Deep Reinforcement Learning for Multi-Period Facility Location: $p_k$-median Dynamic Location Problem

This is the official code for the published paper 'Deep Reinforcement Learning for Multi-Period Facility Location: $p_k$-median Dynamic Location Problem'. <img width="1144" alt="image" src="https://github.com/user-attachments/assets/ddfc30c3-358c-4dcc-ab17-41d6e3a6942d"> The paper was accepted by 32nd ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (SIGSPATIAL 2024) as Full Papar.

Dependencies

All experiments are conducted on the machine with two GTX 3090 GPU and Intel(R) Xeon(R) Silver 4210R CPU @ 2.40GHz, and the code is implemented in Python.

• Python==3.7.0
• pytorch=1.13.1
• SciPy=1.7.3
• tqdm
• tensorboardx

Quick start

python run.py --graph_size 20 --pk 2 3 4 --r 0.32 --run_name 'MultiPM20'
python run.py --graph_size 20 --pk 2 3 4 6 8 --r 0.32 --run_name 'MultiPM50'
python run.py --graph_size 20 --pk 2 4 7 9 10 13 15 --r 0.32 --run_name 'MultiPM100'

Usage

Generating datasets

Generating 1000 MultiPM instances with 20 nodes with $p_k = [2, 3, 4]$

python gen_data.py --graph_size 20 --pk 2 3 4 --dataset_size 1000

Training

For training the MultiPM instances with 20 nodes with $p_k = [2, 3, 4]$ and using rollout as REINFORCE baseline:

python run.py --graph_size 20 --pk 2 3 4 --r 0.32 --run_name 'MultiPM20'

Multiple GPUs

By default, training will happen on all available GPUs. To disable CUDA at all, add the flag --no_cuda. Set the environment variable CUDA_VISIBLE_DEVICES to only use specific GPUs:

CUDA_VISIBLE_DEVICES=0,1 python run.py 

Note that using multiple GPUs has limited efficiency for small problem sizes (up to 50 nodes).

Warm start

You can initialize a run using a pretrained model by using the --load_path option:

python run.py --graph_size 100 --load_path {pre-trained path}

Evalution

To evaluate the Multi instances with 20 nodes by greedy strategy:

python eval.py {dataset_path} --decode_strategy greedy --model {model_path}

To evaluate the Multi instances with 20 nodes by sample1280 strategy:

python eval.py {dataset_path} --decode_strategy sample --width 1280 --model {model_path}

Solution Visualization

<img width="1596" alt="image" src="https://github.com/user-attachments/assets/f99a5efd-7a3b-4124-9b2c-48282a6e6568">

Real Application

<img width="1509" alt="image" src="https://github.com/user-attachments/assets/c04a726c-5b71-4236-b778-62c59183df0c">

Acknowledgements

This repository includes adaptions of the following repositories as baselines:

• https://github.com/wouterkool/attention-learn-to-route/
• https://github.com/kevin031060/CSP_Attention
• https://github.com/CO-RL/SpoNet