Open scheidan opened 4 years ago
Thanks for the pointer. There is already an issue for this work. I plan to take a closer look on a few newer HMC variants and implement them later this month.
UPDATE: I closed the old issue because this one points to the published version of the paper. Let's continue the discussion here.
Nishimura et all (2020) propose an HMC integrator that should work well discontinous likelihoods and hence also for discrete parameters. I was wondering if this could be added to AdvancedHMC as it has a very modular structure.
Reference
Nishimura, A., Dunson, D.B., Lu, J., 2020. Discontinuous Hamiltonian Monte Carlo for discrete parameters and discontinuous likelihoods. Biometrika 107, 365–380. https://doi.org/10.1093/biomet/asz083