Fix Version/s: None
What, if any, are the driving use cases for access to alerts in < 5 min
Another comment, for future reference: #tvs slack Tue Jul 9, Rachel Street has posted a poll about the scientific needs for rapid access to alerts, the results of that poll would inform this study.
Zeljko Ivezic points out that Sara Webb's talk at Hotwired 6 is titled "Science with DWF: The Second to Minute Transient Universe", which would also inform this study. I can't find any current papers by Webb but here are some on the DWF:
Probing the extragalactic fast transient sky at minute timescales with DECam https://ui.adsabs.harvard.edu/abs/2019arXiv190311083A/abstract
The Deeper Wider Faster program: chasing the fastest bursts in the Universe https://ui.adsabs.harvard.edu/abs/2018arXiv180201100A/abstract
Jeff Cooke's talk on DWF from the STScI symposium might also be useful to review https://cloudproject.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=0b814147-17be-478d-b56d-aa3b00e06ff4
MLG Note to Self – in Lynne's talk on the SSSC needs for alerts brokers at the CBW in June 2019, slide 4 says that a needed broker capability was "API or other 'push' notification (e.g., trailed detections) [minutes may count]". Follow-up on the SS use case for alerts on very short timescales.
Work is proceeding on this investigation at the draft (potential dmtn) document:
Early draft of the potential DMTN cleaned up and sent to Eric and Leanne for initial feedback.
A draft document containing some scientific use-cases of <5 minute alerts currently lives in this private lsst-dmsst repository: https://github.com/lsst-dmsst/alert-latency
To summarize: at the time of writing that document, it seems that most alerts-related science goals would be satisfied with a 5 minute alert timescale, but that the quickly evolving fields of fast radio bursts and gravitational wave events – and time domain astronomy more broadly – may soon provide more robust motivation for a 60 second latency.
Since there is currently no tabled proposal to change OTT1, there is no reason to improve this draft to the level of a real DMTN. If that changes in the future, a new ticket can be made to resume this investigation (and possibly add more technical aspects to the cost/benefit of longer latencies, if needed).
Noting here, for future reference: #tvs slack discussion Tue Jul 2, suggestion that low-latency alerts from ToO observations important because "jet-powered (http://arxiv.org/abs/1705.10797) and/or proto-magnetar-powered (http://arxiv.org/abs/1508.07939) UV (and X-ray) transients from binary neutron star mergers are predicted to be detectable only up to ~1000 s following the merger." Om Sharan Salafia
But Eric Bellm points out that it's not yet clear "whether the 60 second latency could be achieved for rapid TOO observations in any case: in the standard survey we expect to have to do a fair amount of pre-caching of slow database queries to get alerts out in time, which is possible because the next pointings are known ahead of time. Potentially for a rapid TOO there wouldn’t be time for those to be executed before the images are taken, which would impose some further delay."