optimization of multiplicity, theta and gh cut #204
Conversation
maxnoe
changed the title
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## main #204 +/- ##
==========================================
+ Coverage 94.33% 95.78% +1.44%
==========================================
Files 58 59 +1
Lines 2738 2845 +107
==========================================
+ Hits 2583 2725 +142
+ Misses 155 120 -35
☔ View full report in Codecov by Sentry. |
maxnoe
force-pushed
the
faster_optimize_cuts
branch
from
62b8b8b
to
46ec05f
Compare
| Bins in reconstructed energy to use for sensitivity computation | ||
| gh_cut_efficiencies: np.ndarray[float, ndim=1] | ||
| The cut efficiencies to scan for best sensitivity. | ||
| theta_cuts: astropy.table.QTable |
There was a problem hiding this comment.
This doc is wrong: it's not the cuts being passed here, but the array of values over which the brute-force optimization scan is done, right? Also, for symmetry, shouldn't there be a similar input for multiplicity?
| ) | ||
|
|
||
| with tqdm(total=n_cuts, disable=not progress) as bar: | ||
| for multiplicity_index, multiplicity_cut in enumerate(multiplicity_cuts): |
There was a problem hiding this comment.
It might be useful to eventually adapt this to use scipy.optimize.brute internally rather than implement the grid-search manually, as that would make it easier to replace it with more efficient optimizers later
There was a problem hiding this comment.
Using the manual grid search, we can be a bit cleverer than brute force: make the cuts in the outermost loop possible rather than computing all cuts for all computations.
| ---------- | ||
| values: ``~numpy.ndarray`` or ``~astropy.units.Quantity`` | ||
| The values on which the cut should be evaluated | ||
| cut_table: ``~astropy.table.Table`` |
No description provided.