Scalar2Vec: Translating Scalar Fields to Vector Fields via Deep Learning

This repository contains the PyTorch implementation for paper "Scalar2Vec: Translating Scalar Fields to Vector Fields via Deep Learning".

Prerequisites

• Linux
• CUDA >= 10.0
• Python >= 3.7
• Numpy
• Pytorch >= 1.0

How to run the code

• First, change the directory path and parameter settings (e.g., batch size, samples etc.) in main.py.
• Second, to train the model, simply call python3 main.py --train train.
• Third, to inference the new data, simply call python3 main.py --train infer.

Citation

@inproceedings{gu2022scalar2vec,
title={Scalar2Vec: Translating Scalar Fields to Vector Fields via Deep Learning},
author={Gu, Pengfei and Han, Jun and Chen, Danny Z and Wang, Chaoli},
booktitle={2022 IEEE 15th Pacific Visualization Symposium (PacificVis)},
pages={31--40},
year={2022},
organization={IEEE}
}

Acknowledgements

This research was supported in part by the U.S. National Science Foundation through grants IIS-1455886, CNS-1629914, DUE-1833129, IIS-1955395, IIS-2101696, and OAC-2104158.