A new implementation of the Astrometic Global Iterative Solution (AGIS) has recently been completed by the Gaia SOC development team at ESAC. AGIS is a core element of the overall Gaia data reduction process. It consists of an iterative, simultaneous adjustment of a large number of astrometric, attitude, instrument calibration, and global physical parameters to obtain the optimum agreement between observation data and an observational model.
In the new ESAC AGIS system the iterative loop is controlled through a stopping criterion that is based on the monitoring of the adjustments to the stellar parallaxes. In each iteration the updates to the parallax values with respect to the previous iteration are histogrammed and the width of the resulting distribution determined. The system is considered converged if this width falls below 1 microarcsec.
The figure shows the evolution of the parallax distribution width with iteration number for a first completed AGIS validation run involving 1.1 million stars simulated over an 18-month observation period. It can be seen that convergence was achieved within 4 iterations.
This satisfactory result demonstrates on the one hand the correctness of the new implementation but also for the first time convincingly validates the fundamental working principles of AGIS on a large scale. Future test runs with more, and increasingly noisier, input data, which are more representative of the real mission, are planned and expected to further corroborate the soundness of the chosen astrometric data reduction approach with AGIS.
For further details see ESAC implementation of the astrometric global iterative solution [GAIA-C3-TN-ESAC-UL-015-1] by U. Lammers, J. Hernandez, J. Hoar, and W. O'Mullane,