Comment on `Two-body random ensembles: from nuclear spectra to random polynomials
Extracted source text ยท arXiv nucl-th/0104005
\documentstyle[aps,prl,twocolumn,epsfig]{revtex}
\draft
\newcommand{\ba}{\begin{eqnarray}}
\newcommand{\ea}{\end{eqnarray}}
\def\ii{\'{\i}}
\setlength{\topmargin}{-1.0truecm}
\begin{document}
\pagestyle{plain}
\title{Comment on `Two-body random ensembles:
from nuclear spectra to random polynomials'}
\author{R. Bijker$^{a}$ and A. Frank$^{a,b}$}
\address{$^{a}$Instituto de Ciencias Nucleares,
Universidad Nacional Aut\'onoma de M\'exico, \\
Apartado Postal 70-543, 04510 M\'exico, D.F., M\'exico
\newline
$^{b}$Centro de Ciencias F\ii sicas,
Universidad Nacional Aut\'onoma de M\'exico, \\
Apartado Postal 139-B, Cuernavaca, Morelos, M\'exico}
\maketitle
%\pacs{PACS number(s): 21.10.Re, 21.60.Ev, 21.60.Fw, 24.60.Lz}
An investigation of the spectroscopy of even-even nuclei with
random one- and two-body interactions has led to the surprising
result that the ground state has $J^P=0^+$ in typically $60-70$
$\%$ of the cases. This holds both for the nuclear shell model
\cite{JBD,BFP} and the interacting boson model \cite{BF1}.
To understand this phenomenon it was suggested to study the
distribution of the lowest eigenvalues \cite{BF2}. To this end,
a technique was developed in \cite{DK} for tridiagonal
Hamiltonian matrices, such as for the one- and two-body
interactions in the vibron model. It relies on the property
that for these matrices it is possible to obtain an estimate for
the lowest eigenvalue as the minimum of a function \cite{Hollenberg}.
The purpose of this comment is to point out that for
the vibron model the appropriate function is not a
quadratic polynomial as used in \cite{DK}, but a more
general form which in the vibrational basis to leading order
in $1/N$, is given by
\ba
f(z,j) = a \, z^2 + b \, z + c + d \, j^2
+ e \, \sqrt{(1-z)^2(z^2-j^2)} ~.
\nonumber
\ea
Here $z=n_p/N$ and $j=J/N$ \cite{DK}.
In the absence of off-diagonal terms ($e=0$), the spectrum
is that of a three-dimensional oscillator characterized by
$n_p=0,1,\ldots,N$ and $J=n_p,n_p-2,\ldots,1$ ($n_p$ odd)
or $0$ ($n_p$ even). In this case, the function $f(z,j)$ reduces
to a quadratic polynomial, and the results of \cite{DK} can be used
to obtain the percentages for ground states with $J=0$, $J=1$ and
$J=N$ to be $64.6$, $5.2$ and $19.8$ $\%$, respectively, in
close agreement with the exact values of $64.4$, $4.1$ and
$24.2$ $\%$ obtained for $N=20$ and 100000 runs.
In the case of only off-diagonal interactions ($a=b=c=d=0$),
the ground state always has $J=0$.
In the general case, both diagonal and off-diagonal terms
are present. The off-diagonal term introduces a deformation
in the system, which gives rise to the occurrence of vibrational
bands with $J=0,1,2,\ldots,J_{\rm max}$. As a consequence, the
$J=1$ ground states tend to disappear in favor of an enhancement of
the $J=0$ ground states. This effect explains qualitatively
the change in the $J=0$, $J=1$ and $J=N$ ground states from
$64.4$, $4.1$ and $24.2$ to $72.1$, $0.5$ and $23.7$ $\%$,
respectively.
In order to carry out a more quantitative analysis, we
have developed a simple method to analyze arbitrary
functions $f(z,j)$ by sampling the two-dimensional space
($z,j$). As an example, in Figure~1 we show the percentages of
ground states with $J=0$ as a function of
$x=\sigma_1/(\sigma_0+\sigma_1)$, where $\sigma_i$ denotes
the width of the Gaussian distributions for the diagonal
($i=0$) and off-diagonal ($i=1$) matrix elements.
The case of equal
widths (discussed in the previous paragraph) corresponds
to $x=0.5$. Figure~1 shows a good agreement between the
values obtained from an analysis of $f(z,j)$ (dashed line)
and the exact results (solid line). The discrepancies observed
for the quadratic polynomials of \cite{DK} (dotted line)
are due to a different treatment of the off-diagonal matrix element.
In conclusion, the approach proposed in \cite{DK} provides an
interesting method to analyze random interactions in terms of
random polynomials which is especially suited for tridiagonal
Hamiltonian matrices. The advantage of this method is that it
foregoes the diagonalization of thousands of matrices.
In this comment we have presented a careful analysis of the
dominance of $J^P=0^+$ ground states in the vibron model, and
found good agreement with the exact calculations.
For cases of interest to nuclear physics, such as the IBM or
the shell model, the Hamiltonian matrix has a more complicated form.
Although it can be brought to tridiagonal form, the lack of an
analytic form for the function $f$ leads to complications comparable
to those of bringing it to diagonal form right away. For these cases
the method becomes difficult to apply (other than by analogy).
Many questions remain open to explain the robust features
observed in randomly interacting many-body systems.
This work was supported in part by CONACyT under projects
32416-E and 32397-E, and by DPAGA under project IN106400.
\begin{references}
\bibitem{JBD}
C.W. Johnson, G.F. Bertsch and D.J. Dean,
Phys. Rev. Lett. {\bf 80}, 2749 (1998).
\bibitem{BFP}
R. Bijker, A. Frank and S. Pittel,
Phys. Rev. C {\bf 60}, 021302 (1999).
\bibitem{BF1}
R. Bijker and A. Frank,
Phys. Rev. Lett. {\bf 84}, 420 (2000).
\bibitem{BF2}
R. Bijker and A. Frank,
Phys. Rev. C {\bf 62}, 014303 (2000).
\bibitem{DK}
D. Kusnezov,
Phys. Rev. Lett. {\bf 85}, 3773 (2000).
\bibitem{Hollenberg}
Lloyd C.L. Hollenberg and N.S. White,
Phys. Rev. B {\bf 54}, 16309 (1996).
\end{references}
\begin{figure}[h]
\centerline{\hbox{
\epsfig{figure=dekay_fig.eps,height=0.3\textwidth,width=0.5\textwidth} }}
\caption[]{Percentage of $J^P=0^+$ ground states as a function
of $x$ (defined in text) for the exact results (solid), for
$f(z,j)$ (dashed) and for the quadratic polynomial (dotted)
\protect\cite{DK}.}
\end{figure}
\end{document}