Skip to content

CalibrationMapReporter optimization

Reed Essick requested to merge calibreporter-optimization into master

NOTE this should be merged after !13 (merged) has been merged. Otherwise we may pick up changes from !13 (merged) that were not intended because this branch is based off "managing-branches-is-hard".

The changes here are contained to

  • fixed a typo in calibration.py: interp_logcdf -> interp_cdf, which better reflects what is actually stored
  • changing how CalibrationMapReporter stores information on disk so that full interpolation arrays are written. This means we do not have to re-compute the arrays when we read the object back from disk, which should make things run faster.

I tested this with the same stuff I used to test !13 (merged), specifically tests/sanitycheck_calibrate and the results appear to be unchanged, meaning the I/O works as expected.

Edited by Reed Essick

Merge request reports