Energy Estimator to Do List


How to generate new BPF fits:

Filling in the TH2s to make the spline fits:

(where is Erica's code? hopefully in the DEVS repo...)
(how to run Erica's code...)

step 1 - run that code...

Generating Splines:

  • The code to generate spline fits exists in the DEVs repository under: devs / trunk / users / mbaird42 / BPF_numu_analysis / spline_fits
  • Look for the script and follow that model. This will dump a bunch of fits into 2 text files. Make sure to check that each fit looks good (see the list below):
  • Take the output of these 2 text files and copy and paste it into CAFAna/Vars/BFEnergyConstants.{h, cxx}. Some manipulation of the text may b required but it should be straight forward. If you didn't change the names of any of the fits, the information in the header file probably didn't;t change.
  • If you changed the names of some of the splines, you may also have to change that in the std::vector<TSpline3>s defined in the header file.
  • Compile and test the new code to make sure that your fits aren't junk. NOTE: We should probably write a standard testing macro that make some suit of plots...

Things to look for when checking that the spline fits are good:

  • does the spline go through (0,0) or at least isn’t negative for some positive x value?
  • does the spline look like it will b linear way out beyond the data?
  • does it go through the most probable values in each vertical strip of the TH2?
  • does it look like the fit to the gaussian for the vertical strips and the switch over to the median value is sensible?
  • NEED to explore the sensitivity of fits to things like number of knots! (not sure how best to asses that...) Need to come up with a way to figure out if a given fit is good or not...

Studies and physics stuff:

  • do a side-by-side speed test of our energy estimator vs. kCCE
(after energy fits etc are done)
  • redo the quantile boundary histos for FHC with CVN-prong cuts (both with and without the PID per quantile cuts) - make sure I am using the official histos (especially for RHC...)
  • determine the quantile boundaries for RHC
  • determine PID cuts per quantile for CVN-prong
  • redo quantile boundaries with the per-quantile cuts

Erica's tasks:

  • fits by CVN selection

Andrew's tasks:

  • turn some cranks...

Other Studies...

  • Probably a study is needed as to how best to select showers. Should we take the max of the electron/photon/pi0 IDs? Should we sum them for a single "showerID"?
  • Look into whether or not it is useful to try to ID prongs as neutrons. Then maybe add all of that functionality for selecting and computing the energy of neutron prongs (individually and in the general HadE functions) - this is probably more important for RHC
  • Investigate mysterious 2nd proton population

To Discuss

  • should we code in the MC muon length correction that Matt Strait came up with (probably yes...)
  • should we use those weird hardcoded high.low gain calE corrections?

Answered discussions:

  • trackoverlapE - how do we use it with the weighted prong hits so as to not double count?
  • should we fit muons with track calE or prong calE?

Ans: (for both of the above) Use calE for muons, subtract off the trackoverlapE from the muon track and put into garbage bin, use weighted calE for other prongs and just let the MC fits get it right on average