How HSE supports Russia's war: trap.org.ua/en/publications/how-the-russian-higher-school-of-economics-supports-the-war-against-ukraine/
HSE seminar series & speakers (incl. many international ones, from top universities): sites.google.com/view/hse-econ-seminar/hse_seminar

I’d start with Dave Miller’s recent paper on the Bayesian Interpretation of GAMs in MEE arxiv.org/abs/1902.01330

The equivalence holds for the case of a Bayesian view of smoothing as that looks like the REML criterion of a mixed effects model

If this is a general “penalties = priors” thing I don’t think that is generally true. There are specific cases where they work out to be equivalent to a particular (possibly improper) prior. This is what underlies mgcv and IIRC also INLA (but I’m less familiar with the latter)

I think that would be using an “old style” or “scheffer style” random effect interpretation sensu Hodges. We don’t tend to think our spline is a draw from a multivariate Gaussian as if it were some weird set of random effects but we use this as a computational convenience, AKA “new style” ranefs

Isn’t this called maximum a posteriori estimation? What mgcv does is an empirical Bayesian edtimate. The “penalty” is an improper Gaussian prior on the smooth coefs and lambda is inversely proportional to a variance parameter. I’m away from my books & other stuff but I think this is in Wood 2011

If you mentor scientific writers, teach a course with written assignments, or anything similar, you know how much work is involved, and how difficult it can be to provide the guidance developing writers need. This book is for you!
2/2

Congratulations Stephen 🎉🍾