On the road again…
I’m in London now, working with David Balding on HapCluster. We had a late dinner yesterday and I am slightly hung over and a bit lacking in sleep, which is not the optimal state of mind to be trying to solve a mixing problem in an MCMC…
May 6th, 2008 at 12:03 pm
Oh well, give David my regards.
The mixing problem could be solved by using ABC instead. :-)
May 6th, 2008 at 12:12 pm
Sometimes, yes, sometimes… not so sure about this particular MCMC, here I probably just need to have a think about the proposal distribution. In any case, I have already improved on it a lot, so I am reasonably happy with it.
Anyway, I usually do not find it easier to think ABC than MCMC when hung over, so it wouldn’t solve the real problem ;-)