Abcpy

In this release we include the following new features and improvements of ABCpy:

* Improved documentation;
* Better explanation of the examples with comments;
* Addition of a Jupyter notebook;
* Addition of convergence diagnostic tools;
* Addition of Wasserstein distance for ABC;
* Small bug fixes related to Pypi installation.

In this release we include the following new features and improvements of ABCpy:

* We now provide tools to learn summary statistics from dataset using neural networks.
These features presently uses extensions of semi-automatic summary statistics learning
and statistics learning based on distance learning.

* We have provided some additional features for the output analysis to create figures of
smoothed posterior distribution.

* We have fixed some bugs.

* Some improvements of the documentation.

In this release we include the following new features and improvements of ABCpy:

* We now allow the nodes of the Bayesian network to be multivariate random variables, previously only uni-variate random variables were allowed.

* We have provided some additional features for the output analysis, to access the posterior mean at different steps, access to distances computed etc.

* We have fixed some bugs related to transition kernels.

* We have corrected the discrete random walk kernel.

* Some improvements of the documentation.

New release enabling MPI-parallelized simulator model using nested-MPI backend.

This release (0.5.4) contains

* bugfixes in PMC algorithm
* better distribution of parallel jobs in PMC
* addition of discrete uniform as a new discrete probabilistic model
* fixing of some deprecation warnings
* updates in ReadMe

This release comes with several small improvements and fixes.

- Fix issue with singular matrix in perturbation kernel
- Add verbose output for inference schemes using Python logging module
- Add exception handling for map phase in MPI backend