Accelerating Materials Research through Genetic Programming
Abstract
Project Summary / Abstract Genetic programming is a powerful machine learning method that has been successfully used in a variety of fields, but to date it has seen little use in fundamental materials research. To fully realize the potential benefits of genetic programming to materials science, it will be necessary to build upon generic genetic programming approaches, addressing specific deficiencies in ways that are specific to materials. We propose to do this by developing two ways to apply genetic programming to problems in materials science and engineering. In the first application, we will construct a genetic programming algorithm for the automated generation of interatomic potential models. This will involve modifying existing genetic programming methods in a way that take into account the symmetry and connectivity of a material’s atomic structure. In the second application, strategies to discover simple relationships between materials properties will be explored. Specifically, a function that will relate the intrinsic electrical breakdown strength of insulators (an important property that is very difficult to compute or measure) to other more easily accessible material properties (such as the band gap, dielectric constant, etc.) will be sought. The results of the proposed research will provide a foundation for the development of sophisticated data-centric computational tools, based on genetic programming, which will be used by researchers to accelerate the process of materials discovery and design.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141512665
Entities
People
- Timothy K. Mueller
Organizations
- Johns Hopkins University
- Office of Naval Research
- United States Navy