ANTARES processes alerts from the Zwicky Transient Facility survey in real-time, using filters written in Python, and streams the results out using Apache Kafka. Our goal is to process alerts from the Rubin Observatory Legacy Survey of Space and Time.
To get started, read the FAQ which will direct you on how to start using ANTARES.
Also take a look at the Pipeline page, which documents how ANTARES is currently processing alerts.
The ANTARES Documentation Site describes how to start developing your own filters.
The current high-level processing of ANTARES retains alerts that pass certain image-quality tests. ZTF assesses whether or not a given alert is likely to be real or an artifact. Alerts that fall below a threshold value for this assessment (
ztf_rb < 0.55) are ignored and
not included in the ANTARES database. In addition, alerts with bad pixels (
ztf_nbad > 0) or a large difference in aperture vs PSF
ztf_magdiff outside of the range -0.1 to 0.1) are also ignored. This portion of the alert stream is evaluated for known
Solar System objects. After this evaluation, alerts with poor seeing (
ztf_fwhm > 5.0 arcsec) or elongated sources (
ztf_elong > 1.2) are
ignored and not included in the ANTARES database. (Note that, if a source is not included in our database, but has a later alert that
passes our criteria, then history from the ZTF alert packet is stored.) ANTARES then runs our other science filters, but only on
alerts with at least two detections. Alerts with only one detection are likely to be unknown Solar System objects, so we wait for a second
detection at the same location. Alerts with one detection are stored in the database so they can be associated with later alerts.