Machine learning takes materials modeling into new era
Electronic structure — the arrangement of electrons in matter — underpins research areas from drug design to energy storage, yet no simulation technique has offered both high fidelity and scalability across time and length scales. The department’s machine-learning method, Materials Learning Algorithms (MALA), replaces traditional electronic-structure simulation and makes such calculations possible at previously unattainable length scales; the results appeared in npj Computational Materials.