14.2.5 Science Alerts
The science_alerts table contains all 2612 Science Alerts triggered in the period 25-07-2014 to 28-05-2017. We checked that none of these fields in the table contain null values and that there are no repeated values of the transit_id and name fields.
There are two source_id values with two entries each in the table. The duplication of the two transient events is known (see Hodgkin et al. 2021). The other 2608 rows in the table have a single source_id value. The source_id 3627227154239012352 has 2 different entries (Gaia16ade and Gaia16aey), and source_id 3735279357356643968 has also 2 different entries (Gaia16acr and Gaia16adx). These different alerts assigned to the same source_id is due to a known and solved issue in the working catalogue. The second table related with science alerts is alerts_mixedin_sourceids, with 94 different entries with no null values in any field.
There are 94 entries in the table, with 94 unique mixed_in_source_id values, corresponding to 86 unique values in alert_source_id. The alert_source_id 381267274834125568 has 8 different entries in alerts_mixedin_sourceids table, corresponding to Gaia17ann, with different values for mixed_in_source_id. Similarly the alert_source_id 2973719876967601408 has 2 different entries in the alerts_mixedin_sourceids table, corresponding to Gaia16afx. The remaining 84 sources in the table have a unique entry in the table. The presence of these sources with unique entry is explained because alert_source_id acts as the index to connect to the primary alert_source_id table - and thus, when there is just a single entry, this means there is a single extra source (in Gaia DR3) to be associated with the primary source.
In order to check if the contents of these two tables are as expected, we selected from the Gaia science alert webpage the alerts from the period 25-07-2014 to 28-05-2017. According to this selection, we could extract 2747 sources. Compared with the 2612 sources included in science_alerts, there are 135 missing sources. 123 out of the 135 are missing because they were detected right at the beginning of operations, during the validation phase. At that time, the nominal transient detectors were completely overwhelmed by artefacts - and so they could not be trusted. For this first phase, the science alerts team started running a temporary detector in parallel, which grouped transits together positionally. 7 out of the 135 alerts have no entry for their primary transit_id in the Gaia DR3 match table. The remaining 5 alerts have a later reference time than the cut-off time (end of data for Gaia DR3, 2017-05-28 08:46:28.954418799 TCB). So this is all as expected.