July 3/25 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, Nathan Heinrich, Ivan Zhenchuk, Muhammad Junaid, Nacer Hamdi JLab - Dave Gaskell Virginia - Richard Trotta CUA - Tanja Horn FIU - Pete Markowitz Junaid ------ Q2=3.85, W=2.62 LT-sep - now using functional forms from Tanja's thesis Eqn 5.9 - initial parameters are same as Table 5.1 Final Values - ratios using this parameterization are not surprisingly worse than with the Bosted global parameterization - high eps: R~0.7 at low -t to R~0.1 at high -t - low eps: R~1.0 at low -t to R~0.2 at high -t - we expect these ratios to converge rapidly to near 1 after a few iterations, the important thing now is the functional form that is tried - still two scripts to write for LT-separations: - SIMC weight recalculation script - functional fits to LT-separated cross sections - also still doing checks on correct error calculation method Nathan ------ PionLT deadtime investigations - had a conversation with Dave last week about the unexpected time shifts that need to be corrected - has all scripts working now to check if there is a difference if TCOIN.param is called, or ONLINE params - can replicate timing shifts between: - explicit call TCOIN.param and implicit call standard.database - explicit call ONLINE param and implicit call standard.database - this means the implicit one is overwriting the explicit one - the weird timing shifts must be coming from standard.database - further to this, Nathan went through with Dave and identified some timing cut parameters that are too tight - if no good time is found within the window, it might be grabbing a garbage time instead Nacer ----- KaonLT Q2=0.50 Lt-sep - also moved to Tanja's thesis parameterization - everything is the same as Junaid, except that mK substitutes for mpi in TT denominator - no plots to show yet, but ratios are generally similar to what Junaid showed, i.e. a decreasing ratio with -t while the Koltenuk model had flat ratios near 2 - discussion about the calculation of errors - MINUIT is the "gold standard" that all errors should be compared to - Nacer is using curve_fit in python https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.curve_fit.html - Dave has confirmed on his own that using absolute_sigma=true in Python gives errors very similar to MINUIT - example call used by Nacer: curve_fit(fit_func, phi_all_fit, xsec_all_fit, sigma=err_all_fit, absolute_sigma =True, p0=[0.04, 0.16, 0.00, -0.01]) - Garth's quick script using PHYSICA fit - Dave agrees that "E2" error is the one to use, agrees well with MINUIT Ivan ---- KaonLT BSA analysis - Q2=3.0, W=3.14 K+ sample that was missed last week - some phi bins are ~2 sigma away from zero, BSA~0.1, a bit bigger than the other K+ Q2 settings, of course statistics are better too - only one bin covering full t-range - Q2=0.5 pi+ 5 t-bins - statistics are fairly good - BSA is somewhat larger for highest t-bin [0.067-0.11] - Nacer: some of the BSA plots are not left-right symmetric about phi=0 - Dave: this is likely statistical fluctuations, not new physics - *NB* Garth: Ivan should take a look at pi+n event selection, PID systematic uncertainties Richard ------- KaonLT high Q2 LT-sep - ran new functional forms for all settings - Good News! the new functional forms give reasonable results for ALL kinematics - adjusting pion subtraction scaling for bad t-phi bins - previously was using a common background subtraction factor for all settings - now applying a separate factor for each t-phi bin, as Nacer does - this has a significant impact for some bins, as the factor varies from ~0.85 to ~0.2, depending on the bin - i.e. the pion background was significantly over-subtracted for some bins previously - the factor is determined by equalizing the integrals in the 0.88