skip to main content

gaia data release 3 documentation

9.6 Data products

9.6.3 Correlation matrix

A correlation matrix as well as internal errors are also given as data product. These quantities are usually extremely small and reflect the final step of the convergence of the iterative process more than external evaluation of the uncertainties on the quantities. In the optimisation of the fitted parameters process only the 30 best solutions from our 10 000 random trials are kept to compute our statistics such as mean, standard deviations and covariance. Therefore the computation of the final covariance matrix will be done on the final set of 30 best parameters. This will explain the very small values obtained since at this step the algorithm should have converged and thus leading to very small differences between the 30 best sets of parameters.