Optimization: roughly 40% of a call to ols_ramp_fit_single is consumed by 2 lines

[braingram]

https://github.com/spacetelescope/stcal/blob/main/src/stcal/ramp_fitting/ols_fit.py#L1657-L1658

These appear to be iterating through pixels. Is there a way to convert this to a vector operation or some other optimization?

Here is a plot of the time taken to run ols_ramp_fit_single generated with snakeviz (note the 40% number was calculated using line_profiler as snakeviz times don't always make sense, note that calc_slope contains a block for calc_slope with an equal measurement of time).

[zacharyburnett]

the link in the discussion is not pinned to a specific commit, so the line numbers have changed since original posting:

stcal/src/stcal/ramp_fitting/ols_fit.py

Lines 1897 to 1900 in 3fa4d42

	# Reset the initial False groups to be True so that the first False is now either a jump or sat
	# Because start was set to be the first True, the initial False values will not be included
	for pixel in range(npix):
	mask_2d[: start[pixel], pixel] = True

[braingram]

This can likely be closed since there is now a c version (and the python version will likely be removed).

I'm leaving it open for now but if anyone feels like it should be closed feel free to close it.