Predicting the failure of two-dimensional silica glasses

Datasets for the paper: Font-Clos F, Zanchi M, Hiemer S, Bonfanti S, Guerra R, Zaiser M, Zapperi S. Predicting the failure of two-dimensional silica glasses. arXiv preprint arXiv:2201.09723. 2022 Jan 24. https://arxiv.org/abs/2201.09723

The datasets can be used together with the codes available at https://github.com/ComplexityBiosystems/2D-silica-ML to reproduce the resuts of the paper above.

In particular, the datasets contain the the following data

• The dataset '20201222-var02-onethousand.zip' contains configurations at fixed disorder (s²= 0.2).
• The dataset '20210209-variable-disorder.zip' contains configurations at variable disorder.
• The dataset 'ml_dataset_fracture_atoms_var02.tar.gz' contains the images for the prediction of the crack path.
• The dataset 'ml_dataset_var02.tar.gz' contains images of silica at fixed disorder (s²= 0.2). for the prediction of disorder, rupture strain and location.
• The dataset 'ml_dataset_variable_var.tar.gz' contains images of silica at variable disorder for the prediction of disorder, rupture strain and location.
• The dataset 'trained_models' contains an example of a ResNet trained for predicting rupture strain at variable disorder (it can be used for testing the Grad-CAM method).