Optimizing large parameter sets in variational quantum Monte Carlo

Abstract

We present a technique for optimizing hundreds of thousands of variational parameters in variational quantum Monte Carlo. By introducing iterative Krylov subspace solvers and by multiplying by the Hamiltonian and overlap matrices as they are sampled, we remove the need to construct and store these matrices and thus bypass the most expensive steps of the stochastic reconfiguration and linear method optimization techniques. We demonstrate the effectiveness of this approach by using stochastic reconfiguration to optimize a correlator product state wavefunction with a pfaffian reference for four example systems. In two examples on the two dimensional Hubbard model, we study 16 and 64 site lattices, recovering energies accurate to 1% in the smaller lattice and predicting particle-hole phase separation in the larger. In two examples involving an ab initio Hamiltonian, we investigate the potential energy curve of a symmetrically dissociated 4x4 hydrogen lattice as well as the singlet-triplet gap in free base porphin. In the hydrogen system we recover 98% or more of the correlation energy at all geometries, while for porphin we compute the gap in a 24 orbital active space to within 0.02eV of the exact result. The numbers of variational parameters in these examples range from 4x10^3 to 5x10^5, demonstrating an ability to go far beyond the reach of previous formulations of stochastic reconfiguration.

5 pages, 4 figures, suggested PACS numbers 02.70.Ss, 71.10.Fd, 31.15.-p

References in corpus (3)

• Full optimization of Jastrow-Slater wave functions with application to the first-row atoms and homonuclear diatomic molecules
• Variational quantum Monte Carlo simulations with tensor-network states
• Spatially inhomogeneous phase in the two-dimensional repulsive Hubbard model

Cited by in corpus (76)

• Density matrix embedding: A simple alternative to dynamical mean-field theory
• Deep neural network solution of the electronic Schrödinger equation
• Ab-Initio Solution of the Many-Electron Schrödinger Equation with Deep Neural Networks
• Neural-Network Quantum States, String-Bond States, and Chiral Topological States
• Density Matrix Embedding from Broken Symmetry Lattice Mean-Fields
• Particle-particle and quasiparticle random phase approximations: Connections to coupled cluster theory
• Origin of high-Tc superconductivity in doped Hubbard models and their extensions: Roles of uniform charge fluctuations
• An exactly size consistent geminal power via Jastrow factor networks in a local one particle basis
• mVMC - Open-source software for many-variable variational Monte Carlo method
• Competition among various charge-inhomogeneous states and d-wave superconducting state in Hubbard models on square lattices
• The Jastrow antisymmetric geminal power in Hilbert space: theory, benchmarking, and application to a novel transition state
• The Shape of Full Configuration Interaction to Come
• TurboRVB: a many-body toolkit for {\it ab initio} electronic simulations by quantum Monte Carlo
• Stripe and superconducting order competing in the Hubbard model on a square lattice studied by a combined variational Monte Carlo and tensor network method
• Excited states with selected CI-QMC: chemically accurate excitation energies and geometries
• Perturbatively selected configuration-interaction wave functions for efficient geometry optimization in quantum Monte Carlo
• Wave function Ansatz (but Periodic) Networks and the Homogeneous Electron Gas
• Multi-reference symmetry-projected variational approaches for ground and excited states of the one-dimensional Hubbard model
• Symmetry-projected variational approach for ground and excited states of the two-dimensional Hubbard model
• Time-dependent many-variable variational Monte Carlo method for nonequilibrium strongly correlated electron systems
• Message-Passing Neural Quantum States for the Homogeneous Electron Gas
• The HANDE-QMC project: open-source stochastic quantum chemistry from the ground state up
• Variational Monte Carlo method for fermionic models combined with tensor networks and applications to the hole-doped two-dimensional Hubbard model
• A machine learning approach to dynamical properties of quantum many-body systems
• Hybrid convolutional neural network and PEPS wave functions for quantum many-particle states
• Striped spin liquid crystal ground state instability of kagome antiferromagnets
• A Projector Quantum Monte Carlo Method for non-linear wavefunctions
• Local density approximation in site-occupation embedding theory
• Direct sampling of projected entangled-pair states
• Scaling of neural-network quantum states for time evolution
• Quantum Monte Carlo with Coupled-Cluster wave functions
• A hybrid approach to excited-state-specific variational Monte Carlo and doubly excited states
• Projector quantum Monte Carlo with matrix product states
• Tailoring CIPSI expansions for QMC calculations of electronic excitations: the case study of thiophene
• Mitigating the Sign Problem Through Basis Rotations
• Complementary First and Second Derivative Methods for Ansatz Optimization in Variational Monte Carlo
• Artificial Neural Network Based Computation for Out-of-Time-Ordered Correlators
• Symmetry projected Jastrow mean field wavefunction in variational Monte Carlo
• Excitons and Dark Fermions as Origins of Mott Gap, Pseudogap and Superconductivity in Cuprate Superconductors --- General Idea and Basic Concept Based on Gap Physics
• Explicitly antisymmetrized neural network layers for variational Monte Carlo simulation
• Accurate quantum Monte Carlo forces for machine-learned force fields: Ethanol as a benchmark
• A Kaczmarz-inspired approach to accelerate the optimization of neural network wavefunctions
• A correlator product state study of molecular magnetism in the giant Keplerate Mo72Fe30
• Variational Monte Carlo simulations using tensor-product projected states
• Accurate Simulation of the Hubbard Model with Finite Fermionic Projected Entangled Pair States
• An accelerated linear method for optimizing non-linear wavefunctions in variational Monte Carlo
• Equations of state and stability of MgSiO${}_3$ perovskite and post-perovskite phases from quantum Monte Carlo simulations
• Non-linear biases, stochastically-sampled effective Hamiltonians and spectral functions in quantum Monte Carlo methods
• Variational description of the ground state of the repulsive two-dimensional Hubbard model in terms of nonorthogonal symmetry-projected Slater determinants
• Charge Order and Superconductivity as Competing Brothers in Cuprate High-$T_{\mathrm{c}}$ Superconductors
• Band-Renormalization Effects and Predominant Antiferromagnetic Order in Two-Dimensional Hubbard Model
• Artificial Intelligence for Science in Quantum, Atomistic, and Continuum Systems
• Charge-transfer excited states: Seeking a balanced and efficient wave function ansatz in variational Monte Carlo
• Minimal matrix product states and generalizations of mean-field and geminal wavefunctions
• Solving the nuclear pairing model with neural network quantum states
• The quantum Gaussian process state: A kernel-inspired state with quantum support data
• Autoregressive neural Slater-Jastrow ansatz for variational Monte Carlo simulation
• Superconductivity in bilayer $t$-$t^{\prime }$ Hubbard models
• A transformed framework for dynamic correlation in multireference problems
• Neural-network Quantum States for Spin-1 systems: spin-basis and parameterization effects on compactness of representations
• Multi-reference symmetry-projected variational approximation for the ground state of the doped one-dimensional Hubbard model
• Tensor Network Computations That Capture Strict Variationality, Volume Law Behavior, and the Efficient Representation of Neural Network States
• Rayleigh-Gauss-Newton optimization with enhanced sampling for variational Monte Carlo
• Deep learning quantum Monte Carlo for solids
• Ab-Initio Potential Energy Surfaces by Pairing GNNs with Neural Wave Functions
• Tensor Network States with Three-Site Correlators
• Time Evolution and Deterministic Optimisation of Correlator Product States
• Neural-network quantum state study of the long-range antiferromagnetic Ising chain
• Using Local Operator Fluctuations to Identify Wave Function Improvements
• Convergence of variational Monte Carlo simulation and scale-invariant pre-training
• Efficiency of neural-network state representations of one-dimensional quantum spin systems
• Neuralized Fermionic Tensor Networks for Quantum Many-Body Systems
• A toolbox for elementary fermions with a dipolar Fermi gas in a 3D optical lattice
• Highly accurate ab-initio thermochemistry via real-space quantum Monte Carlo: Benzene
• Bridging the Gap between Deep Learning and Frustrated Quantum Spin System for Extreme-scale Simulations on New Generation of Sunway Supercomputer
• Assessing Orbital Optimization in Variational and Diffusion Monte Carlo