Polyimide Explorer
A machine-learning implementation that evaluates the thermal and mechanical properties of the “golden plastic” polyimide. Based on it we developed an online interactive platform https://polyimide-explorer-a6356ff113c4.herokuapp.com/. It provides a visualization of more than 77,000 high-performing hypothetical polyimides. Their commercially-available reacting components are linked in the platform to the PubChem database. The machine learning model is also embedded in the platform for an easy application. Please refer to our work “Discovery of Multi-Functional Polyimides through High-Throughput Screening using Explainable Machine Learning” for additional details.
Copolymer Machine Learning
A machine-learning (ML) implementation that incorporates the information of both molecular composition and sequence distribution of copolymers including random, block, and gradient copolymers. Given the molecular composition (SMILES of monomers) and copolymer sequence type (random, block, gradient), the ML model can incorporate both information and establish the structure-property relationship. Please refer to our work “Machine Learning Strategies for the Structure-Property Relationship of Copolymers” for additional details.
Polymer Gas Membrane ML (pgmML)
A machine-learning implementation that learns generalizable, interpretable models between polymer chemistry and membrane gas permeability, which can be used for polymer discovery. Polymer membranes perform innumerable separations with far-reaching environmental implications. Despite decades of research, design of new membrane materials remains a largely Edisonian process. To address this shortcoming, we demonstrate a generalizable, accurate machine learning (ML) implementation for the discovery of innovative polymers with ideal performance. Specifically, multitask ML models are trained on experimental data to link polymer chemistry to gas permeabilities of He, H2, O2, N2, CO2, and CH4. We interpret the ML models and extract valuable insights into the contributions of different chemical moieties to permeability and selectivity. We then screen over 9 million hypothetical polymers and identify thousands that lie well above current performance upper bounds, including hundreds of never-before-seen ultrapermeable polymer membranes with O2 and CO2 permeability greater than 104 and 105 Barrers, respectively. High-fidelity molecular dynamics simulations confirm the ML-predicted gas permeabilities of the promising candidates, which suggests that many can be translated to reality. Please refer to our work “Machine learning enables interpretable discovery of innovative polymers for gas separation membranes” for additional details.
OpenFSI
OpenFSI is a highly efficient and portable fluid-structure interaction (FSI) simulation package based on the immersed-boundary method. The fluid dynamics is accounted for by the software Palabos. The structure solver is implemented within the framework of LAMMPS. In the current version, there are 1D, 2D, and 3D lattice models and 3D shell models in structure solvers. Using these models, we can model a broad FSI problem including the swimming of micro-organisms with tails, flapping of 2D or 3D plates mimicking bird flying and fish swimming, and biological flow with large numbers of blood cells. Please refer to our work “OpenFSI: A highly efficient and portable fluid–structure simulation package based on immersed-boundary method” for additional details.
Highlighted features
- Particle-based lattice model The solid is discretized into a lattice structure, and the mechanical properties such as stretching and bending are described by a series of potential functions that are applied on the lattice nodes.
- Coupling high-performance software packages i.e., LAMMPS and Palabos.
- Highly portable There are broad applications including 2D and 3D.
