Metal-Insulator Transition in the Two-Dimensional Hubbard Model at Half-Filling with Lifetime Effects within the Moment Approach
Extracted source text ยท arXiv cond-mat/9709080
%-----------------------------------------------------------------%
%-----------------------------------------------------------------%
\documentstyle[aps]{revtex}
\begin{document}
\voffset=-2mm
\title{METAL-INSULATOR TRANSITION IN THE TWO-DIMENSIONAL HUBBARD MODEL
AT HALF-FILLING WITH LIFETIME EFFECTS WITHIN THE MOMENT
APPROACH.}
%
\author{\it J.J. Rodr\'{\i}guez-N\'u\~nez}
\address{Departamento de F\'{\i}sica - CCNE,\\
Universidade Federal de
Santa Maria, CAMOBI,\\
97105 - 900 Santa Maria - RS,
Brazil. \\e-m: jjrn@mfis.ccne.br}
\author{\it S. Schafroth}
\address{Physik-Institut der Universit\"at Z\"urich,\\
Winterthurerstrasse 190, CH-8057 Z\"urich,
Switzerland\\e-m: schafrot@physik.unizu.ch}
\date{\today}
\maketitle
%
% Abstract
%
\begin{abstract}
%
We explore the effect of the imaginary part of the
self-energy, $Im\Sigma(\vec{k},\omega)$, having
a single pole, $\Omega(\vec{k},\omega)$, with
{\it spectral} weight, $\alpha(\vec{k})$, and
quasi-particle lifetime, $\Gamma(\vec{k})$, on the
density of states. We solve the set of parameters,
$\Omega(\vec{k},\omega$), $\alpha(\vec{k})$, and
$\Gamma(\vec{k})$ by means of the moment approach (exact sum rules)
of Nolting\cite{1}. Our choice for $\Sigma(k,\omega)$,
satisfies the Kramers -
Kronig relationship automatically. Due to our choice of
the self - energy, the system is not a Fermi liquid for
any value of the interaction, a result which is also
true in the moment approach of Nolting without lifetime
effects. By increasing the value of
the local interaction, $U/W$, at half-filling ($\rho =
1/2$), we go from a paramagnetic metal to a paramagnetic insulator,
(Mott metal - insulator transition ($MMIT$))\cite{2}
for values of $U/W$ of the
order of $U/W \geq 1$ ($W$ is the band
width) which is in agreement with
numerical results for finite lattices and
for infinity dimensions ($D = \infty$).
These results settle down the main weakness
of the spherical approximation of Nolting\cite{1}:
a finite gap for any finite value of
the interaction, i.e., an insulator for
any finite value of $U/W$. Lifetime effects are
absolutely indispensable. Our scheme works better than the
one of improving the narrowing band factor, $B(\vec{k})$,
beyond the spherical approximation of Nolting\cite{1}.\\
\\
Pacs numbers: 74.20.-Fg, 74.10.-z, 74.60.-w, 74.72.-h
\end{abstract}
\pacs{PACS numbers 74.20.-Fg, 74.10.-z, 74.60.-w, 74.72.-h}
% PACS
%
% I. Introduction
%
\section{Introduction}\label{sec:intro}
After the discovery of
the high-$T_c$ materials\cite{3},
the study of correlations has gained interested due
to the fact that there is the belief\cite{4}
that the normal properties of these materials could
be explained in the framework of the Hubbard
model\cite{5,6}, since electron correlations
are strong, i.e., the on-site electron-electron
repulsions $U$ are much larger than the energies
associated with the hybridization of atomic orbitals
belonging to different atoms\cite{7}. The strategy,
according to Anderson\cite{8} to address the problem
of the high-$T_c$ superconductivity, is to try to
find a theory accounting for the normal state properties
of the cuprates and then find an electron mechanism
which destabilizes the normal state towards a
superconducting state. The Hubbard Hamiltonian is a kind
of minimum model\cite{9} which takes into account quantum
mechanical motion of electrons in a solid, and nonlinear repulsion
between electrons. Even thought this model is too simple to
describe solids faithfully, serious theoretical studies have
revealed that to understand the various properties of it is a
very difficult task, since is the simplest many-body Hamiltonian
one can write down and which cannot be reduced to a single-particle
theory\cite{Assa}. Its study will prove useful in developing
various notions and techniques in statistical physics of
many degrees of freedom. Besides cuprate superconductors, the
Hubbard model may also be applicable to describing the metal -
insulating transition in materials like $La_xSr_{1-x}TiO_3$ and
$V_{2-y}O_3$ in which paramagnetic metal, antiferromagnetic
insulator, and a phase antiferromagnetic
metal are clearly separated in temperature -
pressure phase diagram\cite{all}. All these features, found in the
Hubbard model, make it suitable to reproduce experimental data.
Additionally, the study of correlations by itself in the Hubbard
model is a rewarding task since it will shed light on still unsolved points of
the novel materials. For example, at high temperatures ($T_c~30-130~K$) these
HTSC cuprates, which are poor conductors, become superconductors.
This feature is strange indeed because the Coulomb repulsion is
strong. Furthermore, the behavior of these materials at $T~>~T_c$ is even
more puzzling than the superconductivity itself. Contrary to the
predictions of the Fermi liquid theory, the resistivity at $T~>~T_c$
and optimum doping is linear in temperature, i.e., $R~\approx~T$. This suggests a very strong
scattering of elementary excitations. A discussion of the possible
breakdown of Fermi liquid theory is given in Ref.\cite{Alexei}. There is also another approach which
consists in studying toy models, i.e., exactly solvable,\cite{toy}
in order to get some ideas on the one-particle properties of highly correlated electron systems.
In this Letter, we will use the moment approach
(or sum rules) of Nolting\cite{1} for the spectral density,
$A(\vec{k},\omega)$. As it is well known in the literature\cite{ours}, one
of the drawbacks of the moment approach in the spherical approximation - where
$B(\vec{k})$, the narrowing band factor, is not $k$ - dependent -
is that we {\it always} find a gap in the density of states $(DOS)$.
If the chemical potential happens to be in this gap, then we always have an insulator.
It has been argued that the way to cure this unrealistic gap is
to have a better approximation for the narrowing band factor, $B(\vec{k})$,
task which has proven to be hard just to the present\cite{ours}. It is due to that
we have followed a different path which consists in proposing a single pole
structure in the self-energy, $\Sigma(\vec{k},\omega)$.
The model we study is the Hubbard Hamiltonian
\begin{eqnarray}\label{Ham}
H = t_{\vec{i},\vec{j}}
c_{\vec{i}\sigma}^{\dagger}c_{\vec{j}\sigma}
+ \frac{U}{2} n_{\vec{i}\sigma}n_{\vec{i}\bar{\sigma}}
- \mu c^{\dagger}_{\vec{i}\sigma}c_{\vec{i}\sigma}~~,
\end{eqnarray}
where $c_{\vec{i}\sigma}^{\dagger}$ ($c_{\vec{i}\sigma}$) are creation
(annihilation) electron operators with spin $\sigma$. $n_{\vec{i}
\sigma} \equiv c_{\vec{i}\sigma}^{\dagger}c_{\vec{i}\sigma}$.
$U$ is the local interaction, $\mu$ the chemical
potential and we work in the grand canonical ensemble. We have
adopted Einstein convention for repeated indices, i.e., for the $N_s$
sites $\vec{i}$, the $z$ nearest-neighbor sites $(n.n.)$
$\vec{j}$ and for spin up and down ($\sigma = -\bar{\sigma}
= \pm 1$). $t_{\vec{i},\vec{j}} = -t$, for $n.n.$ and zero otherwise.
Let us propose for the self - energy,
$\Sigma(\vec{k},\omega)$, the following single pole Ansatz
\begin{equation}\label{single}
\Sigma(\vec{k},\omega) = \frac{\alpha(\vec{k})}{\omega -
\Omega(\vec{k}) - i\Gamma(\vec{k})}~~~ ; ~~~
\alpha(\vec{k}),\Gamma(\vec{k}) \in \Re ~~~
\end{equation}
With our choice for $\Sigma(\vec{k},\omega)$, we have the real
and imaginary parts of the self - energy satisfy the
Kramers - Kronig relations\cite{CL}.
By definition the one-particle Green function,
$G(\vec{k},\omega)$, in terms of $\Sigma(\vec{k},\omega)$ is given as
\begin{equation}\label{1PGF}
G(\vec{k},\omega) = \frac{1}{\omega - \varepsilon_{\vec{k}}
- \Sigma(\vec{k},\omega)} ~~~ ,
\end{equation}
\noindent where $\varepsilon_{\vec{k}} = - 2t (\cos(k_x)
+ \cos(k_y) - \mu$. Also, we will require the spectral density,
$A(\vec{k},\omega)$ which is defined as
\begin{equation}\label{Akw}
A(\vec{k},\omega) = - \frac{1}{\pi}
\lim_{\delta \rightarrow 0^+} Im
G(\vec{k},\omega + i\delta) ~~~
\end{equation}
Using Eqs. (\ref{single} - \ref{Akw}), we arrive to the
following expression for the spectral density
\begin{equation}\label{11}
A(\vec{k},\omega) = \frac{-1}{\pi}
\frac{\alpha(\vec{k}) \Gamma(\vec{k})}
{\left( (\omega - \varepsilon_{\vec{k}})(\omega -
\Omega_{\vec{k}}) - \alpha(\vec{k}) \right)^2 +
\Gamma^2(\vec{k})(\omega - \varepsilon_{\vec{k}})^2 }
\end{equation}
Using the first three sum rules of Nolting\cite{1}
for the the spectral function of Eq. (\ref{Akw}) we obtain
the following equations
\begin{eqnarray}\label{3mom's}
\int_{-\infty}^{+\infty} A(\vec{k},\omega)d\omega &=&
a_o(\vec{k}) ~~~ \nonumber \\
\int_{-\infty}^{+\infty} \omega A(\vec{k},\omega)d\omega
&=& a_1(\vec{k}) ~~~
\nonumber \\
\int_{-\infty}^{+\infty} \omega^2 A(\vec{k},\omega)d\omega
&=& a_2(\vec{k})
~~~ ,
\end{eqnarray}
\noindent where
the $a_i(\vec{k})$'s,
$i=0,2$ are given
in Ref.\cite{1} (see also Ref.\cite{ours}).
We do not use the forth moment
or sum rule because we are at half - filling. Then,
we assume that at $\rho = 1/2$ the chemical
potential, $\mu = U/2$. This is verified as we will see
from our results. At this moment, we would like
to say once more that the drawback of previous calculations
(included ours) consists in that the density of states which results of
the two pole Ansatz for the one - particle
Green function, in the spherical
approximation of Nolting\cite{1}, always has a gap. This
solution (always a gap) is known in the literature as
the Hubbard-I solution\cite{IS} which has been critized since
many years ago by Laura Roth\cite{LR} among others. We
call the attention to Ref.\cite{oles} where the authors point out to the
fact that the $\vec{k}$-dependence has to be included in
$B(\vec{k})$. However, they study the
negative Hubbard model in the strong coupling
limit and in this limit we have a well developed correlation
gap, anyway. Ole\'s
and coworkers are practicioners of
the two - pole ansatz for the one - particle Green function,
without lifetime effects. Here we are including lifetime effects
as a crucial ingredient in the formulation. Morover, our
results clearly show that the cure the Hubbard-I solution in the
moment approach is not in the $B(\vec{k})$ - term.
In Fig. 1 we present the {\it energy spectrum}
of the self - energy for several values of the Brillouin zone,
for $U/t = 4$. In Fig. 2 we show the $\vec{k}$-dependence
of $\alpha(\vec{k})$ for $U/t = 4, 6, 8, 12$. In Fig. 3 we present
the $\vec{k}$ dependence of $\Gamma(\vec{k})$ for several values
of the Brillouin zone, for $U/t = 4, 6, 8, 12$. In Fig. 4 we
show the density of states, $N(\omega)$ vs $\omega$. $N(\omega)$
is defined as
\begin{equation}\label{dos}
N(\omega) = \frac{1}{N_s} \sum_{\vec{k}} A(\vec{k},\omega) ~~~
\end{equation}
\noindent In all the figures we have worked with a square lattice
with periodicity of $N_s = 32 \times 32$. We conclude that the
{\it spectral weight}, $\alpha(\vec{k})$ and the damping factor,
$\Gamma(\vec{k})$, do not strongly depend on $\vec{k}$,
for small values of $U/W$. For example,
for $U/t = 4.0$, $\Gamma(\vec{k})/t \approx 1.5$.
However, for larger values of $U/W$, $\alpha(\vec{k})$ has some
$\vec{k}$ dependence. On the contrary, $\Gamma(\vec{k})$ shows a
stronger $\vec{k}$ - dependence for larger values of $U/W$. We
also note that $\alpha(\vec{k}) \times \Gamma(\vec{k}) < 0$
in order to give a positive spectral density (see Eq.(\ref{11})).
Taking a look to
the density of states, $N(\omega)$ vs $\omega$, we see
that the correlation gap opens up
for $U/W \geq 1.0$. This is equivalent to the Hubbard-III like
solution\cite{III}. Due to our choice of the self - energy, we do not
have a Fermi liquid\cite{8,III}. Edwards and Hertz\cite{EH} have
studied the breakdown of Fermi liquid theory in the Hubbard model
at $T = 0$. They have drawn a phase boundary between a Fermi liquid
and a non - Fermi liquid ($\rho~vs~U$ in the
paramagnetic phase). According to these authors
for small values of $U$ we have a Fermi liquid behavior
(metallic Fermi liquid) and for larger values of $U$ we have
a non - Fermi liquid metal. In our approach we have a non -
Fermi liquid behavior for any value of interaction. In order to
agree with the authors of reference\cite{EH} we have to include
a $\omega^2$ behavior for frequencies close by to the chemical
potential in addition to the single pole structure chosen in the
present work. A metallic Fermi to metallic non - Fermi liquid
behavior has been obtained by Figueira, Anda and Nogueira\cite{b}.
We should mention that the phase diagram of the Hubbard model
contains the anti - ferromagnetic transition\cite{40} which has not
been considered here.
In short, we have postulated a one - pole Ansatz for the
self - energy (Eq. (\ref{single})) in the moment approach of
Nolting\cite{1}. Our essential new working idea
with respect to the now canonical method of Nolting is that we
start from the self - energy while Nolting proposes a two pole
ansatz for the one - particle Green function from the begining.
In both approaches sum rules for the spectral function,
$A(\vec{k},\omega)$, are imposed.
As function of interaction, we see a
Mott insulator - transition (MMIT) from a metal at weak
interaction (no gap at the zero frequency) to a well developed
gap (an insulator) for large values of the local interaction.
Our formulation is not equivalent to the infinite dimension
calculation of Georges et al\cite{a} since we have included the
$\vec{k}$ - dependence of the selfenergy (we are working in two
dimensions). There is the calculation of Figueira, Anda and
Nogueira\cite{b} where the authors have also neglected the $\vec{k}$ -
dependence of $\Sigma(\vec{k},\omega)$. In particular, the
authors of Refs.\cite{b,a} find a Kondo peak at $\omega = 0$. In
our case, $\rho = 1/2$, the peak-like structure seen at the
chemical potential is most likely due to the van Hove logarithmic
singularity which always goes away with the opening of the correlation
gap. With our precision\cite{201}, we cannot conclude about the
presence or not of the Kondo peak. Nolting
himself\cite{c} has also studied the effect of damping on
magnetism. We leave for the future the comparison with Ref.\cite{d} where
the authors discuss the use of the moment approach for interpolating
$\Sigma(\vec{k},\omega)$ between weak and strong interaction. We include
Figure 5 to present $\Omega_{(m,n)}$ vs $n$ for fixed values of
$m$, for $U/t = 4.0$ and a mesh of $10 \times 10$. It is worth
looking to the band structure for $m = 0$. This curve has
quasi the same structure than the band structure shown in Figure
1, for a bigger mesh ($32 \times 32$). Then, we can say that finite
size effects are minimal at least with respect to the band
structure calculations. (See the discussion of Ref.\cite{201}).
Finally, we would like to say that the non Fermi liquid behavior
that we have in our approach (by construction) has nothing to do
with any microscopic mechanism. Our arguments rely on general
grounds of a self - energy ansatz which is used in the one - particle
Green function to which exact sum rules (moments) are imposed.
Those interested in physical mechanisms of non - Fermi liquid
behavior are encouraged to take a look to the nice account presented
in Anderson's book\cite{TT}.
We would like to thank CONICIT (project F-139) and
the Brazilian Agency CNPq (project 300705/95-96) for finantial support.
Interesting discussions with Prof. H. Beck, Prof. M.S. Figueira, Prof. E.
Anda and Dr.
M.H. Pedersen are fully acknowledged. We thank Mar\'{\i}a Dolores
Garc\'{\i}a Gonz\'alez for a reading of the manuscript.
%
% Acknowledgments
%
%
% References.
% ^^^^^^^^^^^
%
\begin{references}
%
%%%%%%%%%%%%%%%%%%%%%
\bibitem{1}
W. Nolting,
Z. Physik {\bf 225}, 25 (1972); W. Nolting,
{\it Grundkurs: Theoretische
Physik. 7 Viel-Teilchen-Theorie.}
Verlag Zimmermann-Neufang
(Ulmen -1992).
%%%%%%%%%%%%%%
\bibitem{2}
F. Gebhard, {\it The Mott
Metal - Insulator Transition. Models and
Methods.} Springer (1997) and references herein.
%%%%%%%%%%%%%
\bibitem{3}
J. Bednorz and K.A. M\"uller,
Z. Phys. B~{\bf 64}, 189 (1986).
%%%%%%%%%%%%%%%%%%%%%%
\bibitem{4}
P.W. Anderson, Science
{\bf 235}, 1196 (1987);
{\it Frontiers and Borderlines
in Many Particle Physics}.
North Holland, Amsterdam (1988).
%%%%%%%%%%%%%%%
\bibitem{5}
J. Hubbard, Proc. R. Soc.
London A {\bf 276}, 238 (1963).
%%%%%%%%%%%%%%
\bibitem{6}
J. Hubbard, Proc. R. Soc.
London A {\bf 281}, 401 (1964).
%%%%%%%%%%%%%%%%%%%%%%
\bibitem{7}
Peter Fulde, {\it Electron
Correlations in Molecules
and Solids}. Springer-Verlag
(1993). 2nd Edition.
%%%%%%%%%%%%%%%%%%%%
\bibitem{8}
J. Gonz\'alez, M.A. Mart\'{\i}n -
Delgado, G. Sierra and A.H. Vozmediano,
{\it Quantum Electron Liquids and High -
$T_c$ Superconductivity}. Springer (1995).
%%%%%%%%%%%%%%%%%%
\bibitem{9}
Hal Tasaki, {\it The Hubbard
Model: Introduction and
Some Rigorous Results}.
Preprint Sissa (1995); J. Stat.
Phys. {\bf 84}, 535 (1996).
%%%%%%%%%%%%%%%%%%
\bibitem{Assa}
Assa Auerbach, {\it
Interacting Electron and
Quantum Magnetism}.
Springer-Verlag (1994).
This author justifies the
truncations of the Hubbard
model in the "atomic
limit". In this way, one
avoids the high complexity
(such as screening effects)
of the long-range Coulomb
forces.
%%%%%%%%%%%%%%%%%%%%
\bibitem{all}
D.B. McWhan, A. Menth, J.P.
Remeika, W.F. Brinkman, and
T.M. Rice, Phys. Rev. B {\bf 7},
1920 (1973); S.A. Carter,
T.F. Rosenbaum, J.M. Honig,
and J. Spalek, Phys. Rev. Lett.
{\bf 67}, 3440 (1991); Y.
Tokura, Phys. Rev. Lett. {\bf 70},
2126 (1993).
%%%%%%%%%%%%%%%%%%%%%
\bibitem{Alexei}
Alexei M. Tsvelik,
{\it Quantum Field
Theory in Condensed
Matter Physics}. Chapter 12.
Cambridge University Press (1996). Here the
author analyses a non - Fermi liquid case when
when there are long distant forces in metals.
%%%%%%%%%%%%%%%%%%%%%%
\bibitem{toy}
Y. Hatsugai and M. Kohmoto,
J. Phys. Soc. Jpn,
{\bf 61}, 2056 (1992); D.
Lidsky, J. Shiraishi,
Y. Hatsugai and M. Kuhmoto,
cond-mat/9705068; see
also, M.A. Continentino
and M.D. Coutinho-Filho,
Solid State Commun.
{\bf 90}, 619 (1994);
F.S. Nogueira and E.V.
Anda, Int. J. Mod. Phys.
B{\bf 10}, 3075 (1996);
G. Baskaran, Modern
Phys. Letters B{\bf 5},
643 (1991); V.M. Muthukumar
and G. Baskaran, Modern
Phys. Letters B{\bf 8}, 699
(1994); Phys. Rev. Lett.
{\bf 71}, 777 (1993).
%%%%%%%%%%%%%%%%%%%%%%%
\bibitem{ours}
R. Micnas, M.H. Pedersen,
S. Schafroth, T. Schneider,
J.J. Rodr\'{\i}guez - N\'u\~nez
and H. Beck,
Phys. Rev. B {\bf 52}, 16223
(1995); M.H. Pedersen, J. J.
Rodr\'{\i}guez-N\'u\~nez,
H. Beck, T. Schneider
and S. Schafroth,
Z. Physik B {\bf 103},
21-28 (1997); T. Schneider,
M.H. Pedersen and J.J.
Rodr\'{\i}guez - N\'u\~nez,
Z. Phys. B {\bf 100}, 263 (1996).
%%%%%%%%%%%%%%%%%%%%
\bibitem{CL}
P.M. Chaikin and T.C. Lubensky,
{\it Principles of Condensed Matter Physics}.
Chapter 7. Cambridge (1995).
%%%%%%%%%%%%%%%%%%
\bibitem{IS}
Yu. Izyumov and Yu. N. Skryabin,
{\it Statistical
Mechanics of Magnetically
Ordered Systems}. New York:
Plenum Press (1988).
%%%%%%%%%%%%%%%%%%%%%%
\bibitem{LR}
Laura Roth, Phys. Rev.
{\bf 184}, 451 (1969).
%%%%%%%%%%%%%
\bibitem{oles}
Henk Eskes and Andrzej M. Ole\'s,
Phys. Rev. B {\bf 55}, 2032 (1997).
%%%%%%%%%%%%%%%%%%%%%%%
\bibitem{III}
J. Hubbard, Proc. Royal Soc. London A{\bf 281},
401 (1964).
%%%%%%%%%%%%%%%%%%%%%%%%
\bibitem{EH}
D.M. Edwards and J.A. Hertz, Physica B {\bf 163},
527 (1990).
%%%%%%%%%%%%%%%%%%%%%%
\bibitem{b}
M.S. Figueira, E. Anda and F.S. Nogueira (Unpublished);
also, M.S. Figueira, private communication. This calculation
is at half filling.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\bibitem{40}
R.T. Scalettar, D.J. Scalapino, R.L. Sugar and D.
Toussaint, Phys. Rev. B {\bf 39}, 4711 (1989). These
authors discuss the phase diagram of the half - filled
$3D$ Hubbard model ($4^3$) and
$8^3$ lattices. They calculate $T_N$, the N\'eel
temperature, by requiring that the staggered susceptibility,
$\chi$, varies linearly with lattice size.
%%%%%%%%%%%%%%%%%%%%%%
\bibitem{a}
Antoine Georges,
Gabriel Kotliar,
Werner Krauth and
Marcelo J. Rozenberg,
Rev. Mod.
Phys. {\bf 68},
13 (1996).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\bibitem{201}
Our calculations have been done for a finite
$\vec{k}$ - mesh of $32 \times 32$. Most likely we
have to consider larger meshes. However, the main
conclusion of the present paper will remain:
the Mott metal - insulator transition ($MMIT$).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\bibitem{c}
T. Herrmann and W. Nolting, Phys.
Rev. B {\bf 53}, 10579
(1996). These authors discuss
the case of magnetism due
to quasi-particle damping.
%%%%%%%%%%%%%%%%%%%%%%%
\bibitem{d}
M. Potthoff, T. Wegner and W. Nolting,
cond-mat/9704005. Here the authors study
the use of the moment for interpolating
the self-energy.
%%%%%%%%%%%%%%%%%%%%%
\bibitem{TT}
P.W. Anderson, {\it The Theory of Superconductivity in
the High - $T_c$ Cuprates}. Princeton Series in Physics (1997);
see also Ref.\cite{Alexei}
\end{references}
\vspace{1.6cm}
\begin{center}
{\huge FIGURES}
\end{center}
\vspace{0.8cm}
\noindent
Figure 1. $\Omega_{0,n}$ vs n for
$U/t = 4.0$ and $\rho = 0.5$. The $\vec{k}$ vector
is given by $\vec{k} = \pi(0,n)/32$. As we work in
two dimensions, the bandwidth is $W = 8t$. Compare with
$m = 0$ of Figure 5.
\vspace{0.6cm}
\noindent
Figure 2. $\alpha(0,n)~vs~n$ for different values of $U/t$.
Same parameters as in Figure 1.
\vspace{0.6cm}
\noindent
Figure 3. We show $\Gamma(0,n)~vs~n$ for different values of $U/t$.
Same parameters as in Figure 1.
\vspace{0.6cm}
\noindent Figure 4. $N(\omega)~vs~\omega$ for
$U/t = 4, ~12$. We see the opening of the correlation gap for
$U/W \geq 1.0$, signaling the Mott metal - insulator transition.
\vspace{0.6cm}
\noindent Figure 5. $\Omega_{m,n}$ vs $n$ for some fixed
values of $m$. $U/t = 4.0$, $\rho = 0.5$ and our mesh is
$10 \times 10$. Compare $m = 0$ with Figure 1.
\end{document}