Predicting the quantum mechanical properties of large molecular systems

Record for quantum chemistry calculation.


Follow us onFollow Tech Explorist on Google News

Using the world’s powerful supercomputer- Summit supercomputer at the Oak Ridge National Lab in the U.S., scientists from The Australian National University (ANU), have successfully predicted the quantum mechanical properties of large molecular systems. The calculations are so accurate that it surpasses all previous experiments. What’s more, it has broken the world record for the largest Hartree-Fock calculation ever performed, setting new standards in High-Performance Computing.

Such calculations can potentially solve significant energy generation problems, fuel production, water purification, and the manufacturing of medicines, foods, textiles, and consumer goods.

The Hartree-Fock method decides the electronic structure and the energy of a quantum mechanical molecular system.

Scientists ran their calculations for over half an hour using 26,268 NVIDIA V100 Graphics Processing Units (GPUs). They then simulated 20,063 water molecules at a resolution never before possible.

Dr. Giuseppe Barca from the Australian National University (ANU) said, “The new algorithm brings quantum-mechanical to the next level in terms of molecular sizes, enabling us to reach scales so large they belong to the domains of biology.”

“Such computational predictions open entirely new research horizons in areas where experiments are too expensive or simply impracticable. This result sets the benchmark for comparison for years to come.”

Professor Sean Smith, director of the National Computational Infrastructure, said the calculation scale was “massive.”

“GPUs are computationally more powerful and energy-efficient than CPUs, but much more difficult to harness. Using tens of thousands of GPU cores with such efficacy is a computational grand challenge.”

Journal Reference:
  1. Barca et al., Scaling the Hartree-Fock Matrix Build on Summit. Proceedings of the International Conference for High-Performance Computing, Networking, Storage, and Analysis (2020). DOI: 10.5555/3433701.3433808


See stories of the future in your inbox each morning.