Sign In

Communications of the ACM

ACM TechNews

Automating Materials Design

View as: Print Mobile App Share:
New software identified five different families of microstructures, each defined by a shared skeleton (blue).

Researchers at the Massachusetts Institute of Technology Computer Science and Artificial Intelligence Laboratory have developed a new system that allows designers to numerically specify the properties they want their materials to have, and the system gene

Credit: MIT News

Researchers at the Massachusetts Institute of Technology's (MIT) Computer Science and Artificial Intelligence Laboratory say they have developed a system that can numerically specify the properties that designers want imbued in a material, and generate a microstructure that matches the specification.

"Basically, this is a completely automated process for discovering optimal structure families for metamaterials," notes MIT professor Wojciech Matusik.

He says the system first infers a mathematical formula for reconstructing each of the modeled microstructures in a cluster, and then machine-learning techniques are applied to determine correlations between specific values for the variables in the formulae and the measured properties of the resulting microstructures.

Matusik notes every step in the process is automated, including the measurement of similarities, clustering, skeleton extraction, formula derivation, and the correlation of geometries and properties. Matusik also thinks this approach could be equally applicable to any collection of microstructures assessed according to any criteria.

From MIT News
View Full Article


Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


No entries found