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

Physical Review Letters

Beyond the Gutzwiller approximation in the slave-boson approach: Inclusion of fluctuations with the correct continuum limit of the functional integral

Enrico Arrigoni, Giancarlo Calvanese Strinati

Published Nov 8, 1993• 30 citations
Publisher

Abstract

We study the single-band Hubbard model via the functional-integral four-slave-boson formulation introduced by Kotliar and Ruckenstein, and show that a proper treatment of the continuum limit of the functional integral removes the inconsistencies due to an incorrect handling of this limit so far performed in the literature. It follows from our analysis that the Kotliar and Ruckenstein mean-field solution has to be abandoned in order to include fluctuation corrections in a systematic way. We show explicitly how to construct a suitable alternative solution which admits the inclusion of fluctuations.

DOI
10.1103/physrevlett.71.3178

Related papers

Correct continuum limit of the functional-integral representation for the four-slave-boson approach to the Hubbard model: Paramagnetic phase

Enrico Arrigoni, Giancarlo Calvanese Strinati

The Hubbard model with finite on-site repulsion U is studied via the functional-integral formulation of the four-slave-boson approach by Kotliar and Ruckenstein. It is shown that a correct treatment of the continuum imaginary time limit (which is required by the very definition of the functional integral) modifies the free energy when fluctuation (1/N) corrections beyond mean field are considered, thus removing the inconsistencies originating from the incorrect handling of this pathologic limit so far performed in the literature. In particular, our treatment correctly restores the decrease of the average number of doubly occupied sites for increasing U. Our analysis requires us to suitably interpret the Kotliar and Ruckenstein choice for the bosonic hopping operator and to abandon the commonly used normal-ordering prescription, in order to obtain meaningful fluctuation corrections. In this way we recover the exact solution at U=0 not only at the mean-field level but also at the next order in 1/N. In addition, we consider alternative choices for the bosonic hopping operator and test them numerically for a simple two-site model for which the exact solution is readily available for any U. We also discuss how the 1/N expansion can be formally generalized to the four-slave-boson approach, and provide a simplified prescription to obtain the additional terms in the free energy which result at the order 1/N from the correct continuum limit.

arXiv:cond-mat/9501132•Physical review. B, Condensed matter•Jul 15, 1995•14 citations

Fluctuations beyond the Gutzwiller approximation in the slave-boson approach

Th. Jolicœur, J. C. Le Guillou

We study the Hubbard model through a functional-integral formulation proposed by Kotliar and Ruckenstein. A careful use of the U(1${)}^{\ensuremath{\bigotimes}3}$ gauge symmetry of the bosonized action allows us to compute the complete one-loop correction to the free energy in the case of half-filled band. We extract the effective mass of the fermion in the metallic phase and show that the noninteracting limit is not recovered for vanishing on-site repulsion. The bosonic Hamiltonian must be adjusted order by order in perturbation theory.

Physical review. B, Condensed matter•Aug 1, 1991•49 citations

Correct formulation of the 1/Nexpansion for the slave-boson approach within the functional integral

Enrico Arrigoni, Claudio Castellani, Roberto Raimondi +1 more

We consider the slave-boson method within the coherent-state functional integral representation of the partition function, and show how to deal with the continuum imaginary-time limit required by the very definition of the functional integral. We find that a correct treatment of the continuum limit modifies the free energy when fluctuation (1/N) corrections beyond the mean-field solution are considered. Numerical results are presented for a two-level single-site model system (with an infinite Hubbard repulsion), for which the additional terms in the free energy introduced by the correct continuum limit act to validate the 1/N expansion. Our analysis calls for a revision of several outcomes of the slave-boson method with the inclusion of fluctuation corrections.

Physical review. B, Condensed matter•Jul 15, 1994•39 citations

Slave-boson mean field versus quantum Monte Carlo results for the Hubbard model

L. Lilly, A. Muramatsu, Werner R. Hanke

The one-band Hubbard model is considered in the slave-boson formulation first introduced by Kotliar and Ruckenstein. It is shown that a mean-field approximation, where broken-symmetry states appropriate for a bipartite lattice are allowed, leads to a quantitative agreement with quantum Monte Carlo results for local observables over a wide range of interactions (0U\ensuremath{\le}20t) and doping (0n〉\ensuremath{\le}1). Thus, our saddle-point solutions constitute an excellent starting point for a systematic treatment of fluctuations.

Physical Review Letters•Sep 1, 1990•97 citations

Spin-rotation-invariant slave-boson approach to the Hubbard model

T. Li, P. Wölfle, P. J. Hirschfeld

We present a slave-boson representation for the Hubbard model, introducing Bose fields for the empty, singly, and doubly occupied sites and show that the boson for the singly occupied site must transform as a tensor of rank two under spin rotations. This generalizes the formulation of Kotliar and Ruckenstein, which is not manifestly spin-rotation invariant. The paramagnetic saddle point of the corresponding functional integral is identical to Gutzwiller's solution. As an illustration of the method, we calculate the ${T}^{3}$lnT spin-fluctuation contribution to the specific heat and find it to be fully consistent with Fermi-liquid theory. We discuss further applications of this approach.

Physical review. B, Condensed matter•Oct 1, 1989•209 citations

Spin- and charge-fluctuation contribution to mass enhancement

T. Li

We study the one-loop contribution of the spin and charge fluctuation to the boson hopping factor in the slave-boson representation of the Hubbard model introduced by Kotliar and Ruckenstein. Whether one uses a low-lying cutoff or sums over the whole bosonic spectrum, the noninteracting limit for the fluctuation correction of the mass enhancement is not recovered for vanishing Coulomb interaction. We argue that one should be cautious when attempting to extend the loop-expansion calculation, in the functional-integral scheme of Kotliar and Ruckenstein, to the observables in which the high-energy excitations may play an important role.

Physical review. B, Condensed matter•Aug 15, 1993•4 citations
We study the single-band Hubbard model via the functional-integral four-slave-boson formulation introduced by Kotliar and Ruckenstein, and show that a proper treatment of the continuum limit of the functional integral removes the inconsistencies due to an incorrect handling of this limit so far performed in the literature. It follows from our analysis that the Kotliar and Ruckenstein mean-field solution has to be abandoned in order to include fluctuation corrections in a systematic way. We show explicitly how to construct a suitable alternative solution which admits the inclusion of fluctuations.
The Hubbard model with finite on-site repulsion U is studied via the functional-integral formulation of the four-slave-boson approach by Kotliar and Ruckenstein. It is shown that a correct treatment of the continuum imaginary time limit (which is required by the very definition of the functional integral) modifies the free energy when fluctuation (1/N) corrections beyond mean field are considered, thus removing the inconsistencies originating from the incorrect handling of this pathologic limit so far performed in the literature. In particular, our treatment correctly restores the decrease of the average number of doubly occupied sites for increasing U. Our analysis requires us to suitably interpret the Kotliar and Ruckenstein choice for the bosonic hopping operator and to abandon the commonly used normal-ordering prescription, in order to obtain meaningful fluctuation corrections. In this way we recover the exact solution at U=0 not only at the mean-field level but also at the next order in 1/N. In addition, we consider alternative choices for the bosonic hopping operator and test them numerically for a simple two-site model for which the exact solution is readily available for any U. We also discuss how the 1/N expansion can be formally generalized to the four-slave-boson approach, and provide a simplified prescription to obtain the additional terms in the free energy which result at the order 1/N from the correct continuum limit.
We study the Hubbard model through a functional-integral formulation proposed by Kotliar and Ruckenstein. A careful use of the U(1${)}^{\ensuremath{\bigotimes}3}$ gauge symmetry of the bosonized action allows us to compute the complete one-loop correction to the free energy in the case of half-filled band. We extract the effective mass of the fermion in the metallic phase and show that the noninteracting limit is not recovered for vanishing on-site repulsion. The bosonic Hamiltonian must be adjusted order by order in perturbation theory.
We consider the slave-boson method within the coherent-state functional integral representation of the partition function, and show how to deal with the continuum imaginary-time limit required by the very definition of the functional integral. We find that a correct treatment of the continuum limit modifies the free energy when fluctuation (1/N) corrections beyond the mean-field solution are considered. Numerical results are presented for a two-level single-site model system (with an infinite Hubbard repulsion), for which the additional terms in the free energy introduced by the correct continuum limit act to validate the 1/N expansion. Our analysis calls for a revision of several outcomes of the slave-boson method with the inclusion of fluctuation corrections.
The one-band Hubbard model is considered in the slave-boson formulation first introduced by Kotliar and Ruckenstein. It is shown that a mean-field approximation, where broken-symmetry states appropriate for a bipartite lattice are allowed, leads to a quantitative agreement with quantum Monte Carlo results for local observables over a wide range of interactions (0U\ensuremath{\le}20t) and doping (0n〉\ensuremath{\le}1). Thus, our saddle-point solutions constitute an excellent starting point for a systematic treatment of fluctuations.
We present a slave-boson representation for the Hubbard model, introducing Bose fields for the empty, singly, and doubly occupied sites and show that the boson for the singly occupied site must transform as a tensor of rank two under spin rotations. This generalizes the formulation of Kotliar and Ruckenstein, which is not manifestly spin-rotation invariant. The paramagnetic saddle point of the corresponding functional integral is identical to Gutzwiller's solution. As an illustration of the method, we calculate the T3{T}^{3}T3lnT spin-fluctuation contribution to the specific heat and find it to be fully consistent with Fermi-liquid theory. We discuss further applications of this approach.
We study the one-loop contribution of the spin and charge fluctuation to the boson hopping factor in the slave-boson representation of the Hubbard model introduced by Kotliar and Ruckenstein. Whether one uses a low-lying cutoff or sums over the whole bosonic spectrum, the noninteracting limit for the fluctuation correction of the mass enhancement is not recovered for vanishing Coulomb interaction. We argue that one should be cautious when attempting to extend the loop-expansion calculation, in the functional-integral scheme of Kotliar and Ruckenstein, to the observables in which the high-energy excitations may play an important role.