A-coherent-approach-to-quantum-classical-optimization

This repository contains the code used to produce the results of the publication A coherent approach to quantum-classical optimization.

Results and data verification

• The data used to create the figures in the article is located in the pretraining/logger_data folder.

• Within the logger_data folder, there are several subfolders, each containing the data necessary to reproduce the results presented in the article. See the associated readme for more information.

• The code used to generate the figures for the paper can be found in the file plots_figures_paper.ipynb. The code used to generate the figures in the the supplementary information can be found in the file plots_figures_supplementary_information.ipynb

• In the file classical_quantum_optimisation.ipynb we present functional examples of hybrid optimization schemes. We also added two files presenting the operation of the DMRG algorithm and the WI,II MPO called dmrg.ipynb and mpo_time_evolution.ipynb respectively.

Code

The code used is divided into two main modules.

• The first module is located within the qibo_analysis folder. This folder contains the code necessary to conduct the study on the performance of pure Gibbs states as initialization states.

• The second module is located within the variational_algorithms folder. This folder contains the code that implements the combined optimization protocol for tensor networks and VQA. It includes both the new protocol introduced and the state-of-the-art protocol used for comparison.

Installation

To correctly install the dependencies, please create an environment with python = 3.10.

conda create -n pretraining_env python=3.10
conda activate pretraining_env

And install the corresponding packages

pip install -r requirements.txt

Citation

@misc{cáliz2024coherentapproachquantumclassicaloptimization,
      title={A coherent approach to quantum-classical optimization}, 
      author={Andrés N. Cáliz and Jordi Riu and Josep Bosch and Pau Torrente and Jose Miralles and Arnau Riera},
      year={2024},
      eprint={2409.13924},
      archivePrefix={arXiv},
      primaryClass={quant-ph},
      url={https://arxiv.org/abs/2409.13924}, 
}