Things To Do
Implement stochastic graph node for generalized extreme value distribution.
Test Hybrid Monte Carlo sampling algorithm for random effects.
Integrate out random effects precision in above algoritms.
Write up example(s) where inference is made about functional of stochastic nodes.
Document stratagey for choosing updaters.
(done)
Write Java GUI front end for ClassicBUGS -- JavaBUGS
Port ClassicBUGS to Java byte code using Garden Point Compiler.
Make BUGS produce output in Html format, in particular Html tables.
Write code to convert tables produced by WinBUGS into Html tables.
Write code to enable WinBUGS to produce Html tables for results
Package BRugs R functions in standard way.
(done thanks to Uwe & Sibylle)
Test BUGS script language.
(done some)
Get some reversible jump stuff working again.
Investigate delayed rejection in Metropolis algorithm.
(done)
Move GraphStochastic.Node.Constrain to GraphMultivariate.Node.Constrain
Always make block updater for multivatiate node (chain graphs as well as conjugate MV priors)
Alter UpdaterSlice so that no over-relaxation if not unimodal. If bounded support and not unimodal use stochastic binary search to samples (no search length)