How are things going at NERSC? Is the image differencing up and running?
PhoSim v3.6 is out, and promises multi-threading - so ought to be our route to getting the Run 3 “long jobs” completed. How’s this looking so far?
Tom’s early results based on Twinkles Run3 (and four threads), https://github.com/LSSTDESC/Twinkles/issues/420
However, Tony’s tests at NERSC (and using many more threads) show diminishing returns.
Tom: After running a number of single-threaded and multi-threaded phoSim tests, the results are qualitatively similar but not identical. FITS images look about the same although, interestingly, one image has an artifact that may be a reflection or even a meteor. A more quantitative validation is in order, possibly beginning with image subtraction to look at residuals, pushing the test images through DM to see what sources it finds, a look at the background light, etc.
What’s the conclusion about emulating calibration?
Updates on error model work:
The SL WG want to include lensed supernovae - let’s make that happen.