Phase diagrams of the generalized spin-1/2 ladder under staggered field and dimerization: A renormalization group study
Extracted source text ยท arXiv cond-mat/0306364
%\documentstyle[osa,manuscript]{revtex} % DON'T CHANGE
%\documentstyle[epsf,floats,aps,latexsym,twocolumn,prb]{revtex}
\documentclass[twocolumn,floatfix,showpacs,preprintnumbers,amsmath,amssymb,prb]{revtex4}
%\documentclass[preprint,floatfix,showpacs,preprintnumbers,amsmath,amssymb,prb]{revtex4}
\usepackage{graphicx}% Include figure files
\usepackage{dcolumn}% Align table columns on decimal point
\usepackage{bm}% bold math
%
%
\newcommand{\MF}{{\large{\manual META}\-{\manual FONT}}}
\newcommand{\manual}{rm} % Substitute rm (Roman) font.
\newcommand\bs{\char '134 } % add backslash char to \tt font
%
%
\begin{document}
\input{epsf}
%\preprint{mpi-pks/0207xxx}
\title{Phase diagrams of the generalized spin-$\frac{1}{2}$ ladder under
staggered field and dimerization: A renormalization group study}
\author{Y.-J. Wang}
\affiliation{Max-Planck-Institut f\"{u}r Physik Komplexer Systeme,
N\"{o}thnitzer Stra{\ss}e 38, 01187 Dresden, Germany}
%\address{Max Planck Institute for the Physics of Complex Systems,
%Noethnitzer Strasse 38, 01187, Dresden, Germany}
\date{\today}
\begin{abstract}
In the weak-coupling regime of the continuous theories, two sets of one-loop
renormalization group equations are derived and solved to disclose the phase
diagrams of the antiferromagnetic generalized two-leg spin-$\frac{1}{2}$
ladder under the effect of (I) a staggered external magnetic field and (II) an
explicit dimerization. In model (I), the splitting of the SU(2)$_2$
critical line into U(1) and Z$_2$ critical surfaces is observed; while in
model (II), two critical surfaces arising from their underlying critical lines
with SU(2)$_2$ and Z$_2$ characteristics merge into an SU(2)$_1$ critical
surface on the line where the model attains its highest symmetry.
\end{abstract}
\pacs{75.10.Jm, 71.10.Pm, 64.60.Ak, 05.70.Jk}
\maketitle
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Introduction}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
The physics of quantum phase transitions in one-dimensional systems has
attracted much attention in recent years because not only of its
speciality---the conformal symmetries and universalities in (1+1)-critical
models,\cite{gins88, aff88, gnt99} but also of its generality---the quantum
critical phenomena in higher
dimensions.\cite{cont01} A general question is posed as follows: If a
critical system defined in the ultraviolet limit is deformed by more
than one relevant perturbations, what is the fate of the system? Can
the criticality be reached again in the infrared limit because of the
``cancellation'' effect among these perturbations, thus a quantum phase
transition takes place? A typical model of this sort is the so-called
double-frequency sine-Gordon model,\cite{dm98,fgn00} where an Ising
transition was thought to be driven by two competing relevant
perturbations. The description of the phase transition requires a
non-perturbative scheme being able to identify correctly those degrees
of freedom that remain massive and the low-energy ones that eventually
become critical and undergo the transition.
The same strategy was extended to a more complicated non-Abelian
case---a two-leg spin-$\frac{1}{2}$ Heisenberg antiferromagnetic ladder
subject to a site-parity-breaking dimerization field.\cite{wn00} The non-trivial
aspect of the non-perturbative approach adopted in this study of the
strong-coupling limit of the model is that one has to preserve the SU(2)
symmetry, which cannot be spontaneously broken, in the Ising model language
which apparently has only a discrete symmetry. This hidden symmetry was realized
when using very non-local duality transformations for the coupled Ising
models. An effective low-energy Hamiltonian depicting the ensuing quantum
phase transition was derived and the transmutation of all physical fields
at the infrared fixed point was identified. It was unambiguously shown that
a quantum phase transition occurs from the universality class of SU(2) level 2
Wess-Zumino-Novikov-Witten (WZNW) model in the ultraviolet limit to the
universality class of SU(2) level 1 WZNW model in the infrared limit.
Another possible scenario was set up by introducing a staggered magnetic
field, which explicitly breaks both SU(2) and bond-parity symmetries, to the
generalized spin ladder.\cite{wefn02} Depending on which phase the model
belongs to in absence of the staggered field, either U(1) or Z$_2$ criticality
was predicted under the effect of the staggered field. An interesting
new phase that interpolates between the Haldane spin liquid phase
and the spontaneously dimerized phase was found beyond the transitions.
The main features of this intermediate phase are: partially (transverse)
coherent spin excitations and partially (longitudinal) non-vanishing string
order parameter.
As we can see, the physics of the generalized spin-$\frac{1}{2}$ ladder under
the effect of the dimerization or the staggered field is very rich. In this
paper, we shall scrutinize these two models in a more general perspective and
try to understand the phase diagrams as a whole. Our main attention is to be
paid to the overall topology of the phase diagrams; while the physical
properties in various phases are only to be mentioned briefly if available.
The paper is organized as follows. In Sec.~\ref{sec_mod}, we present a short
overview of the continuous version of our models. In Sec.~\ref{sec_rgeq}, we
derive two sets of renormalization group (RG) equations and central charge
formulas. These equations are (numerically) solved and analyzed in
Sec.~\ref{sec_sol} to bring about the phase diagrams of the continuous models.
The implication to the ladder systems is summarized and concluded in
Sec.~\ref{sec_sum}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Models and Their Continuous Field-Theoretic Mappings} \label{sec_mod}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We consider a standard two-leg spin-$\frac{1}{2}$ antiferromagnetic ($J>0$)
Heisenberg ladder ($J_\perp$) generalized to include a four-spin interaction
($V$):\cite{nt97,km98}
\begin{eqnarray}
H_{\text{gen}} &=& J \sum_{a=1,2} \sum_n \mbox{\boldmath $S$}_{a,n} \!\cdot\!
\mbox{\boldmath $S$}_{a,n+1}
+ J_\perp \sum_n \mbox{\boldmath $S$}_{1,n} \!\cdot\!
\mbox{\boldmath $S$}_{2,n} \nonumber \\
&& + V \sum_n (\mbox{\boldmath $S$}_{1,n} \!\cdot\! \mbox{\boldmath $S$}_{1,n+1})
(\mbox{\boldmath $S$}_{2,n} \!\cdot\! \mbox{\boldmath $S$}_{2,n+1}) \;,
\label{ham_gen}
\end{eqnarray}
superimposed by either a staggered magnetic field, model (I),
\begin{equation}
H_{h} = h \sum_{a=1,2} \sum_n (-1)^n S_{a,n}^z \;,
\label{ham_h}
\end{equation}
or a $\pi$-phase (relative) dimerization, model (II),
\begin{equation}
H_{\Delta} = \Delta \sum_{a=1,2} \sum_n (-1)^{n+a}
\mbox{\boldmath $S$}_{a,n} \!\cdot\! \mbox{\boldmath $S$}_{a,n+1} \;.
\label{ham_dlt}
\end{equation}
Throughout the paper, we treat the models only in the weak-coupling regime,
providing
\begin{equation}
|J_\perp|, |V|, |h| \; (\text{or } |\Delta|) \ll J \;. \label{wccnd}
\end{equation}
In this limit, the field-theoretic approach in continuum is of advance and
power in describing the low-energy physics of the models. In fact, the
resulting Hamiltonians are represented by four interacting Majorana (real)
fermions or, equivalently, four coupled Ising models.
According to Refs.~\onlinecite{snt96, nt97, wn00, wefn02}, we have the
following mappings in terms of Majoranas ($\xi_{R,L}$) and Ising variables
($\sigma, \mu$):
\begin{equation}
{\cal H}_{\text{gen}} = {\cal H}_{\text{crit}} + {\cal H}_{\text{mass}}
+ {\cal H}_{\text{marg}} \;,
\end{equation}
where the critical theory
\begin{equation}
{\cal H}_{\text{crit}} = -\frac{iv}{2} \sum_{\mu=0}^3 \big(
\xi^\mu_R \partial_x \xi^\mu_R - \xi^\mu_L \partial_x \xi^\mu_L \big)
\end{equation}
having an O(4) symmetry is inherited from two decoupled Heisenberg chains
with SU(2)$\times$SU(2) symmetry, $v\sim Ja$.\cite{note1} The mass terms result from part
of the interchain coupling ($J_\perp$) and the four-spin interaction ($V$),
both of which have the same scaling dimension, $d=1$:
\begin{equation}
{\cal H}_{\text{mass}} = -im_s\, \xi^0_R \xi^0_L - im_t\,
\mbox{\boldmath $\xi$}_R \!\cdot\! \mbox{\boldmath $\xi$}_L \;,
\end{equation}
where the singlet (associated with $\xi^0$) and triplet
($\mbox{\boldmath $\xi$} = (\xi^1, \xi^2, \xi^3)$) masses are related to
the parameters of the lattice spin model by
\begin{equation}
m_s = -3 c J_\perp -c' V \;, \qquad
m_t = c J_\perp -c' V \;, \label{mrjv}
\end{equation}
respectively. Here $c$ and $c'$ are some {\em positive}
constants. The marginal terms are from the current-current part of the
interchain coupling with scaling dimension $d=2$:
\begin{equation}
{\cal H}_{\text{marg}} = \frac{1}{2} g_1 ( \mbox{\boldmath $\xi$}_R \!\cdot\!
\mbox{\boldmath $\xi$}_L)^2 + g_2 ( \mbox{\boldmath $\xi$}_R \!\cdot\!
\mbox{\boldmath $\xi$}_L) (\xi_R^0 \xi_L^0) \;,
\end{equation}
where $g_1 = \frac{1}{2} J_\perp a -\gamma$ and $g_2 = -\frac{1}{2} J_\perp a
-\gamma$, with $\gamma (>0)$ the marginally irrelevant coupling in the chains.
Obviously, the symmetry of the continuous model ${\cal H}_{\text{gen}}$ is
O(3)$\times$Z$_2$, corresponding to SU(2)$\times$Z$_2$ for the lattice model
(\ref{ham_gen}) (Z$_2$ arising from the interchange of two chains),
unless $J_\perp =0$ when the highest symmetry of O(4) is reached,
corresponding to SU(2)$\times$SU(2).
The field-theoretic counterparts of the staggered field (\ref{ham_h}) and the
dimerization (\ref{ham_dlt}) are represented, respectively, by
\begin{eqnarray}
{\cal H}_h &=& (h/\alpha) \sigma_1 \sigma_2 \mu_3 \mu_0 \;, \\
{\cal H}_\Delta &=& (\Delta/\alpha) \sigma_1 \sigma_2 \sigma_3 \sigma_0 \;,
\end{eqnarray}
where $\alpha$ with a magnitude order of the lattice constant is the
short-distance cut-off of the theory. While ${\cal H}_\Delta$ keeps the O(3)
invariance for the triplet sector intact, ${\cal H}_h$ breaks this symmetry down to
O(2)$\times$Z$_2$. This is a reflection of the oriented field (\ref{ham_h})
lowers SU(2) to U(1)$\times$Z$_2$.
In what follows, we shall address model (I): ${\cal H}_{\text{gen}} +
{\cal H}_h$, and model (II): ${\cal H}_{\text{gen}} + {\cal H}_\Delta$, in RG
approach. Our purpose aims at drawing the phase diagrams by analyzing the
infrared properties of the RG equations.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Derivation of the RG Equations} \label{sec_rgeq}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
It is rather straightforward to establish the one-loop RG equations out of
the operator product expansions\cite{gins88} (OPE's) between the perturbative
operators.\cite{cardy88} For a critical model perturbed by
\begin{equation}
{\cal H}_{\text{pert}} = \frac{v}{\pi\alpha^2} \sum_i \lambda_i {\cal O}_i \;,
\end{equation}
where $\lambda_i$ are dimensionless coupling constants of the corresponding
dimensionless operators ${\cal O}_i$ (with scaling dimension $d_i$), if
${\cal O}_i$ are normalized in such a way that
\begin{equation}
\langle {\cal O}_i(z,\bar{z}) {\cal O}_j(0,0) \rangle
= \delta_{ij} \bigg( \frac{\alpha}{|z|} \bigg)^{2d_i} \;,
\label{norm}
\end{equation}
($z= v\tau + ix$ being the time-space in complex coordinates) and the
short-distance OPE's satisfy
\begin{equation}
{\cal O}_i(z,\bar{z}) {\cal O}_j(0,0)
\sim {c_{ij}}^k \bigg( \frac{\alpha}{|z|} \bigg)^{d_i+d_j-d_k}
{\cal O}_k(0,0) \;,
\end{equation}
the one-loop RG equations for $\lambda_i$ are readily given by
\begin{equation}
\dot{\lambda}_k \equiv \frac{d \lambda_k}{d \ln \!L}
= (2-d_k) \lambda_k - \sum_{ij}
{c_{ij}}^k \lambda_i \lambda_j + O(\lambda^3) \;. \label{rgeq}
\end{equation}
In addition, the central charge satisfying Zamo\-lod\-chi\-kov's decreasing
theorem\cite{zam86} is perturbatively evaluated to this order to be
\begin{eqnarray}
C(\{\lambda\}) &=& C_{\text{UV}} - 3 \sum_k (2-d_k) \lambda_k^2 \nonumber \\
&& + 2 \sum_{ijk} {c_{ij}}^k \lambda_i \lambda_j \lambda_k
+ O(\lambda^4) \;, \label{cnch}
\end{eqnarray}
where $C_{\text{UV}}$ is the central charge of the unperturbed critical system
(the ultraviolet fixed point). In the present case, $C_{\text{UV}} =2$ is the
central charge of two decoupled chains.
Under normalization (\ref{norm}), the OPE coefficients ${c_{ij}}^k$ are
related to the three-point function $\langle {\cal O}_i {\cal O}_j {\cal O}_k
\rangle$, and thus are complete symmetric among the indices
($ijk$).\cite{gins88} Not only this procedure simplifies the calculations,
but also is it essential to give rise to a correct form for the central charge
(\ref{cnch}).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{RG equations for model (I)}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Since for this model,
\begin{equation}
{\cal H}_{\text{pert}}^{\text{(I)}} = {\cal H}_{\text{mass}}
+ {\cal H}_{\text{marg}} + {\cal H}_{h} \;,
\end{equation}
and noticing ${\cal H}_{h}$ breaks the O(3) symmetry in the triplet sector,
we have to divide the triplet mass $m_t$ and the marginal coupling constants
$g_1$ and $g_2$ into doublets and singlets. Denoting the mass bilinear (or
energy density) of the Ising model, $\varepsilon_\mu = i\xi_R^\mu \xi_L^\mu$,
we write down:
\begin{eqnarray}
{\cal H}_{\text{pert}}^{\text{(I)}} &=&
-m_t^d (\varepsilon_1 + \varepsilon_2)
-m_t^s \varepsilon_3 - m_s \varepsilon_0 \nonumber \\
&& - g_1^d \varepsilon_1 \varepsilon_2
- g_1^s (\varepsilon_1 + \varepsilon_2) \varepsilon_3 \nonumber \\
&& - g_2^d (\varepsilon_1 + \varepsilon_2) \varepsilon_0
- g_2^s \varepsilon_3 \varepsilon_0 \nonumber \\
&& + (h/\alpha) \sigma_1 \sigma_2 \mu_3 \mu_0 \nonumber \\
&=& \frac{v}{\pi\alpha^2} \sum_{i=1}^8 \lambda_i {\cal O}_i \;,
\end{eqnarray}
where the dimensionless operators in both fermion and Abelian boson
representations (``$+/-$'' associated with sectors $(1,2)/(3,0)$,
respectively):
\begin{eqnarray}
&& {\cal O}_1 = -\sqrt{2} \pi\alpha (\varepsilon_1 + \varepsilon_2)
= -\sqrt{2} \cos\sqrt{4\pi} \Phi_+ \;, \nonumber \\
&& {\cal O}_2 = - 2\pi\alpha \varepsilon_3
= -\cos\sqrt{4\pi} \Phi_- + \cos\sqrt{4\pi} \Theta_- \;, \nonumber \\
&& {\cal O}_3 = - 2\pi\alpha \varepsilon_0
= -\cos\sqrt{4\pi} \Phi_- - \cos\sqrt{4\pi} \Theta_- \;; \nonumber \\
&& {\cal O}_4 = -(2\pi\alpha)^2 \varepsilon_1 \varepsilon_2
= 4\pi\alpha^2 \bar{\partial} \Phi_+ \partial \Phi_+ \;, \nonumber \\
&& {\cal O}_5 = -2\sqrt{2} (\pi\alpha)^2 (\varepsilon_1 + \varepsilon_2)
\varepsilon_3 \nonumber \\
&& \hspace{1.5em} = -\sqrt{2} \cos\sqrt{4\pi} \Phi_+
\big(\cos\sqrt{4\pi} \Phi_- - \cos\sqrt{4\pi} \Theta_- \big) \;, \nonumber \\
&& {\cal O}_6 = -2\sqrt{2} (\pi\alpha)^2 (\varepsilon_1 + \varepsilon_2)
\varepsilon_0 \nonumber \\
&& \hspace{1.5em} = -\sqrt{2} \cos\sqrt{4\pi} \Phi_+
\big(\cos\sqrt{4\pi} \Phi_- + \cos\sqrt{4\pi} \Theta_- \big) \;, \nonumber \\
&& {\cal O}_7 = -(2\pi\alpha)^2 \varepsilon_3 \varepsilon_0
= 4\pi\alpha^2 \bar{\partial} \Phi_- \partial \Phi_- \;; \nonumber \\
&& {\cal O}_8 = 2 \sigma_1 \sigma_2 \mu_3 \mu_0
= 2 \sin\sqrt{\pi} \Phi_+ \cos\sqrt{\pi} \Phi_- \;.
\end{eqnarray}
The scaling dimensions, $d_{1,2,3} = 1$, $d_{4,5,6,7} = 2$, and
$d_8 = \frac{1}{2}$. Accordingly, the dimensionless coupling constants:
\begin{eqnarray}
&& \lambda_1 \equiv \frac{\eta_1}{\sqrt{2}} = \frac{m_t^d \alpha}{\sqrt{2}v}
\;, \qquad
\lambda_2 \equiv \frac{\eta_2}{2} = \frac{m_t^s \alpha}{2v} \;, \nonumber \\
&& \lambda_3 \equiv \frac{\eta_3}{2} = \frac{m_s \alpha}{2v} \;; \nonumber \\
&& \lambda_4 \equiv \frac{\eta_4}{4} = \frac{g_1^d}{4\pi v} \;, \qquad
\lambda_5 \equiv \frac{\eta_5}{2\sqrt{2}} = \frac{g_1^s}{2\sqrt{2}\pi v} \;,
\nonumber \\
&& \lambda_6 \equiv \frac{\eta_6}{2\sqrt{2}} = \frac{g_2^d}{2\sqrt{2}\pi v}
\;, \qquad
\lambda_7 \equiv \frac{\eta_7}{4} = \frac{g_2^s}{4\pi v} \;; \nonumber \\
&& \lambda_8 \equiv \frac{\eta_8}{2} = \frac{\pi h}{2v} \;. \label{cpct}
\end{eqnarray}
It is easy to calculate all the OPE coefficients by using either
fermion or boson representation. For example, we find\cite{note2}
\begin{eqnarray}
&& \hspace{-2em} {\cal O}_8(z,\bar{z}) {\cal O}_8 \sim \frac{\alpha}{|z|}
+ \frac{1}{\sqrt{2}} {\cal O}_1 -\frac{1}{2} ( {\cal O}_2 + {\cal O}_3 )
\nonumber \\
&& + \bigg( \frac{\alpha}{|z|} \bigg)^{-1}
\bigg[ \frac{1}{2\sqrt{2}} ( {\cal O}_5 + {\cal O}_6 )
- \frac{1}{4} ( {\cal O}_4 + {\cal O}_7 ) \bigg] \;. \label{o8ope}
\end{eqnarray}
Therefore, we read off
\begin{eqnarray}
&& {c_{88}}^1 = \frac{1}{\sqrt{2}} \;, \quad
{c_{88}}^2 = {c_{88}}^3 = -\frac{1}{2} \;, \nonumber \\
&& {c_{88}}^5 = {c_{88}}^6 = \frac{1}{2\sqrt{2}} \;, \quad
{c_{88}}^4 = {c_{88}}^7 = -\frac{1}{4} \;.
\end{eqnarray}
The algebra in this way is closed, and we list all the coefficients in
Table~\ref{tab_coe}.
\begin{table}[ht]
\caption{OPE coefficients ${c_{ij}}^k$ (given in the parentheses, in front are
indices $k$).}
\begin{tabular}{c|cccccccc}
%\colrule\colrule
\hline\hline
$i\backslash j$ & 1 & 2 & 3 & 4 & 5 & 6 & 7 & 8 \\ \colrule
1 & 4(-1) & 5(-1) & 6(-1) & 1(-1) & 2(-1) & 3(-1)
& & 8($\frac{1}{\sqrt{2}}$) \\
2 & 5(-1) & & 7(-1) & & 1(-1) & & 3(-1) & 8(-$\frac{1}{2}$) \\
3 & 6(-1) & 7(-1) & & & & 1(-1) & 2(-1) & 8(-$\frac{1}{2}$) \\
4 & 1(-1) & & & & 5(-1) & 6(-1) & & 8(-$\frac{1}{4}$) \\
5 & 2(-1) & 1(-1) & & 5(-1) & 4(-1) & 7(-1)
& 6(-1) & 8($\frac{1}{2\sqrt{2}}$) \\
6 & 3(-1) & & 1(-1) & 6(-1) & 7(-1) & 4(-1) &
5(-1) & 8($\frac{1}{2\sqrt{2}}$) \\
7 & & 3(-1) & 2(-1) & & 6(-1) & 5(-1) &
& 8(-$\frac{1}{4}$) \\
8 & 8($\frac{1}{\sqrt{2}}$) & 8(-$\frac{1}{2}$) & 8(-$\frac{1}{2}$)
& 8(-$\frac{1}{4}$) & 8($\frac{1}{2\sqrt{2}}$) & 8($\frac{1}{2\sqrt{2}}$)
& 8(-$\frac{1}{4}$) & (${c_{88}}^k$)\footnote{ $\;{c_{88}}^k
= 1(\frac{1}{\sqrt{2}}), \, 2(-\frac{1}{2}), \, 3(-\frac{1}{2}),
\, 4(-\frac{1}{4}), \, 5(\frac{1}{2\sqrt{2}}), \, 6(\frac{1}{2\sqrt{2}}),
\, 7(-\frac{1}{4})$.} \\ %\colrule\colrule
\hline\hline
\end{tabular}
\label{tab_coe}
\end{table}
Now, it follows from Eq.~(\ref{rgeq}) that the RG equations for the
couplings $\eta$ (instead of $\lambda$):
\begin{eqnarray}
\dot{\eta}_1 &=& \eta_1 + \frac{1}{2} (\eta_1 \eta_4 + \eta_2 \eta_5
+ \eta_3 \eta_6 ) - \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_2 &=& \eta_2 + \frac{1}{2} ( 2 \eta_1 \eta_5
+ \eta_3 \eta_7 ) + \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_3 &=& \eta_3 + \frac{1}{2} ( 2 \eta_1 \eta_6
+ \eta_2 \eta_7 ) + \frac{1}{4} \eta_8^2 \;; \nonumber \\
\dot{\eta}_4 &=& 2 \eta_1^2 + \frac{1}{2} (\eta_5^2 + \eta_6^2)
+ \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_5 &=& 2 \eta_1 \eta_2 + \frac{1}{2} (\eta_4 \eta_5 +
\eta_6 \eta_7 ) - \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_6 &=& 2 \eta_1 \eta_3 + \frac{1}{2} (\eta_4 \eta_6 +
\eta_5 \eta_7 ) - \frac{1}{4} \eta_8^2 \;, \nonumber\\
\dot{\eta}_7 &=& 2 \eta_2 \eta_3 + \eta_5 \eta_6
+ \frac{1}{4} \eta_8^2 \;; \nonumber \\
\dot{\eta}_8 &=& \frac{3}{2} \eta_8 - \eta_1 \eta_8
+ \frac{1}{2} (\eta_2 + \eta_3) \eta_8 \nonumber \\
&& - \frac{1}{4} (\eta_5 + \eta_6) \eta_8
+ \frac{1}{8} (\eta_4 + \eta_7) \eta_8 \;. \label{rgm1}
\end{eqnarray}
The perturbative central charge (\ref{cnch}) for the present model in
terms of $\eta$ is
\begin{eqnarray}
C(\{\eta\}) &=& 2 - \frac{3}{4} \Big( 2\eta_1^2 + \eta_2^2 + \eta_3^2
+ \frac{3}{2} \eta_8^2 \Big) \nonumber \\
&& -\frac{3}{2} \Big(\eta_1 \eta_2 \eta_5 + \eta_1 \eta_3 \eta_6
+ \frac{1}{2} \eta_2 \eta_3 \eta_7 + \frac{1}{4} \eta_5 \eta_6 \eta_7 \Big)
\nonumber \\
&& -\frac{3}{16} (4\eta_1^2 + \eta_5^2 + \eta_6^2) \eta_4 \nonumber \\
&& + \frac{3}{8} \Big( 2 \eta_1 - \eta_2 - \eta_3 - \frac{1}{4} \eta_4
\nonumber \\
&& \qquad + \frac{1}{2} \eta_5 + \frac{1}{2} \eta_6
- \frac{1}{4} \eta_7 \Big) \eta_8^2 \;.
\end{eqnarray}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{RG equations for model (II)}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Comparing with the previous model, the only difference now is replacing
${\cal H}_h$ by ${\cal H}_\Delta$, i.e.,
\begin{eqnarray}
&& {\cal O}_8 = 2 \sigma_1 \sigma_2 \sigma_3 \sigma_0
= 2 \sin\sqrt{\pi} \Phi_+ \sin\sqrt{\pi} \Phi_- \;, \\
&& \lambda_8 \equiv \frac{\eta_8}{2} = \frac{\pi\Delta}{2v} \;,
\end{eqnarray}
and Eq.~(\ref{o8ope}) is modified to be
\begin{eqnarray}
&& \hspace{-2em} {\cal O}_8(z,\bar{z}) {\cal O}_8 \sim \frac{\alpha}{|z|}
+ \frac{1}{\sqrt{2}} {\cal O}_1 +\frac{1}{2} ( {\cal O}_2 + {\cal O}_3 )
\nonumber \\
&& - \bigg( \frac{\alpha}{|z|} \bigg)^{-1}
\bigg[ \frac{1}{2\sqrt{2}} ( {\cal O}_5 + {\cal O}_6 )
+ \frac{1}{4} ( {\cal O}_4 + {\cal O}_7 ) \bigg] \;,
\end{eqnarray}
yielding
\begin{equation}
\begin{array}{rl} {c_{88}}^k = & 1\big(\frac{1}{\sqrt{2}}\big)\,, \;
2\big(\frac{1}{2}\big)\,, \; 3\big(\frac{1}{2}\big)\,, \\
& 4\big(\!-\!\frac{1}{4}\big)\,, \; 5\big(\!-\!\frac{1}{2\sqrt{2}}\big)\,, \;
6\big(\!-\!\frac{1}{2\sqrt{2}}\big)\,, \; 7\big(\!-\!\frac{1}{4}\big)
\;. \end{array}
\end{equation}
However, as we have known, the symmetry of model (II) is higher than that of
model (I)---the O(3) symmetry in the triplet sector [i.e., the SU(2) symmetry
in the original lattice model] should remain. In other
words, the division of the triplet into doublet and singlet is futile in the
present case. This fact is reflected in the RG equations also. From
Eqs.~(\ref{rgm1}), by reversing the signs of the coefficients involving
${c_{88}}^2$, ${c_{88}}^3$, ${c_{88}}^5$, and ${c_{88}}^6$, three of the eight
equations are redundant, implying $\eta_1=\eta_2$, $\eta_4=\eta_5$, and
$\eta_6=\eta_7$. As a result, the RG equations degenerate to
\begin{eqnarray}
\dot{\eta}_1 &=& \eta_1 + \frac{1}{2} ( 2 \eta_1 \eta_4
+ \eta_3 \eta_6 ) - \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_3 &=& \eta_3 + \frac{3}{2} \eta_1 \eta_6
- \frac{1}{4} \eta_8^2 \;; \nonumber \\
\dot{\eta}_4 &=& 2 \eta_1^2 + \frac{1}{2} (\eta_4^2 + \eta_6^2)
+ \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_6 &=& 2 \eta_1 \eta_3 + \eta_4 \eta_6
+ \frac{1}{4} \eta_8^2 \;; \nonumber\\
\dot{\eta}_8 &=& \frac{3}{2} \eta_8 - \frac{1}{2} (3 \eta_1 + \eta_3) \eta_8
+ \frac{3}{8} (\eta_4 + \eta_6) \eta_8 \;, \label{rgm2}
\end{eqnarray}
and the central charge to
\begin{eqnarray}
C(\{\eta\}) &=& 2 - \frac{3}{4} \Big( 3\eta_1^2 + \eta_3^2
+ \frac{3}{2} \eta_8^2 \Big) \nonumber \\
&& -\frac{3}{2} \Big(\eta_1^2 \eta_4 + \frac{3}{2} \eta_1 \eta_3 \eta_6
+ \frac{1}{4} \eta_4 \eta_6^2 \Big)
\nonumber \\
&& -\frac{3}{16} (4\eta_1^2 + \eta_4^2 + \eta_6^2) \eta_4 \nonumber \\
&& + \frac{3}{8} \Big( 3 \eta_1 + \eta_3
- \frac{3}{4} \eta_4 - \frac{3}{4} \eta_6 \Big) \eta_8^2 \;.
\end{eqnarray}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Solutions to the RG Equations and the Emerging Phase Diagrams}
\label{sec_sol}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
To analyze the infrared behaviors of our present models, in this section,
we numerically integrate these two sets of RG Eqs.~(\ref{rgm1}) and (\ref{rgm2})
to depict generic phase diagrams for the spin ladder under a staggered field
or dimerization. On account of the fact that for a system in a massive regime,
the role of marginal coupling is usually exhausted by renormalizing the mass
and velocity, it is reasonable to take no thought of these marginal
perturbations at the outset and argue that the overall qualitative feature
remains correct although the position and shape of phase boundary may not be
exact. Let us begin with a known system---the generalized spin
ladder (\ref{ham_gen})\cite{nt97}---as an example. A variant of this model
includes four-spin ring exchange, which is believed to be relevant to some
exotic properties in real ladder materials. \cite{breh99,hn02,mvm02}
In the following we denote the bare ultraviolet couplings by
$\eta^{(0)} \equiv \eta(L=1)$ and the renormalized infrared couplings by
$\eta^{(\infty)} \equiv \eta(L=\infty)$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Phase diagram of the generalized spin ladder}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In this case, the RG equations are trivially $\dot{\eta}_1 = \eta_1$ and
$\dot{\eta}_3 = \eta_3$. Unless the initial value (bare coupling) vanishes,
$\eta_1^{(0)}=0$ or $\eta_3^{(0)}=0$, the system is always renormalized to
some strong-coupling massive phase as $L\to \infty$. Because of relations
in Eqs.~(\ref{cpct}), $\eta_1^{(0)} = \frac{m_t \alpha}{v}$ and $\eta_3^{(0)}
= \frac{m_s \alpha}{v}$, we conclude that $m_t =0$ and $m_s =0$ are phase
transition lines on which the model becomes critical. They belong to the
universality classes of critical SU(2)$_2$ WZNW model and Z$_2$ Ising model,
respectively.\cite{nt97} As a base for our further exploration, we draw the phase
diagram of the generalized spin ladder (\ref{ham_gen}) in Fig.~\ref{fig_pd_gl}.
\begin{figure}[ht]
\begin{picture}(0,200)
\leavevmode\centering\special{
psfile=pd_gl.eps
hoffset=-110
voffset=0
hscale=32
vscale=32}
\end{picture}
\caption{Phase diagram of the generalized spin ladder. There exist two
critical lines L:~SU(2)$_2$ and L:~Z$_2$, and four massive sheets
S:~H$^{(\pm)}$ and S:~sD$^{(\pm)}$. Notice the dimers in the Haldane phases
are only symbolic rather than a true order.}
\label{fig_pd_gl}
\end{figure}
According to Ref.~\onlinecite{nt97}, there are four massive phases divided by
two critical lines. The second and fourth quadrants of the figure are two
Haldane (H$^{(\pm)}$) spin liquid phases, and the first and third quadrants
belong to spontaneously dimerized (sD$^{(\pm)}$) phases. The ``$+/-$'' sectors
are actually related by duality of reversing the signs of the masses.
The axis of abscissas represents a critical
model with an SU(2)$_2$ WZNW universality class (L:SU(2)$_2$), and the axis of
ordinates a Z$_2$ (Ising) one (L:Z$_2$). Although the Haldane phases and the
dimer phases share the feature of non-vanishing topological string order
parameter in common and thermodynamically they are indistinguishable, the
symmetry of the ground state and the behavior of the dynamical susceptibility
are completely different.
(i) S:~H$^{(+)}$ phase [$\eta_1^{(\infty)}=+\infty$, $\eta_3^{(\infty)}=-\infty$]:
Neither parity (site or bond) symmetry nor translational symmetry is broken.
There is a coherent peak for the dynamical spin susceptibility at $q_\perp=\pi$,
$q\sim \pi$, $\omega \sim |m_t|$, characterizing coherent $S=1$ single-magnon
excitations.
(ii) S:~H$^{(-)}$ phase [$\eta_1^{(\infty)}=-\infty$, $\eta_3^{(\infty)}=+\infty$]:
No symmetry breaking (except for the hidden string order). The peak is at
$q_\perp=0$, $q\sim \pi$, $\omega \sim |m_t|$ instead.
(iii) S:~sD$^{(+)}$ phase [$\eta_1^{(\infty)}=+\infty$, $\eta_3^{(\infty)}=+\infty$]:
The discrete site-parity and translational symmetries are spontaneously
broken. The corresponding order parameter is $\langle (-1)^n
(\mbox{\boldmath $S$}_{1,n} \!\cdot\! \mbox{\boldmath $S$}_{1,n+1}
+ \mbox{\boldmath $S$}_{2,n} \!\cdot\! \mbox{\boldmath $S$}_{2,n+1}) \rangle
\ne 0$ (0-phase). No coherent peak in the spin susceptibility (square-root
singularity only).
(iv) S:~sD$^{(-)}$ phase [$\eta_1^{(\infty)}=-\infty$, $\eta_3^{(\infty)}=-\infty$]:
The discrete symmetries are broken with an order parameter $\langle (-1)^n
(\mbox{\boldmath $S$}_{1,n} \!\cdot\! \mbox{\boldmath $S$}_{1,n+1}
- \mbox{\boldmath $S$}_{2,n} \!\cdot\! \mbox{\boldmath $S$}_{2,n+1}) \rangle
\ne 0$ ($\pi$-phase). No coherent peak either.
We also show in the figure the directions of the lattice model parameters
$J_\perp$ and $V$ via relation (\ref{mrjv}), although the exact scales for the
coordinates are not known. We find that the standard ladder with positive,
antiferro- (negative, ferro-) $J_\perp$ is in Haldane H$^{(+)}$ (H$^{(-)}$)
phase. The pure four-spin coupling model with $J_\perp=0$ is in the sD$^{(-)}$
phase when $V>0$ or in the sD$^{(+)}$ phase when $V<0$. This is a special case
with a higher symmetry.
It is also interesting to check in RG sense the flow of the central charge.
For this simple model we have
\begin{equation}
C = 2 - \frac{3}{4} ( 3\eta_1^2 + \eta_3^2 ) \;.
\end{equation}
Considering this is a perturbative one-loop result and should not be taken
seriously when $\eta$ approaches unity, we assume that we can mimic the true
fixed point property by introducing a cut-off in $\eta$. If we are allowed
to choose this cut-off to be $\sqrt{\frac{2}{3}}$, the correct central charges
for both SU(2)$_2$ and Z$_2$ can be recovered, since for the former when
$(\eta_1^*, \eta_3^*)=(0, \sqrt{\frac{2}{3}})$,
$C^* = \frac{3}{2}$, and for the latter
when $(\eta_1^*, \eta_3^*)=(\sqrt{\frac{2}{3}},0)$,
$C^* = \frac{1}{2}$. The existence of the SU(2)$_2$ criticality was recently
confirmed by a direct estimate of the numerical value of the central
charge.\cite{hn02}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Phase diagram of model (I)}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Model (I) is the generalized spin ladder (\ref{ham_gen}) in a staggered field
(\ref{ham_h}). The RG equations regarding the relevant perturbations now become
\begin{eqnarray}
\dot{\eta}_1 &=& \eta_1 - \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_2 &=& \eta_2 + \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_3 &=& \eta_3 + \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_8 &=& \frac{3}{2} \eta_8 - \eta_1 \eta_8
+ \frac{1}{2} (\eta_2 + \eta_3) \eta_8 \;. \label{dfeq_m1}
\end{eqnarray}
Evidently, the equations are invariant under the transformation
$\eta_8 \to -\eta_8$, which corresponds to $h\to -h$ in the original
spin model. We need only deal with the positive case. At first glance
the system has two mathematical fixed points: (a) $(\eta_1^*, \eta_2^*,
\eta_3^*, \eta_8^*) = (0,0,0,0)$ and (b) $(\frac{3}{4}, -\frac{3}{4},
-\frac{3}{4}, \sqrt{3})$. The one at the origin is an unstable fixed
point recounting two decoupled chains as our starting point; while the
fixed point (b) is fallacious. This is due to the fact that $\eta_1$
and $\eta_2$ are not completely independent, i.e., initially $\eta_1^{(0)}
= \eta_2^{(0)} = \frac{m_t\alpha}{v}$, rendering the fixed point (b)
physically unaccessible. However, the values in (b) do have full play in the
the structure of the phase diagram.
\begin{figure}[ht]
\begin{picture}(0,130)
\leavevmode\centering\special{
psfile=pd_md1.eps
hoffset=-120
voffset=0
hscale=28
vscale=28}
\end{picture}
\caption{Phase diagram of the generalized spin ladder in a staggered field.
The massive phases occupying different blocks are separated by interfaces
either second-ordered (enclosed by solid line) or with the transition nature
unknown from the RG equations alone (dashed line).}
\label{fig_pd_md1}
\end{figure}
The emerging phase diagram of model (I) is shown in Fig.~\ref{fig_pd_md1}.
It is notable that all the interesting phases lie in the first three quadrants
of the base and the fourth quadrant is in some sense rather barren. This is
controlled by the ``fixed point'' (b) mentioned above. To make the structure
clearer, we anatomically plot figures in two intersecting planes (see
Fig.~\ref{fig_md1_sec}). Various phases are labeled in the figures, where
``B'' denotes a block or bulk space that is always massive, ``S'' stands for
a surface or sheet as phase boundary separating two different phases, which
can be critical or first-ordered, and ``L'' is an intersectional line of two
sheets.
\begin{figure}[ht]
\begin{picture}(0,260)
\leavevmode\centering\special{
psfile=pd_md1_sec_r.eps
hoffset=-120
voffset=0
hscale=38
vscale=38}
\end{picture}
\caption{Sectional plot of the phase diagram of the generalized spin
ladder in a staggered field: (a) $\eta_3^{(0)}$ is a constant as $-0.2$.
(b) $\eta_1^{(0)} = \eta_2^{(0)}$ is a constant as $0.2$.}
\label{fig_md1_sec}
\end{figure}
Now let us interpret in some detail the obtained phase diagram. We take the
example with a parameter range where two consecutive phase transitions occur
when increasing the staggered field---one belongs to the U(1) universality
class; the other's property is not determined by the RG theory itself
(probably first-ordered). Suppose a generalized spin ladder offers such a
kind of parameters:
\begin{equation}
\begin{array}{l}
m_t \sim \eta_1^{(0)} = \eta_2^{(0)} = 0.01 \;, \\
m_s \sim \eta_3^{(0)} = -0.2 \;,
\end{array}
\end{equation}
where $\eta^{(0)}$'s are bare-couplings as the initial values of the
differential equations (\ref{dfeq_m1}): $\eta^{(0)} \equiv \eta(L=1)$.
Now the system is subjected to a staggered field. It is found when the
field is increased to
\begin{equation}
h_{c1} \sim (\eta_8^{(0)})_{c1} = 0.0354863 \;,
\end{equation}
the system undergoes a second-order phase transition with a U(1) universality
class (or Gaussian fixed point). The behaviors of the RG flows in the vicinity
of this transition is shown in Figs.~\ref{fig_md1_u1foflow} (a), (b), and (c).
\begin{figure}[ht]
\begin{picture}(0,270)
\leavevmode\centering\special{
psfile=u1foflow.eps
hoffset=-130
voffset=-25
hscale=46
vscale=46}
\end{picture}
\caption{Parametric plot of the RG flows across two phase transitions. The
figures in the left column describe the flows in the vicinity of the U(1)
transition, where the solid lines correspond to $\eta_8^{(0)}
= (\eta_8^{(0)})_{c1}$, the dashed lines to $\eta_8^{(0)}= (\eta_8^{(0)})_{c1}
-0.001$, and the dotted lines to $\eta_8^{(0)}= (\eta_8^{(0)})_{c1} +0.001$.
The figures in the right column are in the vicinity of the second
(unidentified) transition, where the solid lines correspond to $\eta_8^{(0)}
= (\eta_8^{(0)})_{c2}$, the dotted lines to $\eta_8^{(0)}= (\eta_8^{(0)})_{c2}
-0.001$, and the dash-dotted lines to $\eta_8^{(0)}= (\eta_8^{(0)})_{c2} +0.001$.
Notice in panel (f) the dash-dotted line for $\eta_3(L)$ eventually
approaches $+\infty$ after an upturn that is not shown in the figure.}
\label{fig_md1_u1foflow}
\end{figure}
Below the U(1) transition the system is in the bulk massive phase B:~H$^{(+)}$
characterized by the couplings renormalized to $(\eta_1^{(\infty)},
\eta_2^{(\infty)}, \eta_3^{(\infty)}, \eta_8^{(\infty)}) = (+\infty, +\infty,
-\infty, 0)$ (the dashed lines in the figures). At the transition, the
infrared property becomes $(\eta_1^{(\infty)}, \eta_2^{(\infty)},
\eta_3^{(\infty)}, \eta_8^{(\infty)}) = (0, +\infty, -\infty, 0)$
(the solid lines in the figures). Since $\eta_1$ is proportional to the split
doublet mass, $m_t^d$, it is bound to attribute this transition to a U(1)
character. Now we are in the bulk phase of B:~I characterized by
$(-\infty, +\infty, -\infty, 0)$ (the dotted lines in the figures). This is
an intermediate phase interpolating between the Haldane phase and the
spontaneously dimered phase. The most distinctive features of this phase are
that it has only transverse coherent spin excitations and longitudinal string
order parameter. (For detailed properties, cf. Ref.~\onlinecite{wefn02}.)
If the staggered field is further increased to reach the point
\begin{equation}
h_{c2} \sim (\eta_8^{(0)})_{c2} = 0.0560487 \;,
\end{equation}
the second transition occurs. The RG flows around this transition is shown in
Figs.~\ref{fig_md1_u1foflow} (d), (e), and (f). At the transition the
couplings are renormalized to $(-\infty, +\infty, -\infty, \sqrt{3})$
(the solid lines in the figures). After the transition the system enters the
B:~H$^{(-)}$ phase characterized by $(-\infty, +\infty, +\infty, \infty)$
(the dash-dotted lines in the figures).
Unfortunately, we can tell very little from the present scheme regarding the
property of this transition. It may be a first-order transition or even a
crossover.
In this way, we are able to classify most of the phases and transitions by
the infrared properties inferred from the RG equations. We summarize the main
results as follows.
\paragraph*{a. Bulks}:
\begin{equation}
\begin{tabular}{l|cccc}
\hline \hline
& $\eta_1^{(\infty)}$ & $\eta_2^{(\infty)}$ & $\eta_3^{(\infty)}$
& $\eta_8^{(\infty)}$ \\ \hline
B:sD$^{(+)}$ & $+\infty$ & $+\infty$ & $+\infty$ & $0$ \\
B:sD$^{(-)}$ & $-\infty$ & $-\infty$ & $-\infty$ & $0$ \\
B:H$^{(+)}$ & $+\infty$ & $+\infty$ & $-\infty$ & $0$ \\
B:H$^{(-)}$ & $-\infty$ & $+\infty$ & $+\infty$ & $\infty$ \\
B:I & $-\infty$ & $+\infty$ & $-\infty$ & $0$ \\ \hline \hline
\end{tabular}
\label{m1blk}
\end{equation}
Comparing with the zero-field case (previous subsection), we find that
three massive phases sD$^{(\pm)}$ and H$^{(+)}$ are essentially passed
from their $h=0$ counterparts directly except for the broken symmetry $\eta_1
\ne \eta_2$. However, B:~H$^{(-)}$ phase is different from S:~H$^{(-)}$ in the
sense that $\eta_2^{(\infty)} \to +\infty$ instead of $-\infty$ and
$\eta_8^{(\infty)} \to \infty$ instead of $0$. Obviously, B:~I is a totally
new phase which has no zero-field counterpart.
\paragraph*{b. Sheets}:
\begin{equation}
\begin{tabular}{l|cccc}
\hline \hline
& $\eta_1^{(\infty)}$ & $\eta_2^{(\infty)}$ & $\eta_3^{(\infty)}$
& $\eta_8^{(\infty)}$ \\ \hline
S:U(1) [H$^{(+)}$-I] & $0$ & $+\infty$ & $-\infty$ & $0$ \\
S:Z$_2$ [sD$^{(-)}$-I] & $-\infty$ & $0$ & $-\infty$ & $0$ \\
S:Z$_2$ [H$^{(+)}$-sD$^{(+)}$] & $+\infty$ & $+\infty$ & $0$ & $0$ \\
S:Z$_2$ [sD$^{(-)}$-H$^{(-)}$] & $-\infty$ & $-\infty$ & $-\infty$ & $\sqrt{3}$ \\
S: [I-H$^{(-)}$] & $-\infty$ & $+\infty$ & $-\infty$ & $\sqrt{3}$ \\
S: [H$^{(+)}$-H$^{(-)}$] & $+\infty$ & $+\infty$ & $-\infty$ & $\sqrt{3}$ \\
S: [sD$^{(+)}$-H$^{(-)}$] & $+\infty$ & $+\infty$ & $+\infty$ & $\sqrt{3}$
\\ \hline \hline
\end{tabular}
\end{equation}
It is perspicuous that the critical surfaces split from the SU(2)$_2$ critical
line and having U(1) and Z$_2$ criticalities correspond to the infrared
fixed points with $(\eta_1, \eta_8) \propto (m_t^d, h) \to (0,0)$ and
$(\eta_2, \eta_8) \propto (m_t^s, h) \to (0,0)$, respectively. The other Z$_2$
criticality is trivially same as and connected to the one when $h=0$,
$(\eta_3, \eta_8) \propto (m_s, h) \to (0,0)$. At least one of the four
unknown boundaries (dashed lines in Fig.~\ref{fig_pd_md1} and
\ref{fig_md1_sec}) can be identified as Z$_2$, i.e., S: [sD$^{(-)}$-H$^{(-)}$].
The reason is to be shown soon.
\paragraph*{c. Lines}:
\begin{equation}
\begin{tabular}{l|cccc}
\hline \hline
& $\eta_1^{(\infty)}$ & $\eta_2^{(\infty)}$ & $\eta_3^{(\infty)}$
& $\eta_8^{(\infty)}$ \\ \hline
L:U(1) [H$^{(+)}$-I-H$^{(-)}$] & $+\frac{3}{4}$ & $+\infty$ & $-\infty$ & $\sqrt{3}$ \\
L:Z$_2$ [sD$^{(-)}$-I-H$^{(-)}$] & $-\infty$ & $-\frac{3}{4}$ & $-\infty$ & $\sqrt{3}$ \\
L:Z$_2$ [H$^{(+)}$-sD$^{(+)}$-H$^{(-)}$] & $+\infty$ & $+\infty$ &$-\frac{3}{4}$ & $\sqrt{3}$ \\
\hline \hline
\end{tabular}
\end{equation}
These lines are intersections of two surfaces. The special values here are
associated with the ``fixed point'' (b) of Eqs.~(\ref{dfeq_m1}).
\begin{figure}[ht]
\begin{picture}(0,170)
\leavevmode\centering\special{
psfile=pd_md1_bas_r.eps
hoffset=-90
voffset=0
hscale=25
vscale=25}
\end{picture}
\caption{A top-view plot of the phase diagram of the generalized spin ladder
in a staggered field. The true phase boundaries in the base ($h=0$) should be
extended as shown by the arrows.}
\label{fig_md1_bas}
\end{figure}
To close our discussion, we show in Fig.~\ref{fig_md1_bas} a top-view of the
phase diagram. The U(1) zone is confined in a wedged area demarcated by the
SU(2)$_2$ line $\eta_1^{(0)} = \eta_2^{(0)} =0$ and $\eta_1^{(0)} =
\eta_2^{(0)} = -\frac{1}{3} \eta_3^{(0)}>0$,
i.e., $m_t= -\frac{1}{3} m_s>0$. The latter condition amounts to $V=0$ and
$J_\perp >0$ via relation (\ref{mrjv}), coinciding with the location of a
standard spin ladder with antiferromagnetic interchain coupling. The adjacent
Z$_2$ region is delimited by condition $\eta_1^{(0)} = \eta_2^{(0)} =
\frac{1}{5} \eta_3^{(0)}<0$. The trivial Z$_2$ region is within the Z$_2$ line
$\eta_3^{(0)} = 0$ and $\eta_1^{(0)} = \eta_2^{(0)} =
- 3 \eta_3^{(0)}>0$. The unidentified sheet have $\eta_1^{(0)} = \eta_2^{(0)}
= \eta_3^{(0)}$ as its boundary in the base. This is a model with $J_\perp=0$,
i.e., only four-spin interchain coupling. However, one might have noticed that
there is something unnatural here. Considering the internal relation between
the B:~H$^{(-)}$ phase in the bulk and the S:~H$^{(-)}$ phase in the base, we
suggest in the true situation the boundary of this sheet should bend over to
connect to the critical lines in the base (as shown by the arrows in
Fig.~\ref{fig_md1_bas}). By combining the ``integrating out'' method and the
study of the bosonized model, we observe that a Z$_2$ surface, connecting two
axes, covers the whole third quadrant in Fig.~\ref{fig_md1_bas}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\subsection{Phase diagram of model (II)}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Now we turn to Model (II), the generalized spin ladder (\ref{ham_gen}) under
dimerization (\ref{ham_dlt}). It follows from Eqs.~(\ref{rgm2}) by retaining
only the relevant couplings:
\begin{eqnarray}
\dot{\eta}_1 &=& \eta_1 - \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_3 &=& \eta_3 - \frac{1}{4} \eta_8^2 \;, \nonumber \\
\dot{\eta}_8 &=& \frac{3}{2} \eta_8 - \frac{1}{2} (3\eta_1 + \eta_3) \eta_8 \;.
\label{dfeq_m2}
\end{eqnarray}
This is a model with an unbroken O(3) symmetry. As before, we have an invariance
$\eta_8 \to -\eta_8$ (i.e., $\Delta \to -\Delta$). In contrast with the previous
model, now the system has really a fixed point $(\eta_1^*, \eta_3^*, \eta_8^*)
= (\frac{3}{4}, \frac{3}{4}, \sqrt{3})$ besides the trivial one at
the origin. This fixed point occurs at $\eta_1 =\eta_3$, or equivalently
$m_t =m_s$, corresponding to the pure four-spin coupling model ($J_\perp =0$)
when the system has a higher SU(2)$\times$SU(2)$\approx$O(4) symmetry. This is
also the condition for the line where the SU(2)$_1$ sheet merges the Z$_2$
sheet into the SU(2)$_2$ sheet. The RG flows around this fix point are shown in
\begin{figure}[ht]
\begin{picture}(0,145)
\leavevmode\centering\special{
psfile=md2_sfl.eps
hoffset=-115
voffset=-15
hscale=40
vscale=40}
\end{picture}
\caption{RG flows to the fixed point $(\eta_1^*, \eta_3^*, \eta_8^*)
= (\frac{3}{4}, \frac{3}{4}, \sqrt{3})$ (solid line) at $\eta_1^{(0)}
= \eta_3^{(0)}=0.1$ and $(\eta_8^{(0)})_c = 0.137410$, and below
($\eta_8^{(0)} = (\eta_8^{(0)})_c -0.001$, dashed line) and above
($\eta_8^{(0)} = (\eta_8^{(0)})_c +0.001$, dotted line) the transition.}
\label{fig_md2_sfl}
\end{figure}
Fig.~\ref{fig_md2_sfl}, where we choose the initial values for the base quantities:
$\eta_1^{(0)} = \eta_3^{(0)}=0.1$. The initial value of the dimerization for
the fixed point is $(\eta_8^{(0)})_c = 0.137410$.
\begin{figure}[ht]
\begin{picture}(0,145)
\leavevmode\centering\special{
psfile=pd_md2.eps
hoffset=-122
voffset=0
hscale=28
vscale=28}
\end{picture}
\caption{Phase diagram of the generalized spin ladder under $\pi$-phase
dimerization.}
\label{fig_pd_md2}
\end{figure}
The phase diagram of this model is plotted in Fig.~\ref{fig_pd_md2}, and
Fig.~\ref{fig_md2_sec} is for the sectional pictures of the same phase
diagram, where we have labeled various phases.
\begin{figure}[ht]
\begin{picture}(0,260)
\leavevmode\centering\special{
psfile=pd_md2_sec_r.eps
hoffset=-120
voffset=0
hscale=38
vscale=38}
\end{picture}
\caption{Sectional plot of the phase diagram of the generalized spin
ladder under dimerization.}
\label{fig_md2_sec}
\end{figure}
As before, the classification of phases and criticalities is associated with
the infrared properties of the system resulted from the RG equations.
\paragraph*{a. Bulks}:
\begin{equation}
\begin{tabular}{l|ccc}
\hline \hline
& $\eta_1^{(\infty)}$ & $\eta_3^{(\infty)}$
& $\eta_8^{(\infty)}$ \\ \hline
B:H$^{(+)}$ & $+\infty$ & $-\infty$ & $0$ \\
B:H$^{(-)}$ & $-\infty$ & $+\infty$ & $0$ \\
B:sD$^{(+)}$ & $+\infty$ & $+\infty$ & $0$ \\
B:D$^{(-)}$ & $-\infty$ & $-\infty$ & $\infty$ \\ \hline \hline
\end{tabular}
\end{equation}
Three massive phases H$^{(\pm)}$ and sD$^{(+)}$ are same as their $\Delta=0$
counterparts. Since the symmetry of phase sD$^{(+)}$ is different from the
explicit dimerization perturbation (\ref{ham_dlt}), it is still spontaneously
dimerized. However, B:~D$^{(-)}$ phase is different from S:~sD$^{(-)}$, because
now the system is explicitly dimerized with the same symmetry of the perturbed
Hamiltonian. In the infrared limit $\eta_8^{(\infty)} \to \infty$ instead of $0$.
It is also noticed from the figures that phase B:~H$^{(-)}$ is restricted to a
rather small region in the phase space. This may not be the real situation (see
discussions at the end of this section).
\paragraph*{b. Sheets}:
\begin{equation}
\begin{tabular}{l|ccc}
\hline \hline
& $\eta_1^{(\infty)}$ & $\eta_3^{(\infty)}$ & $\eta_8^{(\infty)}$ \\ \hline
S:SU(2)$_2$ [sD$^{(+)}$-H$^{(-)}$] & $0$ & $+\infty$ & $0$ \\
S:Z$_2$ [sD$^{(+)}$-H$^{(+)}$] & $+\infty$ & $0$ & $0$ \\
S:SU(2)$_1$ [H$^{(+)}$-D$^{(-)}$] & $+\infty$ & $-\infty$ & $\sqrt{3}$ \\
S: [H$^{(-)}$-D$^{(-)}$] & $-\infty$ & $+\infty$ & $\sqrt{3}$ \\ \hline \hline
\end{tabular}
\end{equation}
Like the case before, the identification of the SU(2)$_2$ and Z$_2$
criticalities is directly due to the fact that in addition to the dimerization
($\eta_8^{(\infty)}$), the triplet ($\eta_1^{(\infty)}$) and singlet
($\eta_3^{(\infty)}$) masses are renormalized to vanish, respectively.
However, the reference of the interface between phases H$^{(+)}$ and D$^{(-)}$
to an SU(2)$_1$ criticality is a bit subtle. This conclusion is drawn from
other studies of the standard ladder under
dimerization.\cite{tnhs95,ms9698,cg99,wn00} In Ref.~\onlinecite{tnhs95}, a
standard spin-$\frac{1}{2}$ ladder with ferromagnetic interchain coupling
under $0$-phase dimerization (the dual case of our present model) was
semiclassically mapped onto an O(3) nonlinear sigma model with some
topological angle. The SU(2)$_1$ criticality is realized when this angle
equals $\pi$. On the other hand, an effective Hamiltonian was explicitly
derived in Ref.~\onlinecite{wn00} which describes the same transition in terms
of a single spin-$\frac{1}{2}$ chain with bond alternation, and the
transmutation of all physical quantities from the ultraviolet fixed point to
the infrared fixed point was established. We have only one unidentified
surface, S: [H$^{(-)}$-D$^{(-)}$], in this model.
\paragraph*{c. Line}:
\begin{equation}
\begin{tabular}{l|ccc}
\hline \hline
& $\eta_1^{(\infty)}$ & $\eta_3^{(\infty)}$ & $\eta_8^{(\infty)}$ \\ \hline
L:SU(2)$_2$ [sD$^{(+)}$-H$^{(+)}$-D$^{(-)}$-H$^{(-)}$] & $\frac{3}{4}$
& $\frac{3}{4}$ & $\sqrt{3}$ \\ \hline \hline
\end{tabular}
\end{equation}
The only critical line in this model represents the fixed point of
Eqs.~(\ref{dfeq_m2}), the RG flow for the point on which is shown in
Fig.~\ref{fig_md2_sfl}. This is a very special line with a higher O(4)
symmetry: it is not only the ridge line where all sheets meet, but also
the common boundary shared by all four phases (blocks) of the model.
\begin{figure}[ht]
\begin{picture}(0,170)
\leavevmode\centering\special{
psfile=pd_md2_bas.eps
hoffset=-95
voffset=0
hscale=25
vscale=25}
\end{picture}
\caption{Top-view plot of the phase diagram of the generalized spin ladder
under dimerization. The true SU(2)$_1$ sheet should connect to the SU(2)$_2$
line in the base, and the unidentified sheet to the Z$_2$ line in the base
(as shown by the arrows).}
\label{fig_md2_bas}
\end{figure}
Since the obtained phase diagram is grounded on the relevant part of the
one-loop RG equations, the position of the phase boundaries may not be so
correct. For instance, considering the intimacy between the bulk B:~D$^{(-)}$
phase and the S:~sD$^{(-)}$ phase in the base, most plausibly the SU(2)$_1$
sheet is connected to the SU(2)$_2$ line of $\Delta =0$, and the boundary for
the unidentified sheet is pushed to connect to the Z$_2$ line in the base (see
Fig.~\ref{fig_md2_bas}). So the space for the bulk B:~H$^{(-)}$ phase is
considerably enlarged. This is supported by the fact that, as we have
mentioned, the ($J_\perp >0$) standard ladder ($V=0$) undergoes the SU(2)$_1$
transition under the ($\pi$-phase) dimerization. So the SU(2)$_1$ sheet cannot
terminate at the location which is a result from the approximate RG
Eqs.~(\ref{dfeq_m2}) as shown in Fig.~\ref{fig_pd_md2}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Summary and Discussions} \label{sec_sum}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Based on the one-loop RG equations and the assumption that the relevant
perturbations govern the relative structure of various phases in the
parameter space, we have speculated from the infrared properties of these
equations the phase diagrams of two related spin ladder models in the
weak-coupling regime.
In model (I), the external staggered field breaks the SU(2) symmetry down to
U(1) and Z$_2$, so that the criticalities at most have the same symmetries
and no room for SU(2) to live. Indeed, we find both U(1) and Z$_2$
criticalities come forth with the introduction of the staggered field (see
Fig.~\ref{fig_pd_md1} and \ref{fig_md1_bas}), which actually evolve from the
underlying SU(2)$_2$ criticality in absence of the field. The present result
confirms the picture we obtained through a somewhat different approach in
Ref.~\onlinecite{wefn02}. In that study, we literally integrated out the fast
degrees of freedom associated with the singlet modes close to the SU(2)$_2$
line and derived an effective action which describes the triplet sector with
anisotropy induced by the staggered field. The vanishing of the split
effective triplet mass manifests the criticality accordingly.
The new observation is that the intermediate phase (B:~I), which interpolates
between the Haldane phase (B:~H$^{(+)}$) and the dimer phase (B:~sD$^{(-)}$),
is restricted to a very limited region, and there should exist another
(unidentified) transition which separates B:~I phase from the large-$h$
(Haldane) phase (B:~H$^{(-)}$) which occupies most of the space in the phase
diagram. From the infrared property (\ref{m1blk}), we see that B:~I phase is
quite unique entirely due to the staggered field, and there is no resemblance
preexisting when $h=0$.
Comparing with model (I), model (II) is relatively simple in the sense that
the SU(2) symmetry is unbroken and the O(3) symmetry in the triplet sector
remains, and the number of RG equations is thus reduced. All four bulk phases
can basically be traced back to their counterparts when $\Delta =0$. None the
less, as we have seen, the phase diagram of this model is by no means
uneventful (see Fig.~\ref{fig_pd_md2} and \ref{fig_md2_bas}). The phenomenon
extremely intriguing is the merging of SU(2)$_1$ and Z$_2$ critical sheets
into SU(2)$_2$ sheet. The information encoded in the RG equations reveals that
this occurs at some critical dimerization strength $\Delta_c(V)$ when the
system has a higher O(4) symmetry or SU(2)$\times$SU(2) in the original
lattice model, i.e., $J_\perp =0$. The merging line with SU(2)$_2$ criticality
is shared by all four massive phases of the model.
The benevolent aspect of our present theory is that we are able to assort
different phases according to their infrared properties from the RG flows,
and the identification of the critical sheets that emerge from the underlying
critical lines in the base should be reliable. These critical sheets,
including two Z$_2$ and U(1) in model (I) and SU(2)$_2$ and Z$_2$ in model
(II), can also be verified to exist\cite{nwup} by means of ``integrating out''
procedure employed in Ref.~\onlinecite{wefn02}, although the terminating
positions of these sheets are unknown by this approach. However, as one may
have perceived, the insolvency of the present theory is that not all of the
transitions can be attributed. Even worse is the phase boundaries resolved
from the one-loop equations may not be accurate. (See our suggestions for
the modification of the phase diagrams in Fig.~\ref{fig_md1_bas} and
\ref{fig_md2_bas}.) Therefore, although the topology of the phase diagrams
we believe has been overall and qualitatively captured, the unwavering
pin-pointing of the phase diagrams has to resort to other, say numerical,
methods. For example, it is predicted here for model (II) of pure four-spin
interaction ($V$), when increasing the dimerization, there is only one
second-order (SU(2)$_2$) transition which connects directly two dimerized
phases with mutually mismatched orders. If adding a bit positive
$J_\perp$ coupling, one would expect two consecutive transitions: Z$_2$ and
SU(2)$_1$ in a row.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\acknowledgments
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
It is a pleasure to acknowledge the helpful discussions with P.\ Fulde,
K.\ Penc, R.\ Narayanan, T.\ Vekua, and M.\ Nakamura. Special thanks are due
to A.\ Nersesyan for calling the problem to my attention. This work was
supported under the visitors program of MPI-PKS.
%\begin{references}
\begin{thebibliography}{99}
\bibitem{gins88} See, for a concise review on conformal field theory,
P.\ Ginsparg, in {\it Fields, Strings and Critical Phenomena}, ed.\ by
E.\ Br\'{e}zin and J.\ Zinn-Justin (Elsevier Science, Amsterdam, 1989).
\bibitem{aff88} See, for a review on spin chains, I.\ Affleck, in
Ref.~\onlinecite{gins88}; J.\ Phys.: Condens.\ Matter {\bf 1}, 3047 (1989).
\bibitem{gnt99} See, for a recent book on bosonization, A.O.\ Gogolin,
A.A.\ Nersesyan, and A.M.\ Tsvelik, {\it Bosonization and Strongly Correlated
Systems} (Cambridge University Press, Cambridge, England, 1999).
\bibitem{cont01} See, e.g., M.A.\ Continentino, {\it Quantum Scaling in Many
Body Systems} (World Scientific, Singapore, 2001).
\bibitem{dm98} G.\ Delfino and G.\ Mussardo, Nucl.\ Phys.\ B {\bf 516}, 675
(1998).
\bibitem{fgn00} M.\ Fabrizio, A.O.\ Gogolin, and A.A.\ Nersesyan,
Phys.\ Rev.\ Lett.\ {\bf 83}, 2014 (1999); Nucl.\ Phys.\ B {\bf 580}, 647
(2000).
\bibitem{wn00} Y.-J.\ Wang and A.A.\ Nersesyan, Nucl.\ Phys.\ B {\bf 583}, 671
(2000).
\bibitem{wefn02} Y.-J.\ Wang, F.H.L.\ Essler, M.\ Fabrizio, and
A.A.\ Nersesyan, Phys.\ Rev.\ B {\bf 66}, 024412 (2002).
\bibitem{nt97} A.A.\ Nersesyan and A.M.\ Tsvelik, Phys.\ Rev.\ Lett.\ {\bf 78},
3939 (1997).
\bibitem{km98} A.K.\ Kolezhuk and H.-J.\ Mikeska, Phys.\ Rev.\ Lett.\ {\bf 80},
2709 (1998).
\bibitem{snt96} D.G.\ Shelton, A.A.\ Nersesyan, and A.M.\ Tsvelik,
Phys.\ Rev.\ B {\bf 53}, 8521 (1996).
\bibitem{note1} Rigorously speaking, the interchain coupling $J_\perp$ shifts
the triplet and singlet velocities differently. However, we neglect this
difference under the condition (\ref{wccnd}).
\bibitem{cardy88} See, e.g., J.L.\ Cardy, in Ref.~\onlinecite{gins88};
A.W.W.\ Ludwig and J.L.\ Cardy, Nucl.\ Phys.\ B {\bf 285}, 687 (1987).
\bibitem{zam86} A.B.\ Zamolodchikov, JETP Lett.\ {\bf 43}, 730 (1986)
[Pis'ma Zh.\ Eksp.\ Teor.\ Fiz.\ {\bf 43}, 565 (1986)].
\bibitem{note2} The definitions and conventions we take may be found, e.g.,
in Appendix A of Ref.~\onlinecite{wefn02}.
\bibitem{breh99} S.\ Brehmer, H.-J.\ Mikeska, M.\ M\"uller, N.\ Nagaosa,
and S.\ Uchida, Phys.\ Rev.\ B {\bf 60}, 329 (1999);
M.\ Matsuda, K.\ Katsumata, R.S.\ Eccleston, S.\ Brehmer, and H.-J.\ Mikeska,
{\it ibid}. {\bf 62}, 8903 (2000);
T.S.\ Nunner, P.\ Brune, T.\ Kopp, M.\ Windt, and M.\ Gr\"uninger,
{\it ibid}. {\bf 66}, 180404(R) (2002);
\bibitem{hn02} K.\ Hijii and K.\ Nomura, Phys.\ Rev.\ B {\bf 65}, 104413 (2002).
\bibitem{mvm02} M.\ M\"{u}ller, T.\ Vekua, H.-J.\ Mikeska,
Phys.\ Rev.\ B {\bf 66}, 134423 (2002);
T.\ Hikihara, T.\ Momoi, and X.\ Hu, Phys.\ Rev.\ Lett.\ {\bf 90},
087204 (2003);
A.\ L\"{a}uchli, G.\ Schmid, and M.\ Troyer, Phys.\ Rev.\ B {\bf 67},
100409(R) (2003).
\bibitem{tnhs95} K.\ Totsuka and M.\ Suzuki, J.\ Phys.: Condens.\ Matter {\bf 7},
6079 (1995).
\bibitem{ms9698} M.A.\ Martin-Delgado, R.\ Shankar, and G.\ Sierra,
Phys.\ Rev.\ Lett.\ {\bf 77}, 3443 (1996);
M.A.\ Martin-Delgado, J.\ Dukelsky, and G.\ Sierra, Phys.\ Lett.\ {\bf 250A},
430 (1998).
\bibitem{cg99} D.C.\ Cabra and M.D.\ Grynberg, Phys.\ Rev.\ Lett.\ {\bf 82},
1768 (1999).
\bibitem{nwup} A.A.\ Nersesyan and Y.-J.\ Wang, unpublished.
%\end{references}
\end{thebibliography}
\end{document}