Searching for the fastest explosions in the Universe

The Universe is a dynamic place. While galaxies and stars take millions or even billions of years to change, there are transient objects in the Universe that change in brightness incredibly quickly. Supernovae - the deaths of massive stars - are bright for a few weeks, but there are many faster things in the Universe. With this project we hope to find new types of fast explosions that have never been identified before!

This project uses data from NASA's Transiting Exoplanet Survey Satellite (TESS), an explorer-class mission that was originally designed to detect planets orbiting their host stars. TESS operates as a "stop and stare" satellite, meaning that it observes one large region of the sky for ~27 days before moving on to another region. During each of these observations, TESS records images somewhere between every 30 minutes to 200 seconds; this means we can investigate how objects change in brightness on these rapid timescales.

Our python-based data analysis pipeline, TESSELATE, removes all stars that don't change in brightness, so that any object left in the final image must have changed in some way. TESSELATE then finds all of these changing objects, extracts their brightness from each image, and plots it through time in a figure called a "light curve". By analysing these light curves, we can then determine what type of object they are.

Discovering the unknown

One of the most exciting things about this project is that we don't know exactly what we might find. It's possible that we won't find anything new at all, which is interesting in its own way. Despite this uncertainty, we have some expectations on what a real event should look like.

Light curves

Light curves display how the brightness of an object changes over time. The apparent brightness of an object is the combination of noise (random fluctuations near zero), and the real signal (generated by actual light). If an event is too faint, the light curve will be dominated by noise, making it very difficult to determine whether the event is real or not.

We expect the light curves of real events to follow certain shapes depending on what type of event generated them. For example, we generally expect that rapid explosive events should rise sharply in brightness and fade away a little slower. Often we will see light curves that look like the one below, where there is an abrupt jump in brightness, followed by a fast decline.

While we may have some expectations of what the light curves of fast explosions look like, the Universe often surprises us, so keep an open mind!

Source shape

In the images themselves, a real event must have a "star-like" appearance; in TESS images, stars look like blocky objects that are about 3 pixels across. In the figures we use in this project, we show the shape of the detected source in the top right. Our detection algorithm should restrict sources which are not star-like, but sometimes it can get confused. When determining if an event is real, check if the shape of the detected source looks suspiciously non-starlike!

We can also check the images to see if the source moves over the space of an hour. If it does move, rather than fading or disappearing, then it is probably an asteroid. While observing asteroids with TESS is scientifically interesting, it isn't the main goal of this project.

Figure caption: A rapid transient detected by our search pipeline. In the light curve (left) for this event we see that is has an abrupt change in brightness, appearing and fading away within an hour. We can also see that from the detection images (right) it looks like a TESS star in the brightest image, and is gone 1 hour later.

Following discovery

Once we have a list of candidates, we need to try and classify them. This process can be challenging as we need to link together many data sets from a global network of telescopes and sky surveys. While looking at the additional data, we're trying to rule out things that we know are common. For example, if we find a catalogue entry for a faint red star at the same location as a candidate identified by TESS, we can be reasonably confident that the event was a flare from a flare star. For positive discoveries, we will be looking for events that coincide with other galaxies, or seemingly nothing.

How your classifications help our search

Since we are exploring a new time domain for transients, we don't know exactly what we will find, nor what they will look like. With your help we will be able to disentangle exciting new discoveries from junk detections, and build a database of fast transient light curves observed by TESS. Eventually, your decisions will help train machine learning algorithms to automatically perform this classification on each new candidate.

We will keep the project updated on what we find together!