Bayesian consistency, or (again) lack thereof

This is a placeholder for some discussion about the circumstances (and the manner) in which Bayesian posterior updates to a given prior will converge to the “true” parameter value. In applied math, the model is almost always mis-specified, and so we have to ask “if the truth is complicated and I apply Bayesian procedures to my simplified model, what kind of mistakes am I likely (or not likely) to make?”


Previous
Previous

Asymptotic behavior of the heat equation in free space

Next
Next

Wilks’ phenomenon, or lack thereof