Kaon_LT Experiment

                 July 24/25 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, Nathan Heinrich, Ivan Zhenchuk, Alicia Postuma,

   Muhammad Junaid, Nacer Hamdi

Virginia - Richard Trotta

FIU - Pete Markowitz

JMU - Gabriel Niculescu

CUA - Tanja Horn

JLab - Dave Gaskell

Ohio - Julie Roche

Gabriel

-------

Sigma/Lambda analysis

- fitting Sigma0 yield, high epsilon KaonLT data

- subtracts Randoms, MT, pion leakthrough

- use MC templates to fit Lambda, Sigma0 peaks

  - allows to shift data by a few MeV to make peaks match

- reran SIMC w/ resmult=3.5 factor from Ali

  - modification in shms_mchut.f

- compare data and MC, for pions due to greater statistics

  - fit MM peak w/ simple Gaussian + bkd function, extract pi+n center and

    width

  - increase in MC peak width agrees well with Ali's values, but still much

    narrower than data

  - even a factor of 3.5 implies a tracking resolution >0.5mm, definitely

    unreasonable to increase it further to better describe the data

- Junaid: it's important to remove NGC from SIMC since it was not there for

  KaonLT running

  - haven't done this yet, yes it's important but it will make the resolution

    discrepancy even worse, not better

- shows plot of pi+n center vs p_SHMS, a correlation is seen, peak shifts from

  ~0.95 at low momentum to ~0.935 at high momentum

  - Dave: we've been looking more closely recently at 2022 SHMS data, getting

    some weird effects starting as low as p_SHMS=4 GeV/c

    - spoke with Mark about it, apparently there are some saturation effects

      start as low as 4 GeV/c, which is a surprise

    - *NB* suggest to keep looking for correlations of MM peak with focal plane

      and target variables, see if we can find a way to correct it

      - it may be clearer to start first looking at the KaonLT Heep data

Richard

-------

Q2=4.4 KaonLT analysis

- looking in more detail at cut differences between high and low epsilon, both

  to better understand where sig_lo>sig_hi is coming from and for systematics

  study

  - sig_lo ~10% larger than sig_hi for many bins

  - the main differences between high and low epsilon are:

    - no right SHMS for low epsilon

    - HGC hole cut larger for high epsilon

      - asked GH about HGC hole cuts

      - Vijay's study was for the low epsilon data with smaller HGC hole

      - had done a quick set of HGC cuts for thesis, now need to do a more

        careful study

- looking at lambda peak, clearly there is an extra bkd not included in the

  pion subtraction

  - Garth: *NB* the difference between data and pion leakthrough curve to left

    of Lambda peak is nonzero but nearly flat, could try fitting a flat bkd fit

    to 1.02<MM<1.07 region and subtract it off

  - Alicia has a better idea: the pion bkd underneath Lambda has a different

    t-dependence than pi+n

    - normalizing to pi+n peak can lead to an undersubtraction of bkd beneath

      Lambda

    - *NB* suggest to scale pion bkd to region left of Lambda (1.02 to 1.08),

      it will oversubtract pi+ region but probably give a cleaner result in

      Lambda region

   - Nacer: usually people try to fit bkd with a polynomial

     - Richard: *NB* will try a bunch of different subtraction methods, we will

       choose a favorite one for the final result and use the others as part of

       the systematics calculation

Nacer

-----

KaonLT Q2=0.5 LT-sep

- doing SIMC weight recalculation script checks

  - using Junaid's new weight script

- Richard: has a script that moves files from volatile to cache and copies each

  instance in a different directory, it's very useful when doing iterations

Ivan

----

KaonLT Q2=0.5 BSA

- plots of MM for each t-phi bin

  - Green=K+ w/dummy subtraction, Red=pion subtraction, Violet=K+ randoms

    Blue=K+ after all subtractions

  - for pion bkd normalization, there are no pi+n events at higher -t, using

    closest t-bin for the normalization

    - get a bit of under/over subtraction in Lambda region

    - there is no bin where the oversubtraction is very large, planning to deal

      with this as a systematic uncertainty

- new BSA uncertainty calculation

  - instead of using bkd-subtracted yields, derive a new formula where pion bkd,

    randoms and Lambda before subtractions are explicitly included

    P=pion bkd, R=randoms

  - Alicia and Ivan independenty derived the same formula

  - now get positive definite results everywhere, even for cases where the bkd

    is large, while previously there were some cases with sqrt(negative number)

- plot of sig_LT' vs t

  - fit BSA with new errors vs phi

  - some outlier points due to sqrt(negative number) fixed, leads to slightly

    different BSA values when combining L,C,R SHMS settings

  - BSA now has fewer fluctuations, sinusoidal curve more noticeable now

  - LT' errors are larger now with new formula

- will refine scripts and continue systematic studies

Alicia

------

u-channel: feeding Pythia into SIMC

- trying for Q2=3, W=3.14 SHMS center setting

- Pythia inputs: xmin=0.15, xmax=0.35, ymin=0.56, ymax=0.66

  - can control the electron with kinematic inputs, but not so much the proton

  - the desired u-channel data is at end of proton momentum distribution

  - added a new flag to keep only p_SHMS>5.4 GeV/c

  - fortunately theta_proton is centered on SHMS acceptance

- Henry Klest had modified pythia6_eic.f to define program inputs and outputs

  - added ePmin - event not saved if proton momentum too low

  - skipping the generation of exclusive events, since these are already in

    SIMC

  - 1M evts run in ~5 min, produce 1k evts after cuts

- an annoyance is that the Pythia script doesn't allow it to run in bkd, need a

  terminal prompt every few minutes to keep iFarm connection from dying

  - have to keep jobs <1M evts

- result: 5M evts in Pythia give 5k evts after cuts, of these 311 make it

  through SIMC

  - mostly not making the proton arm, investigating

  - checking Pythia acceptance variables

    - SHMS_ypfp looks like an issue

    - HMS_ypfp not centered at zero, most events are rejected

- result: 20M Pythia evnts, 19k output, 1233 SIMC events

  - the MM shape looks reasonable, still need to do detailed comparo w/data

Junaid

------

PionLT LT-sep preparations

- using Tanja's functional form

  - in sigL, p7 in exponential and p8,p9 in denominator are all fixed for now

  - fitting all other parameters

- weight recalc script, appplying to SIMC file output by recon_hcana

  - Dave: it's important to use Ratio method to recalculate the weight, to

    retain radiative correction, virtual flux factor, Jacobian

    - Weight=sigCM*Jacobian*GammaFlux*RadCorr

    - Weight_new=sigCM_new/sigCM_old*Weight_old

  - tried one iteration, ratios jumped from ~1 to ~40, clearly an error in the

    weight recalculation script

  - *NB* recon_hcana does not touch the weight or cross section, no other

    corrections are necessary that are not in the above equations

Nathan

------

PionLT coin deadtime investigations

- still debugging weird CT issues

  - finally has a straight CT line vs xfp_HMS

  - now using pTRIG4(HMS-ELREAL) instead of pTRIG3(HMS-3/4)

  - changed time window cuts slightly execpt for HT2, PT2

    - now using a cut on HT2, no cut on PT2

    - still needs to check that PT2 did not screw up SHMS

- Dave: only need cuts on quantities used to calculate CoinTime

  - should use 1 and 3, ELREAL should have same timing as 3/4, the difference

    is only on extra PID info

  - *NB* will try PT3 cuts intead of PT4, and check SHMS

- once has a set of runs with no CT anomalies, will start CoinBlocking study

  - will do a write up for RedMine

Next Meeting

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

- Thur July 31 @ 15:30 Eastern/13:30 Regina

  - PionLT will go first