The Arizona-NOAO Temporal Analysis and Response to Events System (ANTARES) is a joint project between NOAO and the University of Arizona Computer Science and Mathematics Departments to build a software infrastructure system to process alerts generated by astronomical time-domain survey programs. The night sky is an ever-changing place, with variable stars, moving objects in the Solar System, and transient events throughout the Universe. Astronomical imaging surveys can compare an image taken tonight with an earlier reference. The difference between these images reveals the objects that have changed and each change can be thought of as a new "alert." Some of these objects are rare or short-lived, so distinguishing them as quickly as possible is necessary to ensure prompt follow-up observations using a broad range of telescope resources.


Over the history of time-domain astronomy, the number of alerts generated per night has grown dramatically. Most surveys produced alerts at a low enough rate that human interaction could sort through and evaluate which alerts were the most relevant for a given project or which was the most interesting. This is changing now with the commissioning of the Zwicky Transient Facility in 2018, producing a few hundred thousand alerts on typical nights. A few years from now, the Large Synoptic Survey Telescope (LSST) will commence operations. It is expected to produce several million alerts per night.

LSST has a large mirror that enables it to see faint things quickly and a large camera that images ten square degrees at a time. This allows LSST to image thousands of square degrees over a night, comparing each image to a reference and detecting the objects that have changed. It is likely that LSST will generate millions of alerts per night, each night, for the ten-year duration of the survey. A good fraction of these will be known variable stars or moving objects, but hidden among them will be rare and interesting objects that have relatively short lifetimes. Only with additional follow-up will these objects reveal their nature. Without the ability to rapidly sort through these millions of alerts each night and winnow them down to a reasonable number that can be studied in detail, we will lose these rare and potentially extraordinarily interesting objects. We are developing the ANTARES system to process the LSST alert stream and enhance the ability of the community to use LSST for time-domain science.

The traditional path for exploiting a new astronomical resource was to build an instrument that could take advantage of new capabilities. LSST is the epitome of the transition in modern astronomy away from classically scheduled nights to survey systems that produce vast quantities of data. In this new paradigm, ANTARES is the instrument we are building and it looks at the data, not at the sky.

The Time-Domain Ecosystem

The term used for a system that sits between a source and a customer while adding value is a broker and that is how we characterize ANTARES. It has to live in a larger time-domain ecosystem in order to be of most value. The surveys themselves are the alert generators. They take the images, perform image differencing, identify sources that have changed, evaluate whether sources are artifacts or real, and disseminate the alerts.

ANTARES takes these alerts and annotates them with catalog information from objects associated with the alerts across the full electromagnetic spectrum as well as past history of the sources producing the alerts. It uses features, measured or derived quantities or qualities of the alerts, to characterize them into coarse bins. Objects that are relatively mundane or do not require rapid follow up are diverted. These are not lost, but rather stored in a database that could be used as a resource to study variable sources that do not require immediate attention. ANTARES then ranks the remaining sources by evaluating a measure of how rare they are and distributes these alerts.

Downstream from ANTARES are other brokers that can take advantage of the ANTARES annotations for more refined characterization or classification. These could even be copies of the ANTARES infrastructure, but with different filter sets. ANTARES ranked alerts could trigger robotic or manual follow-up observations. Many of these activities that handle alerts subsequent to ANTARES will require other types of software infrastructure to implement them efficiently.

Software systems similar to ANTARES already exist, but they are typically tuned to specific programs and still require a large human involvement. ANTARES aims to be scalable to the LSST rate and volume will also being as general as possible to fulfill the broader goals of the US astronomical community. It will be a public resource.


ANTARES operates by ingesting alerts. It associates the alerts with known objects from multi-wavelength catalogs. If this source has produced an earlier alert, that history is also associated along with any calculated values. Using the annotated information from these associations, ANTARES will derive features, if necessary. Features can be as simple as the change in brightness or the Galactic coordinates or something more complex involving various moments of light curves. These features are compared against a database that we call the Touchstone. It is essentially a distillation of astronomical knowledge into features that can then be used to characterize new alerts. Theoretical models can also be incorporated. Various filtering stages are then applied to evaluate the relative rarity of the alerts. More details are available in our architecture description.

The purpose of ANTARES is to serve the general astronomical community. In order to meet a broad range of astronomical interests, we need community involvement. If you have data that would help to characterize a class of objects that we could include in the Touchstone or you can provide a set of features that distinguish a class of objects, we could use your help. We are now interacting with the public survey of the Zwicky Transient Facility. If you are interested in contributing, please let us know.

The ANTARES team gratefully acknowledges support from the National Science Foundation via INSPIRE Grant CISE AST-1344204, PI:Snodgrass.