Fig. 4: Computational efficiency and molecular dynamics (MD) stability.
From: A Euclidean transformer for fast and stable machine learned force fields
a Number of frames per second (FPS) vs. the averaged stability coefficient (upper panel) and FPS vs. the averaged force mean absolute error (MAE) (lower panel) for four small organic molecules from the MD17 data set as reported in32 for different state-of-the-art MPNN architectures12,36,37,43,50,94,95,96. b Since run times are sensitive to hardware and software implementation details, we re-implement two representative models along the trade-off lines under settings identical to the SO3krates MLFF (using jax), which yields framework-corrected FPS (dashed vs. solid line). We observe speed-ups between 28 (for NequIP) and 15 (for SchNet) in our re-implementations. We find, that SO3krates enables reliable MD simulations and high accuracies without sacrificing computational performance. Gray shaded area indicates the regime of sub-millisecond step time. c MD step time vs. the number of atoms in the system. The smaller pre-factor in the computational complexity compared to SO(3) convolutions (Table 1) results in computational speed-ups that grow in system size. d MD stability observed at temperatures 300 K and 500 K. The transition to higher temperatures results in a drop of stability for the invariant model, hinting towards less robustness and weaker extrapolation behavior. Flexible molecules such as DHA pose a challenge for the invariant model at 300 K already. Bar height is the mean stability over six MD runs and yellow dots denote stability for individual MD runs. Error bars correspond to the 2σ confidence interval. Per-structure error distributions for an invariant and an equivariant SO3krates model with the same error on the test set. Spread and mean of the error distributions are given in Supplementary Table 1.
