Machine learning shortcuts advanced materials design
Their results, say the researchers, could herald a dramatic acceleration in the creation of new technologies for optics, aerospace, energy storage, and potentially medicine while simultaneously saving laboratories money on computing costs.
“We’re shortening the design cycle,” says David Montes de Oca Zapiain, a computational materials scientist at Sandia who helped lead the research. “The design of components grossly outpaces the design of the materials you need to build them. We want to change that. Once you design a component, we’d like to be able to design a compatible material for that component without needing to wait for years, as it happens with the current process.”
The researchers used machine learning to accelerate a computer simulation that predicts how changing a design or fabrication process, such as tweaking the amounts of metals in an alloy, will affect a material. A project might require thousands of simulations, which can take weeks, months or even years to run.
In their work, the researchers clocked a single, unaided simulation on a high-performance computing cluster with 128 processing cores at 12 minutes. With machine learning, the same simulation took 60 milliseconds using only 36 cores – equivalent to 42,000 times faster on equal computers. This, say the researchers, means that they can now learn in under 15 minutes what would normally take a year.
The researchers’ new algorithm arrived at an answer that was 5% different from the standard simulation’s result – a very accurate prediction for the team’s purposes. Machine learning trades some accuracy for speed because it makes approximations to shortcut calculations.
“Our machine-learning framework achieves essentially the same accuracy as the high-fidelity model but at a fraction of the computational cost,” says Sandia materials scientist Rémi Dingreville, who also worked on the project.
The researchers say they are going to use their algorithm first to research ultrathin optical technologies for next-generation monitors and screens. Their research, though, could prove widely useful because the simulation they accelerated describes a common event – the change, or evolution, of a material’s microscopic building blocks over time.
While machine learning has previously been used to shortcut simulations that calculate how interactions between atoms and molecules change over time, the researchers say their results demonstrate the first use of machine learning to accelerate simulations of materials at relatively large, microscopic scales, which they expect will be of greater practical value to scientists and engineers. For instance, scientists can now quickly simulate how miniscule droplets of melted metal will glob together when they cool and solidify, or conversely, how a mixture will separate into layers of its constituent parts when it melts.
Many other natural phenomena, say the researchers, including the formation of proteins, follow similar patterns. And while the researchers have not tested the machine learning algorithm on simulations of proteins, they say they are interested in exploring the possibility in the future.
The research was funded by the U.S. Department of Energy’s Basic Energy Sciences program and was conducted at the Center for Integrated Nanotechnologies, a DOE user research facility jointly operated by Sandia and Los Alamos national labs. For more, see “Accelerating phase-field-based microstructure evolution predictions via surrogate models trained by machine learning methods.”
Related articles:
Machine learning aids search for ‘superhard’ materials
AI material prediction platform launched
Algorithm mines materials science literature for new discoveries
AI finds new materials for supercapacitors
ML algorithm captures 3D microstructures in real time
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :