Operationally meaningful representations of physical systems in neural networks

Code for: H. Poulsen Nautrup, T. Metger, R. Iten, S. Jerbi, L.M. Trenkwalder, H.Wilming, H.J. Briegel, and R. Renner. "Operationally meaningful representations of physical systems in neural networks" (2020).

This repository contains the trained Tensorflow models used in section 5 of the paper as well as code to load, train and analyze them. The code for the example using reinforcement learning (section 6 in the paper) can be found here.

Requires:

• Python 3.6.7
• numpy 1.16.2
• scipy 1.2.1
• matplotlib 3.0.3
• tensorflow 1.13.1
• tensorboard 1.13.1
• tqdm 4.31.1
• jupyter 1.0.0

To use the code:

1. Clone the repository.
2. Add the cloned directory communicating_scinet to your python path. See here for instructions for doing this in a virtual environment. Without a virtual environment, see here.

Generated data files are stored in the data directory. Saved models are stored in the tf_save directory. Tensorboard logs are stored in the tf_log directory.

Some documentation is available in the code. For further questions, please contact us directly.