A note on parametric Bayesian inference via gradient flows (Q2218178)

The authors summarize several recent developments for efficient sampling methods for parameters based on Bayesian inference. The authors reformulated different formulations for gradient flows on the manifold in the parameter space, including strong form, weak form and De Giorgi type duality form. As far as my knowledge goes, the authors' mathematical approach is good. They tried to consolidate all different formulas and used a parametric Bayesian inference via several gradient flows. This work is new and useful for future, in particular for the nonconvex optimization of effective sampling studies.