David Waterman

actually, probably not so bad as the `reshaped` list just rearranges the pointers to the data. However, building `jacobian` without consuming the values probably does double the memory requirement.

NB can track total current memory usage with psutil https://stackoverflow.com/questions/938733/total-memory-used-by-python-process but this does not change over the scope of that function. Possibly have to trigger garbage collection or something?

Memory checks in `dials.integrate` (https://github.com/dials/dials/pull/1221) might be useful here

Just to note here, it is difficult to actually predict the memory usage in refinement as it is not dominated by a single object. The Jacobian was a large part...

This is an interesting idea and I'll think about how best to approach it. There is similar analysis done when the experiments are parameterised. We could figure out which blocks...

This is a bit more broken than I thought due to https://github.com/napari/napari/issues/5111. I'll convert to draft

This might have been improved by #752

The full information about what parameters were refined is written into `dials.refine.debug.log`, but that does not help here. However, for a static crystal model this can be inferred if the...

I came across this looking at inconsistencies in geometry when importing MTZs from DIALS vs those from XDS (both from xia2 runs) in https://github.com/dials/dials/tree/import_mtz. Seems there is still much to...

> Re: BATCH header for MOSFLM above; looks like beam direction is opposite to what is claimed in MOSFLM documentation at http://www.mrc-lmb.cam.ac.uk/harry/mosflm/mosflm_user_guide.html#a3 i.e. > > ``` > /! > Y-axis...