Initial release

The main features implemented in this release are:

• Model Predictive Control implementation and state space system simulation;
• Online training algorithm;
• Online version of the Stochastic Weight Averaging;
• Possibility to run transfer-learning experiments by pre-training a network on a dataset and continue the training on another;
• HPO supported by syne-tune.

This version of the code has been used for the paper "An Online Learning Method for Microgrid Energy Management Control" presented at the 31st Mediterranean Conference on Control and Automation.

Main contributors of this code are:

• Vittorio Casagrande @vittpi
• Martin Ferianc @martinferianc