Skip to content

Releases: darrenjw/jax-smfsb

Version 1.1.3

02 Jan 14:55
Compare
Choose a tag to compare

This version fixes a bug in sim_time_series. The output from this function was not including the initial state of the system. This was inconsistent with implementations of this function in other versions of the library. Everyone is encouraged to upgrade to this release of the library ASAP.

The correctness of the exact stochastic simulation algorithm has now been verified against the discrete stochastic models test suite (DSMTS). See the new dsmts directory for details.

Version 1.1.2

28 Dec 21:55
Compare
Choose a tag to compare

Bumped up NumPy version requirement from < 2 to >= 2, as seems fine. Also added extra demo.

Version 1.1.1

17 Nov 15:22
Compare
Choose a tag to compare

Tidy of code (especially variable naming) using ruff, and improved documentation, including a tutorial.

Version 1.1.0

03 Nov 17:06
Compare
Choose a tag to compare

This release introduces many breaking changes to syntax. Method names have been updated to have a more pythonic style. If you are upgrading to this release from a previous release you will need to update your code. However, all of the changes are purely cosmetic changes to names. There are no changes to the semantics of the code.

Version 1.0.0

08 Sep 15:37
Compare
Choose a tag to compare

First official release. Covers all of the main simulation and inference algorithms from the book.

Version 0.0.9

08 Sep 14:26
Compare
Choose a tag to compare

Added ABC-SMC code. The library is now complete.

Version 0.0.8

05 Sep 20:17
Compare
Choose a tag to compare

Added MCMC, particle filter and particle MCMC functions, for parameter inference.

Version 0.0.7

04 Sep 19:25
Compare
Choose a tag to compare

Added functions and methods for spatial simulation in 1D and 2D (both Gillespie and CLE).

Version 0.0.6

26 Apr 21:02
Compare
Choose a tag to compare

Approximate simulation algorithms now included.

Version 0.0.5

26 Apr 16:20
Compare
Choose a tag to compare

Included some built-in models.