v1.0.1

This release contains the full implementation of the Ising model as a perception-action loop (same as in v1.0.0), designed to compare four different intrinsic utility approaches:

• Predictive information
• Empowerment
• Active inference (intrinsic component only)
• Thermodynamic efficiency

The goal of this simulation is to identify the optimal regime in the parameter space for each utility measure using the Ising model as a common example.

Key features:

• Jupyter notebook for running the simulations and visualising the comparisons, with full descriptions of the calculation process.
• Required packages documented in requirements.txt
• This release includes all the necessary scripts and notebooks for running the simulations and comparing the intrinsic utility measures. The codebase is designed to be extensible for further investigations in related fields.

Updates in this release:

• Added source code for examples in section 2 of the paper (criticality and phase transition). Two examples are given: percolation and Drossel and Schwabl forest fire model.

Ising Model as Perception-Action Loop - Intrinsic Utility Comparison

This release contains the full implementation of the Ising model as a perception-action loop, designed to compare four different intrinsic utility approaches:

• Predictive information
• Empowerment
• Active inference (intrinsic component only)
• Thermodynamic efficiency

The goal of this simulation is to identify the optimal regime in the parameter space for each utility measure using the Ising model as a common example.

Key features:

• Jupyter notebook for running the simulations and visualising the comparisons, with full descriptions of the calculation process.
• Required packages documented in requirements.txt
• This release includes all the necessary scripts and notebooks for running the simulations and comparing the intrinsic utility measures. The codebase is designed to be extensible for further investigations in related fields.