Released a new dataset: Binary 2D morphologies of polymer phase separation along with a python toolbox providing API and utilities to aid the usage of our dataset. Documentation is available here.
In our work on Physics-aware generative models, we introduce generative models for the fast synthesis of binary microstructure images, which can be potentially useful in novel material discovery. For our experiments, our team has created a dataset of 2D binary micro-structures of polymer phase separations. This dataset was generated through the simulation of a time evolving Cahn -Hilliard equation, describing phase separation in binary polymer blends. Several realizations of the equation were done through different values of volume fractions and binary interaction parameters. Morphologies were outputted at constant time intervals.
Our dataset is available on Zenodo and open to access under the CC 4.0 license. Please visit the link: Binary 2D morphologies of polymer phase separation
Along with the dataset, we develop an open-source python toolbox, that provides several tools related to setup, usage and analysis functionality for a smoother interface.
Our python toolbox is available on github here: python toolbox
For enhanced usability, we follow best practices for open data sharing. Our data is stored in easy to access HDF5 format, and we provide detailed API and documentation of our toolbox, too.
For complete reference, please find the documentation and API here: Detailed documentation.
Our submission to Open Data Challenge organized by Materials Research Society (MRS) pertaining to above dataset is chosen as one of the three finalists - and eventually won the second prize competing against the teams from Stanford and MIT! Our presentation is available here: MRS Open data challenge submission.
Please write to Viraj Shah, (email: viraj at iastate dot edu) for any queries reagarding the dataset.
Publications1. R. Singh, V. Shah, B. Pokuri, S. Sarkar, B. Ganapathysubramanian, and C. Hegde, Physics-aware Deep Generative Models for Creating Synthetic Microstructures ,NeurIPS 2018 Workshop on Machine Learning for Molecules and Materials.
The poster is available here: Poster