Closed tyler-a-cox closed 1 year ago
Patch coverage: 100.00
% and no project coverage change.
Comparison is base (
ec4fcf6
) 97.15% compared to head (a63ee4a
) 97.15%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
This PR improves the speed of
smoothcal.solve_2D_DPSS
by usingnucal._linear_fit
to allow for additional filtering options. For the nominal inputs ofsmooth_cal.CalibrationSmoother.time_freq_2D_filter
, the newlu_solve
method is roughly 12x faster than the previously usedpinv
method.For other uses of this function such as full_day_rfi.ipynb and full_data_auto_checker.ipynb where filtering scale is finer than the typical
smooth_cal
run, the newlu_solve
method is significantly faster with comparable results.This PR also deprecates the
XTXinv
keyword argument in favor of a more generalcached_input
, which also stores intermediate data products computed byscipy.linalg.lu_factor
.