I am relatively new to Stan, which I’m getting to like more and more. It’s the best code I’ve seen around and the manual is just wonderful.
I’m trying to implement the “Cruising the simplex” approach by Michael Betancourt:
but I’m having trouble coding it.
Could any body share a working example written in Stan? I have also looked into the alternatives offered in the manual, but I’m not yet experienced enough to implement those either. Any example using those too, will be most helpful.
I’m using a simplex to describe the weights of to be applied to a set of predictions to fit some data. The standard approach on simple, small datasets works fine, but I’m trying to speed up the process as eventually I will explore ~6000 weights.
I’ve looked into the QR parametrisation also, but since I need to impose a positivity constrain, this does not seem to be the optimal solution.
I’d appreciate any suggestion or example you can provide.