• Corpus ID: 247779339

Ab initio calculation of real solids via neural network ansatz

  title={Ab initio calculation of real solids via neural network ansatz},
  author={Xiang Li and Zhe Li and Ji Chen},
Neural networks have been applied to tackle many-body electron correlations for small molecules and physical models in recent years. Here we propose a new architecture that extends molecular neural networks with the inclusion of periodic boundary conditions to enable ab initio calculation of real solids. The accuracy of our approach is demonstrated in four different types of systems, namely the one-dimensional periodic hydrogen chain, the two-dimensional graphene, the three-dimensional lithium… 

Figures and Tables from this paper

Sampling-free Inference for Ab-Initio Potential Energy Surface Networks
This work proposes the PlaNet framework to simultaneously train a surrogate model that avoids expensive Monte-Carlo integration and reduces inference time from minutes or even hours to milliseconds, which can accurately model high-resolution multi-dimensional energy surfaces that previously would have been unobtainable via neural wave functions.
Gold-standard solutions to the Schrödinger equation using deep learning: How much physics do we need?
A novel deep learning architecture is introduced that achieves 40-70% lower energy error at 8x lower computational cost compared to previous approaches and establishes a new benchmark by calculating the most accurate variational ground state energies ever published for a number of different atoms and molecules.


Ab-Initio Solution of the Many-Electron Schrödinger Equation with Deep Neural Networks
Deep neural networks can improve the accuracy of variational quantum Monte Carlo to the point where it outperforms other ab-initio quantum chemistry methods, opening the possibility of accurate direct optimisation of wavefunctions for previously intractable molecules and solids.
Towards an exact description of electronic wavefunctions in real solids
The application of an exact technique, full configuration interaction quantum Monte Carlo to a variety of real solids, providing reference many-electron energies that are used to rigorously benchmark the standard hierarchy of quantum-chemical techniques, up to the ‘gold standard’ coupled-cluster ansatz.
Towards the ground state of molecules via diffusion Monte Carlo on neural networks
Diffusion Monte Carlo (DMC) based on fixed-node approximation has enjoyed significant devel-opments in the past decades and become one of the go-to methods when accurate ground state energy of molecules
Deep-neural-network solution of the electronic Schrödinger equation
High-accuracy quantum chemistry methods struggle with a combinatorial explosion of Slater determinants in larger molecular systems, but now a method has been developed that learns electronic wavefunctions with deep neural networks and reaches high accuracy with only a few determinants.
Ab initio electronic density in solids by many-body plane-wave auxiliary-field quantum Monte Carlo calculations
We present accurate many-body results of the electronic densities in several solid materials, including Si, NaCl, and Cu. These results are obtained using the ab initio auxiliary-field quantum Monte
Fermionic neural network with effective core potential
This work integrates a powerful neural-network based model (FermiNet) with the effective core potential method, which helps to reduce the complexity of the problem by replacing inner core electrons with additional semi-local potential terms in Hamiltonian.
Bulk and surface energetics of crystalline lithium hydride: Benchmarks from quantum Monte Carlo and quantum chemistry
We show how accurate benchmark values of the surface formation energy of crystalline lithium hydride can be computed by the complementary techniques of quantum Monte Carlo (QMC) and
Pushing the frontiers of density functionals by solving the fractional electron problem
The present work offers a solution to a long-standing critical problem in DFT and demonstrates the success of combining DFT with the modern machine-learning methodology, and trains a neural network on molecular data and on fictitious systems with fractional charge and spin.
A shortcut to the thermodynamic limit for quantum many-body calculations of metals
Computationally efficient and accurate quantum mechanical approximations to solve the many-electron Schrödinger equation are crucial for computational materials science. Methods such as coupled