Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:cond-mat/0404493 · The Journal of Chemical Physics

Higher order and infinite Trotter-number extrapolations in path integral Monte Carlo

L. Brualla, K. Sakkos, Jordi Boronat, J. Casulleras

Published Jun 23, 2004• 83 citations
arXivHTMLPDF

Abstract

Improvements beyond the primitive approximation in the path integral Monte Carlo method are explored both in a model problem and in real systems. Two different strategies are studied: The Richardson extrapolation on top of the path integral Monte Carlo data and the Takahashi-Imada action. The Richardson extrapolation, mainly combined with the primitive action, always reduces the number-of-beads dependence, helps in determining the approach to the dominant power law behavior, and all without additional computational cost. The Takahashi-Imada action has been tested in two hard-core interacting quantum liquids at low temperature. The results obtained show that the fourth-order behavior near the asymptote is conserved, and that the use of this improved action reduces the computing time with respect to the primitive approximation.

DOI
10.1063/1.1760512

Related papers

Extrapolation to infinite Trotter number in path-integral Monte Carlo simulations of solid-state systems

Alessandro Cuccoli, Alessandro Macchi, Gaia Pedrolli +2 more

We consider the problem of the extrapolation of path-integral Monte Carlo (PIMC) data to infinite Trotter number P. Finite-P data, being even functions of P, have high-P dependence that is generally well described by a quadratic fit, ${\mathit{a}}_{0}$+${\mathit{a}}_{1}$${\mathit{P}}^{\mathrm{\ensuremath{-}}2}$, where ${\mathit{a}}_{0}$ is the exact quantum value. However, in order to get convergence it is often necessary to run PIMC codes with rather high P values, which implies long computer times and larger statistical errors of the data. It is well known that also for harmonic systems the finite-P data are not exact; nevertheless, they can be easily calculated by Gaussian quadrature. Starting from this observation, we suggest an easy way to correct PIMC data for anharmonic systems in order to take into account the harmonic part exactly, with strong improvement of the extrapolation to P=\ensuremath{\infty}. Lower Trotter numbers are thus required, with the advantages of computer-time saving and much better accuracy of the extrapolated values, without any change in the PIMC code. In order to demonstrate the effectiveness of the approach, we report finite-P data processing for a single anharmonic particle, whose finite-P data are obtained by the matrix-squaring method, and for a chain of atoms with Morse interaction.

Physical review. B, Condensed matter•May 1, 1995•14 citations

Path integral Monte Carlo for dissipative many-body systems

Luca Capriotti, R. VAIA, Alessandro Cuccoli +2 more

We address the possibility of performing numerical Monte Carlo simulations for the thermodynamics of quantum dissipative systems. Dissipation is considered within the Caldeira-Leggett formulation, which describes the system in the path-integral formalism through the inclusion of an influence action that is bilocal and quadratic in the system's coordinates. At a first sight the usual direct approach of discretizing the path integral could seem feasible, but complications arise when one tries to introduce a physically meaningful dissipation kernel: in particular its imaginary-time dependence turns out to be severely singular and difficult to evaluate analytically, in spite of the simple expressions for its Matsubara components. We therefore propose to face the numerical problem using Fourier path-integral Monte Carlo, that can be formulated in two different ways: transforming the continuous paths and then truncating the high Fourier components (with possible improvements upon the truncation procedure), or performing the Fourier transformation upon the discretized paths. The latter choice leads to a simpler formulation and allows for a better control of the extrapolation to the limit of infinite Trotter number. The method is implemented for a single nonlinear particle with Ohmic dissipation and for a phi(4) chain with Drude-like dissipation.

•Jan 1, 2003•2 citations

Hamiltonians, path integrals, and a new renormalization group

Marvin Weinstein

New nonperturbative methods for dealing with Hamiltonian systems are introduced. The derivation of these methods requires identities for rewriting exponentials of sums of operators which are different from the usual Campbell-Hausdorff formula. These identities allow one to derive approximations to ${e}^{\ensuremath{-}\ensuremath{\delta}H}$ which are correct to higher order in $\ensuremath{\delta}$ and which contain fewer terms than the Campbell-Hausdorff formula. This allows one to generate path-integral actions which are more accurate for finite-size steps in time and which can be exploited to improve the rate of convergence of Monte Carlo calculations. To show that these methods allow one to include effects which show up in the stationary-phase approximation to the path integral, e.g., instantons, solitons, etc., I not only derive the connection between the Hamiltonian and path-integral formalism but the relationship between a specific stationary-phase approximation and the corresponding Hamiltonian calculation. My focus, however, is the direct application of these new identities to the study of nonperturbative Hamiltonian dynamics. I show that these methods are easier to apply and give better results than those based upon the naive $t$-expansion, block mean-field, or real-space renormalization ideas. Comparison of these older methods with the computational tools introduced in this paper are discussed in the context of simple examples. It is shown that the new methods allow one to extrapolate answers to finite $t$ without the use of Pad\'e approximants.

Physical review. D. Particles, fields, gravitation, and cosmology/Physical review. D. Particles and fields•Jun 15, 1993•11 citations

Path-integral Monte Carlo simulations without the sign problem: Multilevel blocking approach for effective actions

Reinhold Egger, Lothar Mühlbacher, Chi H. Mak

The multilevel blocking algorithm recently proposed as a possible solution to the sign problem in path-integral Monte Carlo simulations has been extended to systems with long-ranged interactions along the Trotter direction. As an application, results for the real-time quantum dynamics of the spin-boson model are presented.

arXiv:physics/0002031•Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics•May 1, 2000•72 citations

A high-temperature approximation for the path-integral quantum Monte Carlo method

Miroslav Kolář, S. F. O’Shea

A high-temperature approximation for the discretized path-integral quantum Monte Carlo (PIQMC) method is formulated. At higher temperatures, all P fictitious classical particles representing a single quantum particle stay close together, and an efficient approximation is obtained when, in the primitive short-time propagator, an essentially local harmonic approximation is used for the external potential at the common centre of mass of P fictitious particles - the integration over P - 1 dimensions can then be carried out analytically, and a classical formula of the effective-potential type is obtained for the partition function. Also discussed are the proper form and applicability of the virial total-energy estimator for finite systems, and the computation of the temperature dependence of the configurational partition function in a single PIQMC run.

Journal of Physics A Mathematical and General•Jul 7, 1996•9 citations

A Path Integral Monte Carlo Method for the Quasielastic Response

Carlo Carraro

We formulate the quasielastic response of a non-relativistic many-body system at zero temperature in terms of ground state density matrix elements and real time path integrals that embody the final state interactions. While the former provide the weight for a conventional Monte Carlo calculation, the latter require a more sophisticated treatment. We argue that the recently developed Stationary Phase Monte Carlo technique can be used to study the approach to "Y-scaling." We perform calculations for a particle in a potential well in one and three dimensions and compare them to the exact results available for these models. We then derive an eikonal approximation to the Path Integrals. This method is suitably generalized to treat strongly repulsive interactions, and allows comparison to Silver's theory of final state interactions in a straightforward way. We also give an exact prescription to calculate the scaling limit for potentials comprising a hard core. Finally, we study the approach to scaling in a model 4He nucleus, and find good agreement with experimental data.

•Jun 14, 2007•1 citation
Improvements beyond the primitive approximation in the path integral Monte Carlo method are explored both in a model problem and in real systems. Two different strategies are studied: The Richardson extrapolation on top of the path integral Monte Carlo data and the Takahashi-Imada action. The Richardson extrapolation, mainly combined with the primitive action, always reduces the number-of-beads dependence, helps in determining the approach to the dominant power law behavior, and all without additional computational cost. The Takahashi-Imada action has been tested in two hard-core interacting quantum liquids at low temperature. The results obtained show that the fourth-order behavior near the asymptote is conserved, and that the use of this improved action reduces the computing time with respect to the primitive approximation.
We consider the problem of the extrapolation of path-integral Monte Carlo (PIMC) data to infinite Trotter number P. Finite-P data, being even functions of P, have high-P dependence that is generally well described by a quadratic fit, a0{\mathit{a}}_{0}a0​+a1{\mathit{a}}_{1}a1​${\mathit{P}}^{\mathrm{\ensuremath{-}}2}$, where a0{\mathit{a}}_{0}a0​ is the exact quantum value. However, in order to get convergence it is often necessary to run PIMC codes with rather high P values, which implies long computer times and larger statistical errors of the data. It is well known that also for harmonic systems the finite-P data are not exact; nevertheless, they can be easily calculated by Gaussian quadrature. Starting from this observation, we suggest an easy way to correct PIMC data for anharmonic systems in order to take into account the harmonic part exactly, with strong improvement of the extrapolation to P=\ensuremath{\infty}. Lower Trotter numbers are thus required, with the advantages of computer-time saving and much better accuracy of the extrapolated values, without any change in the PIMC code. In order to demonstrate the effectiveness of the approach, we report finite-P data processing for a single anharmonic particle, whose finite-P data are obtained by the matrix-squaring method, and for a chain of atoms with Morse interaction.
We address the possibility of performing numerical Monte Carlo simulations for the thermodynamics of quantum dissipative systems. Dissipation is considered within the Caldeira-Leggett formulation, which describes the system in the path-integral formalism through the inclusion of an influence action that is bilocal and quadratic in the system's coordinates. At a first sight the usual direct approach of discretizing the path integral could seem feasible, but complications arise when one tries to introduce a physically meaningful dissipation kernel: in particular its imaginary-time dependence turns out to be severely singular and difficult to evaluate analytically, in spite of the simple expressions for its Matsubara components. We therefore propose to face the numerical problem using Fourier path-integral Monte Carlo, that can be formulated in two different ways: transforming the continuous paths and then truncating the high Fourier components (with possible improvements upon the truncation procedure), or performing the Fourier transformation upon the discretized paths. The latter choice leads to a simpler formulation and allows for a better control of the extrapolation to the limit of infinite Trotter number. The method is implemented for a single nonlinear particle with Ohmic dissipation and for a phi(4) chain with Drude-like dissipation.
New nonperturbative methods for dealing with Hamiltonian systems are introduced. The derivation of these methods requires identities for rewriting exponentials of sums of operators which are different from the usual Campbell-Hausdorff formula. These identities allow one to derive approximations to ${e}^{\ensuremath{-}\ensuremath{\delta}H}$ which are correct to higher order in $\ensuremath{\delta}$ and which contain fewer terms than the Campbell-Hausdorff formula. This allows one to generate path-integral actions which are more accurate for finite-size steps in time and which can be exploited to improve the rate of convergence of Monte Carlo calculations. To show that these methods allow one to include effects which show up in the stationary-phase approximation to the path integral, e.g., instantons, solitons, etc., I not only derive the connection between the Hamiltonian and path-integral formalism but the relationship between a specific stationary-phase approximation and the corresponding Hamiltonian calculation. My focus, however, is the direct application of these new identities to the study of nonperturbative Hamiltonian dynamics. I show that these methods are easier to apply and give better results than those based upon the naive ttt-expansion, block mean-field, or real-space renormalization ideas. Comparison of these older methods with the computational tools introduced in this paper are discussed in the context of simple examples. It is shown that the new methods allow one to extrapolate answers to finite ttt without the use of Pad\'e approximants.
A high-temperature approximation for the discretized path-integral quantum Monte Carlo (PIQMC) method is formulated. At higher temperatures, all P fictitious classical particles representing a single quantum particle stay close together, and an efficient approximation is obtained when, in the primitive short-time propagator, an essentially local harmonic approximation is used for the external potential at the common centre of mass of P fictitious particles - the integration over P - 1 dimensions can then be carried out analytically, and a classical formula of the effective-potential type is obtained for the partition function. Also discussed are the proper form and applicability of the virial total-energy estimator for finite systems, and the computation of the temperature dependence of the configurational partition function in a single PIQMC run.
We formulate the quasielastic response of a non-relativistic many-body system at zero temperature in terms of ground state density matrix elements and real time path integrals that embody the final state interactions. While the former provide the weight for a conventional Monte Carlo calculation, the latter require a more sophisticated treatment. We argue that the recently developed Stationary Phase Monte Carlo technique can be used to study the approach to "Y-scaling." We perform calculations for a particle in a potential well in one and three dimensions and compare them to the exact results available for these models. We then derive an eikonal approximation to the Path Integrals. This method is suitably generalized to treat strongly repulsive interactions, and allows comparison to Silver's theory of final state interactions in a straightforward way. We also give an exact prescription to calculate the scaling limit for potentials comprising a hard core. Finally, we study the approach to scaling in a model 4He nucleus, and find good agreement with experimental data.