July 10/25 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, Alicia Postuma, Muhammad Junaid,
   Ivan Zhenchuk, Nacer Hamdi
Virginia - Richard Trotta
Ohio - Julie Roche

Richard
-------
Q2=4.4 W=2.74 LT-separations
- now using the same simpler parameterization for all settings
- only showing Center SHMS settings right now
- in the parameterization, t_av is the average t-value over all data, while |t|
  is the central value for a given t-bin

- pion background was adjusted for all t-phi bins
  - previously was doing a fit of the pi+n region to determine the integral for
    normalization, now just summing over channels in the neutron MM-range
  - this gives the correct normalization now, while we noticed some obvious
    over subtraction last week
    - for pion subtraction, using 0.88<mpi<0.94 integration range, in
  - as a result, the oscillations in phi are smaller now

- with this smaller pion subtraction, still has some background underneath
  the Lambda peak that needs investigation
  - will look at improving the HGC hole cut, particularly at 10.6GeV, to reduce
    pion leakthrough
  - some t-phi bins are barely populated for Center setting (not unexpected)
    - however, for these settings, can't normalize to nonexistent pi+n peak
    - *NB* Garth: suggest to fit by hand the pion background in the Delta
      region to the background underneath the Lambda for these cases, taking
      care to avoid negative counts, should use the SIMC peak shape as a guide
      for how much to subtract
  - *NB* Nacer: if there is background remaining after pion subtraction,
    suggest to fit a simple polynomial to subtract it off

Nacer
-----
KaonLT Low Q2 LT-sep
- wrote 2 scripts to fit the functional params
  - sig_uns obtained using Koltenuk ratios, but fitting the parameters using
    Tanja's Fpi-2 parameterization

- one script uses ROOTfit and the other uses Python, comparing them to check
    for consistency
    - results the same, except for sigL param(6) which fit more stably with
      ROOT than Python
    - Nacer is giving initial params in the fit to be about 1, rather than
      using Tanja's thesis values
    - long discussion about the need to properly initialize the fit params, as
      fitting algorithms often find false minima
  - Richard: it is quite possible to end up with Ratios near 1, but in a false
    chi-square minimum
     - this is because the Ratios are based on the integrals of the
       distributions
     - *NB* to really be sure you have the correct fit, you need to compare
       Data and MC distributions of focal plane and physics variables.  They
       will converge when you have the correct fit
     
- Nacer's code made the mistake of feeding degree values into sin(thetacm)
  rather than radians for LT and TT, so the fits weirdly oscillate
  - Garth: there is no guarantee the equations from Tanja's thesis will be what
    eventually gives a good fit to the data, so pay attention to what the fits
    give you, so we can figure out what to eventually adjust

Junaid
------
Q2=3.85 W=2.62 LT-sep
- fitting Tanja's Fpi-2 parameterization as well, using Python
  - in sigL, keeping param(8),param(9) in denominator fixed for now, and only
    allowing the 3 parameters in the numerator to float
  - reran SIMC w/Fpi-2 parameterization
  - initial values in the fit were taken from Tanja's thesis, not ~1 as Nacer
    did
  - has fits, but hasn't looked at the new parameters yet
  - Garth: suggests to take a look at the parameters and see how they have
    changed, that can tell you which params to fix and which to let float or
    set to zero

- working on a Weight recalculation script, meetings planned with Richard and
  Bill next week

Nathan
------
PionLT deadtime investigations
- still following up on weird timing shifts identified a few months ago
- changed pTRIG.ROC1,2  pTRIG4_ROC1,2  PT2  timing window ranges
  - the changes are small, e.g. pTRIG1.ROC1 changed from 5100-6200 to 5000-6000
  - now T.coin.pTRIG1_ROC1_tdcTimeRaw shows a multiple peak structure
  - will make different versions of plot for different multiplicities to see
    what is going on

- with this change, CT vs xfp shows that the weird timing shift has gone away
  - not sure why, maybe ONLINE params are being called somehow?
    - doing a new replay to be sure
  - CT offset has also shifted, shouldn't have, indicates that something else
    has changed that shouldn't have

- still trying to debug code, to be sure correct DBASE are being used
  - not sure if last week's plots (showing no change) are right or not
  - maybe the explicit call to TCOIN.param had a typo?  Unfortunately, did not
    keep a record so no way to know

Alicia
------
u-channel background fitting
- had a discussion with Henry Klest re Pythia generator

- also had a discussion with Bill last week
  - he things it's uncorrelated 3-pi background, suggests to modify the
    phase-space SIMC routine rather than use Pythia
    - *NB* Garth: suggest to look at Appendix in Ambrozewicz thesis to figure
      out what to change in his derivation

Ivan
----
KaonLT K+ and pi+ BSA analysis
- random subtraction is messed up for K+, need to look at it again
- also messed up for pi+
  - the issue was that the reals window is shifted between pi+ and K+ and the
    randoms normalization factor was not calculated correctly
  - this affected pi+ BSA, and since pi+ is also used for background
    subtraction for K+, it also affects K+ BSA
  - anticipating the BSA will be closer to zero, with less randomness when
    randoms fixed, but of course there is no guarantee of this
  - *NB* Garth: suggests to try a separate pion subtraction factor for each
    t-bin, not a global factor at each Q2
    - *NB* same comments regarding background underneath the Lambda that are in
      Nacer comments above apply

- starting preparations for systematic error studies

Next Meeting
-------------
- Thur July 17 @ 15:30 Eastern/13:30 Regina
  - PionLT will go first