Quantum Monte Carlo

Quantum Monte Carlo
Name	Quantum Monte Carlo
Field	Computational physics; Computational chemistry
Introduced	1950s
Notable	John von Neumann, Enrico Fermi, Richard Feynman, David Ceperley, C. J. Umrigar

Quantum Monte Carlo Quantum Monte Carlo refers to a family of stochastic computational techniques for solving quantum many-body problems by sampling probability distributions with random processes. Developed and applied across Los Alamos National Laboratory, Princeton University, Massachusetts Institute of Technology, and University of Illinois at Urbana–Champaign, these methods bridge theoretical frameworks from Paul Dirac and Werner Heisenberg to numerical implementations used in studies associated with Nobel Prize in Physics laureates such as Richard Feynman and Enrico Fermi. Researchers use Quantum Monte Carlo to obtain high-accuracy energies, correlation functions, and response properties where deterministic methods like Hartree–Fock or Density functional theory may be insufficient.

Introduction

Quantum Monte Carlo techniques employ stochastic sampling to evaluate integrals and expectation values arising in the Schrödinger equation, path integral formulations, and second-quantized Hamiltonians. Foundational influences include work by John von Neumann, Stanislaw Ulam, and Nicholas Metropolis who contributed to early Monte Carlo machinery, while conceptual underpinnings draw on path integrals by Richard Feynman and variational principles related to Erwin Schrödinger. Practitioners often contrast QMC with deterministic approaches pioneered at institutions like Bell Labs and IBM Research.

Methods and Algorithms

Common algorithmic families include variational Monte Carlo (VMC), diffusion Monte Carlo (DMC), and path integral Monte Carlo (PIMC), each developed with contributions from groups at University of Illinois at Urbana–Champaign, Rutgers University, and Cornell University. VMC uses trial wavefunctions often parameterized with techniques associated with Hermann A. Haus and optimization methods from Stephen Boyd and Michael I. Jordan-level machine learning. DMC implements projector methods with importance sampling originating in work by David Ceperley and B. J. Alder, and uses fixed-node approximations linked historically to C. J. Umrigar. PIMC samples thermal density matrices inspired by Richard Feynman’s path integral formalism and was advanced in studies at Los Alamos National Laboratory and Argonne National Laboratory. Auxiliary-field QMC (AFQMC) and continuous-time QMC trace back to methods developed at University of Cambridge and Columbia University for lattice models studied by Kenneth G. Wilson and Philip W. Anderson. Sampling techniques employ Markov chain Monte Carlo strategies refined from Nicholas Metropolis and Marshall N. Rosenbluth, while variance reduction and reweighting use ideas popularized at Los Alamos National Laboratory and in statistical mechanics by Lev Landau.

Applications

Quantum Monte Carlo has been deployed for electronic structure problems in molecules and solids studied at Harvard University, California Institute of Technology, and University of California, Berkeley, for simulating superfluidity and Bose–Einstein condensation related to experiments at University of Colorado Boulder and MIT. In condensed matter physics, QMC informs models of high-temperature superconductivity linked to Bednorz and Müller-related research and Hubbard model studies by Kenneth G. Wilson and Philip W. Anderson. In nuclear physics, QMC methods are applied to light nuclei and neutron matter in work associated with Argonne National Laboratory and Oak Ridge National Laboratory. Quantum chemistry applications complement methods developed at ETH Zurich and Max Planck Institute for Solid State Research.

Implementation and Software

Major software packages implementing QMC algorithms include projects originating from research groups at Princeton University, University of Illinois at Urbana–Champaign, and Argonne National Laboratory. Examples developed in academic collaborations and national laboratories parallel software ecosystems like Quantum ESPRESSO and GAMESS in scope, while packages interface with quantum chemistry codes from Gaussian (software) and NWChem workflows. High-performance implementations exploit architectures from NVIDIA GPUs and supercomputers at Oak Ridge Leadership Computing Facility and Argonne Leadership Computing Facility.

Accuracy, Scaling, and Limitations

QMC methods can achieve benchmark accuracy rivaling correlated wavefunction theories developed by John Pople and Walter Kohn but face the fermion sign problem connected conceptually to work by John von Neumann and computational complexity results related to Stephen Cook. Scaling with particle number and basis complexity challenges applications to large systems, prompting hybrid strategies integrating density functional approximations from Walter Kohn and tensor network insights from Guifre Vidal. Approximations such as the fixed-node constraint, importance sampling, and pseudopotentials developed at Los Alamos National Laboratory control errors but introduce biases studied by groups at Argonne National Laboratory.

Historical Development

Monte Carlo sampling traces to Stanislaw Ulam and Nicholas Metropolis in the Manhattan Project era at Los Alamos National Laboratory, while quantum variants emerged with studies by B. J. Alder and David Ceperley on electron liquids and liquids helium. Variational Monte Carlo matured through work by C. J. Umrigar and contemporaries at Cornell University and Princeton University, and diffusion and path integral formulations were advanced at Massachusetts Institute of Technology and University of Cambridge. Cross-pollination with quantum chemistry progressed via collaborations involving John Pople and groups at Harvard University.

Recent Advances and Future Directions

Recent progress links QMC to machine learning techniques from Geoffrey Hinton and Yoshua Bengio for wavefunction ansätze, integrations with tensor network methods advanced by Guifre Vidal, and algorithmic improvements from researchers at Caltech and Stanford University. Efforts toward mitigating the sign problem involve ideas explored at Perimeter Institute and Institute for Advanced Study, while exascale deployments target resources at Oak Ridge National Laboratory and Argonne National Laboratory. Future directions include tighter integration with quantum computing experiments at IBM Research and Google (company) and applications to materials initiatives associated with Materials Project.

Category:Computational physics