Kaon_LT Experiment

                 Jan 19/23 PionLT/KaonLT Analysis Meeting Notes

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

                           (Notes by GH and SJDK)

                    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 - Stephen Kay, Garth Huber, Ali Usman, Alicia Postuma, Nathan Heinrich, 

   Love Preet, Vijay Kumar

JLab - Dave Gaskell

Ohio - Jacob Murphy, Julie Roche

CUA - Richard Trotta, Tanja Horn

FIU - Pete Markowitz

Richard Updates

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

LT-separation code

  - making progress on different parts of code

  - cuts scripts DONE

     - showed some sample t-cuts

     - GH - Why are there -ve -t values?

        - Plot isn't finalised yet, just for testing -t cut selection

        - No selection on the Lambda peak, PID cuts not optimized

     - Diamond cut bug, a value crashes the script

        - JM - It's looking for first/last bin above some minimum, may not be

          finding it

  - calculating yields script in progres

  - combine SIMC, Dummy, Data - combine.py DONE

  - average_kin, calc_unsep Fortran Code DONE

  - LTsep, fitting (C++) IN PROGRESS

  - plotting scripts IN PROGRESS

  - SIMC script from Bill was outdated, meeting with Bill tomorrow, get correct

    version

  - Will add SIMC on top of existing plots

  - Combining Left, Right, Center SHMS settings

     - Bill and DG had a discussion, apparently the code adds data L/R/C and

       simulation L/R/C and combines to get yields

     - this is a surprise, since normally we *compute the Exp/MC ratios* first

       for L/R/C, and then *combine the ratios* in an error weighted average to

       get the combined phi coverge for a setting

     - GH: would rather do the ratios method, as was done for the pion analysis

     - DG: if the model properly describes the data, both methods should give

       the same results, but it is likely the ratios method will cause the MC

       model to converge more quickly to the data

     - TH: both methods were checked in Fpi-2, and gave same results, but K+

       analysis is new and the MC model will start further from the data

     - Should probably get ratios for L/R/C individually and get an error

       weighted average

        - Try both?

TO DO:

- finalize PID, Lumi, SIMC calcs for Heep

   - Need to discuss SIMC EM, PM calcs with DG

- Will post flowchart to Github, slides online too

Ali Updates

-----------

PID studies and HMS Calor/Cher Efficiencies

  - Need to finalise efficiences and PID prior to LT-sep, so a high priority

  - Looking at HMS Calorimeter and Cherenkov

     - Need to get a very clean e- sample

     - Use Cherenkov to get good Calo sample and vice versa

  - DG: What data is this?

  - AU - High eps, 10.6 GeV production data (Autumn 2018)

       Q2=3.0, W=2.32, Phms=6.59, S1Xrate=283kHz, 51uA, Run 4865

       Run Plan predicted e-Rate=63kHz, pi-Rate=0.5kHz

     - Coincidence data, but look at all HMS events

  - 95.10 +/ 0.01% efficiency for HMS Cherenkov

  - 95.86 +/- 0.01% effeciency for HMS calorimeter

  - DG: This is a fairly high HMS momentum setting

     - Thinks the gas pressure was set to give a 4GeV/c pion threshold (hence

       more NPE), since the e-/pi- rate was expected to be favorable

       Pi- passing the Cherenkov cut could explain low calorimeter efficiency

  - Try a setting with a reasonable ratio but *below 4 GeV on HMS*  Compare

  - DG: *p112 of Burcu's thesis* discusses the effect of a HMS Cher light leak

    in the J/Psi data

     - Even with cut at 0 NPE, Burcu saw a 3% inefficiency, which was believed

       to be due to FADC deadtime caused by the PMT always firing

     - NH: comparison of Ali's NPE plot with Burcu's: Burcu has 0PE peak that

       seems to be absent in Ali's plot

        - Don't *think* we're seeing this?  Possibly the light leak opened up

          after the fall 2018 data were taken? 

        - Dave G thinks the light leak was fixed in summer 2019

        - This implies the *spring 19 KaonLT and summer 2019 PionLT data* will

          be *affected*.  Needs investigation, particularly by Vijay.

  - AU: will also look into trimming edges of calorimeter, as discussed by

    PeterB

     - DG, TH: this should be in the reconstruction code automatically (as

       part of tracked variable fiducial cut), no need to do it twice

TO DO:

- check position dep in Calor

- look at different rates

Vijay Updates

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

- working on LT-sep code with RT and Bill, trying to understand scripts

- Need to replay PionLT data again

  - GH - Please keep notes on *issues Ali presented* as they need separate

    investigation for Summer 2019 data

Alicia Updates

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

pi+ BSA Analysis (from KaonLT data)

  - 10.6 GeV data replayed with new error calculations from AU (dummy and LH2)

  - only 10.6 GeV will give good results

     - High stats, full phi coverage and good polarisation

  - Double checked ROOT error propagation, seems to be handling them correctly,

    treated as independent random errors

  - Peter B suggested not doing a dummy subtraction and modifying the fit

  - Doing a full dummy target subtraction, separately for each helicity.  This

    way there is no need to make any assumptions on AL BSA.  There are lots of

    Dummy target statistics, so this seems the best way of handling this

     - still need to correct for Dummy target thickness, over-subtracting right

       now

     - RT: dummy_target_corr = 4.8579

   - see evidence of K+Lambda leakthrough in MM spectrum, need to adjust HGC,

     Aerogel cuts

     - HGC cut at >5 seems to let in some kaon background

     - DG: not concerned about a little K+ contribution, as it is kinematically

       unfavored to get into pi+n peak.  However, agree that it is good to

       understand the cuts and their effects

  - Fits to Asymmetry Plots

    BSA = Asin(phi) / [ 1+ Bcos(phi)+ Ccos(2phi) ]

     - second t-bin (-t=0.23-0.30) looks "overfit", particularly in comparison

       to the other t-bins, which have more sinusoidal fits

     - Tried some new fitting strategies

       1) A varies within dA from 1 parameter fit

       2) Full fit with restriction 0<B<A, 0<C<A

     - method 1 not effective in reducing "overfit"

     - method 2 surprisingly effective, ended up with B,C nearly zero

     - Some justification for enforcing A > B(and C) would be nice, theory

       paper somewhere?

     - JR: Fit with B/C fairly large doesn't really look that bad, how much does

       A actually change here?

        - If A doesn't change much, will be fine

        - Size of A/B/C and their relative errors are very important

        - experience from DVCS analysis is that the simplest (sin) fit

          *overestimates our knowledge of A* compared to true knowledge, we ended

          up listing both results in our publication

     - t-bins will be re-divided after Ali's analysis is further along, they

       are not optimized for statistics per bin yet

     - TH: actually first t-bin also has some similar features (wiggles in

       phi-distribution) as bin 2, so this might be some consistent feature of

       the data that needs to be looked at

     - JR: In future, can you *show how error bars* for the asymmetry are

       calculated? (message Julie)

  - New Plots added

     - LT' cross section plotted versus -t

     - Polar plot of the asymmetry (t/phi, asymmetry is colour scale)

TO DO:

- Need to incorporate different dummy target thickness

- optimize PID cuts and cuts for mean t,epsilon calcs

- Repeat analysis for other (Q2,W) settings

- Run Regge/GPD models for comparison to LT' data

Jacob Updates

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

HMS Optics

  - Finished all of the cuts for optics settings!

     - over 600 cuts for all optics settings set

  - Preliminary calibrations for 5.9 setting next week hopefully, as well as

    update on 5.8

Nathan Updates - Lots of plots!

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

PionLT Ref Times update

  - Following Carlos Yero' procedure

  - Start with ADC Ref Time for HMS/SHMS

  - Multiplicity == 1 cut for all plots, while CY used >2 cut as he had more

    events at ==2 than NH sees

  - HMS 4 peak structure in all COIN runs, 2 peak structures only in SINGLES

    running (ELREAL trigger)

  - DG: If trigger is ELREAL, *two legs could take timing*, EL-LO or EL-HI,

    would explain the two peaks for each set (COIN, SINGLES)

  - DG: normally there are separate timing cuts for COIN and SINGLES events

     - NH prefers to place one set of cuts, including both sets of peaks, as

       there are very few events between the two sets of peaks

    - DG: wide timing cuts may cause issues when looking at other detectors, so

      this needs to be carefully looked at

    - DG: try to *find a run* where HMS rate is high and *see if it looks* the

      same as when HMS rates are low, maybe you can get away with loose cut

  - SHMS timing peak structure

    - need to figure out which peak is COIN trigger

    - presumably COIN comes first (right peak)

    - see shifts in RAW trigger time between 2021,22

       - Brad replaced a faulty delay module in June 2022, some small timing

       - adjustments (see below)

   - NH: will need 2 versions of cuts for SINGLE arm data, depending on whether

     ELREAL or 3/4 trigger is used

       - similarly, all HMS-3/4 runs will need separate timing cuts

- Brad posted an updated trigger flowchart

  - Fall 2018-Summer 2019 Trigger Config  https://redmine.jlab.org/attachments/1731

  - Fall 2021-Summer 2022 Trigger Update  https://redmine.jlab.org/attachments/1730

  - Fall 2021 oscilloscope traces  https://logbooks.jlab.org/entry/3902530

  - Summer 2022 oscilloscope traces  https://logbooks.jlab.org/entry/4004019

TO DO:

- need to look at TDC reference times for Hodo and DC

   - probably need separate cuts for each Hodo TDC

- communicated with Cameron Cotton on NGC calib

   - now has his code, but hasn't looked at it yet

- JM: should start JM, NH, MJ regular separate meetings soon

Stephen Updates

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

- will try to do some reorganization of Hall C widi

   - will have a template for comment by next week

- also do not forget the Hall C Quarterly Analysis meeting next Thursday

Next Meeting

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

Thursday Jan 26 @ 17:00 Eastern/16:00 Regina/14:00 Pacific

- GH will not be able to attend the first ~30 minutes, hopes to join late,

  depending on Air Canada performance