Artificial Intelligence in Russia Simplifies the Search for Materials With Desired Properties
A new algorithm dramatically lowers the computational threshold for discovering polymers, superconductors and catalysts by enabling advanced searches on an ordinary laptop
Evolutionary Search Meets Machine Learning
Russian researchers have introduced USPEX 2.5, an algorithm capable of identifying new materials with specified properties using nothing more than a standard laptop. The key advantage is its hybrid approach: combining evolutionary search for optimal crystal structures with embedded machine-learning models.
USPEX 2.5 generates possible crystalline configurations, optimizes them and selects the ones that best match the desired parameters. Previously, such work required powerful and expensive supercomputers. The new version lowers the barrier to entry, making complex calculations accessible — a major step toward the “democratization of science.”
Saving Time, Expanding Possibilities
The algorithm is particularly valuable for searching for superhard materials, polymers for microelectronics, novel catalysts and high‑temperature superconductors. As Professor Artyom Oganov noted, discovering new materials is like searching for a needle in a haystack — and the new system dramatically reduces the time required.
This is not the only breakthrough in the field. Novosibirsk State University has developed a software suite integrating atomistic‑modeling methods capable of predicting material properties with high accuracy. Together, these innovations broaden the community of researchers able to design new materials and reaffirm Russia’s position among global leaders in AI‑driven materials science.
The growing number of such developments forms a foundation for future technological breakthroughs — progress that may eventually lead to major scientific discoveries.
