Heat a crystal, and its atoms start to jiggle like plucked strings. An X-ray beam passing through the crystal scatters off these vibrations, painting on the detector not only sharp spots but also faint, smeared patterns. These patterns are like fingerprints: from them, you can reconstruct how strongly atoms pull and push on each other.

Previously, such reconstruction was like picking a lock with a hundred pins. The program had to repeatedly solve cumbersome equations, and each step took a long time. The new work takes a different route: instead of solving the equations directly, it adjusts the forces so that the calculated pattern matches what's seen in the image. It does this using a fast way to generate random atomic vibrations, bypassing the bottleneck of old methods.

This approach makes analyzing X-ray images fast and automatic. Now we can study en masse how heat propagates in different crystals — from semiconductors to battery materials. It's a step toward getting a complete picture of a material's thermal properties from a single image.