    peak), particularly for Left SHMS setting, less on Right setting, and
    negligible for Center SHMS setting
  - what remains to left of Lambda peak is fit with a Chebyshev 2nd order
    polynomial and subtracted through whole MM region
  - one improvement is that the Sigma region is also fit with a Chebyshev
    polynomial and this is subtracted too
    - the Sigma subtraction allows the MM integration range for the yield to be
      expanded, now using 1.10<Mk<1.16
  - comparing Chebyshev and flat line fits to check for over-subtraction
  - running an interation to get new Exp/MC Ratios, still not ready yet

- Nacer: have you tried doing a fit without the pion subtraction?
  - doing a pion subtraction makes the background underneath the Lambda more
    flat, otherwise the piDelta region is a broad peak underneath the Lambda
  - will consider doing fit w/o pion subtraction as a systematic check
  - a bkd fit has to be done separately for each t-phi bin, the statistics for
    some settings will be poor
  - the goal is to complete the systematics studies by mid-September

Nacer
-----
KaonLT Q2=0.5 LT-separations
- trying to improve the functional form fits for sigT,L,etc.
  - if fit all functional forms all at once, get bad Exp/MC ratios
  - one suggestion was to remove LT,TT from the fit and focus first on T,L

- Iteration 01 shows what happens with LT,TT included
  - obtain ratios 0.2-0.5, some oscillation w/phi

- Remove LT,TT and keep T,L at initial parameter values
  - ratios ~0.3, flatter, as expected since no phi-dependence in the model
  - then do Rosenbluth fit to sig_UNS obtained from these ratios
    - extract T=p1/Q2+p2/Q4 and complicated expression for
      L=p5*Q2*exp((p6-p7)*ln(Q2)*|t|)/(1+p8*Q2+p9*Q4)^2
    - in the parameter fit, fix p7,p8,p9, all small numbers <1
      - p5 is fit, and is a large number
    - ratios up to 25, but fairly flat w/phi except for highest t-bin

- Iteration 02 w/o LT,TT
  - ratios still ~25
  - ratios don't change much in Iterations 03,04 either

- Dave: the parameter fits are done only over the t-range of the central values
  of the bins, while SIMC is itself evaluated over the full acceptance of the
  diamond
  - the MC yields are going crazy because of events outside the fitting range
  - *NB* need to modify the plotting range, so we can see what the functions
    are doing over the full acceptance of the data, and not just the region
    covered by the central values of the bins
  - *NB* one option is to add IF statements to the weight recalculation script
    to keep the weight to sensible values outside the fitting range
  - *NB* another option is to turn off some terms in the sigL expression, so
    that it is simpler and less likely to be pathological outside the fitting
    range
- Garth: *NB* suggest to overlay the kinematic and focal plane distributions of
  data and MC, sizable differences are evidence of this problem

Alicia
------
Referee comments for pi+n BSA PLB
- Fig 4: the plot shows Q2-xB coverage, confusion about the t-dependence of the
  data points
  - no uncertainty shown for CLAS kinematics, mistaken impression that our
    points are less precise than CLAS
  - shows various versions of new plot, prefers the one explicitly showing the
    CLAS-12 binning ranges
  - Tanja: need to make KaonLT data more prominent on the plot
    - Garth will try making violet points smaller and play with the binning
      range shading

- Fig 6: models not evaluated at same kinematics as data, differences in t_min
  cause confusion
  - referees would also like to see a greater range of CLAS data to be shown
  - rerunning models exactly at experimental kinematics
    - PARTONS jobs take a lot of time
    - now have CKY model executable, so we can do this too
    - CKY has also reverse-engineered KM and VR models, doing checks and have
      promised to send us the codes
  - new plots extend to t-min properly, conclusions unchanged

  - added more CLAS data based on 2 criteria:
    - try to find CLAS settings where there is significant overlap in binning
      range with KaonLT data
    - also need close central kinematics to ours
    - we originally had fairly tight criteria, in response to referees now have
      broader criteria

 - to continue models to higher -t, past KaonLT data, using the highest -t
   KaonLT kinematics since the t-dependence is flatter there

- Fig 7: Q2-dependence plot
  - referees want GK predictions and more CLAS points
  - added 2 more CLAS points: Q2~4.0, widened t-range to 0.4+/-0.08 from 0.06

- Tables: referee would like a sig_LT' table added, for both our data and CLAS
  used in the study

- Referees other comments
  - we have no mention of Radiative Corrections
    - Dave: the best way to study this would be to add the observed asymmetry
      to the SIMC pi+ model, and see if the predicted asymmetry changes when
      run RAD on/off
    - Garth: suggest a simple change to the model, assume sigLT'/sig0=0.1
      everywhere except within 0.1 of t_min

  - comment on whether quark-hadron duality might explain why both Regge and
    GPD models describe some aspects of the data
    - Tanja: we're not trying to validate a specific model, keep comments to
      our empirical study assisted with models, quark-hadron duality is outside
      the scope of the paper

  - comment about higher twist effects
    - there is no GPD calculation yet that includes higher twist effects, we
      encourage the development of models incorporating higher twist

Ivan
----
- working on Co-Op report, putting together plots
- last meeting will be next week

Nathan's slides (shown by GH)
-----------------------------
PionLT Coin RefTimes update
- has a set of cuts that improve performance
 - cuts pushed to GitHub and tested for production data
 - CoinTime offsets had to be redone (thanks Junaid)

- Before: t_coin_trigNames="pTRIG1_ROC1 pTRIG4_ROC1 pTRIG1_ROC2 pTRIG4_ROC2"
  pTRIG1_ROC1: 5100 – 6200
  pTRIG4_ROC1: 5600 – 6200
  PT2: 4700 – 6000
  pTRIG1_ROC2: 5800 – 7000
  pTRIG4_ROC2: 6300 – 7000

- After: t_coin_trigNames="pTRIG1_ROC1 pTRIG3_ROC1 pTRIG1_ROC2 pTRIG3_ROC2"
  hT2: 2400 - 4600
  hTRIG3_ROC1: 3400 – 4600
  pTRIG1_ROC1: 4600 – 6200
  pTRIG1_ROC2: 5400 – 7000
  pTRIG3_ROC2: 4200 – 6500

- important to check replay scripts do not contain
  gHcParms->Load("PARAM/TRIG/tcoin.param");

- plots of HMS xfp vs CoinTime and SHMS xfp vs CoinTime look good
- significant improvement in CTime.CoinTime_RAW_ROC1,2
  - planning to use ROC2 since it's much cleaner

- working on a more formal write-up for HallC docDB and will present results at
  Quarterly Analysis Meeting
  - would like to better understand why the new cuts work better

  - Dave: Nathan was previously cutting on the wrong variable, now that he's
    using the correct variable things work better now

Next Meeting
-------------
- Thur Aug 21 @ 15:30 Eastern/13:30 Regina
  - PionLT will go first

_ Note: we will need to do a poll soon, to pick a time for September