Electronic polarizability of finite chains is accurately calculated from the total energy variation of the system produced by small but finite static electric fields applied along the chain direction. Normalized polarizability, that is, polarizability divided by chain length, diverges as the second power of length for metallic systems but approaches a constant value for insulating systems. This behaviour provides a very convenient way to characterize the wave-function malleability of finite systems as it avoids the need of attaching infinite contacts to the chain ends. Hubbard model calculations at half filling show that the method works for a small U = 1 interaction value that corresponds to a really small spectral gap of 0.005 (hopping t = -1 is assumed). Once successfully checked, the method has been applied to the long-range hopping model of Gebhard and Ruckenstein showing 1/r hopping decay (Gebhard and Ruckenstein 1992 Phys. Rev. Lett. 68 244; Gebhard et al 1994 Phys. Rev. B 49 10926). Metallicity for U values below the reported metal-insulator transition is obtained but the surprise comes for U values larger than the critical one (when a gap appears in the spectral density of states) because a steady increase of the normalized polarizability with size is obtained. This critical size-scaling behaviour can be understood as corresponding to a molecule which polarizability is unbounded. We have checked that a real transfer of charge from one chain end to the opposite occurs as a response to very small electric fields in spite of the existence of a large gap of the order of U for one-particle excitations. Finally, ab initio quantum chemistry calculations of realistic poly-acetylene chains prove that the occurrence of such critical behaviour in real systems is unlikely.
Electronic polarizability of finite chains is accurately calculated from the total energy variation of the system produced by small but finite static electric fields applied along the chain direction. Normalized polarizability, that is, polarizability divided by chain length, diverges as the second power of length for metallic systems but approaches a constant value for insulating systems. This behaviour provides a very convenient way to characterize the wave-function malleability of finite systems as it avoids the need of attaching infinite contacts to the chain ends. Hubbard model calculations at half filling show that the method works for a small U = 1 interaction value that corresponds to a really small spectral gap of 0.005 (hopping t = -1 is assumed). Once successfully checked, the method has been applied to the long-range hopping model of Gebhard and Ruckenstein showing 1/r hopping decay (Gebhard and Ruckenstein 1992 Phys. Rev. Lett. 68 244; Gebhard et al 1994 Phys. Rev. B 49 10926). Metallicity for U values below the reported metal-insulator transition is obtained but the surprise comes for U values larger than the critical one (when a gap appears in the spectral density of states) because a steady increase of the normalized polarizability with size is obtained. This critical size-scaling behaviour can be understood as corresponding to a molecule which polarizability is unbounded. We have checked that a real transfer of charge from one chain end to the opposite occurs as a response to very small electric fields in spite of the existence of a large gap of the order of U for one-particle excitations. Finally, ab initio quantum chemistry calculations of realistic poly-acetylene chains prove that the occurrence of such critical behaviour in real systems is unlikely.
The diagrammatic valence-bond technique is adopted to investigate one-dimensional half-filled Hubbard chains with long-range e-e interactions and alternating charge-transfer integrals. Exact results for finite-size systems with different boundary conditions are carefully extrapolated to the infinite-chain limit. In addition to various ground-state properties, such as bond-charge-density wave and bond-order amplitude, we calculate the optical gap and the corresponding oscillator strength. We introduce adiabatic phonons able to modulate on-site energies and investigate the stability of the system with respect to site-charge-density wave distortion. We also evaluate the response of the electronic system to the perturbation induced by adiabatic phonons which modulate charge-transfer integrals, and calculate the corresponding screening of vibrational frequencies.
The one-band two-dimensional Hubbard Hamiltonian is analyzed using a real-space renormalization-group block method. The renormalization-group method, previously applied to the one-dimensional case has been extended to two dimensions. It is also shown how to avoid the proliferation of new terms during the process if the shape of the blocks and the symmetries of the kept states are chosen conveniently. We characterize the ground state by its energy per site, the gap of charge excitations, the double occupancy, and the effective hopping. The system shows an insulating behavior in the case of half filling for all nonzero interaction parameter, U, if the hopping is isotropic (ty=tx); that is to say, the gap opens for all positive U, as in the one-dimensional case. If the hopping anisotropy, \ensuremath{\alpha}=ty/tx, is different from the unity and zero the ground state is a conductor up to a critical value (U/t)c1, which depends on \ensuremath{\alpha}. We have analyzed in the same way another filling, n=7/9. The system behaves as a conductor, paramagnetic up to a value (U/t)c2\ensuremath{\sim}2 and weakly ferromagnetic for higher values of U.
We investigate three different types of quantum phase transition occurring in quasi one-dimensional systems theoretically and numerically. First, we study the band-insulator to Mott-insulator transition occurring in charge-transfer complexes, for which the half-filled one-dimensional ionic Hubbard model is considered to be the prototype model. The study is carried out by first deriving an effective spin-one model, and then studying the model numerically using the density matrix renormalization group. We perform a careful finite-size scaling analysis of the mass gaps, order-parameters, and relative susceptibility. We confirm the existence of two quantum critical points. Analysis of the critical exponents confirms that the band-insulator-to-spontaneously-dimerized phase transition belong to the 2D Ising class. The spontaneously dimerized phase undergoes a phase transition to the Mott-insulator which is an infinite-order. Second, we investigate the Mott metal-insulator transition for the half-filled Hubbard model with both nearest-neighbor t and next-nearest-neighbor t′ hopping terms. We study the model using the bosonization approach and density matrix renormalization group simulations. An effective low-energy Hamiltonian that describes the insulator-metal transition is derived. We present results of density matrix renormalization group calculations of spin and charge distribution in various sectors of the phase diagram. The numerical results support the picture derived from the effective theory and give evidence for the complete separation of the transitions involving the spin and the charge degrees of freedom. Finally, we investigate quantum phase transitions phases in low-dimensional fermionic and spin models that go from uniform to spatially inhomogeneous, i.e., dimerized, trimerized, or incommensurate, phases. We propose a new approach based on studying the length dependence of the von Neumann entropy and its corresponding Fourier spectrum for finite segments in the ground state of finite chains. Peaks at a nonzero wave vector are indicators of oscillatory behavior in decaying correlation functions and also provide significant information about certain relevant features of the excitation spectrum; in particular, they can identify the wave vector of soft modes in critical models.
The finite-cell-scaling method is applied to the one-dimensional Hubbard model. In the half-filled-band case and at zero temperature, the numerical results obtained for the ground-state energy differ from the exact solution by approximately 1%. For the range 1<tU<5, where U is the Coulomb interaction and t is the transfer integral, the value of the normalized gap $\frac{\ensuremath{\Delta}}{t}$ agrees well with the exact result while it differs significantly for 0.2<tU<0.9. For the latter range, we show that the gap varies as ${(\frac{U}{t})}^{\frac{1}{2}}\mathrm{exp}[{(a+\frac{\mathrm{bU}}{t})}^{\ensuremath{-}1}]$, where a and b are negative constants which depend on the size of the system, instead of ${(\frac{U}{t})}^{\frac{1}{2}}\mathrm{exp}(\frac{\ensuremath{-}2\ensuremath{\pi}t}{U})$ as in the exact solution.
The one-dimensional Hubbard model is studied using a real-space renormalisation-group block method previously applied to other quantum systems. The authors primarily consider the half-filled band for which the ground state is found to be insulating for all non-zero values of the Hubbard interaction, in agreement with the exact solution. When the method is applied to first order in perturbation, they find that the insulating gap for small U/t behaves like exp (-At 2 /U 2 ) where U is the Hubbard interaction and t is the nearest-neighbour transfer integral. When the method is extended to second order, the authors find that for small U/t the gap behaves like (U/t) alpha with alpha approximately=11.2. The exact asymptotic behaviour of the gap ( approximately (U/t) 1/2 exp(-At/U)) is not recovered, but a good semiquantative agreement with the exact solution is obtained for both the gap and the ground state energy. Extensions to non-half-filled bands and to higher dimensions are discussed. The validity of the renormalisation group method is also discussed.
We investigate the localization of two interacting particles in a one-dimensional random potential. Our definition of the two-particle localization length, \ensuremath{\xi}, is the same as that of von Oppen et al. [Phys. Rev. Lett. 76, 491 (1996)]. \ensuremath{\xi}'s for chains of finite lengths are calculated numerically using the recursive Green's function method for several values of the strength of the disorder, W, and the strength of interaction, U. When U=0, \ensuremath{\xi} approaches a value larger than half the single-particle localization length as the system size tends to infinity and behaves as $\ensuremath{\xi}\ensuremath{\sim}{W}^{\ensuremath{-}{\ensuremath{\nu}}_{0}}$ for small W with ${\ensuremath{\nu}}_{0}=2.1\ifmmode\pm\else\textpm\fi{}0.1.$ When $U\ensuremath{\ne}0,$ we use the finite size scaling ansatz and find the relation $\ensuremath{\xi}\ensuremath{\sim}{W}^{\ensuremath{-}\ensuremath{\nu}}$ with $\ensuremath{\nu}=2.9\ifmmode\pm\else\textpm\fi{}0.2.$ Moreover, data show the scaling behavior $\ensuremath{\xi}\ensuremath{\sim}{W}^{\ensuremath{-}{\ensuremath{\nu}}_{0}}g(b|U|{/W}^{\ensuremath{\Delta}})$ with $\ensuremath{\Delta}=4.0\ifmmode\pm\else\textpm\fi{}0.5.$