Circopt-RL-ZXCalc

This repository contains the code used to produce the results of the publication Reinforcement Learning Based Quantum Circuit Optimization via ZX-Calculus.

Results and data verification

• The data to create the figures of the paper are inside the folder results/5x60_non_clifford. The json data files were generated with agent_test.py. The code used to generate the figures is in the file results.ipynb. The code used to create Table 2 from the paper can be found in the file benchmark.py.

• The optimized circuits can be found in the rl-zx/results/circuits path:

  • Original: Contains the original circuits.
  • gflow-cflow-opt: Contains the optimized circuits using the combination of Staudacher et al. and Holker
  • NRSCM: Contains the optimized circuits using the algorithm from Nam et al.
  • rl-zx-opt: Contains the optimized circuits using the RL agent presented in this paper.

• In rl-zx/results/dataare the jsonfiles used to generate Figure 11.

Code

• The agent architecture can be found in the file rl_agent.
• The environment can be found in rl-zx/gym-zx/envs with the name zx_env.py

Citation

@misc{riu2024reinforcement,
      title={Reinforcement Learning Based Quantum Circuit Optimization via ZX-Calculus}, 
      author={Jordi Riu and Jan Nogué and Gerard Vilaplana and Artur Garcia-Saez and Marta P. Estarellas},
      year={2024},
      eprint={2312.11597},
      archivePrefix={arXiv},
      primaryClass={quant-ph}
}