Grad DFT: a software library for machine learning enhanced density functional theory
Authors:
Pablo A. M. Casares,
Jack S. Baker,
Matija Medvidovic,
Roberto dos Reis,
Juan Miguel Arrazola
Abstract:
Density functional theory (DFT) stands as a cornerstone method in computational quantum chemistry and materials science due to its remarkable versatility and scalability. Yet, it suffers from limitations in accuracy, particularly when dealing with strongly correlated systems. To address these shortcomings, recent work has begun to explore how machine learning can expand the capabilities of DFT; an…
▽ More
Density functional theory (DFT) stands as a cornerstone method in computational quantum chemistry and materials science due to its remarkable versatility and scalability. Yet, it suffers from limitations in accuracy, particularly when dealing with strongly correlated systems. To address these shortcomings, recent work has begun to explore how machine learning can expand the capabilities of DFT; an endeavor with many open questions and technical challenges. In this work, we present Grad DFT: a fully differentiable JAX-based DFT library, enabling quick prototy** and experimentation with machine learning-enhanced exchange-correlation energy functionals. Grad DFT employs a pioneering parametrization of exchange-correlation functionals constructed using a weighted sum of energy densities, where the weights are determined using neural networks. Moreover, Grad DFT encompasses a comprehensive suite of auxiliary functions, notably featuring a just-in-time compilable and fully differentiable self-consistent iterative procedure. To support training and benchmarking efforts, we additionally compile a curated dataset of experimental dissociation energies of dimers, half of which contain transition metal atoms characterized by strong electronic correlations. The software library is tested against experimental results to study the generalization capabilities of a neural functional across potential energy surfaces and atomic species, as well as the effect of training data noise on the resulting model accuracy.
△ Less
Submitted 11 December, 2023; v1 submitted 22 September, 2023;
originally announced September 2023.
Generative models for sampling of lattice field theories
Authors:
Matija Medvidovic,
Juan Carrasquilla,
Lauren E. Hayward,
Bohdan Kulchytskyy
Abstract:
We explore a self-learning Markov chain Monte Carlo method based on the Adversarial Non-linear Independent Components Estimation Monte Carlo, which utilizes generative models and artificial neural networks. We apply this method to the scalar $\varphi^4$ lattice field theory in the weak-coupling regime and, in doing so, greatly increase the system sizes explored to date with this self-learning tech…
▽ More
We explore a self-learning Markov chain Monte Carlo method based on the Adversarial Non-linear Independent Components Estimation Monte Carlo, which utilizes generative models and artificial neural networks. We apply this method to the scalar $\varphi^4$ lattice field theory in the weak-coupling regime and, in doing so, greatly increase the system sizes explored to date with this self-learning technique. Our approach does not rely on a pre-existing training set of samples, as the agent systematically improves its performance by bootstrap** samples collected by the model itself. We evaluate the performance of the trained model by examining its mixing time and study the ergodicity of generated samples. When compared to methods such as Hamiltonian Monte Carlo, this approach provides unique advantages such as the speed of inference and a compressed representation of Monte Carlo proposals for potential use in downstream tasks.
△ Less
Submitted 5 January, 2021; v1 submitted 2 December, 2020;
originally announced December 2020.
