GMLS-Nets (PyTorch Implementation)
Package provides machine learning methods for learning features from scattered unstructured data sets using Generalized Moving Least Squares (GMLS). Provides techniques which can be used to generalize approaches, such as Convolutional Neural Networks (CNNs) which utilize translational and other symmetry, to unstructured data. GMLS-Nets package also provides approaches for learning differential operators, PDEs, and other features from scattered data.
Quick Start
Method 1: Install for python using pip
pip install -U gmlsnets-pytorch
For use of the package see the examples page. For getting the latest version use pip install --upgrade gmlsnets-pytorch. More information on the structure of the package also can be found on the documentation page.
If previously installed the package, please update to the latest version using pip install --upgrade gmlsnets-pytorch
Manual Installation
Method 2: Download the gmlsnets_pytorch-1.0.0.tar.gz file above, then uncompress
tar -xvf gmlsnets_pytorch-1.0.0.tar.gz
For local install, please be sure to edit in your codes the path location of base directory by adding
sys.path.append('package-path-location-here');
Note the package resides in the sub-directory ./gmlsnets_pytorch-1.0.0/gmlsnets_pytorch/
Packages
Please be sure to install PyTorch package >= 1.2.0 with Python 3 (ideally >= 3.7). Also, be sure to install the following packages: numpy>=1.16, scipy>=1.3, matplotlib>=3.0.
Use
For examples and documentation, see
Additional Information
If you find these codes or methods helpful for your project, please cite:
GMLS-Nets: A Framework for Learning from Unstructured Data, N. Trask, R. G. Patel, B. J. Gross, and P. J. Atzberger, arXiv:1909.05371, (2019), [arXiv].
@article{trask_patel_gross_atzberger_GMLS_Nets_2019,
title={GMLS-Nets: A framework for learning from unstructured data},
author={Nathaniel Trask, Ravi G. Patel, Ben J. Gross, Paul J. Atzberger},
journal={arXiv:1909.05371},
month={September}
year={2019}
url={https://arxiv.org/abs/1909.05371}
}
For TensorFlow implementation of GMLS-Nets, see https://github.com/rgp62/gmls-nets.
Acknowledgements We gratefully acknowledge support from DOE Grant ASCR PHILMS DE-SC0019246.