This release allows nimbleSMC
to work with
nimble
versions 1.0.0 and later (i.e., AD-enabled
NIMBLE).
Improve efficiency of ensemble Kalman Filter (issue #2).
Check and error out when detect deterministic latent nodes (issue #5).
Avoid numerical problems from underflow caused by exponentiating weights in various filters (PR #4).
Switch citation to newly-published JSS paper.
Make various cleanups to testing.
Move continuous integration testing to GitHub Actions.
Sequential Monte Carlo functionality was moved from the core nimble package as of nimble version 0.10.0.