[Enhancement]: normalization mean and var in sinergym output - Githubissues

ugr-sail / sinergym

Gym environment for building simulation and control using reinforcement learning

https://ugr-sail.github.io/sinergym/

MIT License

131 stars 35 forks source link

[Enhancement]: normalization mean and var in sinergym output #409

Closed AlejandroCN7 closed 5 months ago

AlejandroCN7 commented 6 months ago

Improvement 🔧

As described in version 3.3.1, functionality has been added to save calibration configurations for adaptive observation normalization.

However, its logging and usage in the Sinergym tool can be improved for better usability, hence the following enhancements are proposed:

Save the mean and variance in a text file in the Sinergym output if the normalization wrapper has been used.
Add an assert to provide a more precise notification if the wrapper has not been correctly indicated (in case a previous calibration is to be used).
Improve the example notebook to better demonstrate its use.
Mean and var cloud be indicated as list or numpy instead of only numpy.

Checklist

[x] I have checked that there is no similar issue in the repo (required)
[x] I have read the documentation (required)

:pencil: Please, don't forget to include more labels besides enhancement if it is necessary.