Anders Niklasson
Theoretical Division, Los Alamos National Laboratory, USA
Tensor cores, along with tensor processing units, represent a new form of hardware acceleration specifically designed for deep neural network calculations in artificial intelligence (AI) applications. AI optimized hardware provides extraordinary computational speed and energy efficiency, but with the caveat that it is designed for tensor contractions (matrix-matrix multiplications) using only low-precision floating point operations.
In spite of this, we demonstrate how tensor cores can be applied with high efficiency to the challenging and numerically sensitive problem of quantum-based Born-Oppenheimer molecular dynamics, which requires highly accurate electronic structure optimizations and conservative force evaluations. The interatomic forces are calculated on-the-fly from an electronic structure that is obtained from a generalized deep neural network, where the computational structure naturally takes advantage of the exceptional processing power of the AI hardware and allows for high performance in excess of 100 Tflops on the tensor cores of a single Nvidia A100 GPU.
Stable molecular dynamics trajectories are generated using the framework of extended Lagrangian Born-Oppenheimer molecular dynamics (XL-BOMD). XL-BOMD is based on a backward error analysis or a shadow Hamiltonian approach that combines computational efficiency with long-term stability, even when using approximate charge relaxations and force evaluations that are limited in accuracy by the numerically noisy conditions caused by the low precision floating-point operations of the AI hardware. A canonical ensemble simulation scheme is also presented, where the additional numerical noise in the calculated forces is absorbed into a Langevin-like dynamics.