cond-mat0512191/ta50.tex
1: \documentclass[aps,prb,showpacs,twocolumn]{revtex4}
2: \usepackage{graphicx}
3: 
4: \newcommand{\be}{\begin{eqnarray}}
5: \newcommand{\ee}{\end{eqnarray}}
6: \newcommand{\bmat}{\left(\begin{array}}
7: \newcommand{\emat}{\end{array}\right)}
8: \newcommand{\no}{\nonumber}
9: \newcommand{\e}{\epsilon}
10: 
11: \begin{document}
12: \title{Mesoscopic conductance fluctuations in a coupled quantum dot system}
13: \author{Kazutaka Takahashi}
14: \affiliation{Department of Physics, Tokyo Institute of Technology,  
15:  Tokyo 152--8551, Japan}
16: \affiliation{Condensed Matter Theory Laboratory, RIKEN, 
17:  Saitama 351--0198, Japan}
18: \author{Tomosuke Aono}
19: \affiliation{Department of Physics, Ben--Gurion University of the Negev,
20:  Beer--Sheva 84105, Israel}
21: \date{\today}
22: 
23: \begin{abstract}
24:  We study the transport properties of 
25:  an Aharonov--Bohm ring containing two quantum dots.
26:  One of the dots has well--separated resonant levels, while
27:  the other is chaotic and is treated by random matrix theory.
28:  We find that the conductance through the ring 
29:  is significantly affected by mesoscopic fluctuations.
30:  The Breit--Wigner resonant peak is changed to 
31:  an antiresonance by increasing the ratio of the level broadening 
32:  to the mean level spacing of the random dot.
33:  The asymmetric Fano form turns into a symmetric one 
34:  and the resonant peak can be controlled by magnetic flux.
35:  The conductance distribution function clearly shows 
36:  the influence of strong fluctuations.
37: \end{abstract}
38: \pacs{
39: 73.21.La,  % Quantum dots
40: 73.23.-b,  % Electronic transport in mesoscopic systems
41: 05.60.Gg,  % Quantum transport
42: 05.45.Gg   % Control of chaos, applications of chaos
43: }
44: \maketitle
45: 
46: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47: %\section{Introduction}
48: 
49:  The quantum dot~\cite{Beenakker, Alhassid} (QD) is an ideal system to 
50:  study the phase coherence of quantum mechanical wave functions.
51:  Such effects can be explained by the interference of different pathways, 
52:  induced by elastic scattering from an irregular boundary and/or
53:  impurities. 
54:  When a dot is connected to leads, there is a strong overlap between 
55:  the dot and leads and we must treat the whole as a quantum system.
56:  In fact, a recent numerical calculation of a chaotic dot 
57:  shows that there are scars that connect between the leads.~\cite{SI}
58: 
59:  The coexistence of a direct path and discrete levels 
60:  in the dot induces a prominent effect, the Fano effect.~\cite{Fano}
61:  It was shown in QD systems~\cite{KAKI, JMHG} that asymmetric Fano peaks 
62:  can be controlled by gate voltages and magnetic fields.
63:  The work in Ref.~\onlinecite{KAKI} demonstrated that  an Aharonov-Bohm 
64:  (AB) ring is a suitable system to study this effect.
65:  A dot is embedded in one of the arms, 
66:  and the ring geometry is utilized as an interferometer.
67:  This effect has also been observed in other experiments, including
68:  a microwave cavity~\cite{RLBKS} and optical absorption.~\cite{FCSWP}
69: 
70:  The situation changes drastically 
71:  if we take into account mesoscopic fluctuations.
72:  A random distribution of levels in the dot leads to 
73:  sample--to--sample fluctuations of the conductance.
74:  In contrast to bulk systems, such fluctuations cannot be neglected 
75:  in QD systems, and the conductance distribution has 
76:  a broad non--Gaussian shape.~\cite{PEI}
77:  It was observed in a QD system~\cite{HPMBDH} 
78:  and in a chaotic microwave cavity.~\cite{HZOAA}
79:  Clerk {\it et al.} developed a statistical theory 
80:  of Fano peaks assuming a random distribution of peaks.~\cite{CWB}
81:  These authors focused on resonance--to--resonance fluctuations of 
82:  the asymmetric Fano form, rather than conductance fluctuations.
83: 
84:  In this paper we develop a statistical theory 
85:  of the AB ring with two QDs in the arms.
86:  In addition to a resonant dot in one of the arms, 
87:  a random dot is connected to the other arm 
88:  and is treated by random matrix theory (RMT).~\cite{Mehta} 
89:  The orbit through the resonance is correlated to 
90:  those through the random levels, 
91:  and strong fluctuations can be observed in the conductance.
92:  We show that the Breit--Wigner and Fano resonant forms 
93:  are no longer maintained in the averaged conductance.
94: 
95:  We focus on the following two issues.
96:  First, we examine the properties of those orbits that 
97:  contribute to the conductance.
98:  Motivated by the work of Ref.~\onlinecite{CWB}, we take into account 
99:  the direct nonresonant path in the random dot as well as 
100:  the Breit--Wigner resonant path in the regular dot.
101:  We show that these two channels give qualitatively different 
102:  contributions to the conductance in the presence of random levels.
103:  Second, the level broadening of the random dot.
104:  For an open dot, the broadening can be larger than 
105:  the mean level spacing due to strong coupling to the 
106:  leads.~\cite{Beenakker, Alhassid}
107:  We systematically change the broadening from small to large values
108:  to elucidate how this parameter affects the results.
109: 
110: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
111: %\section{Model}
112: 
113:  Our system is defined by the internal Hamiltonian 
114:  for the two QDs and their couplings to the left and right leads.
115:  It is depicted in the upper left inset in Fig.~\ref{xeg}(a).
116:  We assume that one of the dots (dot 1) has regular resonant levels 
117:  and the level spacing is much larger than the level broadening.
118:  In this case, each of the levels can be treated independently.
119:  The other dot (dot 2) is relatively large and 
120:  has many levels distributed randomly.
121:  As is known from scattering theory 
122:  (see, e.g., Refs.~\onlinecite{Beenakker, Alhassid}, 
123:  and \onlinecite{VWZ}),  
124:  the $S$ matrix of the system is written as 
125: \be
126:  S = 1-2\pi i w^\dag\frac{1}{E-H+i\pi ww^\dag}w, \label{S}
127: \ee
128:  where $H=H_1\oplus H_2$ denotes the Hamiltonian 
129:  for the QDs, $w$ is the dot--lead coupling, and 
130:  $E$ is the (Fermi) energy.
131:  We adopt a single--level Hamiltonian $H_1=E_1$ for dot 1.
132:  $H_2$ for dot 2 is a member of the Gaussian unitary 
133:  ensemble~\cite{Mehta} and its size $N$ 
134:  is taken to be infinity to find the universal limit.
135:  We assume that the $S$ matrix is a $2\times 2$ matrix, 
136:  which means the left and right leads have a single channel respectively.
137:  The conductance, measured in units of $2e^2/h$, 
138:  is calculated from $g = \langle |S_{12}|^2\rangle$.
139: 
140:  In the present RMT approach, 
141:  there is no need to know the full form of $w$.
142:  The matrix $w$ appears in the $S$ matrix formula (\ref{S})
143:  as $w^\dag G w$, where $G=(E^+-H)^{-1}$.
144:  After the averaging, the matrix structure of $G$ is lost and 
145:  $2\times 2$ matrices $\gamma_{i} = \pi w^{(i)\dag} w^{(i)}$ 
146:  appear in averaged quantities.
147:  Here $i=1,2$ label the dot and $w^{(1)}$($w^{(2)}$) is 
148:  a $1\times 2$ ($N\times 2$) matrix.
149:  Assuming a symmetric coupling with respect to the left and right leads, 
150:  we arrive at the general form
151: \be
152:  \gamma_1 = \frac{\Gamma_1}{2}\bmat{cc} 1 & e^{i\varphi} \\
153:  e^{-i\varphi} & 1 \emat, \ 
154:  \gamma_2 = \frac{N\Gamma_2}{2}\bmat{cc} 1 & ae^{-i\varphi} \\
155:  ae^{i\varphi} & 1 \emat, 
156: \ee
157:  where $\Gamma_i$ is the level broadening of the dot $i$
158:  and $\varphi$ is the AB flux.~\cite{GIA-KK}
159:  The real parameter $a$ ($0\leq a\leq 1$) represents the strength of 
160:  the nonresonant direct path since the off--diagonal part of $\gamma$ 
161:  contributes to the conductance directly.~\cite{VWZ}
162: 
163: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
164: %\section{Averaged S matrix}
165: 
166:  The averaged S matrix is calculated from the relation
167:  $(1-\langle S\rangle)/(1+\langle S\rangle)
168:  =(1-S_1)/(1+S_1)+(1-\langle S_2\rangle)(1+\langle S_2\rangle)$, 
169:  where $S_{1,2}$ are the $S$ matrices of the dot $1,2$ respectively.
170:  In RMT the magnitude of the Gaussian fluctuations 
171:  determines the mean level spacing $\Delta$.
172:  Then the averaged Green's function of the dot 2 is given by  
173:  $\langle G\rangle = -i\pi/N\Delta$, 
174:  where we take the limit $N\to\infty$ and neglect the real part of $G$.
175:  In this limit we obtain
176:  $\langle S_2\rangle = (1-\pi\gamma_2/N\Delta)/(1+\pi\gamma_2/N\Delta)$
177:  while $S_1=(E-E_1-i\gamma_1)/(E-E_1+i\gamma_1)$.
178:  $\langle S \rangle$ is determined by the following four parameters:
179:  $a$, $\varphi$, $\e = (E-E_1)/\Gamma_1$, and $X = \pi\Gamma_2/\Delta$.
180:  $\e$ measures the distance from the resonance point of the regular dot
181:  and $X$ represents the ratio of the level broadening to 
182:  the level spacing for the random dot.
183:  When $X \gg 1$, the level spectrum is continuous.
184:  The use of RMT implies our consideration is restricted 
185:  to the energy scale much smaller than the Thouless energy.
186: 
187:  If we disregard quantum fluctuations, the conductance is given by
188:  $g_0=|\langle S_{12}\rangle|^2$, 
189:  which we call the principal part of $g$.
190:  It is given by the Fano form 
191: \be
192:  g_0 &=& \frac{a^2 X^2}{\left[1+(1+a)X/2\right]^2
193:  \left[1+(1-a)X/2\right]^2} \no\\
194:  & & \times\frac{\left|E-E_1+q\Gamma_1\right|^2}
195:  {(E-E_1)^2
196:  +\frac{\left[1+(1-a \cos 2\varphi)X/2\right]^2}
197:  {\left[1+(1+a)X/2\right]^2
198:  \left[1+(1-a)X/2\right]^2}\Gamma_1^2}, \label{g0}
199: \ee
200:  where $q = ie^{-2i\varphi}/a X$ is the Fano parameter.
201:  This parameter is complex and becomes purely imaginary 
202:  when the AB flux $\varphi$ is zero, 
203:  which is contrasted with the experiments for clean 
204:  systems~\cite{KAKI, JMHG} where a real $q$ has been 
205:  observed in the absence of a magnetic field.
206:  We also note that the result of Eq.(\ref{g0}) holds 
207:  regardless of the choice of the universality class 
208:  because the resonance is not treated 
209:  randomly, in contrast with the approach in Ref.~\onlinecite{CWB}.
210:  When $a=0$, $|q|\to\infty$ and the result reduces to
211:  the standard Breit--Wigner form.
212:  The presence of the random dot leads to a reduction of the level 
213:  broadening and the conductance by a factor of $1/(1+X/2)^2$.
214:  
215: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
216: %\section{Nonlinear $\sigma$ model and transmissiona coefficients}
217: 
218:  We now consider mesoscopic fluctuations of the conductance 
219:  $\delta g = g-g_0$.
220:  We calculate these using the method of supersymmetry,~\cite{VWZ} 
221:  which allows us to derive the nonlinear $\sigma$ model
222: \be
223:  F = \frac{1}{2}\mbox{str}_8 \ln\left(1+\frac{T/2}{1-T/2}
224:  \frac{\Lambda\sigma+\sigma\Lambda}{2}\right), 
225: \ee
226:  where the $4\times 4$ supermatrix $\sigma$ parametrizes 
227:  the saddle point manifold and satisfies $\sigma^2=1$.~\cite{Efetov}
228:  $\mbox{str}$ denotes the supertrace 
229:  and $\Lambda=\mbox{diag}(1,-1)$ in retarded--advanced space.
230:  The $2\times 2$ matrix $T$ defined by 
231:  $T = 1-\langle S\rangle\langle S\rangle^\dag$
232:  is called the transmission coefficients.
233:  This form of the $\sigma$ model is well known
234:  as a standard model of a single random dot.
235:  The only difference is in the $T$ matrix.
236:  The fact that the $\sigma$ model is written 
237:  in terms of $T$ only demonstrates the universality of 
238:  the correlation functions of the S matrix elements.
239:  For instance, $\delta g$ in Eq.(\ref{deltag}) is a function of $T$.
240:  This is to be contrasted with the result of Eq.(\ref{g0}), 
241:  where such $T$ universality does not hold.
242:  In the present model, we find the eigenvalues of $T$ at $a=0$ 
243: \be
244:  T_1 = \frac{2X}{\left(1+X/2\right)^2}, \qquad
245:  T_2 = \frac{2X}{\left(1+X/2\right)^2+1/\e^2}. \label{T12}
246: \ee
247:  We see that $T_1$ is independent of $\e$.
248:  
249: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
250: %\section{Random S matrix approach}
251: 
252:  We present the analytical results for the conductance
253:  together with numerical ones. 
254:  Numerical calculations are performed 
255:  using the random $S$ matrix model,~\cite{randomS}
256:  where the internal structure of the $S$ matrix is disregarded and 
257:  randomness is imposed directly on $S$. 
258:  For the distribution of $S$, we use the generalized circular unitary 
259:  ensemble (CUE) based on the Poisson kernel~\cite{pker} 
260: \be
261:  P(S)d\mu = 
262:  \frac{\det(1-\langle S\rangle\langle S\rangle^\dag)^2}
263:  {\left|\det(1-S\langle S\rangle^\dag)\right|^4} d\mu,
264: \ee
265:  where $d\mu$ is the measure of the CUE.~\cite{Mehta}
266:  This is the maximally randomized distribution 
267:  under the condition that the average value is $\langle S\rangle$.
268:  Following the previous approach for regular systems~\cite{GIA-KK}
269:  we separate the system into the dots 1 and 2, 
270:  and the left and right forks that connect the dots to leads.
271:  Choosing the fork $S$ matrix in a symmetric form 
272:  we find the transmission through the ring expressed 
273:  by the $S$ matrix $S_{1,2}$ for each dot.
274:  $S_2$ is treated statistically using a Poisson kernel.
275:  This random $S$ matrix approach is equivalent to  
276:  the random Hamiltonian approach if $\langle S\rangle$ is
277:  chosen properly and $N\to\infty$.~\cite{Brouwer}
278:  We use the same expressions for $S_1$ and $\langle S_2\rangle$ 
279:  as in the random Hamiltonian approach.
280: 
281: %%%%%%%%%%%%%%%%%%%%%%%%
282: \begin{figure}[tb]
283: \begin{center}
284:  \includegraphics[width=0.9\columnwidth]{xg.eps} 
285: 
286:  \medskip
287: 
288:  \includegraphics[width=0.9\columnwidth]{eg.eps}
289:  \caption{(Color Online)
290:  Conductance vs $X=\pi\Gamma_2/\Delta$ (a) 
291:  and $\e=(E-E_1)/\Gamma_1$ (b) at $a=0$. 
292:  The thick (thin) lines are the total conductance $g$ 
293:  (the principal part $g_0$).
294:  Upper left inset in (a): Sketch of the sample. 
295:  The parameters characterizing each dot are shown in the figure.
296:  Upper right inset in (a) and (b): Fluctuation part $\delta g=g-g_0$.
297: }
298:  \label{xeg}
299: \end{center}
300: \end{figure}
301: %%%%%%%%%%%%%%%%%%%%%%%%
302: 
303: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
304: %\section{Conductance}
305: 
306:  First we consider the $X$ dependence of the conductance at $a=0$.
307:  When $\e\to\infty$, where the regular dot is detached 
308:  from the random dot, we have $T_1=T_2=2X/(1+X/2)^2$, 
309:  and recover the known result~\cite{Efetov2}
310: \be
311:  g_0=0, \quad \delta g = \frac{T_1}{3}+\frac{T_1^2}{6}.
312:  \label{deltag}
313: \ee
314:  At the resonance point $\e=0$,
315:  $T_1=2X/(1+X/2)^2$, $T_2=0$, and we obtain 
316: \be
317:  g_0 = \frac{1}{\left(1+X/2\right)^2}, \quad 
318:  \delta g = \frac{T_1}{4}=\frac{X/2}{\left(1+X/2\right)^2}. \label{grp}
319: \ee
320:  In Fig.~\ref{xeg}(a), $X$ dependence of the conductance 
321:  is shown for several values of $\e$.
322:  $g_0$ shows a peak at $X=0$
323:  while $\delta g$ takes a maximum at $X=2$, as shown by 
324:  the thin lines and the inset in Fig.~\ref{xeg}(a), respectively.
325:  As $\e\to\infty$ 
326:  $g_0$ ($\delta g$) is monotonically decreasing (increasing) 
327:  and the result rapidly approaches Eq.(\ref{deltag}).
328: 
329:  $\e$ dependence of the conductance is shown in Fig.~\ref{xeg}(b).
330:  A resonance peak appears at $\e=0$, 
331:  reflecting transport through the regular dot.
332:  This peak structure, however, changes qualitatively as a function of $X$.
333:  For small $X$ the peak is convex and 
334:  the peak height decreases on increasing $X$.
335:  When $X=2$, $g$ is independent of $\e$.
336:  Increasing $X$ further, we find that the peak turns into 
337:  an antiresonance and $g$ decreases monotonically.
338:  The result for $X=2$ corresponds to that of 
339:  the CUE because $\langle S_2\rangle=0$, and 
340:  the Poisson kernel $P(S_2)$ becomes unity.
341: 
342: %%%%%%%%%%%%%%%%%%%%%%%%
343: \begin{figure}[tb]
344: \begin{center}
345:  \includegraphics[width=0.9\columnwidth]{fanog.eps}
346:  \caption{(Color Online)
347:  Conductance vs $\e$ at $a=0.7$ and $\varphi=\pi/8$.
348:  The thick (thin) lines are the total conductance $g$
349:  (principal part $g_0$).
350:  Inset: Fluctuation part $\delta g$.}
351:  \label{fanog}
352: \end{center}
353: \end{figure}
354: %%%%%%%%%%%%%%%%%%%%%%%%
355: 
356: % direct reaction model
357: 
358:  The obtained results are for a model with a resonance at $a=0$. 
359:  Now the question is whether they are characteristic of
360:  the resonance model.
361:  For comparison, we discuss the direct reaction model 
362:  with $a\ne 0$ and $\e\to\infty$.
363:  Both models are similar in that 
364:  the conductance $g_0$ becomes finite in the absence of fluctuations.
365:  $g_0$ becomes maximum at $a=1$, 
366:  which is similar to the situation at $\e=0$ in the resonance model.
367:  In the direct reaction model, 
368:  the eigenvalues of the $T$ matrix are given by 
369: \be
370:  T_{1,2} = \frac{2(1\pm a)X}{\left[1+(1\pm a)X/2\right]^2}.
371: \ee
372:  $T_2$ goes to zero as $a\to 1$ as in the limit $\e\to 0$ in Eq.(\ref{T12}). 
373:  In this limit, we find $T_1=4X/(1+X)^2$, $T_2=0$, and 
374: \be
375:  g_0 = \frac{X^2}{\left(1+X\right)^2}, \quad 
376:  \delta g = \frac{T_1}{4}=\frac{X}{\left(1+X\right)^2}.
377: \ee
378:  Note that $T_1$ dependence of $\delta g$ is the same as in Eq.(\ref{grp}) 
379:  but the $X$ dependence is different.
380:  We have found numerically that 
381:  the peak is maintained for an arbitrary value of $X$ and 
382:  that there is no antiresonance, in contrast to the resonance model.
383: 
384: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
385: %\section{Fano effect}
386: 
387: %%%%%%%%%%%%%%%%%%%%%%%%
388: \begin{figure}[b]
389: \begin{center}
390:  \includegraphics[width=0.9\columnwidth]{pg.eps}
391:  \caption{(Color Online)
392:  Conductance distribution at $a=0$.}
393:  \label{pg}
394: \end{center}
395: \end{figure}
396: %%%%%%%%%%%%%%%%%%%%%%%%
397: 
398:  As we have shown in Eq.(\ref{g0}), when we consider both the resonance 
399:  and direct reaction channels, the Fano effect appears in $g_0$.
400:  In Fig.~\ref{fanog}, a typical numerical result of the conductance is shown 
401:  at $a=0.7$ and $\varphi=\pi/8$.
402:  We observe that an asymmetric form is obtained for the fluctuation part 
403:  as well as for the principal part.
404:  However, these asymmetric peaks work to compensate each other,
405:  resulting in a symmetric $g$ 
406:  that has a form similar to the previous result at $a=0$.
407:  The difference is that the conductance as a whole is enhanced 
408:  due to the direct path contribution.
409:  Far from the resonance point, 
410:  the conductance is independent of $\varphi$,
411:  while it is sensitive at the resonance point.
412:  The resonance (antiresonance) is amplified at $\varphi=0$ ($\pi/4$)
413:  where the Fano parameter $q$ is pure imaginary (real).
414:  When $q$ takes a complex value (when $0<\varphi<\pi/4$), a dip 
415:  in the resonance peak is formed at the intermediate values of $X$ 
416:  (See the plot of $X=2$ in Fig.~\ref{fanog}).
417: 
418: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
419: %\section{Conductance distribution function}
420: 
421:  As we have shown, $\delta g$ are of the order of $g_0$ and
422:  the mere calculation of the averaged conductance is not enough 
423:  to characterize the system.
424:  We show the numerical results for the conductance distribution function  
425:  $P(g)=\langle\delta (g-|S_{12}|^2)\rangle$ in Fig.~\ref{pg}.
426:  For $\e\gg 1$ we find the results in Refs.~\onlinecite{PEI} and 
427:  \onlinecite{randomS} expressed in terms of $T_1$ in Eq.(\ref{T12}).
428:  For finite value of $\e$, the $T$ universality does not hold and 
429:  the results essentially depend on $X$ and $\e$.
430:  When $X<2$, the peak representing the resonant conductance 
431:  moves from $g=0$ to 1 as $\e$ decreases.
432:  When $X>2$, the resonant conductance is suppressed 
433:  and the distribution function shows 
434:  the strong influence of the chaotic scattering.
435:  When $\e=0$, two peaks appear at $g=0$ and $1$.
436:  The peak at $g=0$ is larger (smaller) than that at $g=1$ 
437:  when $X>2$ ($X<2$), which clearly shows the coexistence of 
438:  the contributions from both regular and chaotic dots.
439:  At $X=2$, $P(g)$ is always symmetric and
440:  is consistent with the results of the averaged conductance.
441:  The curve for $\e=0$ is well fitted by the function 
442:  $P(g)=1/\pi\sqrt{g(1-g)}$.
443: 
444: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
445: %\section{Dephasing}
446: 
447:  We remark on the effect of dephasing.
448:  It can be simply studied by introducing 
449:  an imaginary part to the energy $E\to E+i/2\tau$.
450:  The substitution of this to Eq.(\ref{g0})
451:  leads to the reduction of the conductance.
452:  The effect on the fluctuation part is to add 
453:  $F_\tau = (\Delta\tau)^{-1}\mbox{str}\,\sigma\Lambda$
454:  to the $\sigma$ model.
455:  Numerically we have observed that the fluctuation part is strongly 
456:  suppressed by this effect while the principal part shows 
457:  a small reduction.
458:  This means that the resonance is preserved at any $X$.
459:  We also anticipate that an asymmetric Fano resonance can be observed, 
460:  since the cancellation of the asymmetry between 
461:  the principal and fluctuation parts becomes incomplete.
462:  The quantitative estimate of the dephasing effects 
463:  using the method in Ref.~\onlinecite{BB}
464:  is necessary to compare with the experimental results, as was 
465:  done in Refs.~\onlinecite{HPMBDH} and \onlinecite{HZOAA} 
466:  for the single--dot systems.
467:  A detailed study will be reported elsewhere.
468: 
469: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
470: %\section{Universality class}
471: 
472:  Another interesting problem 
473:  is the ensemble dependence of the result.
474:  Our numerical calculations using the orthogonal and symplectic 
475:  ensembles show that the distribution at 
476:  the resonant point (the lowest graph in Fig.~\ref{pg}) is 
477:  independent of the ensemble.
478:  It implies the distribution is determined by some universal mechanism.
479: 
480: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
481: %\section{Conclusions}
482: 
483:  In conclusion, we have developed a statistical theory 
484:  for an AB ring system with regular and chaotic QDs.
485:  The conductance and its distribution are strongly influenced by 
486:  the mesoscopic fluctuations of the chaotic dot and 
487:  the position of the resonance peaks.
488: 
489:  We are grateful to A. Furusaki and S. Iida for useful discussions.
490:  T.A. acknowledges support from 
491:  the Japan Society for the Promotion of Science.
492: 
493: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
494: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
495: \begin{thebibliography}{99}
496: 
497: \bibitem{Beenakker}
498:  C.W.J. Beenakker, Rev. Mod. Phys. {\bf 69}, 731 (1997).
499: 
500: \bibitem{Alhassid}
501:  Y. Alhassid, Rev. Mod. Phys. {\bf 72}, 895 (2000).
502: 
503: \bibitem{SI}
504:  P.G. Silvestrov and Y. Imry, Phys. Rev. Lett. {\bf 85}, 2565 (2000).
505: 
506: \bibitem{Fano}
507:  U. Fano, Phys. Rev. {\bf 124}, 1866 (1961).
508: 
509: \bibitem{KAKI}
510:  K. Kobayashi, H. Aikawa, S. Katsumoto, and Y. Iye,
511:  Phys. Rev. Lett. {\bf 88}, 256806 (2002);
512:  Phys. Rev. B {\bf 68}, 235304 (2003).
513: 
514: \bibitem{JMHG}
515:  A.C. Johnson, C.M. Marcus, M.P. Hanson, and A.C. Gossard, 
516:  Phys. Rev. Lett. {\bf 93}, 106803 (2004).
517: 
518: \bibitem{RLBKS}
519:  S. Rotter, F. Libisch, J. Burgd\"orfer, U. Kuhl, and H.-J. St\"ockmann, 
520:  Phys. Rev. E {\bf 69}, 046208 (2004).
521: 
522: \bibitem{FCSWP}
523:  J. Faist, F. Capasso, C. Sirtori, K.W. West, and L.N. Pfeiffer, 
524:  Nature (London) {\bf 390}, 589 (1997).
525: 
526: \bibitem{PEI}
527:  V.N. Prigodin, K.B. Efetov, and S. Iida, 
528:  Phys. Rev. Lett. {\bf 71}, 1230 (1993);
529:  Phys. Rev. B {\bf 51}, 17223 (1995).
530: 
531: \bibitem{HPMBDH}
532:  A.G. Huibers, S.R. Patel, C.M. Marcus, P.W. Brouwer, 
533:  C.I. Duru\"oz, and J.S. Harris, Jr., 
534:  Phys. Rev. Lett. {\bf 81}, 1917 (1998).
535: 
536: \bibitem{HZOAA}
537:  S. Hemmady, X. Zheng, E. Ott, T.M. Antonsen, and S.M. Anlage, 
538:  Phys. Rev. Lett. {\bf 94}, 014102 (2005); 
539:  S. Hemmady, X. Zheng, J. Hart, T.M. Antonsen, Jr., E. Ott, 
540:  and S.M. Anlage, cond-mat/0512131 (unpublished).
541: 
542: \bibitem{CWB}
543:  A.A. Clerk, X. Waintal, and P.W. Brouwer, 
544:  Phys. Rev. Lett. {\bf 86}, 4636 (2001).
545: 
546: \bibitem{Mehta}
547:  M.L. Mehta, {\it Random Matrices} (Academic Press, New York, 2004).
548: 
549: \bibitem{VWZ}
550:  J.J.M. Verbaarschot, H.A. Weidenm\"uller, and M.R. Zirnbauer,
551:  Phys. Rep. {\bf 129}, 367 (1985).
552: 
553: \bibitem{GIA-KK}
554:  Y. Gefen, Y. Imry, and M.Ya. Azbel, 
555:  Phys. Rev. Lett. {\bf 52}, 129 (1984); 
556:  B. Kubala and J. K\"onig, 
557:  Phys. Rev. B {\bf 65}, 245301 (2002); 
558:  {\bf 67}, 205303 (2003).
559: 
560: \bibitem{Efetov} K.B. Efetov, Adv. Phys. {\bf 32}, 53 (1983); 
561:  {\it Supersymmetry in Disorder and Chaos} 
562:  (Cambridge University Press, Cambridge, U.K., 1997).
563: 
564: \bibitem{randomS}
565:  H.U. Baranger and P.A. Mello,
566:  Phys. Rev. Lett. {\bf 73}, 142 (1994);
567:  R.A. Jalabert, J.-L. Pichard, and C.W.J. Beenakker,
568:  Europhys. Lett. {\bf 27}, 255 (1994);
569:  P.W. Brouwer and C.W.J. Beenakker,
570:  Phys. Rev. B {\bf 50}, R11263 (1994).
571: 
572: \bibitem{pker}
573:  L.K. Hua, {\it Harmonic Analysis of Functions of Several Complex 
574:  Variables in the Classical Domains} 
575:  (AMS, Providence, RI, 1963);
576:  P.A. Mello, P. Pereyra, and T.H. Seligman,
577:  Ann. Phys. (NY) {\bf 161}, 254 (1985);
578:  see also Ref.~\onlinecite{Beenakker}.
579: 
580: \bibitem{Brouwer}
581:  P.W. Brouwer, Phys. Rev. B {\bf 51}, 16878 (1995).
582: 
583: \bibitem{Efetov2}
584:  K.B. Efetov, Phys. Rev. Lett. {\bf 74}, 2299 (1995).
585: 
586: \bibitem{BB}
587:  P.W. Brouwer and C.W.J. Beenakker, Phys. Rev. B {\bf 55}, 4695 (1997).
588: 
589: \end{thebibliography}
590: 
591: \end{document}
592: