MCMC is not working for some known and some unknown reasons. Need to fix.
Implement Desi's suggestions:
(1) Try to parameterize rates so that they are in the log scale
(2) Investigate priors and hyper-parameters because I'm looking at smaller range of values than eems.
(3) To test whether parts other than the birth-death is working, fix the number of tiles and check whether you can recover the true demographic surface. Try uniform grid (ntiles = 1) and barrier (ntiles = 3).
MCMC is not working for some known and some unknown reasons. Need to fix.
Implement Desi's suggestions: (1) Try to parameterize rates so that they are in the log scale (2) Investigate priors and hyper-parameters because I'm looking at smaller range of values than eems. (3) To test whether parts other than the birth-death is working, fix the number of tiles and check whether you can recover the true demographic surface. Try uniform grid (ntiles = 1) and barrier (ntiles = 3).