...
Attendees: Andy Lawrence, Roy Williams, Terry Sloan, Cosimo Inserra, Stephen Smartt, Sara Casewell, Dave Young, Micheal Fulton, Julien Peloton, Ken Smith, Jakob Nordin, Gareth Francis, Matt Nicholl, Eric C Bellm, Stelios Voutsinas, Meg Schwamb, George Beckett, Nic Wolf
Apologies:
Notes from discussion
...
Broad overview of classification in FINK
In Fink all alerts are processed by science modules where each module is focussed on a specific science objective.
Modules can be cross-match, filters, or more complex, even ML
Most modules are provided by the community
Alert classification is produced based on the module processing.
Most classifications are matched with known objects or solar system objects.
In June 2021, 45% of ZTF alerts are unclassified ie unknown.
Use the classification process to identify candidates for follow-up
Can of course focus on the unclassified objects instead.
Adapting user code to work in the cloud-based environment often needs to be done by fink team
Trainings sets not often representative, looking at active learning to remedy this.
Combining classification labels is being investigated
Multi-messenger science with AMPEL (Jakob Nordin)
...
)
Will cover topics discovered during Ampel development cycle and what we considered doing.
Until now reproducibility unnecessary in astronomy due to instrument growth but after LSST not sure if that will still be case.
Most alerts are faint and hence junk
Reusable software has not been a priority in science, particularly due to PhD cycle, but with projects running for 10+ years this can no longer be the case.
Reproducibility hard to do due to the black box (isolated) nature of the various steps in studies/pipelines.
Users create an analysis schema that runs across the four different information tiers in Ampel.
Users prefer a UI but the Ampel approach is analysis(code) focussed
Q Roy to Julien - How good are your classifications? How many false +ve/-ve?
Julien answer - difficult question to answer, using TNS since Nov 2020 about 300 candidates have been reported. Half of these were spectroscopically followed-up. More than 80% of these were true SNs. This is a good result but this not the only thing to look at. Now trying to make reliable detection faster. Currently have to wait 7 days but would prefer 3 days
.Jakob asked Dave - Sherlock is very nice, but how do people see maintaining and updating software over next 15 years.
Dave noted default algorithm now sits with code (individual users can adjust, but default is version controlled). When you run Sherlock, the version of the classifier is stored for reproducability.
Jakob asked whether Dave would be about to update it
Dave hoped so, but believed someone could be trained up to manage it in around two weeks, though noting code documentation needs to be improved.
Dave asked Julein Julien if Fink classifier is based on light curve data alone.
Yes, for Type 1a Supernovae.
Dave concerned that Type 1a SNe cannot be classified with just two or three data points
Julien agrees, but believes representative training set can help.
In chat, Q to Ken from AndyL, “if Cassandra is widely used in industry, how come we don’t know how to use it? Is it because in Industry its used internally, not by independent users?”
Andy asked Ken about Cassandra and noted concern that not sure how users will interface with Cassandra. Can we learn from Facebook use?
Matter of choice of technology, and past experience of relational databases. Use of Cassandra has been learned from scratch.. Facebook, etc. also use MySQL, etc. Ken has been sampling community use of databases, to see how well supported Cassandra might be. Cassandra just means outside of database. Could be file system, or something else.
Andy believes need to consider user requirements in more detail. Ken asked whether Dave had a sense of how many objects could classify per second (the Q in Zoom chat “ do you have a feel about how may transients Sherlock can classify per second?”)
Dave noted recent speed tests, which achieved ~10k per second.
In zoom chat Q Eric to all : “most of the examples we’re discussing here are the explosive transients. Is that reflecting the actual or desired user communities of Lasair, FINK, Ampel? What are the ambitions ( or not) to support variables/AGN/solar system science?”
Eric noted focus on extragalactic transients, which is a mature field, but asked whether there was any ambition/ interest in supporting solar system, variable star, or AGN communities?
Meg noted that this is her interest in Lasair, and ability to exploit Cassandra for Solar System alerts. Ambition to create Sherlock peer to classify solar-system alerts, and intent to pursue funding for this.
AndyL notes same issue for variable stars, but believes Lasair should be focused on Transients. Mistake to try to reverse engineer for other applications. Integration with DRs and RSP should allow users to undertake variability science, along side Lasair-focused transients. This is an unsolved problem.
Sara stated in Zoom chat “There is interest from the White Dwarf community in using Lasair to look for drop outs/eclipses in WD lightcurves.
Roy noted difference magnitudes in ZTF make it difficult to compute proper variables.
Eric noted Rubin would provide more rigorous forced photometry, which would help.
Meg noted less urgent need for turn-around for solar-system science. Next night is good enough.
Answer from Ampel side from Jakob in Zoom chat - “Our current user base mainly involve extragalactic groups, so AGN modelling is include there. So it is not that we do not want to do it, but we do not have a lot of feedback from eg solar system groups regarding their LSST needs. would be happy to get that, though.”
Stephen noted intent to help solar-system find objects in Lasair, but then link that to the DR light curve in Rubin, as useful.
JulienP Q in Zoom chat : “ For Fink: the LSST-FR community is, for historical reason highly focussed on explosive transients (incl. SN and MMA). But things change. We now support more and more the SSO science, with a large group of experts joining the team. However, we clearly lack of variable star experts (although we would love working on this as they seem to make most of the stream).”
...