ran profiling on DC2 imsim data in , in ~215 seconds.
(very little association, so an under-estimate of performance)
discussion at Pipelines team meeting identified some DRP-specific measurements during ISR that we can remove. Otherwise no single big slow pieces. Think about minimizing IO for calibration frames/preload.
Colin: what about using multiple cores? Eric: good point, future work
DECam HiTS with fakes and Saha Bulge reprocessing both got held up by middleware changes (, , others.). Close to resumption now.
working on closed-loop comcam simulated out of focus images–mirror support loop now converges. now waiting for T&S to give some simulated pointings. hope to wrap this up with crowded fields, ready to integrate into T&S code. Next look at rotational dependence
working on analyzing AuxTel data. Translating zernikes into moving the secondary mirror for collimation
Krzysztof Suberlak : series of out of focus and in-focus images. trying to understand reproducability run-to-run.
trying to look at differential zernike as a function of hexapod movement, since don't have a good AT model
Eric: how big are the physical shifts that these zernike changes represent? Andy: about a mm shifts
Chris Stubbs asked about dependences on free seeing
Colin: how to think about the units of the zernike coefficients? Andy: for AT, we don't know how stable it is. don't know if scatter is mechanical hysteresis, or mirror shape, etc.
Eric: anything DM can help with? Andy: DM astrometry is working well. AT pointing is only good to 40" or so, so blind crossmatch is tough. Not using butler exactly as DM would like–trying to pre-calculate catalogs needed prior to observations. Next is masking & deblending donuts, unclear how much existing DM helps. Want to move T&S zernike code to galsim, but requires a version bump.
DP0.2 pilot runs are running, lots of data on disk–getting the V&V team geared up to look at it, prepare for full-scale processing. No results to share yet. Hopes to get Tim Morton's plot viewer onto the IDF
Eric: we do "QA" and V&V lots of ways–any sense of how this is going to proceed from a tooling process? Colin: using this as forcing function to explore/consolidate.