* Deprecate the "set_data" method of the ABCSMC class. Use the "new" method instead. * Add a "load" method to the ABCSMC class for easier resuming stored ABCSMC runs. * Add an example to the documentation how to resume stored ABC-SMC runs. * Rename the acceptance_rate parameter form ABCSMC.run to min_acceptance_rate for clarity. Usage of acceptance_rate is deprecated. * Various documentation improvements, correcting typos, clarifications, etc.
0.7.2
-----
Easier early stopping models via the IntegratedModel class. Also has now examples.
0.7.1
-----
* Minor refactoring for better Windows compatibility. But runs in serial on Windows
0.7
..........
0.7.0
-----
* ABCSMC.run gets a new parameter "acceptance_rate" to stop sampling if the acceptance rate drops too low. * History.get_all_populations returns a DataFrame with columns "t", "population_end_time", "samples", "epsilon", "particles". That is "nr_samples" got renamed to "samples" and "particles" is new.
0.6.4
-----
Performance improvement. Use MulticoreEvalParallelSampler as default. This should bring better performance for machines with many cores and comparatively small population sizes.