0.3.0
This release focuses on stability and performance improvements. Specifically:
- We now use the stable version
0.1.0
ofbiodivine-lib-param-bn
. This brings additional improvements (compared to0.2.0
) in the SBML parser, as well as other bugfixes and API improvements. - This is the first version to use fully symbolic representation of asynchronous graphs.
- Internal algorithms have been significantly improved to a point where we can handle any fully instantiated real-world network that we know of. Large parameterized models (20+ variables) can expect several orders of magnitude speed-up as well.
- Progress tracking and cancellation support during computation have been reworked and should be more reliable.
- We show a two-state approximation for large attractors that cannot be otherwise visualized.
- Adding Github Actions build check. In the future we should really consider the option to automatically build artifacts for releases using Github Actions.
- We incorporate a large (125+) collection of real world models that can be used for testing (already used in
biodivine-lib-param-bn
for differential SBML testing) and benchmarking (includes simple python scripts to run such benchmakrs). The integrity of the benchmark set is validated using a custom rust executable.