Structural optimization in fingerprint space

Structural optimization has been a crucial component in computational materials research, and structure predictions have relied heavily on this technique in particular. By introducing an extra fingerprint space, we propose a new structural optimization approach that prevents configurations from being stranded in low-symmetry, high-energy conformations. Using this strategy, the chance of achieving low-energy designs has been significantly increased. This performance boost is anticipated to be advantageous for structure search methods that rely on the local optimization of structures. Therefore, the work provides a path toward the objective of predicting the crystal structure of complex systems.