possibility:meteor.diffmap --internal

[tjlane]

Our code in meteor provides everything that is in rs-booster.diffmaps, but in a tested/production format except the "internal" difference map functionality.

I'd propose we just add the ability to make internal diffmaps to meteor, so that we can fully deprecate the corresponding code in rs-booster. This will give people who want to make internal diffmaps access to meteor's goodies.

Then, we could even parrot all the meteor.* scripts as rs.*, so there is a seamless user interface. We should think about how to do that in a way that ensures no conflict if people install different packages from the rs-station universe.

Feedback wanted @DHekstra @kmdalton @alisiafadini (anyone else?)

Cross-ref rs-station/rs-booster#18

[kmdalton]

i've got no qualms with this as long as meteor doesn't have a lot of heavy dependencies. it looks to me that skimage is the only one to be potentially concerned about.

[tjlane]

tbf we could excise the one (denoising) function from skimage that we use and eliminate that dependency -- it looks straightforward

[DHekstra]

• regarding using meteor functions to run rs.diffmap etc: yes. I'd like to keep the rs-booster functions around for now but would be OK with re-routing the rs entry points.
• specific to internal difference maps, I note that a typical complication is that there are systematic differences in scales between the two halves of reciprocal space. so far, the best we have come up with is to use either local scaling (as implemented in SOLVE or in VALDO (currently buried here)). Open to suggestions. TV might well be super useful here. (side note: would be very interesting to have a TV loss during scaling).
• is it possible to supply meteor.diffmap with native phases rather than a native model (see the --refmtz flag in rs.diffmap)? the rationale is that we've seen significant variations in difference maps depending on, for example, the treatment of solvent.
• also pinging @JBGreisman here for his perspective.

[kmdalton]

I made a new conda env and installed rs only using pip.

pip install --dry-run rs-booster
...
Would install contourpy-1.3.0 cycler-0.12.1 fonttools-4.54.1 kiwisolver-1.4.7 matplotlib-3.9.2 packaging-24.1 pillow-11.0.0 pyparsing-3.2.0 rs-booster-0.1.2 seaborn-0.13.2

pip install --dry-run meteor-maps
...
Would install imageio-2.36.0 lazy_loader-0.4 meteor-maps-0.3.0 networkx-3.4.2 packaging-24.1 pillow-11.0.0 scikit-image-0.24.0 structlog-24.4.0 tifffile-2024.9.20

[tjlane]

I am splitting out a few things to keep the discussion here focused on internal diffmaps.

• is it possible to supply meteor.diffmap with native phases rather than a native model (see the --refmtz flag in rs.diffmap)? the rationale is that we've seen significant variations in difference maps depending on, for example, the treatment of solvent.

Yes, see #77.

I made a new conda env and installed rs only using pip.

pip install --dry-run rs-booster
...
Would install contourpy-1.3.0 cycler-0.12.1 fonttools-4.54.1 kiwisolver-1.4.7 matplotlib-3.9.2 packaging-24.1 pillow-11.0.0 pyparsing-3.2.0 rs-booster-0.1.2 seaborn-0.13.2

pip install --dry-run meteor-maps
...
Would install imageio-2.36.0 lazy_loader-0.4 meteor-maps-0.3.0 networkx-3.4.2 packaging-24.1 pillow-11.0.0 scikit-image-0.24.0 structlog-24.4.0 tifffile-2024.9.20

Let's continue discussion over dependencies in #78.

[tjlane]

@DHekstra if you want to support this internal diffmap effort, an example dataset and desired (approximate) normal diffmap output would be v useful!

[DHekstra]

I'll leave that to @hkwang when he has some time (beam time this week!).