1/1
Machine Learning Materials Datasets
Version 5 2018-09-11, 20:29
Version 4 2018-09-11, 17:37
Version 3 2018-09-06, 22:00
Version 2 2018-08-29, 05:10
Version 1 2018-08-28, 18:15
dataset
posted on 2018-09-11, 20:29 authored by Dane MorganDane MorganThree datasets are intended to be used for exploring machine learning applications in materials science. They are formatted in simple form and in particular for easy input into the MAterials Simulation Toolkit - Machine Learning (MAST-ML) package (see https://github.com/uw-cmg/MAST-ML).
Each dataset is a materials property of interest and associated descriptors. For detailed information, please see the attached REAME text file.
The first dataset for dilute solute diffusion can be used to predict an effective diffusion barrier for a solute element moving through another host element. The dataset has been calculated with DFT methods.
The second dataset for perovskite stability gives energies of compostions of potential perovskite materials relative to the convex hull calculated with DFT. The perovskite dataset also includes columns with information about the A site, B site, and X site in the perovskite structure in order to perform more advanced grouping of the data.
The third dataset is a metallic glasses dataset which has values of reduced glass transition temperature (Trg) for a variety of metallic alloys. An additional column is included for majority element for each alloy, which can be an interesting property to group on during tests.