Here today, gone tomorrow: searching for transient sources in astrophysics

Coming from a particle/nuclear physics background, when I started working on IceCube one of the bigger mental adjustments I had to make was to get used to the idea of transient sources.  When an accelerator is running, its particle output is more-or-less constant.  Not so with astrophysical objects. Many (not all) of the most interesting astrophysical objects vary considerably in output (by a factor of 10 or more), over different time scales.  Depending on the source, periods of increased emission may or may not repeat, on either regular or irregular time scales.

In fact, IceCube's most statistically significant signal, from the source TXS0506 was based partly on a search for transients, where we found a transient lasting about 7 months, as I discussed in a previous post. Transients can come over a wide range of length scales, from millisecond long bursts of radio waves called Fast Radio Bursts, up to sources that probably change on time scales longer than we have been observing them.     

In IceCube, time-varying sources add additional complexity to source searches, since searching over a wide range of time scales, degrees of repeatability, etc. can lead to a large increase in the number of trial factors: the more ways you slice and dice the data, the more likely you are to get a statistically significant result.  It is critical to keep track of the number of different observations (positions in the sky, possible pulse start times and lengths etc.) to know if an observation is really statistically significant.  For some sources, we can use radio, optical or X-rays to tell us the best places to look, reducing the number of trials factors

IceCube has recently released a paper on a search for time-varying sources.   The paper included two types of searches.  The first was an all-sky search that looked for emission on different time scales, from about 1/10 second to 100 days.  This suffered from a large trials factor, for the reasons noted above. 

The second search examined one object of particular interest: 3C279, which is a quasi-stellar object.  Despite the 'quasi-stellar' name, it is a distant galaxy containing a massive black hole which powers the emission of powerful particle jets, which were recently imaged  by the Event Horizon Telescope - the image above is from their web page.  3C279 is known to exhibit strong variability in radio, optical and X-ray emission.  These factors made it an attractive place to search for neutrino emission, despite the long distance (5 billion light years).   We used gamma-ray data (using photons with energies above 100 MeV) from the Fermi telescope to select time periods when 3C279 was particularly active.  By focusing on the active periods from a single source, we were able to make a much more sensitive search.

Unfortunately, we did not find anything using either approach.   We are, however, reducing the number of ways that Nature can hide the cosmic-ray accelerators that we know must exist.   We use the non-detection of neutrinos to put limits on how 3C279 could work as an accelerator.