NEWS
bayesSSM 0.7.1.9000
bayesSSM 0.7.1 (2025-12-06)
- Small improvements to README and vignettes for clarity.
bayesSSM 0.7.0 (2025-08-26)
- Add
auxiliary_filter and resample_move_filter for auxiliary and
resample-move particle filters.
See all supported filters in ?particle_filter.
- Refactor
pmmh and internals to support multiple particle filters.
- Add
pf_wrapper argument to pmmh to specify which particle filter to use.
- Improve argument checking by switching to the
checkmate package,
providing clearer and more consistent error messages.
- Make minor edits to README and vignettes for clarity.
bayesSSM 0.6.1 (2025-06-23)
- Improved reproducibility of
pmmh: Setting a seed now ensures consistent
results regardless of the number of cores used.
- Improvement performance of
pmmh slightly.
- Rewrote resampling step in
particle_filter to C++ for improved
performance.
- Added new article
detailed-overview which goes over some theory,
implementation details, and tips for using the package effectively.
- Improved the documentation of the package, including the README and
articles, to provide clearer explanations and examples.
bayesSSM 0.5.0 (2025-05-21)
- The
particles argument in init_fn, which is passed to particle_filter
and pmmh, is deprecated. Please use num_particles instead.
A warning will be issued if particles is used.
- Added support for time dependency in functions. You can now use
t in
transition_fn and likelihood_fn when passing them to particle_filter
and pmmh. This allows for time-varying transition and likelihood functions.
- Fixed a bug in
particle_filter in likelihood calculation causing it to be
shifted by a constant.
- Improved robustness of
pmmh when encountering very low log-likelihood
values.
- Added scaling for the proposal covariance when using
"logit" in pmmh.
- Improved documentation: updated package description,
clarified text in the README and vignette, and added more unit tests.
bayesSSM 0.4.7 (2025-04-23)