Kaon_LT Experiment

                 May 7-8/26 PionLT/KaonLT Analysis Meeting Notes

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

                                (Notes by GH)

                     Today: KaonLT will be discussed first

Thursday: Present

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

Regina - Garth Huber, Nathan Heinrich, Alicia Postuma, Dex Yadlowski,

   Nacer Hamdi, Nermin Sadoun, Muhammad Junaid, Vijay Kumar

FIU - Pete Markowitz

Virginia - Richard Trotta

JLab - Dave Gaskell

CUA - Chi Kin tam, Tanja Horn, Sameer Jain

Glasgow - Kathleen Ramage, Rachel Montgomery

Richard

-------

KaonLT Q2=4.4 W=2.74 LT-sep analysis

- refining the empirical fit used underneath the Lambda MM peak

- shape is a Chebyshev order 2 polynomial

  - changed the MM range where the fit is applied

- *NB* still need to compare data to SIMC (tail matching) to be sure radiative

  tail is not over subtracted

- added various checks in the background fit:

  - mostly based on the the Data/MC Ratios, e.g. #t-phi bins that have Data/MC

    ratio > 3 sigma from unity

  - also a kinematic fit score: compare data to SIMC shape for different

    variables

  - Garth: the kinematic score makes sense, but the first 3 criteria are based

    on the SIMC model, concerned that this will amplify the model dependence.

    Is there a less model-dependent way of implementing a "smoothness"

    criterion, to make sure the amount of subtracted background does not vary

    suddenly between 2 bins?

  - Nacer: what about trying Gabi's background subtraction method,

    i.e. implementing a Q-factor?

    - Richard: would need more clarification from Gabi on it before

      implementing

  - Garth: will you use the variations from this study to determine also the

    background subtraction systematic uncertainty at the same time?

    - *NB* Yes, that is the plan.  Will also cross check results w/ Chi Kin

- did a background fit study with #t-bins varied from 2 to 4, and #phi-bins

  from 8 to 10, to see which binning gave most reliable results according to

  the above criteria

  - Nacer: his worry is that you try to get Data/MC ratios near 1 by varying

    the background, instead of optimizing the Lambda MM peak fit region

    - Richard: not yet comparing background subtracted MM to SIMC, will need to

      modify the code to do that

  - each binning is scored according to different metrics, with weighting to

    obtain a composite score:

    - kinematic fit: 15%

    - 3 Data/MC ratio criteria: 20-35% each, totaling to 80%

    - #valid Data/MC ratio bins: 5%

  - Garth: how was this weighting determined?

    - Just by intuition

    - *NB* Garth suggests to increase the kinematic fit score to 50%, to reduce

      model dependence, and more closely tie the subtraction to the data

- Dave: what is the physical origin of the background you are trying to

  eliminate?  The origin should inform the strategy used to remove it

  - points out worse PID for high epsilon data, due to lack of RFtime cut

  - ending up with nearly zero bkd subtraction at low epsilon after the

    improved HGC hole cuts

  - high epsilon studies with these criteria are still running

- Tanja: how didyou validate the background subtraction?

  - the above criteria vs background scale factor

Chi Kin

-------

KaonLT Heep offset optimization

- spoke w/ Dave and confirmed the phi offset is applied incorrectly in hcana

  - both offsets need to be applied: hphi_offset+h.oopcentral_offset

    - *NB* will make modifications to hcana and push changes

  - Dave also said oopcentral offset needs to be applied in SIMC input file

    - this will affect exactly where events are generated

    - *NB* needs to confirm w/ Dave on correct way to do this

    - htheta_offset, hdelta_delta offset were set to zero in the SIMC input files

      he found

- *NB* Garth: please send me the Heep information from your new analysis, so he

  can recalculate the HeepCheck offsets for comparison to yours

Alicia

------

KaonLT u-channel PID cut corrections

- using PID cuts for protons:

  Aerogel < 4npe

  RFtime

  HGC < 2npe + hole-cut

- Aerogel "Efficiency": caused by knock-on electrons from protons in Aerogel

  - Garth suggests to call a knock-on correction instead of an Efficiency to

    avoid confusion

- RFTime "Efficiency": Junaid corrects for valid events from tails of RFtime

  peak that are removed by the timing window cut

- For Aerogel Efficiency following the method from Bill's Thesis Sec 5.3.8

  - used HeepCoin events for clean proton sample

  - applied PID, acceptance, CoinTime, Missing Energy Cuts

  - Bill obtained 92.7 +/- 1.6% efficiency, applied as a global factor to omega

    data

  - redoing that study here

    - is the result rate dependent?

      - Plot vs SHMS-3/4 rate from 2.2-2.9 kHz

        - no clear dependence, but HeepCoin data cover a narrower range of

          rates than physics data

    - sees a beam energy dependence (proxy for SHMS momentum dependence):

      93.5% (6.2 GeV) to 90.1% (10.6 GeV)

      *NB* Garth: it would be worthwhile to get Nacer's HeepCoin data, to get a

      wider SHMS momentum range

      - it appears one could get a point as low as 2.5 GeV/c SHMS momentum

   - Nacer: the 10.6 GeV Heep CoinTime peak seems very broad, the 8.2 GeV peak

     is much narrower

- Does Alicia's Geant4 proton loss code simulate the effect of knock-on

  electrons?

  - shmsPA counts Cherenkov photons, not NPE

  - seems that the Geant4 code cannot be used to calculate this.  On the other

    hand, it also means the number from the code will not double-correct for

    this effect

  - can try to use the Physics Data to determine this effect, and compare to

    Heep result

    - Garth: not sure the physics data are sufficiently clean to reliably do

     this

Friday: Present

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

Regina - Garth Huber, Nathan Heinrich, Alicia Postuma, Dex Yadlowski,

   Nacer Hamdi, Nermin Sadoun

York - Stephen Kay

CUA - Chi Kin Tam, Sameer Jain, Tanja Horn

Glasgow - Kathleen Ramage

JMU - Gabriel Niculescu

Virginia - Richard Trotta

JLab - Dave Gaskell

Nacer

-----

KaonLT Low Q^2 LT-sep analysis

- replay with CoinBlocking correction is completed, now included in Qeff

  calculation

  - had to resubmit jobs due to SciComp issues

- observes no big changes after CoinTime blocking corr added

  - doing more iterations

  - plans to show separated cross sections next week, then on to systematic

    uncertainty studies

Nathan

------

PionLT Q2=5.0 W=2.95 LT-sep analysis

- preliminary diamond plots shown, will need to make adjustments to 2 corners

- t-resolution plots (SIMC t-recon vs ti) looks very bad

  - looked good straight from SIMC, before recon_hcana

  - *NB* Gabriel: it is likely a -t vs +t issue, in SIMC ti is actually -t, in

    recon_hcana t is +t, so need to add instead of subtract t-recon from ti to

    get the t-resolution

Sameer

------

PionLT Q2=1.6 LT-sep analysis

- ran first few scripts in Junaid's framework, essentially following same steps

  as Nathan

  - need to draw diamond

  - only replayed SHMS_center setting so far, need to run others before looking

    at t-binning, etc

Nermin

------

PionLT Lumi analysis to determine LD2 boiling correction

- 9.2GeV beam Lumi set #1 Carbon

  - showed Relative Yield vs Current and vs ELREAL_rate

  - TLT plots: Red=EDTM Blue=CPULT*ELLT where ELLT is calculated from

    combinatoric equation

  - Runs 12143-12154:

    HMS: scaler, NoTrack, Track yields look flat

    SHMS: get some slope

    - we compared to Nathan's result, they look similar

    - Nathan: Scaler Yield boiling slope indicates some small error in ELLT

      NoTrack Yield exhibits anti-boiling, indicating an error in CPULT in the

      opposite direction

      Track Yield has same anti-boiling effect, indicating that TrackEff is

      probably okay (i.e. the anti-boiling was not worse)

    - Dave: ELCLEAN would be cleaner than ELREAL, but recalls that Nathan tried

      this and it did not help

    - Nathan: maybe its an issue with just this setting, suggests to look at

      other settings to see if the problem persists

- 9.2GeV Lumi set #2

  - plots look similar

  - comparison to Nathan's analysis indicates he had same issues

- 6.4 GeV Lumi Carbon Runs 16738-46

  - HMS: slope is a bit bigger than Nathan's

    - suggestion is to look again at PID cuts

  - SHMS: will try to make the slope for this a bit better too

- 6.4 GeV LD2 Lumi scan

  - point at 80uA is a bit low compared to trend

  - Nathan thinks there are likely ELLT issues for >200kHz rate

- 9.2 GeV LD2 Lumi #1

  - CPULT looks weird vs rate, jumps around a lot, non-monotonic

    - Nathan: this needs more investigation

      - particularly please check if the 45uA run has something wrong, such as

        an error with ELLT calculation

    - Dave: which BCM is being used in this study?

      - Nathan: it is supposed to be BCM2, with the non-linearity correction

        turned on

    - *NB* Nathan: please recheck the current cuts used in this study, it could

      be that they are either too loose or incorrect

      - this is the first time these runs have been replayed, so that seems a

        likely issue

    - Gabriel: based on the scatter or slope of the Carbon data, we can

      evaluate a systematic uncertainty that we apply to the LD2 Lumi study

- 9.2 GeV LD2 Lumi #2

  - HMS shows anti-boiling

  - *NB* Nathan: suggests to look first at Current Cuts before the PID cuts

Kathleen

--------

PionLT LD2 analysis

- will start soon on Lumi study

- Rachel and Kathleen discussed, will recheck Nermin's result and then analyze

  the scan at 7.9 GeV

Next Meetings

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

- Thurs: May 14 @ 13:00 Eastern/11:00 Regina

  - PionLT will go first

- Fri: May 15 @ 12:00 Eastern/10:00 Regina

  - we will continue where we left off

*NOTE THE NEW TIMES AND ZOOM INFO*