The repository contains code for optimizing annealing schedules in a hybrid quantum-classical framework. It is based off of https://github.com/yutuer21/quantumzero, and accompanies a reusability report for Nature Machine Intelligence. We extend the results to include a BFGS gradient method comparison, and to include a MaxCut problem set.
The best starting point for this codebase is to look at the Demo.ipynb jupyter notebook. It contains demonstrations of how the different components of the code work. The run-simulation.py file is then used as a standalone to run and produce data for the plots in the accompanying Nature Machine Intelligence Reusability Report.
This repository also includes a second version of the MCTS code (indicated by _v2 in method names), though the results for the report were all obtained using the original (for reusability reasons).