cyy603 / FJSP_with_AC

Code from the paper An actor-critic algorithm with policy gradients to solve the job shop scheduling problem using deep double recurrent agents."

JSSP_actor-critic_Agasucci_Monaci_Grani

This repository contains the source code and the data related to the paper An actor-critic algorithm with policy gradients to solve the job shop scheduling problem using deep double recurrent agents by Marta Monaci, Valerio Agasucci and Giorgio Grani.

@article{monaci2024actor,
title = {An actor-critic algorithm with policy gradients to solve the job shop scheduling problem using deep double recurrent agents},
journal = {European Journal of Operational Research},
volume = {312},
number = {3},
pages = {910-926},
year = {2024},
issn = {0377-2217},
doi = {https://doi.org/10.1016/j.ejor.2023.07.037},
author = {Marta Monaci and Valerio Agasucci and Giorgio Grani}
}

Installation

The neural network models are implemented using Pytorch.

Requirements

The file requirements.txt reports the list of packages that must be installed to run the code. You can add a package to your environment via pip or anaconda using either pip install "package" or conda install "package". The code can be runned either on GPUs or CPUs.

Configuration and running

In order to test the saved model, you just need to run main_test.py.

Depending on the set of instances you want to test the model on, you have simply to modify the set_to_test parameter as follows:

• Taillard benchmark set (insert 'TaiBenchmarkSet)
• Taillard generated set (insert 'TaiGeneratedSet')
• Gaussian Set (insert 'GaussianSet')
• Poisson set (insert 'PoissonSet')

To generate new sets of instances:

• according to the Taillard method¹ (insert 'TaillardGenerator' and modify the related parameters as you prefer)
• according to the Gaussian method (insert 'GaussianGenerator' and modify the related parameters as you prefer)
• according to the Poisson method (insert 'PoissonGenerator' and modify the related parameters as you prefer)

Results

The output of the experiments can be found in folder Results_TEST/, where:

Statistics_set_to_test.xlsx will report the statistics of the experiments performed on the related set of instances.

Team

Contributors to this code:

• G Grani
• Marta Monaci
• Valerio Agasucci

License

• MIT License
• Copyright 2022 © Marta Monaci, Valerio Agasucci, Giorgio Grani

Footnotes

1. Taken from https://github.com/zcaicaros/L2D ↩