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. B, Condensed matter

Finite-size effects on the optical conductivity of a half-filled Hubbard ring

Charles Allen Stafford, Andrew J. Millis, B. Sriram Shastry

Published Jun 1, 1991• 78 citations
Publisher

Abstract

We use the Bethe-ansatz equations to calculate the total and zero-frequency spectral weight in the optical conductivity of the half-filled one-dimensional Hubbard model as a function of the lattice size L and the on-site repulsion U. The zero-frequency spectral weight \ensuremath{\pi}D scales as ${\mathit{L}}^{1/2}$exp(-L/\ensuremath{\xi}) as L\ensuremath{\rightarrow}\ensuremath{\infty}. Near U=0, \ensuremath{\xi} varies as the inverse of the Lieb-Wu charge gap. In the strongly correlated regime (U\ensuremath{\gg}t), ${\ensuremath{\xi}}^{\mathrm{\ensuremath{-}}1}$=ln(U/t)-1.48. $D--- is negative when L is a multiple of 4, corresponding to a negative inductance. We give a physical explanation of our results in terms of a simple model of ring exchange. The finite-size corrections to the total spectral weight scale as ${\mathit{L}}^{\mathrm{\ensuremath{-}}2}$. We discuss the implications of our results for exact diagonalization calculations of the optical conductivity.

DOI
10.1103/physrevb.43.13660

Related papers

Drude weight, optical conductivity, and flux properties of one-dimensional Hubbard rings

Richard M. Fye, Marcio Jose Martins, Douglas J. Scalapino +2 more

We study the Drude weight, optical conductivity, and flux-periodicity properties of one-dimensional Hubbard chains using Bethe-ansatz, Lanczos, and exact-diagonalization techniques. We find that the Drude weight D is unexpectedly negative for half-filled Hubbard rings with N=4n sites reflecting a strong paramagnetic response, while it is positive (diamagnetic) for half-filled rings of N=4n+2 sites. In both cases at half filling, D vanishes exponentially with N on a length scale set by the inverse of the gap for small U/t. Near half filling, we find that D approaches a constant, positive, diamagnetic value as N increases, indicating metallic behavior. Combining Bethe-ansatz results near half filling with the known U=0 and U=\ensuremath{\infty} limits and Lanczos finite-size extrapolations, we obtain a general picture of D in the thermodynamic limit as a function of band filling and the ratio U/t. We note similarities with the low-frequency integrated spectral weight of certain hole-doped and electron-doped high-${\mathit{T}}_{\mathit{c}}$ compounds. We investigate the finite-frequency optical conductivity, finding structure at \ensuremath{\omega}\ensuremath{\approxeq}U and, for one hole off half filling, at \ensuremath{\omega}\ensuremath{\propto}${\mathit{t}}^{2}$/U. We find that minima in the energy of a Hubbard ring enclosing a flux \ensuremath{\Phi} occur with changes of half a flux quantum rather than a full flux quantum when the Coulomb repulsion is turned on. Lastly, we discuss the relationship of the Drude weight computed from open chains to that obtained from rings, and also the effect of arbitrary phases at the ring boundary.

Physical review. B, Condensed matter•Oct 1, 1991•164 citations

Mobile impurity approach to the optical conductivity in the Hubbard chain

Thomas Veness, Fabian H. L. Eßler

We consider the optical conductivity in the one-dimensional Hubbard model in the metallic phase close to half-filling. In this regime, most of the spectral weight is located at frequencies above an energy scale ${E}_{\mathrm{opt}}$ that tends towards the optical gap in the Mott insulating phase for vanishing doping. Using the Bethe ansatz, we relate ${E}_{\mathrm{opt}}$ to thresholds of particular kinds of excitations in the Hubbard model. We then employ a mobile impurity model to analyze the optical conductivity for frequencies slightly above these thresholds. This entails generalizing mobile impurity models to excited states that are not the highest weight with regards to the SU(2) symmetries of the Hubbard chain, and that occur at a maximum of the impurity dispersion.

arXiv:1601.01251•Physical review. B./Physical review. B•May 2, 2016•9 citations

Transport properties of the one-dimensional Hubbard model at finite temperature

Christoph Karrasch, Dante Marvin Kennes, Joel E. Moore

We study finite-temperature transport properties of the one-dimensional Hubbard model using the density matrix renormalization group. Our aim is twofold: First, we compute both the charge- and the spin-current correlation function of the integrable model at half filling. The former decays rapidly, implying that the corresponding Drude weight is either zero or very small. Second, we calculate the optical charge conductivity ${\ensuremath{\sigma}}_{r}eg(\ensuremath{\omega})$ in the presence of small integrability-breaking next-nearest-neighbor interactions (the extended Hubbard model). The dc conductivity is finite and diverges as the temperature is decreased below the gap. Our results thus suggest that the half-filled, gapped Hubbard model is a normal charge conductor at finite temperatures. As a test bed for our numerics, we compute ${\ensuremath{\sigma}}_{r}eg(\ensuremath{\omega})$ for the integrable XXZ spin chain in its gapped phase.

arXiv:1408.1891•Physical Review B•Oct 3, 2014•56 citations

Dynamic Properties of Interacting Electrons.

L. Tan

The spectral weight functions and the optical conductivity of a highly correlated electronic system described by a simple Hubbard model are studied. The calculations are carried out for a small system, a cubic cluster of eight sites, by means of exact diagonalization using symmetrized basis functions. The spectral weight functions of a single hole (or electron) placed in an otherwise half-filled band are obtained. The density of states at different interaction strengths, ranging from weak to strong couplings, is studied. We observe the spreading of the spectral weight associated with some single-particle eigenstates over a large range of energies, and the appearance of satellite structure at higher excitation energies. The nature of band narrowing is also discussed. The frequency dependent (optical) conductivity is studied for the cases of a half-filled band, and for one hole and two hole dopings away from half filling. The interaction strength is varied between the weak- and strong-coupling limits. The formation of a Hubbard gap is observed. Results are related to aspects of the metal-insulator transition in bulk systems.

Civil War Book Review•Jan 1, 1993•0 citations

Finite-temperature transport in finite-size Hubbard rings in the strong-coupling limit

Nuno M. R. Peres, R. G. Dias, Pedro D Sacramento +1 more

We study the current, the curvature of levels, and the finite temperature charge stiffness, $D(T,L),$ in the strongly correlated limit $U\ensuremath{\gg}t$ for Hubbard rings of L sites, with U the on-site Coulomb repulsion and t the hopping integral. Our study is done for finite-size systems and any band filling. Up to order t, we derive our results following two independent approaches, namely, using the solution provided by the Bethe ansatz and the solution provided by an algebraic method, where the electronic operators are represented in a slave-fermion picture. We find that, in the $U=\ensuremath{\infty}$ case, the finite-temperature charge stiffness is finite for electronic densities n smaller than 1. These results are essentially those of spinless fermions in a lattice of size L, apart from small corrections coming from a statistical flux, due to the spin degrees of freedom. Up to order t, the Mott-Hubbard gap is ${\ensuremath{\Delta}}_{\mathrm{MH}}=U\ensuremath{-}4t,$ and we find that $D(T)$ is finite for $n<1,$ but is zero at half filling. This result comes from the effective flux felt by the holon excitations, which, due to the presence of doubly occupied sites, is renormalized to ${\ensuremath{\Phi}}^{\mathrm{eff}}=\ensuremath{\varphi}{(N}_{h}\ensuremath{-}{N}_{d}{)/(N}_{d}{+N}_{h}),$ and which is zero at half filling, with ${N}_{d}$ and ${N}_{h}$ being the number of doubly occupied and empty lattice sites, respectively. Further, for half filling, the current transported by any eigenstate of the system is zero and, therefore, $D(T)$ is also zero.

arXiv:cond-mat/9911303•Physical review. B, Condensed matter•Feb 15, 2000•34 citations

Conductivity in one-dimensional highly correlated electron systems

Norio Kawakami, Sung-Kil Yang

The dc conductivity for the t-J model is exactly obtained at t=J and for the Hubbard model in one dimension. Our calculation is carried out for arbitrary band filling by use of the finite-size-scaling method based on the Bethe-ansatz solution under twisted boundary conditions. The effective transport mass defined as the inverse of the conductivity serves as a sensitive probe of the metal-insulator transition. The effective mass is extremely enhanced near half-filling both for the t-J model at t=J and the repulsive Hubbard model, which is a direct evidence of the metal-insulator transition due to the electron correlation. We observe that the large antiferromagnetic coupling J in the t-J model increases the effective mass. On the other hand, the conductivity in the attractive Hubbard model exhibits quite different properties from the repulsive models.

Physical review. B, Condensed matter•Oct 15, 1991•57 citations
We use the Bethe-ansatz equations to calculate the total and zero-frequency spectral weight in the optical conductivity of the half-filled one-dimensional Hubbard model as a function of the lattice size L and the on-site repulsion U. The zero-frequency spectral weight \ensuremath{\pi}D scales as L1/2{\mathit{L}}^{1/2}L1/2exp(-L/\ensuremath{\xi}) as L\ensuremath{\rightarrow}\ensuremath{\infty}. Near U=0, \ensuremath{\xi} varies as the inverse of the Lieb-Wu charge gap. In the strongly correlated regime (U\ensuremath{\gg}t), \ensuremathξ\ensuremath−1{\ensuremath{\xi}}^{\mathrm{\ensuremath{-}}1}\ensuremathξ\ensuremath−1=ln(U/t)-1.48. D−−−isnegativewhenLisamultipleof4,correspondingtoanegativeinductance.Wegiveaphysicalexplanationofourresultsintermsofasimplemodelofringexchange.Thefinite−sizecorrectionstothetotalspectralweightscaleasD--- is negative when L is a multiple of 4, corresponding to a negative inductance. We give a physical explanation of our results in terms of a simple model of ring exchange. The finite-size corrections to the total spectral weight scale as D−−−isnegativewhenLisamultipleof4,correspondingtoanegativeinductance.Wegiveaphysicalexplanationofourresultsintermsofasimplemodelofringexchange.Thefinite−sizecorrectionstothetotalspectralweightscaleas{\mathit{L}}^{\mathrm{\ensuremath{-}}2}$. We discuss the implications of our results for exact diagonalization calculations of the optical conductivity.
We study the Drude weight, optical conductivity, and flux-periodicity properties of one-dimensional Hubbard chains using Bethe-ansatz, Lanczos, and exact-diagonalization techniques. We find that the Drude weight D is unexpectedly negative for half-filled Hubbard rings with N=4n sites reflecting a strong paramagnetic response, while it is positive (diamagnetic) for half-filled rings of N=4n+2 sites. In both cases at half filling, D vanishes exponentially with N on a length scale set by the inverse of the gap for small U/t. Near half filling, we find that D approaches a constant, positive, diamagnetic value as N increases, indicating metallic behavior. Combining Bethe-ansatz results near half filling with the known U=0 and U=\ensuremath{\infty} limits and Lanczos finite-size extrapolations, we obtain a general picture of D in the thermodynamic limit as a function of band filling and the ratio U/t. We note similarities with the low-frequency integrated spectral weight of certain hole-doped and electron-doped high-Tc{\mathit{T}}_{\mathit{c}}Tc​ compounds. We investigate the finite-frequency optical conductivity, finding structure at \ensuremath{\omega}\ensuremath{\approxeq}U and, for one hole off half filling, at \ensuremath{\omega}\ensuremath{\propto}t2{\mathit{t}}^{2}t2/U. We find that minima in the energy of a Hubbard ring enclosing a flux \ensuremath{\Phi} occur with changes of half a flux quantum rather than a full flux quantum when the Coulomb repulsion is turned on. Lastly, we discuss the relationship of the Drude weight computed from open chains to that obtained from rings, and also the effect of arbitrary phases at the ring boundary.
We consider the optical conductivity in the one-dimensional Hubbard model in the metallic phase close to half-filling. In this regime, most of the spectral weight is located at frequencies above an energy scale Eopt{E}_{\mathrm{opt}}Eopt​ that tends towards the optical gap in the Mott insulating phase for vanishing doping. Using the Bethe ansatz, we relate Eopt{E}_{\mathrm{opt}}Eopt​ to thresholds of particular kinds of excitations in the Hubbard model. We then employ a mobile impurity model to analyze the optical conductivity for frequencies slightly above these thresholds. This entails generalizing mobile impurity models to excited states that are not the highest weight with regards to the SU(2) symmetries of the Hubbard chain, and that occur at a maximum of the impurity dispersion.
We study finite-temperature transport properties of the one-dimensional Hubbard model using the density matrix renormalization group. Our aim is twofold: First, we compute both the charge- and the spin-current correlation function of the integrable model at half filling. The former decays rapidly, implying that the corresponding Drude weight is either zero or very small. Second, we calculate the optical charge conductivity ${\ensuremath{\sigma}}_{r}eg(\ensuremath{\omega})$ in the presence of small integrability-breaking next-nearest-neighbor interactions (the extended Hubbard model). The dc conductivity is finite and diverges as the temperature is decreased below the gap. Our results thus suggest that the half-filled, gapped Hubbard model is a normal charge conductor at finite temperatures. As a test bed for our numerics, we compute ${\ensuremath{\sigma}}_{r}eg(\ensuremath{\omega})$ for the integrable XXZ spin chain in its gapped phase.
The spectral weight functions and the optical conductivity of a highly correlated electronic system described by a simple Hubbard model are studied. The calculations are carried out for a small system, a cubic cluster of eight sites, by means of exact diagonalization using symmetrized basis functions. The spectral weight functions of a single hole (or electron) placed in an otherwise half-filled band are obtained. The density of states at different interaction strengths, ranging from weak to strong couplings, is studied. We observe the spreading of the spectral weight associated with some single-particle eigenstates over a large range of energies, and the appearance of satellite structure at higher excitation energies. The nature of band narrowing is also discussed. The frequency dependent (optical) conductivity is studied for the cases of a half-filled band, and for one hole and two hole dopings away from half filling. The interaction strength is varied between the weak- and strong-coupling limits. The formation of a Hubbard gap is observed. Results are related to aspects of the metal-insulator transition in bulk systems.
We study the current, the curvature of levels, and the finite temperature charge stiffness, D(T,L),D(T,L),D(T,L), in the strongly correlated limit $U\ensuremath{\gg}t$ for Hubbard rings of L sites, with U the on-site Coulomb repulsion and t the hopping integral. Our study is done for finite-size systems and any band filling. Up to order t, we derive our results following two independent approaches, namely, using the solution provided by the Bethe ansatz and the solution provided by an algebraic method, where the electronic operators are represented in a slave-fermion picture. We find that, in the $U=\ensuremath{\infty}$ case, the finite-temperature charge stiffness is finite for electronic densities n smaller than 1. These results are essentially those of spinless fermions in a lattice of size L, apart from small corrections coming from a statistical flux, due to the spin degrees of freedom. Up to order t, the Mott-Hubbard gap is ${\ensuremath{\Delta}}_{\mathrm{MH}}=U\ensuremath{-}4t,$ and we find that D(T)D(T)D(T) is finite for n<1,n<1,n<1, but is zero at half filling. This result comes from the effective flux felt by the holon excitations, which, due to the presence of doubly occupied sites, is renormalized to ${\ensuremath{\Phi}}^{\mathrm{eff}}=\ensuremath{\varphi}{(N}_{h}\ensuremath{-}{N}_{d}{)/(N}_{d}{+N}_{h}),$ and which is zero at half filling, with Nd{N}_{d}Nd​ and Nh{N}_{h}Nh​ being the number of doubly occupied and empty lattice sites, respectively. Further, for half filling, the current transported by any eigenstate of the system is zero and, therefore, D(T)D(T)D(T) is also zero.
The dc conductivity for the t-J model is exactly obtained at t=J and for the Hubbard model in one dimension. Our calculation is carried out for arbitrary band filling by use of the finite-size-scaling method based on the Bethe-ansatz solution under twisted boundary conditions. The effective transport mass defined as the inverse of the conductivity serves as a sensitive probe of the metal-insulator transition. The effective mass is extremely enhanced near half-filling both for the t-J model at t=J and the repulsive Hubbard model, which is a direct evidence of the metal-insulator transition due to the electron correlation. We observe that the large antiferromagnetic coupling J in the t-J model increases the effective mass. On the other hand, the conductivity in the attractive Hubbard model exhibits quite different properties from the repulsive models.