Ad blocker detected: Our website is made possible by displaying online advertisements to our visitors. Please consider supporting us by disabling your ad blocker on our website.
Microsoft researchers unveiled MatterGen, a new AI model that can generate stable, novel materials across the periodic table.
The details: First announced a year ago, MatterGen, when given prompts with design requirements and constraints, can “generate materials with desired chemistry, mechanical, electronic or magnetic properties.”
According to the researchers, materials generated by MatterGen are twice as likely to be novel and stable when compared to previous generative models. As a proof of concept, the researchers successfully synthesized a novel material based on MatterGen.
Similar to AI image generators, MatterGen is a diffusion model. Operating on a 3D geometry of materials, MatterGen works by adjusting the positions, elements and periodic lattice of a random structure. Trained on more than 600,000 materials, the researchers say MatterGen achieves state-of-the-art performance in generating novel, stable and diverse materials.
Why it matters: Material innovation, according to Microsoft, has “propelled technological innovation for centuries.” Referencing the discovery of lithium cobalt oxide in the 1980s, as well as advancements in silicon chips and superconductors, the researchers wrote that areas like healthcare, energy storage and carbon capture are all reliant on the rapid design of functional new materials.
