A better way to model the behavior of metal alloys

MIT researchers created a technique using machine-learning models to accurately simulate the behavior of metal alloys and other complex materials. The approach captures chemical arrangements across materials, improving predictions without requiring physical testing. This reduces costs and time in materials innovation for aerospace, energy, and computing.