v0.0.7

What's Changed

• internal refactoring and variables naming convention by @LegrandNico in #82
• add support for categorical state nodes by @LegrandNico in #83
• add binary softmax response function by @LegrandNico in #88
• CPC tutorial and documentation by @LegrandNico in #89

Full Changelog: v0.0.6...v0.0.7

v0.0.6

What's Changed

• nodes can store additional parameters by @LegrandNico in #73
• children/parents coupling strength stored in nodes' parameter by @LegrandNico in #74
• explicit variable names in the update functions and avoid duplicating parameters by @LegrandNico in #75
• update function for multivariate descendency by @LegrandNico in #76
• plot_nodes method by @LegrandNico in #70
• support for multiple binary children by @LegrandNico in #79
• improve code coverage by @LegrandNico in #80
• v0.0.6 by @LegrandNico in #81

Full Changelog: v0.0.5...v0.0.6

v0.0.5

What's Changed

• typo by @LegrandNico in #59
• documentation on networks by @LegrandNico in #60
• restore the theory tutorial by @LegrandNico in #61
• multivariate input values for probabilistic networks by @LegrandNico in #62
• test the exact values of all parameters after node updates by @LegrandNico in #64
• value coupling to continuous parent with multiple children by @LegrandNico in #63
• remove numba from dependencies by @LegrandNico in #66
• add a second input_data method for models with time varying update sequences by @LegrandNico in #67
• add environment.yml by @LegrandNico in #68
• change numpy versions in requirements by @LegrandNico in #69
• document the use of custom response functions by @LegrandNico in #71
• v0.0.5 by @LegrandNico in #72

Full Changelog: v0.0.4...v0.0.5

v0.0.4

What's Changed

• data sience exercises by @LegrandNico in #48
• fix typo in math equation by @LegrandNico in #49
• typo by @LegrandNico in #50
• remove volatility update for continuous inputs by @LegrandNico in #53
• improve plot trajectories design by @LegrandNico in #54
• refine target for code coverage by @LegrandNico in #55
• use pre-commit for CI by @LegrandNico in #57
• add the total_gaussian_surprise response function by @LegrandNico in #56

Full Changelog: v0.0.3...v0.0.4

v0.0.3

What's Changed

• Add network plotting functionalities by @LegrandNico in #44
• use the first level surprise as the default response function by @LegrandNico in #46
• add tutorials and documentation by @LegrandNico in #45
• new version by @LegrandNico in #47

Full Changelog: v0.0.2...v0.0.3

v0.0.2

What's Changed

• v0.0.1 by @LegrandNico in #32
• packaging by @LegrandNico in #33
• cleaning and updating documentation by @LegrandNico in #34
• PIP badge in README by @LegrandNico in #35
• remove the pure Python implementation by @LegrandNico in #36
• Update docs: Replace theory part with references by @lilianAweber in #37
• Refactor package internals to handle multi parents/children situations by @LegrandNico in #39
• Use mypy's type checking in git action and pre-commit by @LegrandNico in #40
• Github action to publish on PIP automatically by @LegrandNico in #41
• TestPYPI only when tagged by @LegrandNico in #42
• v0.0.2 by @LegrandNico in #43

New Contributors

• @lilianAweber made their first contribution in #37

Full Changelog: v0.0.1...v0.0.2

v0.0.1

This is the first release of pyhgf, a Python package for nodalized, generalized and multilevel Hierarchical Gaussian Filters.

The package contains pure Python and JAX code to fit binary and continuous HGF (two-level and three-level) and tools for visualization.

The model is fully compatible with PyMC(>=5.0.0). The JAX implementation can update any node structure.

The pure Python implementation was used for development and testing and will be removed in 0.02.

The documentation can be found at: https://ilabcode.github.io/pyhgf/index.html#