- HMC, etc
- SMC:
- ESS estimation
- fix documentation
- better abstraction for population
- implement adaptive approximate bayesian computation
- look at population genetics example
lmh_incremental: benchmark implementation of trace against other implementations of integer setslmh_incremental: can we do withoutundo?smc: can we perform resampling directly in log space?
- implement 'colored balls'
- implement 'Bird migration problem'
- Provide an
Owl-backed set of samplers
- Resampling wrt dirac distributions?
- More random walks (Levy processes, discrete random walks - eg on graphs, etc)
- Change printf into proper logging (to be active when
DAGGER_TEST_ARTIFACTSis set) - Improve resampling coverage (esp for population count = 1)
- Some
asserts deserve to be cleaner errors insmc_inference