search

icecube / skymap_scanner

A distributed system that performs a likelihood scan of event directions for IceCube real-time alerts using CPU cluster(s) and queue-based message passing.
5 stars 2 forks source link

Support realtime 2023a data format (`SplitInIcePulses`) #244

Closed mlincett closed 9 months ago

mlincett commented 10 months ago

With these changes, we set the name of the base pulse series to SkyScanBasePulses we support the change in the input pulses name occurred with the realtime 2023a format.

As part of the prepare_frames, a pulses_name argument is used as the source to copy the pulses from. When pulses_name is None, we use a map from the realtime JSON "version" to the pulses name ("SplitInIcePulses" for 2023a) and "SplitUncleanedInIcePulses" for 2021a. The latter is also the default.

In general, an arbitrary pulses_name is supported in extra_json_message(). In the future, this can be used to read an arbitrary series of pulses instead of the mandated one.

Out of simplicity, I have decided to copy over the pulses to a new "static" name rather than use it as an attribute of the reco object.

tianluyuan commented 9 months ago

@mlincett I merged in the latest updates from main and updated the baseGCD url to point to prod-exe. Now that the Dockerfile is also pulling in the baseGCDs from prod-exe this might be redundant, but perhaps it's ok for now? Otherwise I think this is good to go and I'll go ahead and mark it as ready for review.

Was there anything else you were working on?

mlincett commented 9 months ago

@mlincett I merged in the latest updates from main and updated the baseGCD url to point to prod-exe. Now that the Dockerfile is also pulling in the baseGCDs from prod-exe this might be redundant, but perhaps it's ok for now? Otherwise I think this is good to go and I'll go ahead and mark it as ready for review.

Was there anything else you were working on?

Thank you for picking this up! I believe it should be good to go.

(We will not have single-pixel tests for the new format but I don't see that as strictly necessary for now.)

Initially, I meant to rename the input pulse series name of the scanner to something neutral and different from "SplitUncleanedInIcePulses", but this would require replacing all the single-pixel test data (or finding some workaround). I guess we can ignore that for now.