Jul 31/24 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, Nacer Hamdi, Ali Usman, Alicia Postuma, Nathan Heinrich,
Muhammad Junaid, Vijay Kumar, Zach Sullivan
Virginia - Richard Trotta
Ohio - Julie Roche
CUA - Tanja Horn
JLab - Dave Gaskell (last half)
Nathan & Zach
-------------
- Zach had a problem w/ python script in ALMA9. Nathan and Junaid have not
been able to fix it
- Junaid thinks there has been a change in python syntax, but can't figure it
out, stuck for now
- *NB* Richard will take a look, probably an incompatibility between versions
of python and pyroot that needs to be fixed
- the same problem is likely to also affect Richard's analysis, so this is a
high priority for fixiing
Junaid
------
PionLT Heep Study - 9 beam energies
- investigating two issues
1) some EM distributions are broad
- for all 2021 data, EM peak is sharp
- for all 2022 data, EM peak is broader
- checked raster parameters, there are multiple raster calibs for 2021, but
only one for 2022 that Mark Jones posted, based on run 14809 @ 7.950 GeV
- no raster issues were noted for 2022 run, so presumably a single raster
calib is sufficient
- GH and Richard: *NB* try looking at EM versus raster_x, raster_y
- if there is a problem, EM will be correlated (tilted line) with respect
to these variables
- the plots that Richard recommends are:
2D: (S)HMS EM vs delta
(S)HMS PM vs delta
SHMS delta vs HMS delta
raster_x/y vs PM/EM
raster_x vs raster_y
1D: BPM target x/y
2) calculation of Heep COIN error
- checked error calculations, was using weighted SIMC yield instead of raw
SIMC counts in the error calculation
- this makes a huge difference, MC errors now much smaller
Nacer
-----
Heep error checks
- worked with Junaid, found the problem by checking numbers step-by-step
Comparing Heep models in progress, to establish uncertainty in the MC
prediction
Comparison of Nacer and Richard Heep yield ratios
- it is important that we understand why the Heep ratios deviate from unity,
and perhaps understanding the difference between Nacer and Richard's ratios
will shed some light on this
- *NB* Richard will re-run some scripts to recheck and understand the
discrepancy
Alicia
------
BSA paper update
- B-G Yu happy with latest changes, paper is ready for submission
- filled out fields on PRL server, will be submitted later today
Ali
---
piDelta BSA analysis
- Garth asked to show BSA plots for all settings
Q2=2.1, W=2.95
- 2 t-bins, data very smoothly follow sine-curve fits
- BSA amplitude ~0.12, small errors
Julie: *NB* can you please make residual plots of data-fit? The fit actually
looks too good, Chi-square<1
Q2=3.0, W=3.14, 2 t-bins
- errors a bit lager, fluctuation about sine curve gives Chi-square<1
Q2=3.0, W=2.32, 1 t-bin
- right SHMS setting had mroe stats than left, so errors vary with phi-bin
Q2=4.4, W=2.74, 1 t-bin
- errors a bit bigger, BSA amplitude ~0.15
Q2=5.5, W=3.02, 1 t-bin
- errors bigger, Chi-square clearly <1 for ALL plots
- errors are statistical only, systematics not yet included
Julie: *NB* can you please send me exactly how you calculate your error bars?
- maybe your errors are over-estimated, do not follow 2/3-rule
Next steps
- work on systematic uncertainties in progress
- syst unc due to 2 types of fits, 2 studies
- syst unc due to cut-dependence, 2 studies
Richard
-------
KaonLT LT-iteration fitting
- updated sigT functional form
- added t-dependence, t-trend of Ratios more uniform now, but more iterations
needed as Ratios ~0.4
- Tanja found an old Fpi-1 log entry by Jochen Volmer, dated Mar 16/2000:
- he tried a functional form for sigT of the type:
a+b*(W-Wc)+c*(Q2-Q2c)+d*(W-Wc)*(Q2-Q2c)
- this seems like a very nice function that could describe the variation of
the cross section across the diamond, where the diamond center is at
(Q2c,Wc)
- tried fitting it as sig_UNS at high,low epsilon, just to see what results
- b & c terms have significant t-dependence, d more constant
- curiously, b & c have opposite t-dependences, but they're in orthogonal
directions on the diamond, so not sure if this results in cancellation or not
- looks promising, GH suggests to try this form for sigL and sigT, perhaps
could use a linear t-dependence for b & c, given their strong t-dependence
Vijay
-----
Low Q2 PionLT volume cut systematic uncertainties for L,T,LT,TT
- varied delta, xptar, yptar by +/-10%, 6 cut changes in total per spectrometer
- SHMS-delta
- no variation in sigL,sigT
- SHMS-xptar,yptar
- sigL varies 0.1-0.8%, depending on t-bin
- sigT varies 0.1-1.2%
- HMS-xptar,yptar
- variations are similarly small
- HMS-delta
- variations are much larger, sigL 28-90%, sigT 40-90%
- *NB* we need to better understand what is going on here, unexpected
sensitivity
Dave: *NB* you might need to increase the SIMC event generation limits
- plot delta,xptar,yptar for +10% limits and check
- another thing to keep in mid is that the HMS delta matrix elements are less
understood past 8%, so the +10% cut is past the well-understood acceptance
- 7.2% (-10%) cut is more reasonable, but your results look similar for
8.8%
- *NB* suggest to look at Vladis Tvaskis' PhD thesis:
- Vladis and Henk found that a small correction to HMS-delta was
necessary, affected things at the few percent level
- if this correction is not implemented in the analysis, then we should
look at adding it
- *NB* please make plots of focal plane and physics variables for delta-10%
cut, so we can see if there are any large discrepancies between them that
could explain the unexpected variation
Next Meetings
--------------
- Note the special time:
Wed Aug 7 @ 15:00 Eastern/13:00 Regina
- KaonLT will go first
- Dave and Julie can't attend then
- Garth will be at Exclusive Reactions meeting @ Trento, will connect from
hotel room
- Thur Aug 15 @ 15:00 Eastern/13:00 Regina
- Dave will take notes
- PionLT will go first