# Optics Meeting Apr 3 2024 1100AM ET

From Moller Wiki

Back to Main Page >> Optics Meetings

# Agenda

- Fitting Update [Vassu]
- mySQL database update [Bill [1]]
- Radiative cuts and hole ID (Tyler)

# Attendance

- David, Paul S, Tyler, Bill, Vassu, Sayak, Andrew, Bill, Klaus, Hanjie, Zuhal, Juliette

# Minutes

- (Bill) MySQL database update - working on Table 2, which will be detector specific properties from construction (leakage currents, efficiencies, gains). Suggested detector naming convention (UVA-1, UVA-2, SBU-1...). Leakage currents stored per GEM foil. Probably don't need to have separate naming convention for individual GEM foils, as won't be able to swap foils between GEMs. Leakage, efficiency an dgaion will be each stored for each of the 10 HV sectors for each GEM. Will check with UVA colleagues for input on this scheme.
- (Tyler) Has implemented radiative-tail automated cut algorithm, tested for case of pass-3, single 12C target. It makes 1D radial location histogram for each hole (using MC truth information for relevant hole ID), defines region +- 50 mm around bin with max counts, gets RMS and mean r for these hits, and cuts rad. tail by rejecting hits with R > mean + 2 RMS. Seems to work very well. But how to identify hole locations without using MC truth info? Testing use TSPectrum2 class find holes by looking at 2D intensity pattern; when told to look for a max of 21 holes, for this test case it find correct hole's image central location each time. Seems promising. Isolation of nearby holes might be improved by looking also at phi'. could Tyler suggests factorizing the code so that the Hole ID location (and perhaps the radiative-tail cut?) are before the fitting algorithm is invoked, in independent procedures. We also need to look at what happens for real data where each tracking run will be at only one "rotation" angle, so will only be able to "see" a subset of all the sieve holes. What about Mollers polluting 12C elastics? For Pass 1 and Pass 2 energies, a simple cut to remove large radius hits should cut the Mollers; pass 1 may be dicier (Vassu). We don't yet have a simple way to get simulated data that represents just the values from a single "rotation". However, we are almost there, since the GEM active areas are now in Bill's version of remoll. Bill just needs to get the exact rotation angles for the intermediate positions from Vassu (Vassu will send to Bill). Rotations are labelled A,B,C and D, where A and D have the GEMS centered on the centers of open sectors.
- (Vassu) Generated simulations with 5 different (individual) 12C foil positions. Do separate fits for each foil position, including good radiative tail cuts (2 MeV delta E). Now see the individual theta fit parameters (coefficients for each term) that vary smoothly with v_Z of the 12C foil (linear or quadratic variation). Good! When plotting individual theta residuals for given v_Z, the residuals no longer show bias to one side. This works even with reduced fit that doesn't use phi or phi. Now we are back to the question of how we can extract v_Z from the data. David suggests revisiting (now with radiative cuts applied) a fit of v_z as a function of the 4 GEM variables. May be highly correlated with theta, but perhaps not 100% correlated. Might try for 12C elastic data (with 5 discrete foil positions) or perhaps with Moller electrons (from lH2) at 5-pass, where have continuous v_z distribution, but most of the data will have the Moller kinematic correlation of E' with theta.
- (Vassu) Looking into using a ML neural net as an alternative fitting routine; no results yet, very CPU intensive at present compared to the chi-squared fit.

# Meeting link information

See email invitation, or contact David Armstrong, Kate Evans, or Jennifer McAllister for Zoom link