Kaon_LT Experiment

                Apr 2/26 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, Nermin Sadoun, Muhammad Junaid, Nacer Hamdi,

   Alicia Postuma

Virginia - Richard Trotta

CUA - Chi Kin Tam, Tanja Horn, Sameer Jain

Ohio - Julie Roche

JLab - Dave Gaskell

FIU - Pete Markowitz

Nermin

------

PionLT LD2 analysis

- HMS calorimeter efficiencies

  - went through Junaid's slides on the efficiency analysis

  - HMS momentum used for the analysis is same as used for this setting

  - will use his value

- reading up on scientific background of the measurement and the LT-sep method

- went through Nathan's Lumi report, will follow up with him on what runs to

  use for LD2 analysis

Nacer

-----

KaonLT applying t-shifts

- had some issues w/ LT-sep after applying t-shift

  - changed t-binning, events move to higher -t

  - used this opportunity to better align t-centers for Lambda, Sigma0 bins

- waiting on new replay jobs, had some batch scripting issues

  - will apply Sameer's CoinBlocking correction after replay is done

Richard

-------

More on replay batch jobs

- there have been some SWIF2 updates

- had a meeting w/ Casey, found some ways to make replays faster

  - designate data output to node instead of cache

  - then transfer data from node to tape

  - replays run much faster this way

- updated batch submission script at:

  https://github.com/trottar/lt_analysis/blob/test/run_farm.sh 

  You'll need to follow child scripts, but RT will create a cleaner template

  version and post to redmine once all replays/apply cuts skim files are

  complete

Junaid

------

- no report, will be focusing on thesis writing, will work in parallel on

  systematic studies

Sameer

------

PionLT Q2=1.6 W=3.08 Low epsilon SHMS center (2022) data

- checked stability of Q2 distribution vs run number

  - fit a Gaussian to distribution to get mean

  - Tanja: the Q2-distribution is not actually Gaussian, you should not fit the

    full distribution, either fit just the top of the distribution, or (perhaps

    even better in this case) just use the histogram mean in the stability

    study

- had a question about 2 runs in this setting

  - FADC Mode-10 run: these are short runs where Mode-10 data were taken

    for all detectors, not just Cherenkovs, as a diagnostic check

  - HMS-3/4 run: HMS-ELREAL is typically used as the pre-trigger

    - however, since P_HMS is low for this setting, we took additional data so

      we can determine the ELREAL efficiency at low momentum

  - don't use either of these as part of physics setting

    - *NB* it's important to read up on what Mode-10 is, and the trigger

      configuration

    - *NB* Sameer will have to use the HMS-3/4 run for the ELREAL efficiency

      study as part of his physics analysis

      - this will affect the LT-separation, since it's only an issue for low

        epsilon data, but not high epsilon, for this setting

- HMS, SHMS acceptance cuts

  - showed plots and asked some questions about what cuts to use

  - we discussed this also at a Friday meeting a few weeks ago (see the notes)

  - Junaid: the cuts are based on the region of well understood magnetic

    optics.  For a good introduction to how the optics are calibrated, see

    Carlos' Thesis

    "Cross Section Measurements of Deuteron Electro-Disintegration at Very High

     Recoil Momenta and Large 4-Momentum Transfers (Q2)"

     https://misportal.jlab.org/sti/publications/16338/attachments/7617/cyero_dissertation.pdf

  - will use same acceptance cuts as Nermin and Kathleen here

Alicia

------

KaonLT u-channel studies

- had a productive meeting w/ Bill last week, who suggested:

  1) determine Aerogel Efficiency using Heep data to correct for lost protons

    due to knock-on electrons

    - Heep will have small pi/K contamination, which is good, but rates will be

      lower than data, and the efficiency may be rate-dependent

  2) constrain rho/omega ratio in MM fits

    - some variation is expected, but not by an order of magnitude

  3) should concentrate more on the MM shape study now, and fine-tune the u-bin

    borders later

  4) add a bin exclusion criterion for #Events near m_omega, in addition to

    Total #Events in u-phi bin

  5) modify u-phi plot to use -u' as radius

    - u_min is calculated from central kinematics, still gives some events with

      -u'<0, and of course ROOT cannot plot a negative radius, giving an

      unrealistic cutoff at the center of the u-phi plot

    - Garth perefers to use -u+offset as the radius to avoid this issue

- modified fitting algorithm

  - was fitting each u-phi bin independently

  - based on suggestion (2) now add an intermediate step

    - first look at u-bin integrated over phi, and determine the overall shape

      study fit parameters

    - then fit each u-phi bin separately, using the above values as the

      starting parameters in the fit

  - seems to give slightly better results than before

  - the pion leakthrough seems to have a fairly strong phi-dependence for

    certain u-bins, which is a bit surprising

  - not yet including any K leakthrough.  A clean K leakthrough sample from

    data requires pion subtraction, and wants to avoid the systematic error

    from using pion leakthrough twice

    - since the main K leakthrough is K+Lambda events, instead will use SIMC

      (normalized to data) for this channel

    - will use Tanja's sigfactorized2007 model, some issues with implementation

      that GH needs to look at

    - for lowest -u bin, no pion leakthrough used, pion sample is restricted to

      low MM only and no evidence of pi+n peak in data, distribution is mostly

      misidentified protons as pi+

    - mid -u bin has pi+n peak but doesn't extend much higher in MM

    - high -u bin extends further in MM, includes pi+Delta0 region

  - Garth: suggests to write on each u-phi MM plot which exclusion criterion

    failed if a bin is excluded.  Bill did this, and it was an extremely useful

    diagnostic to see what decision was used to exclude a bin

- fitting Pythia to MM data

  - Bill noticed a discrepancy at the high edge of MM distribution

  - Garth: this is well past the omega region, suggests to restrict MM range

    used in fit to exclude the far right edge of distribution

  - will investigate u-shift to data, to see if it affects the Pythia

    distribution shape, and improves the agreement w/ data

Next steps:

- will attempt Richard's event-by-event background weighting (last week) for

  kinematic comparison plots

- will do a new replay w/ Sameer's CoinBlocking correction

- modify Garth's t-shift code for u-shift

- investigate Aerogel eff w/ Heep data

- modify SIMC collimator routine for proton punchthrough

  - Dave: just replace pion cross section with proton interaction length from

    PDG

Chi Kin

-------

KaonLT Q2=3.0 W=3.14

- after applying t-shift, sigL becomes very small in 1st t-bin and sigT becomes

  large to compensate

  - not a lot of iterations done though

- looked at MM shift again vs t-bin, so see if this is a cause

  - only see shift (~10 MeV) in high epsilon setting, low epsilon MM shift is

    near zero

  - t-dependence of MM shift is small

- some discussion of kinematic offsets used

  - Richard: all spectrometer angle and momentum offsets are global, only the

    beam energy offsets differ by setting (since there is an uncertainty in the

    arc-energy-measurement)

- also some discussion about magnetic optics offsets

  - in hms_10p6_GeV.param, Junaid used

    hphi_offset =  2.8e-4

    htheta_offset = 0

  - some confusion on whether original 2018 ME are used, or NPS refit of 2018

    ME.  Part of the issue is that both ME use the same file name, which is BAD

    practice.  Tanja will follow-up with NPS students to have the file name

    changed

- checked HGC hole cut at high and low epsilon

  - found that x,y axes were reversed, fixed now

  - found that Vijay's hole cut only worked for low epsilon data, hole cut is

    too small for high epsilon data

    - Garth: the issue is that unfortunately the HGC optics were not properly

      aligned for the high epsilon data.  Need to draw a bigger hole cut to

      account for the enlarged inefficient region

  - this would have created extra bkd in the high epsilon data, possibly

    skewing the results

   Next Meetings

------------------

- Thurs: Apr 9 @ 16:00 Eastern/14:00 Regina

  - KaonLT will go first

- Fri: Apr 10 @ 11:00 Eastern/9:00 Regina

  - we will continue where we left off