This PR implements the option for our pmcmc sampler to utilise a Metropolis-within-Gibbs based approach (where start_date is updated using Gibbs Sampling and the rest of the parameters using a Metropolis-Hastings based update as per the original algorithm).
It's passing the tests and I've confirmed the two algorithms produce similar results when it comes to parameter inference. Still need to write some tests, but creating the PR so I can begin assessing coverage.
This PR implements the option for our pmcmc sampler to utilise a Metropolis-within-Gibbs based approach (where start_date is updated using Gibbs Sampling and the rest of the parameters using a Metropolis-Hastings based update as per the original algorithm).
It's passing the tests and I've confirmed the two algorithms produce similar results when it comes to parameter inference. Still need to write some tests, but creating the PR so I can begin assessing coverage.