1:
2: \documentclass
3: [aps,prl,twocolumn,showpacs,floats,amsmath]{revtex4}%
4: %\documentclass
5: %[aps,prl,preprint,showpacs,superscriptaddress,floats,amsmath]{revtex4}%
6: \usepackage{graphicx}
7: \usepackage{amsmath}
8: \usepackage{amsfonts}
9: \usepackage{amssymb}%
10: \usepackage{epsf}
11:
12: %\documentclass[aps,prl,twocolumn]{revtex4}
13: %\documentclass[aps,preprint,prl]{revtex4}
14:
15: \newcommand{\be}{\begin{equation}}
16: \newcommand{\ee}{ \end{equation}}
17: \newcommand{\ba}{\begin{eqnarray}}
18: \newcommand{\ea}{ \end{eqnarray}}
19:
20:
21: \begin{document}
22:
23: \title{Gaussian Decoherence from Spin Environments}
24: \author{W.H. Zurek}
25: \author{F.M. Cucchietti}
26: \author{J.P. Paz}
27: \affiliation{Theoretical Division, MS B213, Los Alamos National
28: Laboratory, Los Alamos, NM 87545}
29: \date{February 10, 2004}
30:
31: \begin{abstract}
32: We examine an exactly solvable model of decoherence -- a spin-system
33: interacting with a collection of environment spins. We show that in
34: this model (introduced some time ago to illustrate environment--induced
35: superselection) generic assumptions about the coupling strengths lead to
36: a universal (Gaussian) suppression of coherence between pointer states.
37: We explore the regimes of validity of these results and discuss their relation to the
38: spectral features of the environment and to the Loschmidt echo (or fidelity).
39: Finally, we comment on the observation of such time dependence in spin echo
40: experiments.
41: \end{abstract}
42:
43: \pacs{03.65.Yz;03.67.-a}
44:
45: \maketitle
46:
47: A single spin--system ${\cal S}$ (with states $\left\{ \left|0\right>, \left|1\right> \right\}$)
48: interacting with an environment ${\cal E}$ of many independent spins
49: ($\left\{ \left| \uparrow_k \right>, \left| \downarrow_k \right> \right\}$,
50: $k=1..N$) through
51: the Hamiltonian
52: \be
53: {\cal H_{SE}} =\left(\left|0\right>\left<0\right| -
54: \left|1\right>\left<1\right| \right) \sum_{k=1}^N \frac{g_k}{2}
55: \left( \left| \uparrow_k \right> \left< \uparrow_k \right| -
56: \left|\downarrow_k \right>\left<\downarrow_k \right| \right)
57: \label{hamiltonian0}
58: \ee
59: may be the simplest solvable model of decoherence in spin systems.
60: It was introduced
61: some time ago \cite{Zurek82} to show that relatively
62: straightforward
63: assumptions about the dynamics can lead to the emergence of a preferred
64: set of pointer states due to environment--induced superselection (einselection)
65: \cite{Zurek82,deco}.
66: Such models have gained additional importance in the past decade because of
67: their relevance to quantum information processing \cite{QIP}.
68:
69: The purpose of
70: our paper is to show that -- with a few additional natural and simple
71: assumptions --
72: one can evaluate the exact time dependence of the reduced density matrix, and
73: demonstrate that the off--diagonal components display a Gaussian (rather than
74: exponential) decay.
75: In effect, we exhibit a simple soluble example of a situation where the usual
76: Markovian \cite{Kossakowski} assumptions about the evolution of a quantum open
77: system are not satisfied at all times.
78: Apart from their implications for decoherence, our results are also relevant to quantum error
79: correction \cite{ErrorCorrection} where precise precise knowledge of the
80: dynamics is essential to select an efficient strategy.
81:
82: To demonstrate the Gaussian time dependence of decoherence we first write down a
83: general solution for the model given by Eq.~(\ref{hamiltonian0}).
84: Starting with:
85: \be
86: \left| \Psi_{\cal SE}(0)\right> =
87: (a \left|0\right>+b \left|1\right>) \bigotimes_{k=1}^N
88: \left( \alpha_k \left| \uparrow_k \right> +
89: \beta_k \left|\downarrow_k \right> \right),
90: \label{initialstate}
91: \ee
92: the state of ${\cal SE}$ at an arbitrary time is given by:
93: \be
94: \left| \Psi_{\cal SE}(t)\right> =
95: a \left|0\right> \left|{\cal E}_0 (t)\right>
96: +b \left|1\right> \left|{\cal E}_1 (t)\right>
97: \label{phit}
98: \ee
99: where
100: \ba
101: \left|{\cal E}_0 (t)\right> & = & \bigotimes_{k=1}^N
102: \left( \alpha_k e^{i g_k t/2} \left| \uparrow_k \right>
103: + \beta_k e^{-i g_k t/2} \left|\downarrow_k \right> \right) \nonumber \\
104: &=& \left|{\cal E}_1 (-t)\right>.
105: \label{environ}
106: \ea
107: The reduced density matrix of the system is then:
108: \ba
109: \rho_{\cal S} & = & {\rm Tr} _{\cal E} \left| \Psi_{\cal SE}(t)\right>
110: \left< \Psi_{\cal SE}(t)\right| \nonumber \\
111: & = & |a|^2 \left|0\right>\left<0\right|+ a b^{*} r(t)
112: \left|0\right>\left<1\right| \nonumber \\
113: & + & a^{*} b r^{*}(t) \left|1\right>\left<0\right| + |b|^2 \left| 1 \right> \left< 1 \right|,
114: \label{reducedrho}
115: \ea
116: where the {\it decoherence} factor
117: $r(t)=\left<{\cal E}_1 (t)|{\cal E}_0 (t)\right>$
118: can be readily obtained:
119: \ba
120: r(t)&=&\prod_{k=1}^N
121: \left( |\alpha_k|^2 e^{i g_k t} + |\beta_k|^2 e^{-i g_k t} \right).
122: \label{roft}
123: \ea
124:
125: It is straightforward to see that $r(0)=1$ and for $t>0$ it decays rapidly
126: to zero, so that the typical fluctuations of the off-diagonal terms of
127: $\rho_{\cal S}$ will be small for large environments, since:
128: \be
129: \left<|r(t)|^2 \right>=2^{-N} \prod_{k=1}^N \left( 1+(|\alpha_k|^2 -
130: |\beta_k|^2)^2 \right),
131: \label{rsoft}
132: \ee
133: Here $\left<...\right>$ denotes a long time average \cite{Zurek82}.
134: Clearly,
135: $\left<|r(t)|^2 \right> \underset{N\rightarrow \infty}{\longrightarrow} 0$,
136: leaving $\rho_{\cal S}$ approximately diagonal in a mixture of the pointer states
137: $\left\{ \left|0\right>, \left|1\right> \right\}$ which retain preexisting
138: classical correlations.
139:
140: This much was known since \cite{Zurek82}. The aim of this paper is to show
141: that, for a fairly generic set of assumptions, the form of $r(t)$ can be
142: further evaluated and that -- quite universally -- it turns out to be
143: approximately Gaussian in time. Thus, the simple model of Ref. \onlinecite{Zurek82}
144: predicts a universal (Gaussian) form of the loss of quantum coherence, whenever
145: the couplings $g_k$ of Eq.~\ref{hamiltonian0} are sufficiently concentrated near
146: their average value so that their standard deviation
147: $\left<(g_k-\left<g_k\right>)^2\right>$ exists and is finite. When this condition
148: is not fulfilled other sorts of time dependence become possible. In particular,
149: $r(t)$ may be exponential when the distribution of couplings is a Lorentzian.
150:
151: To obtain our main result we carry out the multiplication of Eq.(\ref{roft}), re--expressing
152: $r(t)$ as a sum:
153: \ba
154: r(t) &=& \prod_{k=1}^N |\alpha_k|^2 e^{i t \sum_{n} g_n}
155: + \sum_{l=1}^N |\beta_l|^2 \prod_{k\ne l}^N |\alpha_k|^2 \times \nonumber \\
156: & & e^{i t(-g_l +\sum_{n\ne l} g_n )}
157: + \sum_{l=1}^N\sum_{m\ne l}^N |\beta_l|^2 |\beta_m|^2 \times \nonumber \\
158: & &\prod_{k\ne l,m}^N |\alpha_k|^2
159: e^{\left[i t (-g_l-g_m + \sum_{n\ne l,m}^N g_n )\right]}
160: +...
161: \label{rexpansion}
162: \ea
163: There are $\binom{N}{0}$, $\binom{N}{1}$, $\binom{N}{2}$, ... etc. terms in the
164: consecutive sums above.
165: The binomial pattern is clear, and can be made even more apparent by assuming
166: that $\alpha_k=\alpha$ and $g_k=g$ for all $k$. Then,
167: \be
168: r(t)= \sum_{l=0}^N \binom{N}{l} |\alpha|^{2(N-l)} |\beta|^{2l} e^{ig(N-2l)t},
169: \label{binomial}
170: \ee
171: i.e., $r(t)$ is the binomial expansion of $r(t)=\left(|\alpha|^2 e^{igt}+
172: |\beta|^2 e^{-igt}\right)^N$.
173:
174: We now note that, as follows from the Laplace-de Moivre theorem \cite{Gnedenko},
175: for sufficiently large $N$ the coefficients of the binomial expansion
176: of Eq. (\ref{binomial}) can be approximated by a Gaussian:
177: \ba
178: \binom{N}{l} |\alpha|^{2(N-l)} |\beta|^{2l} \simeq %\nonumber\\ & &
179: \frac{\exp{\left[-\frac{(l-N|\beta|^2)^2}{2 N |\alpha \beta|^2}\right]}}
180: {\sqrt{2 \pi N |\alpha \beta|^2}} .
181: \label{gaussian}
182: \ea
183: This limiting form of the distribution of the eigenenergies
184: of the composite ${\cal SE}$ system
185: immediately yields our main result: $r(t)$ is approximately Gaussian
186: since it is a Fourier transform of an approximately Gaussian distribution of
187: the energies resulting from all the possible combinations of the couplings
188: with the environment.
189:
190: The set of all the resulting energies must have an
191: (approximately) Gaussian distribution. This behavior is generic, a result of the
192: law of large numbers \cite{Gnedenko}:
193: these energies can be thought of as being the terminal points of an $N$--step random walk.
194: The contribution of the $k$--th spin of the environment to the random energy is $+g$ or
195: $-g$ with probability $|\alpha|^2$ or $|\beta|^2$ respectively (Fig. 1 a).
196:
197: The same argument can be carried out in the more general case of Eq.~(\ref{rexpansion}).
198: The ``random walk'' picture that yielded the distribution of the couplings
199: remains valid (see Fig. 1 b).
200: However, now the individual steps in the random walk are not all equal. Rather,
201: they are given by the set $\left\{g_k \right\}$ (see Eq.~\ref{hamiltonian0})
202: with each step $g_k$ taken just once in a given walk.
203: There are $2^N$ such distinct random walks, each contributing with the weight given by
204: the product of the relevant $|\alpha_k|^2$ and $|\beta_k|^2$ to
205: the sum of Eq.~(\ref{rexpansion}). This exponential proliferation of the contributing
206: coupling energies allows one to anticipate rapid convergence to the universal Gaussian
207: form of the decoherence factor $r(t)$.
208:
209: \begin{figure}
210: \centering \leavevmode
211: \epsfxsize 3.2in
212: \epsfbox{fig1.eps}
213: %\vspace{0.25 cm}
214: \caption{The distribution of the energies obtains from the random walks with the
215: steps given by the coupling size and in the direction ($+g_k$ or $-g_k$) biased
216: by the probabilities $|\alpha_k|^2$ and $|\beta_k|^2$ as in Eq.~(\ref{Ldosrw})
217: (although in these examples we set $|\alpha_k|^2=1/2$).
218: (a) When all the couplings
219: have the same size $g_k=g$ (Eq.~(\ref{binomial})),
220: a simple Newton's triangle leads to an approximate
221: Gaussian for the distribution of energies. (b) When the couplings
222: differ from step to step (Eq.~(\ref{rexpansion})),
223: the resulting distribution still
224: has a approximately Gaussian envelope for large $N$. }
225: \label{Figure1}
226: \end{figure}
227:
228: Indeed, we can regard the energies resulting from the sums of $g_k$'s as a
229: random variable. Its probability distribution is given by products of the
230: corresponding weights.
231: That is, the typical term in Eq.~\ref{rexpansion} is of the form:
232: \be
233: p_W e^{iE_W t} \equiv \left( \prod_{k\in W^+} |\alpha_k|^2e^{i g_kt}\right)
234: \left(\prod_{k\in W^-}|\beta_k|^2e^{-i g_kt} \right).
235: \ee
236: The resulting terminal energy is
237: \be
238: E_W=\sum_{k\in W^+} g_k-\sum_{k\in W^-} g_k,
239: \ee
240: and the cumulative weight $p_W$ is given by the corresponding product of
241: $|\alpha_k|^2$ and $|\beta_k|^2$. Each such specific random walk $W$
242: corresponding to a given combination of right ($k\in W^+$) and left ($k\in W^-$)
243: ``steps" (see Fig. 1) contributes to the distribution of energies only once.
244: The terminal points $E_{W}$ may or may not be degenerate:
245: As seen in Fig. 1, in the degenerate case, the whole
246: collection of $2^N$ random walks ``collapses" into $N+1$ terminal
247: energies. More typically, in the degenerate case (also displayed in Fig. 1),
248: there are $2^N$ different terminal energies $E_W$.
249: In both cases, the ``envelope" of the distribution $P(E_W)$ should be Gaussian,
250: as we shall argue below.
251:
252: We note
253: that the decoherence factor $r(t)$ can be viewed as the characteristic function
254: \cite{Gnedenko} (i.e., the Fourier transform)
255: of the distribution of eigenenergies $E_W$. Thus,
256: \be
257: r(t)=\int e^{iEt} \eta(E) dE,
258: \label{rLDOS}
259: \ee
260: where the strength function $\eta(E)$, also known as the local density of states (LDOS)
261: \cite{LDOS} is defined in general as
262: \be
263: \eta(E)=\sum_\lambda |\left< \Psi_{\cal SE}(0)| \phi_\lambda \right>|^2 \delta(E-E_\lambda).
264: \ee
265: Above $\left|\phi_\lambda\right>$ are the eigenstates of the full Hamiltonian
266: and $E_\lambda$ its eigenenergies. In our particular model (Eq.~\ref{hamiltonian0})
267: the eigenstates are associated with all possible random walks in the set $W$,
268: and therefore
269: \be
270: \eta(E)=\sum_W p_W \delta(E-E_W).
271: \label{Ldosrw}
272: \ee
273: The discussion of decoherence in our model is thus directly related to
274: the study of the characteristic function of the distribution of
275: coupling energies $\eta(E)$.
276: Moreover, since the $E_W$'s are sums of $g_k$'s (that we assume independent of
277: each other), $r(t)$ is itself a product of characteristic functions of the distributions of
278: the couplings $\{g_k\}$, as we have already seen in the example of
279: Eq. (\ref{roft}). Thus, the distribution of $E_W$ belongs to the
280: class of the so--called {\it infinitely divisible distributions} \cite{Gnedenko,breiman}.
281:
282: \begin{figure}
283: \centering %\leavevmode
284: \epsfxsize 3.2in
285: \epsfbox{fig2.eps}
286: %\vspace{0.25 cm}
287: \caption{(Left panels) Assumed distribution of the couplings $g_k$.
288: (Center panels) Resulting distribution of the
289: eigenenergies $E_W$ (center panels) for $N=6$ ($E_W<0$) and $N=24$
290: ($E_W>0$). In the case of $|\alpha_k|^2=1/2$ this distribution is in effect the
291: ``strength function" (local density of states). (Right panels)
292: Decoherence factor $r(t)$ for different initial
293: conditions with $N=6$ (dashed lines), $N=24$ (thin solid lines) and the average
294: (bold line).}
295: \label{Figure2}
296: \end{figure}
297:
298: The behavior of the decoherence factor $r(t)$
299: -- characteristic function of an infinitely divisible distribution --
300: depends only on the average and variance of the distributions of couplings
301: weighted by the initial state of the environment
302: \cite{Gnedenko,breiman}.
303: The remaining task is to calculate $\eta(E)$, which can be obtained through the
304: statistical analysis of the weighted random walk picture described above. If we
305: denote $x_k$ the random variable that takes the value $+g_k$ or $-g_k$ with probability
306: $|\alpha_k|^2$ or $|\beta_k|^2$ respectively, then its mean value
307: $a_k$ and its variance $b_k$ are
308: \ba
309: a_k&=&(|\alpha_k|^2-|\beta_k|^2)g_k, \nonumber \\
310: b_k^2&=&g_k^2-a_k^2=4|\alpha_k|^2|\beta_k|^2g_k^2.
311: \ea
312: The behavior of the sums of $N$ random variables $x_k$ (and thus, of their characteristic
313: function) depends on whether the so--called Lindeberg condition holds
314: \cite{Gnedenko}. It is expressed in terms of the cumulative variances
315: $B_N^2=\sum b_k^2$, and it is satisfied when the probability of the large
316: individual steps is small; e.g.:
317: \be
318: P(\underset{1\le k \le N}{{\rm max}} |g_k-a_k| \ge \tau B_N)
319: \underset{N\rightarrow\infty}{\longrightarrow} 0,
320: \ee
321: for any positive constant $\tau$. In effect, Lindeberg condition demands that $B_N$ be
322: finite: when it is met, the resulting distribution
323: of energies $E=\sum x_k$ is Gaussian
324: \be
325: P\left( \frac{E-{\overline E}_N}{B_N}<x\right)
326: \underset{N\rightarrow\infty}{\longrightarrow}
327: \int_{-\infty}^{x} e^{-s^2/2}ds,
328: \ee
329: where ${\overline E}_N=\sum_k a_k$. In terms of the LDOS this implies
330: \be
331: \eta(E)\simeq \frac{1}{\sqrt{2\pi B_N^2}}
332: \exp{\left(\frac{-(E-{\overline E}_N)^2}{2 B_N^2}\right)},
333: \ee
334: an expression in excellent agreement with numerical results already for modest
335: values of $N$.
336: This distribution of energies yields a corresponding approximately
337: Gaussian time--dependence of $r(t)$, as seen in Fig. 2. Moreover, at least
338: for short times of interest for, say, quantum error correction, $r(t)$ is
339: approximately Gaussian already for relatively small values of $N$. This
340: conclussion holds whenever the initial distribution of the couplings has a
341: finite variance. The general form of $r(t)$ after applying the Fourier transform
342: of Eq. (\ref{rLDOS}) is
343: \be
344: r(t)\simeq e^{i {\overline E}_N t} e^{-B_N^2 t^2/2}.
345: \ee
346:
347: It is also interesting to investigate cases when Lindeberg condition is not
348: met. Here, the possible limit distributions are given by the stable (or L\'{e}vy)
349: laws \cite{breiman}.
350: One interesting case
351: that yields an exponential decay of the decoherence factor corresponds to the
352: case of the Lorentzian distribution of couplings (see Fig. 3).
353: Further intriguing questions concern the robustness of our conclusion under the
354: changes of the model. We shall address this issue elsewhere
355: \cite{CookPreparation} but, for the time being, we only note that the addition
356: of a strong self--Hamiltonian proportional to $\sigma_x$ changes the nature of the time
357: decay \cite{Dobrovitski}. On the other hand, small changes of the environment
358: Hamiltonians (like for instance dipolar interactions)
359: seem to preserve the Gaussian nature of $r(t)$.
360:
361: It is interesting to notice that the Fourier transform of the strength function
362: $\eta(E)$ is also related to the Loschmidt echo
363: \cite{LETheo} (or fidelity) in the so called Fermi Golden
364: rule regime. The fact that the purity and the fidelity have closely related
365: decay rates has been recently shown \cite{LEdeco}
366: for the case of a bath composed of non--interacting harmonic
367: oscillators. In this sense our results could be interpreted as an extension
368: of the discussion of Ref. \cite{LEdeco} to spin environments.
369:
370: The connection with fidelity is more easily seen if we write a
371: generalized version of the Hamiltonian (\ref{hamiltonian0}),
372: \ba
373: {\cal H_{SE}} =\frac{1}{2}\left(\left|0\right>\left<0\right| \otimes
374: {\cal H}^0_{\cal E} + \left|1\right>\left<1\right| \otimes
375: {\cal H}^1_{\cal E} \right) .
376: \label{hamiltonianCnot}
377: \ea
378: The decoherence factor is then the overlap of the initial state of the
379: environment $\left| \Psi_{\cal E}(0) \right>$ evolved with two different
380: Hamiltonians,
381: \ba
382: r(t)= \left< \Psi_{\cal E}(0) \right|e^{i{\cal H}^0_{\cal E} t/2}
383: e^{-i{\cal H}^1_{\cal E} t/2}
384: \left| \Psi_{\cal E}(0) \right>,
385: \ea
386: which clearly has the form of the amplitude of the Loschmidt echo for the
387: environment with the two states of the system as the perturbation. In the
388: particular model that we are treating, ${\cal H}_{\cal E}^0=-{\cal H}_{\cal E}^1$ and thus
389: \ba
390: r(t)&=& \left< \Psi_{\cal E}(0) \right|e^{-i{\cal H}_{\cal E}^1 t}
391: \left| \Psi_{\cal E}(0) \right>.
392: \label{autocorrelation}
393: \ea
394: This expression is the survival probability of the initial state of the
395: environment under the action of the Hamiltonian ${\cal H}_{\cal E}^1$,
396: which has been shown to be the Fourier transform of the strength function
397: \cite{Heller}. This connection provides another way to understand
398: Eq.(\ref{rLDOS})
399:
400: \begin{figure}
401: \centering \leavevmode
402: \epsfxsize 3.2in
403: \epsfbox{fig3.eps}
404: %\vspace{0.25 cm}
405: \caption{Same as Fig. 2 but for a Lorentzian distribution of the couplings
406: ${g_k}$. In this case $r(t)$ decays exponentially.
407: The histogram and the dashed line in $r(t)$ correspond to $N=20$,
408: the straight thin line is a particular case for $N=100$ and the thick line is
409: the average. We note that the convergence is slower than in the Gaussian case
410: of Fig. 2, because realizations of ${g_k}$ are more likely to have
411: one or two dominant couplings. Therefore, although the average shows a clear
412: exponential decay, fluctuations are noticeable even for large $N$.
413: Notice also that the logarithmic scale confirms the long time saturation of
414: $r(t)$ at $\sim 2^{-N/2}$, Eq~(\ref{rsoft}).}
415: \label{Figure3}
416: \end{figure}
417:
418: Possible experimental applications of our considerations are in nuclear
419: magnetic resonance, but also in other situations where two-level systems
420: interact with spin environments. We note that a Gaussian time dependence has
421: been seen in the NMR setting \cite{DiffusionNMR} but it is usually explained by
422: spin diffusion models (which have rather different character and employ a
423: different set of assumptions).
424: Moreover, there is a substantial body of work \cite{Dobrovitski,deRaedt,Loss}
425: on decoherence due to spin environments,
426: stimulated in part by the interests of quantum computation.
427: The relation between the decoherence factor and the strength function might
428: prove useful in the physical setting of strongly
429: interacting fermions, where it has been shown that the strength function takes a
430: Gaussian shape \cite{Kota}.
431: It is our hope that
432: the simple analytic model described here will assist in gaining further insights
433: into these fascinating problems.
434:
435: We acknowledge fruitful discussions with R. Blume-Kohout and G. Raggio. We also
436: acknowledge partial support from ARDA/NSA grant. JPP received also partial support
437: from a grant by Fundaci\'on Antorchas.
438:
439: \begin{thebibliography}{99}
440:
441: \bibitem{Zurek82} W.H. Zurek, Phys. Rev. D {\bf 26}, 1862 (1982)
442:
443: \bibitem{deco} W.H. Zurek, Phys. Today {\bf 44}, 36 (1991);
444: J. P. Paz and W. H. Zurek, in {\it Coherent matter waves, Les Houches
445: Session LXXII}, R Kaiser, C Westbrook and F David eds., EDP Sciences
446: (Springer Verlag, Berlin, 2001) 533-614;
447: W.H. Zurek, Rev. Mod. Phys. {\bf 75}, 715 (2003).
448:
449: \bibitem{QIP} M. A. Nielsen and I. L.~ Chuang, \textit{Quantum computation
450: and quantum information} (Cambridge University Press, Cambridge, New York, 2000).
451:
452: \bibitem{Kossakowski} A. Kossakowski, Bull. Acad. Pol. Sci., Ser. Sci., Math.
453: Astron. Phys. {\bf 21}, 649 (1973); G. Lindblad, Commun. Math. Phys. {\bf 48}, 119 (1976)
454:
455: \bibitem{ErrorCorrection} J. Preskill, Phys. Today {\bf 52} (6), 24 (1999).
456:
457: \bibitem{Gnedenko} B.V. Gnedenko, {\it The Theory of Probability}, Fourth edition
458: (Chelsea, New York,1968), see Chap. VIII.
459:
460: \bibitem{LDOS} G.Casati, B.V. Chirikov, I. Guarneri and F.M. Izrailev, Phys.
461: Rev. E {\bf 48} R1613 (1993); Phys. Lett. A {\bf 223}, 430 (1996).
462:
463: \bibitem{breiman} L. Breiman, {\it Probability}, Classics in Applied Mathematics
464: (SIAM, Philadelphia, 1992).
465:
466: \bibitem{CookPreparation} F.M. Cucchietti, W.H. Zurek and J.P. Paz, in
467: preparation.
468:
469: \bibitem{Dobrovitski} V.V. Dobrovitski, H.A. De Raedt, M.I. Katsnelson and B.N.
470: Harmon, quant-ph/0112053.
471:
472: \bibitem{LETheo} R.A. Jalabert and H.M. Pastawski, Phys. Rev. Lett. {\bf 86}, 2490 (2001);
473: Ph. Jacquod, P. G. Silvestrov, and C. W. J. Beenakker, Phys. Rev. E \textbf{64}, 055203 (2001);
474: F.M. Cucchietti, H.M. Pastawski, and R.A. Jalabert, cond-mat/0307752.
475:
476: \bibitem{LEdeco} F.M.~Cucchietti, D. A. R.~Dalvit, J.P.~Paz, and W. H.
477: Zurek, Phys. Rev. Lett. \textbf{91}, 210403 (2003).
478:
479: \bibitem{Heller} E. Heller in {\it Chaos and Quantum Physics}, Proceedings of
480: Session LII of the Les Houces Summer School, edited by A. Voros and M.J.
481: Giannoni (North-Holland, Amsterdam, 1990).
482:
483: \bibitem{DiffusionNMR} A. Abragam, {\it The principles of nuclear magnetism},
484: Clarendon Press, Oxford (1978); T.T.P. Cheung, Phys. Rev. B {\bf 23}, 1404
485: (1981).
486:
487: \bibitem{deRaedt} V.V. Dobrovitski and H. A. De Raedt, Phys. Rev. E {\bf 67},
488: 056702 (2003); H.A. De Raedt and V.V. Dobrovitski, quant-ph/0301121.
489:
490: \bibitem{Loss} J. Schliemann, A.V. Khaetskii and D. Loss, Phys. Rev. B {\bf 66},
491: 245303 (2002).
492:
493: \bibitem{Kota} V.K.B. Kota, Phys. Rep. \bf{347}, 223 (2001); V.V. Flambaum and
494: F.M. Izrailev, Phys. Rev. E {\bf 61}, 2539 (2000).
495:
496: \end{thebibliography}
497:
498: \end{document}
499:
500:
501:
502:
503:
504: