Question about the computation of the MAP of the posterior distribution

Hi UQ World!
I’m performing Bayesian inversion in the case of vectorial input (i.e., M>1). After the MCMC sampling, UQLab provides me the Maximum A Posterior (MAP) estimate (i.e., the modes) of the (multivariate) posterior distribution. I would like to ask you how the MAP estimate is computed starting from the posterior samples.

Hi Giovanni

The map estimator returned by UQLab is the point with the maximum posterior probability encountered by the MCMC algorithm (i.e. prior times likelihood). It is computed inside uq_postProcessInversion.

Best, Paul

1 Like

Thank you so much for the information :grinning: