Hungarian researchers achieve breakthrough in quantum-chemical calculations

In the course of the work, the collaborating researchers successfully investigated two exceptionally complex molecular systems: FeMoco and cytochrome P450. FeMoco plays a key role in the conversion of atmospheric nitrogen into ammonia, while cytochrome P450 is an important liver enzyme involved in the breakdown of a wide range of chemical compounds. Achieving an accurate description of these systems is one of the greatest challenges in computational chemistry.

In this research, the team exploited the capabilities of the NVIDIA Blackwell architecture, which can efficiently handle simulations of this complexity. The researchers employed a mixed-precision computational approach: where approximation was sufficient, they used faster, lower-precision calculations, while ensuring maximum accuracy in the critical steps.

The solution is based on the Density Matrix Renormalisation Group (DMRG) method, which has been further developed by Örs Legeza. This approach makes it possible to study systems containing many interacting electrons, which are particularly important, for example, for understanding catalysis and the behaviour of semiconductors.

The researchers’ findings show that hardware originally designed for artificial intelligence is also capable of tackling the most difficult problems in quantum chemistry with high accuracy.

In the longer term, this could mean that quantum-chemical calculations that currently require supercomputers may become routine, which could accelerate the development of new catalysts, semiconductor materials, and pharmaceuticals.

“Our study shows that AI-oriented hardware is not only fast, but is also capable of handling strongly correlated quantum-chemical problems at the limits of computational feasibility with high accuracy,” said Sotiris Xantheas, a computational chemist at PNNL.

“By achieving chemical accuracy with the mixed-precision DMRG method, we have opened a practical path towards the application of next-generation Blackwell systems in catalysis, bioinorganic chemistry, and materials science,” Örs Legeza emphasised.

The research was supported by the Hungarian National Research, Development and Innovation Office (NKFIH), the Hans Fischer Senior Fellowship Programme of the Technical University of Munich, and the SPEC initiative of the US Department of Energy.

The original American press release is available HERE.