Kaon_LT Experiment

                 Feb 2/23 PionLT/KaonLT Analysis Meeting Notes

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

                           (Notes by GH and SJDK)

                    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 - Stephen Kay, Garth Huber, Ali Usman, Alicia Postuma, Nathan Heinrich,

   Vijay Kumar

JLab - Dave Gaskell

Ohio - Jacob Murphy

CSULA - Yeranuhi Ghandilyan, Konrad Aniol

CUA - Richard Trotta, Tanja Horn

FIU - Pete Markowitz

Vijay Updates

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

Update on Q2=0.38 low epsilon Analysis

- CT peak, one peak either side of prompt peak looks very different (more

  counts), one is 2 left of Prompt, other is 3 right of Prompt

  for LEFT-2 SHMS setting only

   - Possibly a different particle leaking in, but then why does this give

     peaks on both sides?

   - Should create 2D distributions to check this

      - Plot CT vs MM

      - CT vs Beta too

   - TH: some weird random peaks were seen in Fpi-2 as well, but only on one

     side, it is weird here to have weird peaks on both sides of prompt peak

   - DG: yes, we can place cuts to avoid these weird peaks in the random

     subtraction, but we should spend some effort to understand why they're

     weird first

- -ve events in MM distribution after Dummy and Random subtraction, to right of

  neutron peak

      - Green - central

      - Red - left1

      - Black - left2

   - After dummy + random subtraction - over subtraction?

      - Wrong charge assumed in dummy subtraction or similar?  *Check*

- MM peak is also very off, wrong MM used?

   - neutron peak is at 0.89, this is off a lot!

   - Vijay thinks it is MMpi, could easily be MMK if using Richard's analysis

     though  *Check*

- Diamond cut on low epsilon data

   - Cut shown for central setting

   - DG: How was this done?

      - Script maybe overcomplicating it?

      - It should be a relatively straightforward set of cuts

      - TH: Often it is nearly impossible to be sure exactly what you're doing

        if a complicated script is used

      - Each plot you make, you should be able to *understand* exactly what is

        done to the data

   - Some ragged blue region, it looks like SHMS+HMS acceptance cuts not

     applied

   - important to set the diamond cut after applying other physics analysis

     cuts, apply to ALL three epsilon data, including low epsilon, for

     consistency

- RefTime cuts for this data?

   - Originally said not set but then said it was, not 100% clear

      - Think it's set ok?  *Verify*

- t-binning

   - Need same t bins for all three epsilon settings

   - Lowest t bin will have nothing from left2

   - Similarly highest bin will be only left2!

   - Look at all statistics for three epsilons to determine t binning

      - Apply diamond cut to mid/high epsilon, and other physics cuts too, such

        as MM, etc.  Want the t-coverage and statistics to correspond

        reasonably well to what will be seen in the final physics analysis

- Data/SIMC comparison

   - No scaling applied to SIMC, and normalization comparison looks

     reasurringly good, as pion model should work very well at these

     low Q2

   - Pion model should be much better than kaon model to begin with

   - SIMC distribution broader than data in xptar

      - implies on some aperture in the simulation is "off"

      - Which though? Very hard to determine

      - This is a mystery

      - Dave G has already updated target geometry in SIMC so it isn't that

      - DG has no other suggestions, but notes that the same is seen in SIDIS

   - SHMS yptar distributions SIMC, Data have same width, but a slight shift

     indicating a small offset issue?

Richard Updates

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

- Diamond cuts

   - Q2=5.5 looks good

         - Acceptance cut is applied, less ragged edges than what Vijay showed

         - script crashes for Q2=2.11, investigating

- t binning

   - Apply all cuts

      - PID

      - CT

   - Only one epsilon shown, will have to show both epsilons after diamond cuts

     to determine optimal binning

- t-distribution goes through zero (not allowed), lots of events at t>0

    - Centre/left look OK

    - Right PID looks bad

         - Still clearly getting pi/p leakthrough

         - Large delta background under lambdas that needs to be removed

      - Try to get a clean pion sample, normalise this to residual neutron peak

        with kaon PID

         - Subtract it off

         - Basically infer the shape of the pion background from selecting out

           clean pions

      - Likely have to similarly get a clean proton sample, and normalize to

        residual omega peak, to remove proton leakthrough

   - Something seems to have gone very wrong for the right setting

     *Investigate*

      - DG: maybe its an error pointing to a wrong calibration file?

   - On plots, don't supress -ve counts

      - Want to see where y=0 actually is, preferably by drawing a line there,

        then can clearly see what's y<0

- SIMC rotation matrix differs from what Carlos has

   - Missing conversion from cartesian to spherical co-ords

   - Definition of what theta/phi for central spectrometer mean differ too

   - DG: Would it be easier to take output of SIMC and pass it through a

     converter/script to adjust the values

      - Copy what Carlos did and run this on the SIMC data

      - Some modifications needed

      - Should probably take values, convert them and save them to a new variable

      - Could just be added to repository as a standard tool

- In Bill's LT-sep code, definitely seem to be missing a script still

   - Vijay noticed that there seems to be this missing link

   - Should be called "average_ratios.f", or something similar, which combines

     the different SHMS settings together to get full phi coverage

- Average the ratios or sum the yields?

   - Either way should work, see notes from 2 weeks ago (23/01/19)

Alicia Updates

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

BSA pi+n Analysis from KaonLT

- Fixed dummy target subtraction with correct thickness factor

- widened MM window and NPE threshold in SHMS Aerogel

- Q2=3.0, W=2.32, 5 t-bin results

   - Asymmetry fits look a bit better now

   - Bin 2 (0.21<-t<0.29) still has a different phi-dependence than the rest

   - simple fit (A) vs full fit (A,B,C) differs by 2 sigma for Bin 2

   - The last t bin is very wide, please *remove* high t tail

- comparison of preliminary LT' with VR Regge model

   - VR model has a known issue that is acknowledged in their paper

      - Does particularly bad at high -t (> 0.5)

   - VR curve does not extend to as low -t as the data

   - GH: this is because the curve is calculated with central kinematics only

      - need to compare VR to data at actual <Q2>,<W> for each t-bin

      - lowest -t bin will have a lower -t_min than the central kinematics, and

        the VR curve will then extend as low as the data

   - even with these issues, the VR comparo to data looks reasonable up to -t=0.5

- Q2=3.0, W=2.32 AL data

   - asymmetry with dummy target data is consistent with zero to 1 sigma

- Q2 = 4.4, W=2.74 first results

   - again see the second bin doing its own thing

   - difference between simple and full fits is only 1 sigma though

   - AL BSA is consistent with zero within errors

- Q2 = 2.115 results

   - *need* finer t-binning, lots of stats

   - full and partial fits different by <1 sigma, bin 2 issue seems largely

     absent here

- Looking at t-bin #2 more carefully

   - starting to get K+ leakthrough in bins 4,5.  Bin 2 looks fairly clean

   - background between pi+N and K+Lambda peaks also grows in Bins 3,4,5

   - DG: Delta contribution seems to have a very different -t dependence to the

     neutron contribution, see a kink at MM=1.05

      - GH: piDelta has different t_min than pi+n, so that's why its absent in

        Bins 1,2

   - GH: wonders how big the pi+n radiative effects are for the higher t bins,

     and whether this explains the extra counts left of piDelta

      - we *need* to run SIMC for these settings and look at the growth of the

        MM radiative tail with t-bin to better understand this

      - will make a nice plot for the paper too

   - DG -For weird second tbin, have you looked at asymmetry before combining

     settings?

      - AP: Yes, overlayed, hard to interpret as statistics poorer

      - Nothing strange immediately jumping out

      - Can *try* at low Q2, where there are more stats, but the weird effect

        seems smaller there

Jacob Updates

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

new HMS saturation curve including higher momentum data

- Saturation correction in B sets in above 5.1 GeV/c

- shows 12% effect at P=6.8 GeV/c

- DG: Isn't this removed in the magnet setting program?

   - checks, the program is unchanged since 2018-Aug-26

   - JM: using B/I from readback, so this is the real effect    

- Looking at Q1,Q2,Q3 saturation, expect to have more to show next week

Nathan Updates

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

Detector time cuts update

- Made an "open" cuts file

   - Wide enough that whole distribution can be seen

- Summing over PMTs since the're stable

   - Doesn't change much between PMTs

- Plots shown from run 13090, but other runs look the same

   - Dashed lines with suggested cuts

   - HMS Cherenkov looks odd

      - All events at zero for H.cer.goodAdcTdcDiffTime_PMT2

   - similarly variables for SHMS calorimeter all zero

      - GH: unfortunately, hcana initializes all variables to zero, instead of

        an error code of -9999, so it's not possible to distinguish a "true

        zero" from an "error code zero".  Suspects this means the variable is

        actually not present, which is not the same as being present but zero

      - Maybe variable has been renamed to something else?  *Investigate*

   - HMS Cherenkov distributions otherwise look fine

   - HMS Calorimeter double peaks are seen

      - most detectors have the right peak larger

      - some detectors have both peaks

      - others have left peak larger

      - making cut wide enough to include both peaks for now

      - GH: the bkd is down by 100X from the peak, so there is a bit more noise

        leaking in with wider cut, but probably okay

      - RT: yes, thinks so too

Next Meeting

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

Thur Feb 9 @ 17:00 Eastern/16:00 Regina/14:00 Pacific

- PionLT will go first