search

ezmsg-org / ezmsg

Pure-Python DAG-based high-performance SHM-backed pub-sub and multi-processing pattern
https://ezmsg.readthedocs.io/en/latest/
MIT License
9 stars 5 forks source link

BandPower - Fixes #44, #45, and #46 #53

Closed cboulay closed 4 months ago

cboulay commented 8 months ago

I don't expect this to be merged without major revisioin, but I hope it's a good starting point for discussing adding BandPower to sigproc.

griffinmilsap commented 5 months ago

Whats the status on this PR; I see it's still draft -- is this something you're building with/working on?

cboulay commented 5 months ago

Yeah I still want to work it out a bit. Using the revised Window will help. I've had many more hours with ezmsg since I opened this PR and I'm confident I can make a cleaner implementation, or at least one that's more consistent in its naming conventions and parameterization.

cboulay commented 4 months ago

Refactored everything using generators. I also moved modify and aggregate into locations that I thought were more natural / discoverable.

I don't have great ways to test spectrograms and bandpower. For bandpower, I just checked that the ordinal magnitudes were the same as the ordinal magnitudes of the specification in the raw signal simulation.

In my tests I have functions called _debug_plot. I encourage you to do interactive debugging on the unit test and then call the commented out calls to _debug_plot.

griffinmilsap commented 4 months ago

this is massive. it looks to me like this affects synth, window, axisarray, and introduces a few new features all together. The feature set it adds looks REALLY nice to have, but there's only so much time in the day. I'm going to review this by merging it into dev, running some of my existing software, and seeing if anything breaks; but just as a heads up I don't have the bandwidth to review every line of this PR and expect we'd have some followup fixes as necessary.

cboulay commented 4 months ago

Yes of course, I'm available to make updates as issues arise.