Described here. Already started implementation: 09d1347
Would be better to use adaptive covariance MCMC rather than the simple Metropolis described here.
Key questions:
- How many MCMC steps to run each active point for? Presumably a hyper-parameter. Read paper more thoroughly.
Described here. Already started implementation: 09d1347
Would be better to use adaptive covariance MCMC rather than the simple Metropolis described here.
Key questions: