GPU-Accelerated Simulations of Thermal Transport using Machine Learning Force Fields

Controlling thermal conductivities of materials is important for a wide range of applications, from thermoelectrics for clean energy generation to electronic devices and thermal barrier coatings. The thermal conductivity is commonly estimated using molecular dynamics simulations within the Green-Kubo formulation. This requires a force field that is both 1) an accurate estimate of the interatomic interactions and 2) fast enough to allow simulations with sufficiently large length and time scales. An accurate force field can also be used to accelerate the calculation of force constants for transport simulations via the Boltzmann transport equation.

   

In this work, we employ the Gaussian Process-based FLARE force field, which automatically learns the interactions of more complex materials than empirical force fields. The resulting model can then be mapped to a low-dimensional, computationally efficient model. Through GPU-acceleration with LAMMPS and the Kokkos library, we achieve excellent performance and obtain well-converged estimates for the thermal conductivity with Green-Kubo molecular dynamics. Our newly developed transport code, Phoebe, computes the thermal conductivity with force constants calculated by FLARE. The data efficiency of FLARE reduces the total number of DFT calculations required, and Phoebe has been highly optimized to solve the Boltzmann transport equation efficiently on GPUs using the Kokkos and cuSOLVER libraries.