Choose timezone
Your profile timezone:
Powered by Indico
v3.3.6-pre
How do we (practically) incorporate an E field uncertainty into the event selection
There are limitations to the maps we have in hand (Ajib, including x direction, extrapolation, and format), but assume for now we accept those limitations.
Assume that once we can represent the effect on the event selection, we can propagate it to the analysis (using ‘unisim’ approach, or using a thoughtful parameterization in relevant kinematics)
Default: Option 1: Take a new map in the MC and resimulate events
=> Consensus to use this approach. Need to decide and discuss details.
=> Need to also discuss which variation(s) we want to consider.
=> Kendall to discuss these plans with Mike Mooney
Variations?
Timescale?
Method?
Concerns?
Option A) Use an alternative map (Ajib’s)
We have spatial distortions for certain regions
Some regions in Y can maybe be neglected for xsec-- use default SCE values from Mike and/or linear extrapolation.
Discuss: Z distortions
Option B) CRT if available - Richie?
RD: The tracks are parallel/colinear, then the measurement of crossing track will come back with nonsensical answers (both negative and positive swings, with centered correctly. Hurts precision of the study. Compared to Ajib, tracks perpendicular to plane-- didn’t see it.
=> Next steps for Richie-- run a little more MC to see if the answer is sensible, and more data-- if the results are similar can consider it fair game to be used
<getting a baby>
=KM MM
Distortion near surface ok
TJ: AJibs maps measure distortion with cosmics → better map
KM: which kind of variations do want to consider? Difference in maps == 1 sigma variation? Important have an end to end solution in a first step
We also need to propagate variations in the E field components
3 components Ex,Ey,Ez . measured Ex in central region of TPC
=> Kendall get Ajib’s slides and get proposal written here for the few options
Ignore the Ex/Ez pieces (Ex as a function of x, y,z)
RD: If you plot those, then get two big circles-- which DO show that dependance-- so then that’s scary because there’s some radial dependance
!! and this is a limitation for non-beam analyses!! Ex is Ex for the beam muon study.
Or use default map
Or use uB method-- but not trivial work
!work alert! Format has to match
Option C) Use default (Mike et al?) map?
Q: How long do we think it will take to get the nominal map (few months? More?)
Current knowledge of timeline for an E field map?
GC: understanding from Mike, is that it will be a while to have new maps available. Can try to push him to produce something faster. => Kendall to email to confirm this.
Timeline for the production?
TY: This production does have a slightly modified SCE map, it’s small. JC: difference is the interpolation issue shown last time? TY: Yes, only around the cathode.
=> Not discussed, come back to this after wards
Alternative option 2: Weight MC to correct for a new map?
AP: For Efield Mike’s maps has fractional variation from Nominal value, so if we change nominal value we will get a different Efield map.
Downside: I think spatial distortions are fixed numbers as a function of coordinates so does not change with Efield.
Q: But, we have to do this anyway right now? Any plan for correlated uncertainties , and what timescale?
Crazy idea: We have done mock data challenges with cross section models by weighting final state kinematics. If we could make a space of relevant true variables, could such a system be created for E field? Let’s discuss!
Alternative option 3: Approximate effect on energy
This may be similar to 2). This depends on how and where the simulation uses E field. Would require checks with regenerated MC to understand the limits of this assumption
Main uses: E field changes recombination, E field changes drift velocity. How do those affect calorimetry//tracking? What else?
Would also have to consider how E field changes would modify the energy estimator -- separate correction//treatment?
Are track distortions purely covered by spatial map changes?
Major concerns:
AP: I think spatial distortion has a big effect on energy measurement (possibly higher effect than recombination) as it effects ‘dx’ in dQ/dx (or dE/dx). If we want to just vary Efield we can can replace EfieldX in Mike’s map using my map and keep everything else the same. I guess we will need his permission if want to modify some plots in his map although anyways we will need to produce duplicate root files, without affecting his.
JC: I predict the reconstruction effects are large-- E field distortions will move particles out of slices
=> Next step: articulate how E field issues couple to reconstruction and efficiency
i) How will we assess reconstruction, efficiency uncertainties (in data-data or data-MC?)
ii) How big are the differences?
Alternative option 4: Your idea here!