Alvaro Faria (Open University, UK) and I have a project suitable for a PhD student looking at using logarithmic pooling to combine predictive distributions from a collection of models already fitted to a set of data. The main idea is working from a fixed set of posterior draws for each model, – fitted in Stan, say – and then finding the optimal weights which give the best predictive performance. The studentship would be based in the UK and application deadline is March 5th 2021.
If this sounds like something you would like to take a look at, feel free to drop me a line.