Kaon_LT Experiment

                 Mar 5-6/26 PionLT/KaonLT Analysis Meeting Notes

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

                                (Notes by GH)

                    Today: PionLT will be discussed first

Thursday: Present

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

Regina - Garth Huber, Nacer Hamdi, Muhammad Junaid, Alicia Postuma,

   Nathan Heinrich, Nermin Sadoun

CUA - Chi Kin Tam, Sameer Jain

Virginia - Richard Trotta

Ohio - Julie Roche

FIU - Pete Markowitz

CSULA - Konrad Aniol

Junaid

------

Q2=3.85 W=2.62 LT-sep

- added another t-bin at higher -t, divided last bin into 2

  - now have 7 t-bins

  - sigL follows a fairly monotonic trend (minus fluctuations), dropping with

    -t

  - Data/MC ratios have some oscillations for last 3 t-bins at low epsilon,

    high epsilon ratios are more consistently near unity

- met with Sameer for 2 hrs this week on LT-sep code

- comparing physics yields w/ Nathan, using same SIMC parameters and the new 7

  t-bins

Next steps:

- will evaluate systematic uncertainties, which will vary with t-bin

Nermin

------

PionLT LD- PID studies

- new NGC calibration using 15 runs from 12027-46

  - shows Poisson fits for PMTs 1-4

  - gain factors are a bit different than last week (only 3 runs), the

    difference is up to ~20%, depending on PMT

- Discussion: should the gain factor vary run-by-run?

  - need to calculate the error bar on the gain parameters

  - do a study of gain params vs run number

  - if the gains are reasonably consistent within uncertainties, then it is

    good to combine them to get an overall gain factor with smaller uncertainty

  - if there is a sudden change in a gain param, some running condition changed

    and can only combine runs before and after the shift

- replayed data w/ new NGC calibration

  - use of the NGC for e-/pi- rejection is not looking optimistic

Nathan

------

- was ill much of the week, did some data analysis comparisons w/ Junaid

Sameer

------

KaonLT CoinTime Blocking studies

- studies completed for Q2=2.115 to 5.0

- working on Q2=0.50 now

  - waiting on new data replay from Nacer (with extra ROOT branches)

  - should have results next week

- asks about lack of labeled CoinLumi runs in KaonLT

  - this was something we did only for the first time in PionLT

  - KaonLT took various physics data were the beam current was intentionally

    varied by a factor of 2-3x, these runs can be used

  - *NB* Richard will give Sameer the run numbers for these studies

  - Sameer can then apply his CoinTime Blocking correction to the yields to see

    if the dependence is flat

- Junaid: it would be useful to put the CoinTime Blocking factor into the hcana

  Report Files

  - Nathan will send Sameer a template example

  - Richard would like to include this in his next replay, so it needs to be

    completed soon

- *NB* Richard will instruct Sameer on how to post slides on RedMine,

  apparently his account is not yet added to the RedMine group

Nacer

-----

KaonLT Q2=0.5 model optimizations (K+Lambda)

- last week, Richard noticed that sigT looked like a pole term, while sigL did

  not

  - tried reversing them, sigL now pole-like, sigT not pole-like

  - did some iterations, confirmed the fit params are different

  - differences in separated cross sections are tiny, can only see the

    difference when looking at the numbers, plots are barely different

    i.e. model dependence is small, which would be good news

  - Data/MC ratios are a bit better with sigT pole-like, so will retain that

    version

- turned MM-shift on/off and investigated effect on separated cross sections

  - did 1 iteration

  - sigT varies ~5%

  - sigL varies ~10% in lowest t-bin, large t-bins have similar effect in

    magnitude, but a larger fractional effect (>100%) due to very small cross

    section

  - Garth: it is good news that sigL at low -t is only affected 10%, but this

    is likely an over-estimate of the systematic error

    - *NB* rather than MM-shift on/off, try applying the t-shift calculated by

      GH from MM-shift

- preparing ROOT files for Sameer's CoinTime Blocking correction

  - asked about the difference between Autumn18 and Winter18 RefCut files

  - Nathan: need to produce ROOT files with the correct branches, to confirm

    the cuts on RawCoinTime are appropriate

    - *NB* Nathan will give Nacer plotting scripts for RawCoinTime plots w/o

      overflows at 10^38

Garth

-----

Shift in t implied by MM peak shift correction

- modified his relativistic kinematics program to do this calculation

  - MMshift = difference between correct and expt values

    MMexpt=MMcorrect+MMshift (i.e. opposite sign of the MMshift applied to data)

  - keeping Q2, W, and meson CM-angle constant, then calculate the change in

    meson kinematics implied by the MMshift

    - meson momentum and meson lab angle changed slightly

  - then calculate t

    tEXP=tCORRECT+tSHIFT (i.e. apply the opposite sign of tSHIFT to data)

- GH supplied tSHIFT values for Chi Kin, Nacer, Junaid MMshifts

  - Chi Kin: can the shift be applied on event-by-event basis?

  - No, one calculation per SHMS setting, not per t-bin

  - agrees that this is slightly inaccurate, as the true shift will vary with

    t-bin, but the dependence appears to be small

  - *NB* GH will post a cleaned up version of the Fortran code

Friday: Present

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

Regina - Garth Huber, Alicia Postuma, Muhammad Junaid, Nacer Hamdi, Nermin

   Sadoun

FIU - Pete Markowitz

York - Stephen Kay

Virginia - Richard Trotta

CUA - Chi Kin Tam

JLab - Dave Gaskell

JMU - Gabriel Niculescu

Glasgow - Kathleen Ramage, Rachel Montgomery

Alicia

------

u-channel proton PID studies, Q2=3.0 W=3.14 centerSHMS high epsilon

- applying HGC, Aerogel, RF cuts (when available)

- MM cut is not applied until after shape study in physics analysis

- shows some preliminary efficiencies per detector

- HGC has a hole cut

  - *NB* Garth: need to treat the hole cut like an acceptance cut, it should be

    applied in both numerator and denominator in efficiency calc

    - the reason why it must be included in denominator is that the

      inefficiency is not uniform, and scaling up the efficient regions to

      account for the inefficient region will produce a new error

- Aerogel eff is 79%, which is suspicously low

- Discussion: accuracy of proton efficiencies in report file?

  - *NB* Junaid suggests the PID cuts used should be rechecked

  - Dave: a <NPE cut is used for Cherenkov proton selection, normally an

    efficiency would not be needed (since it's a lack of a big signal), but

    protons can generate knock-on electrons that would be excluded by the cut

  - *NB* need to confirm that the efficiency calc correctly accounts for the

     protons lost due to knock-on electrons failing cut

- asks about how to calculate purity of the proton sample

  - Garth: the important part is the yield uncertainty due to pi+

    contamination, not the absolute purity itself

    - since this is handled partly by subtracting pi+ sample in the shape

      study, it is probably most productive to have one overall systematic from

      the shape study which includes pi+ impurity contributions

    - you can calculate the pi+ contribution from the shape study, likely a

      referee would ask about it

- shows plots of proton track efficiency vs run number

  - Junaid: it would be good to also plot vs rate

    - will have to check several rates to see which one has the clearest

      correlation, it will probably be S1X

Chi Kin

-------

KaonLT LT-sep: looking at -tmin vs -t (where -tmin is calc event-by-event)

- first shows SIMC data, when applying all cuts except MM, there are a small

  number of events that violate physical conditions (-t<-tmin)

  - when apply MM cut, the unphysical events go away, clearly these are due to

    the radiative tail

- then shows Expt data, ~10% of events have -t<-tmin, seems a resolution effect

  - Dave: resolution mismatch between Expt and SIMC is the potential issue here

  - Gabriel: should not throw these events away, they have good MM, so they're

    valid events

- then applies Garth's tSHIFt for Q2=4.4 W=2.74

  - this moves t-distribution by the correct amount, no events with -t<-tmin

    remain, the problem is gone

  - it's a bit surprising the calculation works so well, will apply to other

    settings too

Richard

-------

- setting up for bit KaonLT data replay

  - goal is to ahve Q2=3.0 W=3.14 setting replayed by end of weekend

Kathleen

--------

PionLT LD+ PID cuts Q2=1.6 W=3.08 Ebeam=6.395

- last week it was suggested to try SHMS left, center before SHMS right

  - things look very similar to SHMS right

  - refining cuts for HMS Hcal, Hcer

Next Meetings

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

- Thurs: Mar 12 @ 16:00 Eastern/14:00 Regina

  - KaonLT will go first

- Fri: Mar 13 @ 11:00 Eastern/9:00 Regina

  - we will continue where we left off

- NOTE: Regina group needs to note the summer schedule times!