Kaon_LT Experiment

               Apr 30-May 1/26 PionLT/KaonLT Analysis Meeting Notes

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

                                (Notes by GH)

                     Today: PionLT will be discussed first

Thursday: Present

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

Regina - Garth Huber, Alicia Postuma, Nermin Sadoun, Nacer Hamdi,

   Nathan Heinrich

CUA - Chi Kin Tam, Sameer Jain

Virginia - Richard Trotta

FIU - Pete Markowitz

Ohio - Julie Roche

Poll for Meetings in Summer

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

We agree upon the following:

  Thursdays 1-2pm Eastern

  Fridays 12-1pm Eastern

  - GH will send Zoom invitations, the Zoom number will change

Nathan

------

PionLT Q2=5.0 W=2.95 low-eps Center-SHMS

- MM histo shows evidence of both K+ and proton leakthrough

  - K+ leakthrough is far from pi+n peak, not a problem since it can be removed

    with MM cut

  - small proton leakthrough is underneath pi+n peak, some additional studies

    should be done to reduce this

- comparison of pi+n peak w/ SIMC shows evidence of an MM offset

  - SIMC and data diamonds also show a small shift between data and SIMC

  - *NB* maybe the htheta offset was applied with wrong sign?  Nathan will

    check

- an HMS momentum offset is applied at 10 GeV, but not for low and mid epsilon

 data

   - GH thinks Junaid says this has to be removed

   - *NB* Nathan will check with Junaid on this

Nacer

-----

KaonLT Low Q2 LT-sep after t-shift applied

- all iFarm jobs finally done

- adjusted t-bins a bit, t-shift moved events higher in -t, so lower limit of

  1st t-bin now a bit too low

Lambda:

- new t-binning:

  - changed from 9 to 8 t-bins

  - t-range:  new=0.074-0.128  old=0.070-0.128

  - now have ~10k events for low t-bins, before the lowest bins had ~5k

- changed functional forms:

  L=p0*|t|*exp(-p1*|t|)/(|t|_mK**2)**2

  T=p2*exp(-|p3*t|)

  LT=p4*exp(-p5*|t|)*sin(theta)

  TT=p6*exp(-p8*|t|)*sin^2(theta)

- after these changes, the Rosenbluth fits look reasonable

  - L: end up with monotonic increase, no bins <0

    T: monotonic decrease

  - Data/MC Ratios: low epsilon ratios look very nice for bins 1-6

    high epsilon ratios look even better for bins 1-8

  - did 20 iterations

    - after the first few iterations the ratios became a bit worse, initially

      the parameters were in a false chi-square minimum

    - then had to do some parameter push by hand, after that the ratios

      improved after more iterations as the fitting found a better minimum

  - Data vs SIMC kinematic histogram comparisons

    - SHMS_xpfp has a mismatch on one side, otherwise looks good

    - *NB* Richard and Chi Kin will also check their distrubtions

  - low-eps Center-SHMS has a left-right asymmetry (2 sigma?) in phi

    distribution that is not shown by SIMC

Sigma0:

  - new t-binning:

    - stay with 5 t-bins

    - t-range:  new=0.194-0.156  old=0.100-0.156

    - first 3 Sigma0 t-bins have same limits as last 3 Lambda t-bins

  - using same new functional forms as for Lambda

  - Rosenbluth fits look good, given the poorer statistics

  - L: consistent w/ zero

    T: flat dependence ~0.2 ub/GeV^2

  - Data/MC Ratios: low epsilon has larger fluctuations than Lambda, but

      reasonably flat

    high epsilon ratios look better, but highest t-bin has left-right phi

      asymmetry

  - Data vs SIMC kinematic histogram comparisons

    - SHMS_xpfp Data is wider than SIMC

      SHMS_ypfp is similar, Data wider than SIMC

Next steps:

- include CoinTime blocking factors

Sameer

------

working on replaying data for Junaid's LT-sep framework

- busy with 2 final exams next week

Chi Kin

-------

KaonLT high Q2 LT-sep

- some discussion on determining new Heep offsets given that the optics ME have

  changed

  - Garth is happy to do HEEPcheck analysis on result of new replay

  - *NB* Chi Kin asks Richard to replay the Heep data for him

- some discussion on how offsets should be applied in recon_hcana for SIMC data

  - what is the most consistent way to apply offsets in SIMC?

  - suggests a modification to the code

  - Richard: doesn't think the results will change

  - Nacer: suggest to try the code both ways and see if there's a difference

Friday: Present

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

Regina - Garth Huber, Nathan Heinrich, Alicia Postuma, Nacer Hamdi,

   Vijay Kumar, Nermin Sadoun

Virginia - Richard Trotta

CSULA - Konrad Aniol

JMU - Gabriel Niculescu

CUA - Chi Kin Tam

Glasgow - Kathleen Ramage, Rachel Montgomery

FIU - Pete Markowitz

Kathleen

--------

PionLT LD2 efficiency study for Q2=1.60 W=3.08 6.40GeV data

- HMS tracking efficiencies ~99.6%, SHMS trackeff ~98%

- HMS Cherenkov and calorimeter efficiencies

  - the efficiencies are low, due to pi- contamination

  - Gabriel: suggests to look at recent Hall C papers and theses on inclusive

    electron scattering, which should have a lot of info on electron PID

  - Garth: suggests to contact Junaid for his report on Calorimeter & Cherenkov

    efficiencies

Nermin

------

PionLT LD2 Luminosity study

- has mostly gotten Nathan's scripts working

- looked at 2 sequences of runs:

  6.395 GeV runs 16716-26

  9.2 GeV runs 12158-66

6.395 GeV:

- discussion on Live Time vs Current plots

  - the 6.395 GeV singles data was taken with 1 spectrometer at a time, the

    EDTM livetime is reliable for this setting

  - the 9.2 GeV data were taken with prescaled singles for both spectrometers

    together, for these the EDTM is NOT reliable, the CPULT needs to be

    calculated from the hodoscope rates and used instead

  - it seems the "simple" CPULT calculation was used by mistake

  - *NB* need to switch to Nathan's combinatoric CPULU calculation using

    individual hodoscope plane rates

    - Nathan: you will find the equation in scaler.py

- Relative Yield Scaler, NoTrack, Track vs Rate

  - there is a small non-linearity at >150 kHz that the correct CPULT

    calculation will hopefully fix

    - even the scaler yield is affected, since the ELLT extraction is affected

      by the CPULT calculation

9.2 GeV:

- getting significant anti-boiling, which indicates the used efficiencies are

  too low

Next steps:

- will also look at Carbon Lumi runs, to check that reproduces Nathan's earlier

  result

Richard asks a question for Nathan about why EDTM doesn't work for dual

spectrometer singles data

- his idea is that for Coin Data the RefTime is taken from the SHMS

  - when Coin DAQ is used, what RefTime is used for HMS singles events?

  - maybe sometimes the wrong RefTime is used for HMS singles events?

- this seems a sensible explanation

  - *NB* we should follow up with Hanjie and Alexandre about this when at JLab

    this summer

Vijay

-----

PionLT Low Q2 LT-sep

- working on systematic uncertainties

- implementing GH's latest comments on paper draft

Richard

-------

KaonLT Q2=4.4 W=2.74 LT-sep

- replayed all settings w/ all corrections applied

  - Chi Kin's HGC hole cuts

  - CoinTime blocking

  - t-shifts applied

  - Diamond cuts adjusted to ensure optimal overlap for all settings

    - high-epsilon Left-SHMS diamond seems different than others, overlap

      causes a large section of the diamond to be removed

    - *NB* Chi Kin says he doesn't see this in his data.  Richard will follow

      up with Chi Kin

    - the only diamond where Chi Kin sees an issue is Q2=3.0 W=2.32, where one

      corner is cut off, Richard also sees that

- revised Empirical fits (still in progress)

  - *NB* Garth: please compare the Lambda peak shape after background

    subtraction to SIMC to be sure the radiative tail is not over subtracted

  - discussion about the empirical fits

    - Low-epsilon: very unclear any empirical fit is needed now that the proper

      HGC hole cuts are used

      - *NB* since the K+ data are now much cleaner, it is worthwhile to look

        again at the RFtime cut for these data

      - Alicia: expects the RFtime can be used to clean up some protons, but

        not pi+, which are already handled by the pion sample subtraction

Alicia

------

KaonLT u-channel analysis

- LT-sep code setup

  - code now runs, based on Nacer's code, modified for u-channel

  - plans to rerun shape study after every few iterations

- verifiying corrections to Normalized Yield

  - presents a nice table of all efficiencies to be applied, and where they

    come from

  - *NB* Garth: suggests to plot both proton and pi+ TrackEffs vs Rate, to be

    sure the difference between them makes sense, only the pi+ TrackEffs have

    been looked at in detail so far

  - will try to determine proton efficiencies (e.g. event loss due to knock-on

    electron events in Cherenkov) from Heep data, following Bill's thesis Sec

    5.3.8

    - *NB* Garth: not sure if the KaonLT Heep data span a large enough range in

      rate, it would be good to check this.  If needed, we would have to figure

      out how to extrapolate to the rate for physics data

- Alicia raises a good issue that the KaonLT CoinLumi data (from physics

  settings) needs to be looked at again with latest corrections, to confirm the

  yield is flat with rate

  - Garth: these are part of the physics data sample already, just need to

    divide out the runs taken with different currents, with proton PID cuts

    used for the physics analysis

  - Nathan: it would be good to look also at pi+, to see if the <cut to select

    proton PID introduces a rate dependence due to knock-on electrons

  - *NB* Alicia will look at these runs for proton and pi+ PID cuts

  - Richard and Chi Kin should do for these same runs with K+ PID cuts

- MM offset study and calculation of u-shifts

  - will leave to a later stage of analysis

Next Meetings

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

- Thurs: May 7 @ 13:00 Eastern/15:00 Regina

  - KaonLT will go first

- Fri: May 8 @ 12:00 Eastern/10:00 Regina

  - we will continue where we left off

*NOTE THE NEW TIMES AND ZOOM INFO*