update ockham factor, remove hack, and don't submit all flashers

This commit contains the following updates:

- remove hack to get rid of low energy events in plot-energy since while
  writing the unidoc I realized it's not necessary now that we add +100 to
  multi-particle fits
- update Ockham factor to use an energy resolution of 5%
- update submit-grid-jobs to submit jobs according to the following criteria:
    - always submit prompt events with no data cleaning cuts
    - submit 10% of prompt flasher events
    - submit all other prompt events
    - submit followers only if they have no data cleaning cuts
- update submit-grid-jobs to place the nhit cut of 100 on the calibrated nhit

1 files changed, 1 insertions, 17 deletions

diff --git a/utils/dc b/utils/dc
index b551925..0c14053 100755
--- a/utils/dc
+++ b/utils/dc
@@ -26,7 +26,6 @@ import emcee
 from scipy.optimize import brentq
 from scipy.stats import truncnorm
 from matplotlib.lines import Line2D
-from sddm.dc import get_proposal_func, estimate_errors, metropolis_hastings, EPSILON
 from sddm.plot import despine
 from sddm.dc import *
 from sddm.plot_energy import *
@@ -372,28 +371,13 @@ if __name__ == '__main__':
     # so we can tag retrigger events
     ev['dt'] = ev.groupby(['run'],as_index=False)['gtr'].transform(lambda x: np.concatenate(([1e9],np.diff(x.values))))
 
-    # This is a bit of a hack. It appears that many times the fit will
-    # actually do much better by including a very low energy electron or
-    # muon. I believe the reason for this is that of course my likelihood
-    # function is not perfect (for example, I don't include the correct
-    # angular distribution for Rayleigh scattered light), and so the fitter
-    # often wants to add a very low energy electron or muon to fix things.
-    #
-    # Ideally I would fix the likelihood function, but for now we just
-    # discard any fit results which have a very low energy electron or
-    # muon.
-    #
-    # FIXME: Test this since query() is new to pandas
-    fits = fits.query('not (n > 1 and ((id1 == 20 and energy1 < 20)  or (id2 == 20 and energy2 < 20)  or (id3 == 20 and energy3 < 20)))')
-    fits = fits.query('not (n > 1 and ((id2 == 22 and energy1 < 200) or (id2 == 22 and energy2 < 200) or (id3 == 22 and energy3 < 200)))')
-
     # Calculate the approximate Ockham factor.
     # See Chapter 20 in "Probability Theory: The Logic of Science" by Jaynes
     #
     # Note: This is a really approximate form by assuming that the shape of
     # the likelihood space is equal to the average uncertainty in the
     # different parameters.
-    fits['w'] = fits['n']*np.log(0.1*0.001) + np.log(fits['energy1']) + fits['n']*np.log(1e-4/(4*np.pi))
+    fits['w'] = fits['n']*np.log(0.05/10e3) + np.log(fits['energy1']) + fits['n']*np.log(1e-4/(4*np.pi))
     fits['w'] -= fits['n']*100
 
     # Note: we index on the left hand site with loc to avoid a copy error