Mar 14/24 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, Nacer Hamdi, Ali Usman, Muhammad Junaid, Nathan Heinrich,
Vijay Kumar, Alicia Postuma
CUA - Richard Trotta
York - Stephen Kay
CSULA - Konrad Aniol
JLab - Dave Gaskell
Ali
---
MM shape study for DeltaPi BSA
- reran MC's with new resolution parameter
- starting on settings with lower P_HMS, where delta correction is not needed
- shows Q2=3.0, W=3.14, high stats setting, positive helicity
- initial fit of MCs to data
- new 10 MeV/bin instead of previous 4 MeV/bin (lower statistical errors/bin)
- with new MM binning, it is now more clear that strength of DeltaPi peak is
underestimated for some phi-bins
- since SIDIS and rho bkds are so similar, using a fixed ratio of 90%
SIDIS/10% rho for now
- cyan=sum of bkds, pi+n tail+SIDIS+rho
- optimistic that 2 t-bins can eventually be used for this setting (data
shown is over full t-range)
- GH: since they are so similar, why don't you try fitting just one of rho or
SIDIS to see which one is better, and just use that one?
- DG: this would be an unphysical choice, but it probably makes no difference
to your analysis
- Nacer: why are you fitting each phi bin separately?
- we have confidence that SIMC can properly describe the shapes for the
different processes, but not their amplitudes, so we need to fit each phi
bin separately
Alicia
------
MM cut study after resolution correction implemented in SIMC
- Dave suggested calculating Data/SIMC ratio for different MM cut values
- UofR group meeting also suggested to compare w/ w/o HGC cut
- HGC gets rid of KLambda leakthrough but not the shape of the region
between pi+n and KLambda
- ratios do not in general have a plateau vs MM cut
- Q2=3, W=3.14 data shown
- t-bins 1,2 show a decrease in ratio vs MM
- t-bin 3 shows a plateau
- t-bins 4-6 show an increase vs MM
- behavior is similar for Q2=4.5, 5.5
- since the ratio is not flat, will instead try placing MM cut in region where
expt yield is flat
- Q2=5.5, flat region is MM=0.99-1.03, so will try MM cuts 1.01+/-0.02
- Q2=4.4, Q2=3.0 W=3.14, flat region looks similar, so same MM cut
- GH: can't think of anything better to try, given the situation
- Dave: have you looked at BSA vs MM cut?
- a preliminary study was done last year, the final study will be done with
these MM cuts
- Dave: if SIMC handles the radiative corrections properly, any cut below 2pi
threshold should give the same answer
- GH: not sure about this, the pion scattering in the collimator also
populates this region, in Fpi-1,2 we had to use a cut well below 2pi
threshold to avoid being too sensitive to this region
- Dave: there is no SHMS collimator routine in SIMC
- GH: wasn't aware of no routine was included at all.
*TASK for GH:* find the HMS collimator code used in Fpi-2 analysis, and get
instructions on how to modify it for SHMS collimator geometry
- Nacer: can an RF cut be applied to help clean up this MM region?
- no, only a small subset of 10.6 GeV data has the RF information available,
unfortunately
HMS Delta-correction
- P_HMS=6.59, Q2=3, W=2.32 data
- hcana uncorrected MMpi vs hsxpfp shows line with tilt on top half
- using reconstructed MMpi script to determine correction
- applying if hsxpfp>0: hsdelta+=3.0*hsxpfp
- resulting MM distribution is narrowed
- P_HMS=5.292, Q2=2.1, W=2.95 data
- hcana uncorrected MMpi vs hsxpfp shows line with tilt on bottom half
- correction: if hsxpfp<0: hsdelta-=4.0*hsxpfp
- having difficulty implementing this correction in hcana
- recompiling hcana on iFarm gives complaint about missing PODD variables
- Stephen: please *send* the PODD error and he will look at his notes to see
if he can fix
- Nathan: thinks it might be better to implement correction in the kinematics
file
CKY Regge model
- received LT' predictions from BG Yu for the BSA paper
Richard
-------
Subtracting pion leakthrough
- last week's results had HGC hole cut inverted, taking only events in the hole
- now the pi+ dist comes all the way across the MM acceptance
- Q2=2.1 setting, high epsilon
- pi+ subtraction does not clean up region left of KLambda peak
- planning to re-optimize PID cuts after graduation
Unsep and Separated Q2=3.0, W=2.32 xsect
- shows ratios vs phi,t at high, low epsilon
- looking at what's responsible for Data/SIMC ratios being much larger at
high eps than low eps
- trying various initial values of model parameters
sigL=g(W)*(p1+p2*logQ2)exp(p3+p4*logQ2)*(-t+0.2)
-- changed last part to just -t to get better t-dependence
sigT=g(W)*p5/(1+p6*Q2)
- Dave: maybe remove the Q2-dependence and only have a simple t-dependence
- Dave: the *real issue*, however, is that if you have sigL<0 from the
beginning, there's a problem with the data itself, not the iteration
- GH: exactly what MM region is integrated for these yields?
- MM cut is applied after pi-subtraction
- GH: if we can't clean up the region left of KLambda with cuts or pion
subtraction, then we will have to consider fitting a polynomial and
subtracting it
- Dave: yes, we will need to sort this out
- Richard shows MM plots for same setting at high and low epsilon
- RF cut can be used only at low epsilon, region left of KLambda is
noticeably cleaner than high epsilon
- GH notices that pion subtraction peak is much narrower than the pi+n MM
leakthrough peak
- Ali's pion sample was analyzed as if pions, not K, hence the peak is narrow
- normalizing to the height of the pi+n peak will result in too little pion
leakthrough being subtracted, as the integrals won't match
- *Ali* will give Richard his cuts and RT will make a branch of his analysis
for the pion subtraction
Vijay
-----
PionLT low Q2 lt-analysis
- found bug in script, error bars vs phi now more reasonable
- shows results after 1 iteration for Q2=0.38, all 3 epsilons
- some discussion about phi-dependence of lowest t-bin
- the low epsilon setting seems to have a different phi-dependence than mid
and high eps, which is weird
- Dave: this could indicate a residual offset issue, lowest t-bin is always
more sensitive to the offsets
- Vijay will investigate further
- t-bins 2-7 have more consistent phi-dependence at low, mid, high epsilon
- 8th t-bin has poor statistics, might have to be removed from analysis
- shows separated cross sections after 1st iteration, and model fits for
iteration #2
- t-bins 1,8 look like outliers for sigT,L after iteration #1
- if only t-bins 2-7 included in fit, things are more consistent
- will have to see what this looks like after more study
Nacer
-----
KaonLT low Q2 event selection
- won't show any plots, since the meeting is getting long
- tweaking some cuts, changing the RF windows
- RF cut is very effective, needs fine tuning
- GH: since the RF cut is so effective, you can probably get by with a
combination of slightly looser PID cuts, to avoid cut-dependence to the
result
Nathan
------
Setting up PionLT Luminosity analysis
- in contact with Dave Mack regarding BCM calibrations
- DM did our first calibs from 2021 run, NH will follow up with him on run#'s
- Dave: if you can generate ROOT trees for him to analyze, it would be even
better
Weird hcana issue
- hcana suddenly did not know what replay variables were
- spent a day debugging
- restored things, but no idea what happened, as could not find any
difference from how things were set up before
Going through Richard's python scripts and finding what's missing
- will plan on meeting RT in early April, after his defense
Junaid
------
Setting up PionLT Heep study
- had some issues with lt-sep package
- merged cut files for 2021,22
- added ngcer
- updated RF time offsets
PID cuts
- HMS cal-etot-track-norm>0.7
- HMS cer NPE>1.5
- CoinTime, extremely clean, practically no randoms
- should be able to just place a cut around prompt peak, no random
subtraction
Next Meeting
------------
- Thur Mar 21 @ 15:00 Eastern/13:00 Regina
- PionLT will go first