EMQAOA-DARBO

Overview

This repository includes the codes and results for the manuscript: Quantum approximate optimization via learning-based adaptive optimization published on Communications Physics link

Installation and usage

This repository requires to install two open-sourced packages:

• ODBO packge: The installation direction is provided in the corresponding main page. NOTE: Tencent has deactivate the Tencent public repo ODBO written in the paper. Please see my personal repo linked, and to get the exact version as the published paper version, go to the tencent_migrate branch.

• TensorCircuit or TC: pip install tensorcircuit

Content list

Files

• DARBO_optimization_ideal_example.ipynb: This is a simple example to illustrate the methods & to run a test MAX-CUT on a random graph with a circuit depth of 4.

• EMQAOA_DARBO_run.ipynb: This is the notebook to illustrate the EMQAOA-DARBO on the real hardware. This collects the hardwared data shown in the manuscript. Note: For non-Tencent-Quantum-Lab user, this set of codes cannot be run directly due to the unavailable access to the Tencent hardware. If you would like to have a try, please contact Tencent Quantum Lab to check the possible options for usage.

• si_more_stats.xlsx: This is a supplemental excel to summarize the optimized losses and $r$ values for different optimizers and different cases.

Folders

• codes: contains all the python codes that run the experiments collected in this work. (Please aware that all BO methods are formulated as a maximization problem (max -loss), and we save the -loss at each iteration. For other optimizers, we save loss at each iteration.)

• graph: contains the graphs used in this work.

• initialization: contains the presaved (& different) initialized parameters to make sure all different optimizers running from the same initial guesses.

• results: each subfolder contains the collected results for the corresponding

• plotting: contains a jupyter notebook to generate all the plots used in the paper. for_plotting folder contains the .txt summary for the results extracted from the raw results.

Please cite us as

@article{cheng2023darbo,
  title={Quantum approximate optimization via learning-based adaptive optimization},
  author={Cheng, Lixue and Chen, Yu-Qin and Zhang, Shi-Xin and Zhang, Shengyu},
  doi = {10.1038/s42005-024-01577-x},
  journal = {Communications Physics},
  number = {1},
  pages = {83},
  volume = {7},
  year = {2024},
}