Kaon_LT Experiment

                 Jun 6/24 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, Muhammad Junaid, Ali Usman, Nathan Heinrich, 

   Zach Sullivan, Nacer Hamdi, Vijay Kumar

York - Stephen Kay

CUA - Casey Morean, Tanja Horn

Virginia - Richard Trotta

CSULA - Konrad Aniol

Nathan and Zach

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

PionLT Luminosity Studies

- Zach now has Nathan's scripts working in his account

- Carbon HMS Runs 16727-16737

  - Accept/Total scaler count is low for <15uA, will need to change current

    cuts for low current runs.  Then will need to recheck low current data

  - CPULT is very low ~81%, while TLT ~98%

    - Richard: suggested looking more carefully at what trigger is used

    - Nathan: things using ELREAL instead of ELCLEAN, will check

- LH2 HMS Runs 16703-16712

  - Again a problem w/ CPULT ~79%

  - No boiling seen up to 60uA, then suddenly down to 92% at 80uA (no Track),

    same for Tracked analysis

  - Scaler analysis looks good however

- Carbon SHMS Runs 16738-16746

  - CPULT looks good, gives hint on what may be wrong for HMS analysis

- LH2 SHMS Runs 16759-16764

  - boiling looks better here, so need to look again at HMS, they should be the

    same within errors

  - Scaler analysis: 98% at 80uA

  - No-Track: 95% at 80uA

  - Track: 94% at 80uA

- Next Steps:

  1) Zach will work on code for Efficiencies for PID cuts used in analysis

    - will look to see if there is any rate dependence

  2) Look at what trigger is used for CPULT calc

  3) Adjust hcana current cuts for low current runs

  - then replay data and look at results

Junaid

------

PionLT Detector Eff Study

- runs with HMS p<4.2 GeV/c

  - Cer_eff uses npeSum >1.5 cut, tighter cut on Calorimeter

  - Cal_eff uses tight cut on Cerenkov instead

- Obtain CerEff=0.9981 CalEff=0.9981

  - KaonLT had CerEff much less than CalEff: CerEff=0.9717 CalEff=0.9929

  - interesting that CerEff are so different between the two experiments

  - Tanja: when you applied tight cut on Cerenkov, did you do that to define

    the sample on the calorimeter?

  - Ali: CerEff is likely different from KaonLT due to different cut value used

    (1.5 vs 2.0 npe).  KaonLT used nepSum>2 because that was needed in the

    Physics analysis.

    - *NB* You should do a quick check of 3-4 physics runs at

      different settings to make sure you don't need a higher cut

PionLT Heep Study

- finalized OOP offsets, corrected beam energies

- there was some confusion on implementing momentum offsets

  - verified how the offset is applied in hcana, offsets need to be entered as

    %P.  This means the P-offsets applied by Vijay are not done correctly.  The

    effect will be small since his P-offsets are small, but should be checked

    as a systematic.

- Heep comparison plots (before vs. after offsets)

  - PMZ, EM worse

  - PMX, PMY, W improved

    - Richard: similar behavior to what he saw

    - confirmed that simc-reconstruction script was used

- Table of shifts between Data and SIMC (after offset vs. expected)

  - PMZ shifts in opposite direction than predicted by GH's program

  - need to look closer at whether to ADD or SUBTRACT offset

  - *NB* Garth: The HeepCheck program says dp_m(par) is correlated only with

    dp_p so the sign of the dp offsets should be checked

Vijay

-----

- Met with Richard, Ali yesterday to discuss systematics, Richard will give

  more details

  - Vijay is working on MM Cut-Dep study

    - will modify cuts by +/-2,4,6,8 MeV from standard 980 MeV MM cut

    - needs to subtract obtained cross section from that obtained with standard

      cut and tabulate the differences vs. t-bin

Richard

-------

- gives some info on using Globus for file transfer from JLab

KaonLT LT-separations

- simplified the functional form for sigL, sigT:

  - sigL=(p1+p2*log(Q2))*exp(p3*|t|)

  - sigT=(p5*(|t|/Q2-1)*exp(p6*|t|)

- after 1 iteration, sigT did not change too much for Q2=3.0, W=3.14, so

  optimistic that on right track

- tried to work on Q2=2.1 L/T-separations

  - the low epsilon diamond is quite small, corresponding to a relatively large

    delta-epsilon

  - this means there will be smaller error magnification for sigL, but at the

    cost of low epsilon data have poor statistics.  Hopefully they will balance

    out in the final analysis

  - low epsilon stats are too low to be sure the functional form is good, etc.

  - will go to higher Q2 and then work down from there

Summary of Random Systematics Discussion w/ Ali, Vijay

Acceptance - vary geometrical cuts

PID - vary PID cuts

  - HGC will drive K+ studies

  - separate studies needed here for every analysis, since backgrounds and cuts

    vary

Tracking - compare effs for different algorithms, tracking parameters

Kinematics - adjust offsets

Radiative Corr - turn on/off radcor flag in SIMC

Model Dep - vary input model

  - Garth: in pi-/pi+ analysis we set LT=TT=0 and extracted new cross sections

    without iterating

  - Tanja: Fpi-2 took a more gentle approach, used different models with

    iterating

Nacer

-----

KaonLT Heep Study

- presented Data and SIMC cuts, corrections for 2 beam energies

- 3.8 GeV Data/MC Yield ratio = 1.02 +/- 0.005 (stat)

- 4.9 GeV ratio = 1.046 +/- 0.004

  - both results very similar to Richard's, although analysis is different

  - *NB* Ali: The CerEff is calculated for >2npe cut, while Nacer is using

    >1.5npe cut.  The cut and efficiency need to correspond to each other, so

    this needs to be checked.

  - We see from Junaid's results that the difference in CerEff is about 3%, so

    this might explain most of the ratio difference from Unity.  Nacer and

    Junaid will discuss.

Ali

---

piDelta MM Shape Study for additional settings

- Q2=3.0, W=3.14

  - exluding (t,phi) bins from fit if low statistics or if MM fit fails

  - Center SHMS, high -t: Two bins (135<phi<180, -180<phi<-135) have no piDelta

    contribution from MM to Data, are excluded

  - The data for these two bins are mostly at higher MM, only a small fraction

    of events near Delta MM

- Q2=2.1, W=2.95

  - two t-bins: 0-0.27, 0.25-0.55.  Higher -t data are excluded 0.55-1.0

- Will have BSA soon for all 5 settings

  - then will compare LT' to Alicia's pi+n (rebinned) and CLAS pi-Delta++

- Also did some systematic uncertainty studies, both related to MM fit

  dependence

  - MM cut dependence is yet to do

 Next Meeting

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

- Next meet: Thur June 13 @ 15:00 Eastern/13:00 Regina

  - KaonLT will go first