Exact Molecular Hessians from Local MLIPs in O(N) Time Using Sparse Differentiation

This arXiv paper presents a method for computing exact molecular Hessians from machine learning interatomic potentials (MLIPs) in O(N) time using sparse differentiation. The Hessian of the energy with respect to nuclear positions is fundamental in atomistic modeling, but traditional approaches require O(N²) cost. The work enables high-accuracy Hessian calculations for larger molecular systems that were previously computationally prohibitive.