1: \documentstyle[prl,twocolumn,aps,epsf]{revtex}
2:
3: \begin{document}
4: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname @twocolumnfalse\endcsname
5: \draft
6: \tolerance 500
7: \title{Aging effects in the quantum dynamics of a dissipative free particle: non-ohmic case}
8: \author{Alain Mauger$^{1}$ and No\"elle Pottier$^{2}$}
9:
10: \address{$^1$ Laboratoire des Milieux D\'esordonn\'es et H\'et\'erog\`enes,
11: CNRS UMR 7603, Universit\'{e} Paris 6,\\
12: 4, place Jussieu, 75252 Paris Cedex 05, France}
13:
14: \address{$^2$ Groupe de Physique des Solides, CNRS UMR 7588,
15: Universit\'{e}s Paris 6 et Paris 7,\\
16: 2, place Jussieu, 75251 Paris Cedex 05, France}
17: \maketitle
18:
19: \begin{abstract}
20: We report new results related to the two-time dynamics of the coordinate of a quantum free particle, damped through
21: its interaction with a fractal thermal bath (non-ohmic coupling $\sim\omega^\delta$ with $0<\delta<1$ or $1<\delta<2)$.
22: When the particle is localized, its position does not age. When it undergoes anomalous diffusion, only its displacement may be
23: defined. It is shown to be an aging variable. The finite temperature aging regime is self-similar. It is described by a
24: scaling function of the ratio ${t_w/\tau}$ of the waiting time to the observation time, as characterized by an exponent
25: directly linked to $\delta$.
26:
27: \end{abstract}
28:
29: \pacs{PACS numbers: 05.30.-d, 05.40.+j, 02.50.Ey}
30:
31: \vskip2pc]
32:
33: In complex out-of-equilibrium systems such as structural glasses and spin glasses, the trend towards equilibrium is
34: accompanied by aging effects: for instance the two-time correlation functions of certain out-of-equilibrium dynamic variables
35: may not be invariant by time translation, even in the limit of a large waiting time (or age). The fluctuation-dissipation
36: theorem, which is valid for dynamic variables at equilibrum, is not verified in this case. Actually the study of aging effects
37: and of the related violation of the fluctuation-dissipation theorem (FDT) is a fundamental problem of the physics of
38: dissipative out-of-equilibrium systems${\protect\cite{Bouchaud}}$. Several aspects of this problem still stand as largely open
39: questions, especially in the case of quantum dissipative systems.
40:
41: In order to be able to discuss these questions at any temperature, one has to work within a quantum framework, the time
42: scale ${\hbar/kT}$ playing a crucial role in the low-temperature dynamics. Since aging effects are also encountered, not
43: only in complex systems, but also in simpler systems, nor disordered nor frustrated$\protect\cite{Cugliandolo}$, it is very
44: natural, to begin with, to carry out the study of quantum aging in this latter type of systems.
45:
46: One archetype of a simple quantum dissipative system displaying aging effects is a particle coupled to a thermal bath but
47: otherwise free, that is evolving in the absence of potential. Aging effects on the displacement correlation function and
48: the corresponding violation of the quantum FDT have recently been discussed$\protect\cite{Pottier}$ for a specific model of
49: dissipation, namely the so-called ohmic model. It corresponds to a particle undergoing standard quantum Brownian
50: motion${\protect\cite{Hakim,Aslangul1}}$. The damped ohmic equation of motion can be given a non retarded form in the
51: classical limit, in which diffusion is normal. However, the ohmic model does not allow to handle all the physical situations
52: of interest, for instance those in which the damped motion is described by a truly retarded equation even in the classical
53: limit and in which either localization or anomalous diffusion phenomena are taking place.
54:
55: In this Letter, we examine aging effects in this simple system using for the dissipation a versatile model able to generate
56: various damped equations of motion, either instantaneous or retarded in the classical limit. The dissipation is introduced
57: {\sl via\/} a linear coupling of the particle to a thermal bath having a continuous distribution of modes of bandwidth
58: $\omega_c$ (Caldeira and Leggett model${\protect\cite{Caldeira,Leggett}}$). The central ingredient of the model is the product
59: of the density of modes of the bath $g(\omega)$ times the squared coupling constant
60: $|\lambda(\omega)|^2$, a product assumed to vary as $\omega^\delta$ at frequencies $\omega\ll\omega_c$. In the ohmic model, the
61: dissipative exponent $\delta$ is equal to 1. The algebraic cases $0<\delta<1$ and $\delta>1$ are known,
62: respectively, as the subohmic or superohmic dissipation models${\protect\cite{Schramm,Grabert,Aslangul2,Weiss}}$.
63:
64: In the following we first show that for $0<\delta<2$ the particle velocity equilibrates at large times and does not age. Then,
65: turning to the study of the particle coordinate, we discuss the domain of ($\delta,T$) parameters for which aging is
66: taking place, or not. Actually the question is a non-trivial one. Indeed the two following properties have been
67: demonstrated$\protect{\cite{Schramm,Grabert,Aslangul2,Weiss}}$. First, at $T=0$ for $0<\delta<1$, the particle is localized,
68: in the sense that the mean square displacement $\Delta x^2(t)=\langle[x(t)-x(0)]^2\rangle$ tends towards a constant at
69: infinite time. Second, at $T=0$ for $1\leq\delta<2$, and also at finite $T$ for $0<\delta<2$, $\Delta x^2(t)$ diverges at
70: infinite time, the diffusion being anomalous except at finite $T$ for $\delta=1$. As a result, we find clear-cut behaviors:
71: as far as the coordinate is concerned, the localized particle does not age, while the diffusing one ages. In this latter
72: situation we provide analytic expressions for the effective temperature and the associated fluctuation-dissipation ratio
73: allowing to write a modified FDT at finite temperature. We show in particular that the aging regime which emerges at large
74: times is self-similar.
75:
76: In the Caldeira and Leggett model, the particle is coupled linearly to a set of harmonic oscillators in thermal equilibrium at
77: temperature $T$. The Hamiltonian of the particle-plus-bath system reads, in obvious notations,
78: %%%%%%%%%%%%%%%%%%%%%%%
79: \begin{eqnarray}
80: \nonumber H={p^2\over 2m}-x\sum_\nu\lambda_\nu(b_\nu&+&b_\nu^\dagger)+\\
81: &&\sum_\nu\hbar\omega_\nu
82: b_\nu^\dagger b_\nu+ x^2\sum_\nu{\lambda_\nu^2\over\hbar\omega_\nu},
83: \label{hamil}
84: \end{eqnarray}
85: %%%%%%%%%%%%%%%%%%%%%%%
86: where $\lambda_\nu$ is a real coupling constant. For $\omega>0$, the quantity $2\pi\,g(\omega)\,|\lambda(\omega)|^2={1\over
87: 2}m\hbar\omega K(\omega)$ is modelized as
88: %%%%%%%%%%%%%%%%%%%%%%%
89: \begin{equation}
90: \label{modelization}
91: K(\omega)=2\gamma{\bigl({\omega\over\gamma}\bigr)}^{\delta-1}f_c\bigl({\omega\over\omega_c}\bigr),
92: \end{equation}
93: ä%%%%%%%%%%%%%%%%%%%%%%
94: where $\gamma$ is a coupling frequency and $f_c$ a high-frequency cut-off function of
95: typical width $\omega_c$. The definition of $K(\omega)$ is extended to $\omega<0$ by imposing that it must be an even function
96: of $\omega$. The particle position operator obeys the retarded equation of motion
97: %%%%%%%%%%%%%%%%%%%%%%%
98: \begin{equation}
99: \label{eqofmotion}
100: \ddot x(t)+\int_{t_i}^tdt'k(t-t')\dot x(t')=
101: -x(t_i)k(t-t_i)+{1\over m}F(t),
102: \end{equation}
103: %%%%%%%%%%%%%%%%%%%%%%%
104: where $t_i$ denotes the initial time at which the coupling is switched on. In Eq. (\ref{eqofmotion}), the inverse Fourier
105: transform~$k(t)$ of $K(\omega)$ plays the role of a memory kernel, and $F(t)$ is a linear combination of
106: bath operators, acting as a stationary random force of correlation function
107: %%%%%%%%%%%%%%%%%%%%%%%
108: \begin{equation}
109: C_{FF}(t)=m\int_{-\infty}^\infty{d\omega\over 2\pi}\,\Re {\rm e}\,\tilde K(\omega)\hbar\omega\coth{\beta\hbar\omega\over
110: 2}e^{-i\omega t}
111: \end{equation}
112: %%%%%%%%%%%%%%%%%%%%%%%
113: with $\Re {\rm e}\,\tilde K(\omega)={1\over 2}K(\omega)$ and $\beta=(k_BT)^{-1}$.
114:
115: It has been demonstrated that, for $0<\delta<2$ the total mass of the particle and of the bath oscillators
116: diverges, while for $\delta>2$ it remains finite and can be considered as a
117: renormalized mass$\protect{\cite{Schramm,Grabert,Aslangul2,Weiss}}$. For $0<\delta<2$ ($\delta\neq
118: 1$),\break the average particle velocity $\langle v(t)\rangle$ relaxes towards zero at large $t-t_i$ like $(t-t_i)^{\delta-2}$
119: (the average is taken over the variables of the initial state of the particle and over the bath variables). For
120: $\delta=1$, the relaxation is exponential. The initial value $\langle v(t_i)\rangle$ being forgotten for $0<\delta<2$, this
121: situation may in this sense be qualified of ergodic. For $\delta>2$, the dynamics is governed at large times by a kinematical
122: term involving the renormalized mass and $\langle v(t_i)\rangle$ is never
123: forgotten$\protect{\cite{Schramm,Grabert,Aslangul2,Weiss}}$.
124:
125: In the following, we limit ourselves to the ergodic case $0<\delta<2$. Then the velocity equilibrates at
126: large times and does not age. This property, already obtained in the ohmic case$\protect{\cite{Pottier}}$, thus generalizes
127: to non-ohmic models with $0<\delta<2$. The two-time velocity correlation function depends
128: only on the time difference and can be computed {\sl via\/} Fourier analysis and the Wiener-Khintchine theorem:
129: %%%%%%%%%%%%%%%%%%%%%%%
130: \begin{equation}
131: C_{vv}(t)={1\over m}\int_{-\infty}^\infty{d\omega\over 2\pi}\,{\Re{\rm
132: e}\,\tilde K(\omega)\over|\tilde K(\omega)-i\omega|^2}\hbar\omega\coth{\beta\hbar\omega\over 2}e^{-i\omega t}.
133: \end{equation}
134: %%%%%%%%%%%%%%%%%%%%%%%
135: With the modelization (\ref{modelization}) for $K(\omega)$ and a lorentzian cut-off function
136: $f_c={\omega_c^2/(\omega_c^2+\omega^2)}$, one has:
137: %%%%%%%%%%%%%%%%%%%%%%%
138: \begin{eqnarray}
139: \nonumber &\Im{\rm m}\,\tilde K(\omega)=\omega
140: {\bigl({|\omega|\over\gamma}\bigr)}^{\delta-2}&f_c\bigl({\omega\over\omega_c}\bigr)\times\\
141: &&\Bigl[\cot{\delta\pi\over
142: 2}+{\bigl({\omega_c\over|\omega|}\bigr)}^{\delta-2}{1\over\sin{\delta\pi\over 2}}\Bigr].
143: \end{eqnarray}
144: %%%%%%%%%%%%%%%%%%%%%%%
145:
146: Then, setting
147: %%%%%%%%%%%%%%%%%%%%%%%
148: \begin{equation}
149: C_{vv}(\omega)={1\over m}{\Re{\rm
150: e}\,\tilde K(\omega)\over|\tilde K(\omega)-i\omega|^2}\hbar\omega\coth{\beta\hbar\omega\over 2},
151: \end{equation}
152: %%%%%%%%%%%%%%%%%%%%%%%
153: one may attempt to define the coordinate spectral density as $C_{xx}(\omega)={C_{vv}(\omega)/\omega^2}$. If convergent,
154: the integral $\int_{-\infty}^\infty{(d\omega/2\pi)}\,C_{xx}(\omega)$ represents $\langle x^2(t)\rangle$,
155: a quantity which must be independent of $t$. Checking the small-$\omega$\break behavior of the integrand with the chosen
156: modelization for the memory kernel, one sees that this is only possible at $T=0$ for $0<\delta<1$. In this case, the particle
157: is localized and it makes sense to define its position in an absolute way as $x(t)=\int_{-\infty}^t v(t')dt'$. The two-time
158: position correlation function
159: %%%%%%%%%%%%%%%%%%%%%%%
160: \begin{equation}
161: C_{xx}(t,t')={1\over 2}\langle\{x(t),x(t')\}_+\rangle,
162: \end{equation}
163: %%%%%%%%%%%%%%%%%%%%%%%
164: where the symbol $\{.,.\}_+$ stands for the anticommutator, only depends on $t-t'$ and does
165: not age. In particular one has
166: %%%%%%%%%%%%%%%%%%%%%%%
167: \begin{equation}
168: \Delta x^2(t)=2\int_{-\infty}^\infty{d\omega\over 2\pi}\,C_{xx}(\omega)(1-\cos\omega t),
169: \end{equation}
170: %%%%%%%%%%%%%%%%%%%%%%%
171: that is, in terms of the integrated velocity correlation function $D(t)=\int_0^t du\,C_{vv}(u)$,
172: %%%%%%%%%%%%%%%%%%%%%%%
173: \begin{equation}
174: \Delta x^2(t)=\Delta x^2(\infty)-2\int_t^\infty D(t')\,dt'.
175: \end{equation}
176: %%%%%%%%%%%%%%%%%%%%%%%
177: The asymptotic behavior at large times of $D(t)$ allows
178: to characterize the relaxation of the mean square displacement towards $\Delta x^2(\infty)$.
179:
180: In other cases, that is at $T=0$ for $1\leq\delta<2$ and at finite $T$ for $0<\delta<2$, the integral
181: $\int_{-\infty}^\infty{(d\omega/2\pi)}\,C_{xx}(\omega)$ diverges and $\langle x^2(t)\rangle$ and $C_{xx}(t,t')$ are infinite.
182: The particle diffuses. Since it is then no more possible to define an absolute position, we focus the interest on the
183: displacement $x(t)-x(t_0)$ ($t\geq t_0$). This quantity does not equilibrate with the bath, even at large times: it ages. The
184: mean square displacement $\Delta x^2(t)$ diverges at infinite time and $D(t)$ represents the time-dependent diffusion
185: coefficient (in an extended sense when diffusion is anomalous, that is at $T=0$ for $1<\delta<2$, and at finite $T$ for
186: $0<\delta<1$ and $1<\delta<2$).
187:
188: Before introducing the appropriate FDT violation factor, let us study in details the
189: behavior of $D(t)$. Since $0<\delta<2$, one can restrict the study to the infinite bath bandwidth limit
190: $\omega_c\to\infty$ in which case calculations are more simple. One has:
191: %%%%%%%%%%%%%%%%%%%%%%%
192: \begin{eqnarray}
193: \label{coeffdiffusion}
194: \nonumber &D(t)=&{\hbar\over m\pi}\gamma^{\delta-2}\int_0^\infty d\omega
195: \coth{\beta\hbar\omega\over 2}\sin\omega t\times\\
196: &&\bigl[\omega^{\delta-1}+\omega^{3-\delta}(|\omega|^{\delta-2}\cot{\delta\pi\over
197: 2}-\gamma^{\delta-2})^2\bigr]^{-1}.
198: \end{eqnarray}
199: %%%%%%%%%%%%%%%%%%%%%%%
200: At finite $T$, it is interesting to discuss on the same footing the classical counterpart of $D(t)$, namely
201: $D^{\rm cl}(t)$ deduced from $D(t)$ by replacing $\coth{(\beta\hbar\omega/2)}$ by ${2/\beta\hbar\omega}$ in
202: formula~(\ref{coeffdiffusion}). Several important features of $D(t)$ and $D^{\rm cl}(t)$ can be obtained by contour
203: integration.
204:
205: At $T=0$, $D(t)$ is found to be the sum of a pole contribution, which exists only for $0<\delta<1$,
206: given by the oscillating function
207: %%%%%%%%%%%%%%%%%%%%%%%
208: \begin{equation}
209: D(t)_{\rm pole}\sim{\hbar\over m}{1\over 2-\delta}e^{-\Lambda t}\sin\Omega t,
210: \end{equation}
211: %%%%%%%%%%%%%%%%%%%%%%%
212: where $\Omega$ and $\Lambda$ are known functions of $\delta$ and $\gamma$,
213: and of a cut contribution behaving at large times as a power-law,
214: %%%%%%%%%%%%%%%%%%%%%%%
215: \begin{equation}
216: \label{Dasympt}
217: D(t)_{\rm cut}\sim{\hbar\over m\pi}(\gamma t)^{\delta-2}\sin^3{\delta\pi\over
218: 2}\Gamma(2-\delta),
219: \end{equation}
220: %%%%%%%%%%%%%%%%%%%%%%%
221: where $\Gamma$ denotes the Euler Gamma function.
222:
223: At finite $T$, $D^{\rm cl}(t)$ is also found to be the sum of an
224: oscillating function, which exists only for $0<\delta<1$,
225: %%%%%%%%%%%%%%%%%%%%%%%
226: \begin{equation}
227: D^{\rm cl}(t)_{\rm pole}\sim{kT\over m\gamma}{2\over
228: 2-\delta}{\bigl(\sin{\delta\pi\over 2}\bigr)}^{1\over 2-\delta}e^{-\Lambda t}\sin(\Omega t-\phi)
229: \end{equation}
230: %%%%%%%%%%%%%%%%%%%%%%%
231: with $\phi={\pi\delta/2(2-\delta)}$, and of a cut contribution behaving at large times as a power-law,
232: %%%%%%%%%%%%%%%%%%%%%%%
233: \begin{equation}
234: \label{Dclasympt}
235: D^{\rm cl}(t)_{\rm cut}\sim{kT\over m\gamma}(\gamma t)^{\delta-1}{\sin{\delta\pi\over 2}\over\Gamma(\delta)}.
236: \end{equation}
237: %%%%%%%%%%%%%%%%%%%%%%%
238:
239: The behaviors of $D(t)$ and $D^{\rm cl}(t)$ at several different temperatures are illustrated on Fig.~\ref{Figure1} for
240: $\delta=0.5$ and on Fig.~\ref{Figure2} for $\delta=1.5$. Interestingly enough, for any given~$\delta$, the curves corresponding
241: to different bath temperatures do not intersect. Actually, it can be shown that, at any fixed time $t$, $D(t)$, like $D^{\rm
242: cl}(t)$, is a monotonously increasing function of $T$.
243:
244: For times $t\ll t_{\rm th}$ ($t_{\rm th}={\hbar/2\pi k_BT}$) and for any value of $\delta$, the curves for $D(t)$
245: at finite $T$ nearly coincide with those at $T=0$, as it should.
246:
247: At intermediate times, and for $0<\delta<1$, an oscillation due to the
248: pole contribution takes place in $D(t)$ (and also in $D^{\rm cl}(t)$ but with a smaller amplitude). For certain values of
249: $\delta$ and $T$, this oscillation may even result in negative values of $D(t)$ during a finite time interval.
250:
251: At large times
252: $t\gg\gamma^{-1},t_{\rm th}$ and for any finite $T$, the curves of $D(t)$ and $D^{\rm cl}(t)$ join together: when
253: \hbox{$0<\delta<1$}, $D(t)$ describes a subdiffusive regime, and, when $\delta>1$, a superdiffusive
254: one. At $T=0$, for $0<\delta<1$, $D(t)$ describes the relaxation of $\Delta x^2(t)$ towards $\Delta x^2(\infty)$
255: and, for $1<\delta<2$, a subdiffusive regime.
256: %%%%%%%%%%%%%%%%%%%%%%%
257: \begin{figure}
258: \vspace{-4mm}
259: \epsfxsize=\columnwidth
260: \centerline{\epsffile{Figure1.eps}}
261: \vspace{0mm}
262: \caption{\label{Figure1}
263: In full lines, $D(t)$ plotted as a function of $\gamma t$ for
264: $\delta=0.5$ and for bath temperatures $T=0$, $k_BT={\hbar\gamma/2\pi}$, $k_BT={\hbar\gamma/\pi}$ (at
265: $T=0$, $D(t)$ is not a diffusion coefficient, but characterizes the relaxation of $\Delta x^2(t)$ towards its limit value). In
266: dashed lines, the corresponding $D^{\rm cl}(t)$.}
267: \end{figure}
268: %%%%%%%%%%%%%%%%%%%%%%%
269:
270: %%%%%%%%%%%%%%%%%%%%%%%
271: \begin{figure}
272: \vspace{-4mm}
273: \epsfxsize=\columnwidth
274: \centerline{\epsffile{Figure2.eps}}
275: \vspace{0mm}
276: \caption{\label{Figure2}
277: In full lines, the quantum diffusion coefficient $D(t)$ plotted as a function of $\gamma t$ for
278: $\delta=1.5$ and for bath temperatures $T=0$, $k_BT={\hbar\gamma/2\pi}$, $k_BT={\hbar\gamma/\pi}$. In dashed lines, the
279: corresponding
280: $D^{\rm cl}(t)$.}
281: \end{figure}
282: %%%%%%%%%%%%%%%%%%%%%%%
283:
284: These features of $D(t)$, especially the large times ones, are essential for describing the aging properties of the
285: displacement correlation function, as defined by
286: %%%%%%%%%%%%%%%%%%%%%%%
287: \begin{equation}
288: C_{xx}(t,t';t_0)={1\over 2}\langle\{[x(t)-x(t_0)],[x(t')-x(t_0)]\}_+\rangle.
289: \end{equation}
290: %%%%%%%%%%%%%%%%%%%%%%%
291:
292: In the classical case, a modified FDT can be written
293: as$\protect{\cite{Bouchaud,Cugliandolo}}$
294: %%%%%%%%%%%%%%%%%%%%%%%
295: \begin{equation}
296: \chi_{xx}(t,t')=\beta\Theta(t-t')X^{\rm cl}(t,t';t_0){\partial C_{xx}(t,t';t_0)\over\partial t'},
297: \end{equation}
298: %%%%%%%%%%%%%%%%%%%%%%%
299: where $\chi_{xx}(t,t')$ is the displacement response function. For a diffusing particle, the fluctuation-dissipation
300: ratio~$X^{\rm cl}(t,t';t_0)$ can be obtained from $D^{\rm cl}(\tau)$ and $D^{\rm
301: cl}(t_w)$$\protect{\cite{Pottier}}$, where $\tau=t-t'$ denotes the observation time and $t_w=t'-t_0$ the waiting time:
302: %%%%%%%%%%%%%%%%%%%%%%%
303: \begin{equation}
304: \label{X}
305: X^{\rm cl}(\tau,t_w)={D^{\rm cl}(\tau)\over D^{\rm cl}(\tau)+D^{\rm cl}(t_w)}.
306: \end{equation}
307: %%%%%%%%%%%%%%%%%%%%%%%
308: For any $\tau$ and $t_w$, one can define an effective inverse temperature as $\beta_{\rm eff}^{\rm cl}(\tau,t_w)=\beta X^{\rm
309: cl}(\tau,t_w)$. Since $X^{\rm cl}$ does not depend on $T$, the bath temperature is rescaled by a factor
310: ${1/X^{\rm cl}}$ larger than 1, due to those fluctuations of the particle displacement which take place during the
311: waiting time. At large times ($\tau,t_w\gg\gamma^{-1},t_{\rm th}$), one can use in formula (\ref{X}) the asymptotic
312: expressions of $D^{\rm cl}(\tau)$ and $D^{\rm cl}(t_w)$ as given by Eq. (\ref{Dclasympt}). Eq. (\ref{X}) then displays the
313: fact that, in a subohmic or superohmic model of exponent $\delta$ ($0<\delta<1$ or $1<\delta<2$), a self-similar aging regime
314: takes place at large times, as pictured by
315: %%%%%%%%%%%%%%%%%%%%%%%
316: \begin{equation}
317: \label{selfsimilar}
318: X^{\rm cl,ag}(\tau,t_w)={1\over 1+{\bigl({t_w\over\tau}\bigr)}^{\delta-1}}.
319: \end{equation}
320: %%%%%%%%%%%%%%%%%%%%%%%
321: Interestingly enough, $X^{\rm cl,ag}$ and $T_{\rm eff}^{\rm
322: cl,ag}=(k_B\beta_{\rm eff}^{\rm cl,ag})^{-1}$ are functions of ${t_w/\tau}$, solely parametrized by $\delta$. They do not
323: depend on the other parameters of the model ({\sl i.e.\/} $\gamma$ or $\omega_c$). For
324: $\delta=1$, one retrieves the results $X^{\rm cl,ag}={1/2}$ and $T_{\rm eff}^{\rm
325: cl,ag}=2T$$\protect{\cite{Cugliandolo,Pottier}}$. For any other value of $\delta$, $X^{\rm cl,ag}$ and $T_{\rm eff}^{\rm
326: cl,ag}$ are algebraic functions of ${t_w/\tau}$. The limits $\tau\to\infty$ and $t_w\to\infty$ do not commute.
327:
328: In the quantum case, the effective temperature $T_{\rm eff}=(k\beta_{\rm eff})^{-1}$ can be obtained
329: from the following equation$\protect{\cite{Pottier}}$:
330: %%%%%%%%%%%%%%%%%%%%%%%
331: \begin{equation}
332: \label{Teff}
333: D_{T_{\rm eff}}(\tau)=D(\tau)+D(t_w).
334: \end{equation}
335: %%%%%%%%%%%%%%%%%%%%%%%
336: Eq.~(\ref{Teff}) also allows to define $T_{\rm eff}$ at $T=0$ for $1\leq\delta<2$.
337: Since $D(t)$ is a monotonously increasing function of $T$, Eq.~(\ref{Teff}) yields for
338: $T_{\rm eff}(\tau,t_w)$ a uniquely defined value.
339:
340: The curves representing $\beta_{\rm eff}(\tau,t_w)$ as a function of~$\tau$ for $\delta=0.5$ and $\delta=1.5$ at a given
341: finite temperature and for a given $t_w\gg\gamma^{-1},t_{\rm th}$ are plotted on Fig.~\ref{Figure3}.
342: Quantum effects do not persist beyond times $\tau\sim t_{\rm th}$. Thus, for times \hbox{$\tau\gg{\gamma}^{-1},t_{\rm th}$,}
343: $X^{\rm cl,ag}(\tau,t_w)$ as given by formula~(\ref{selfsimilar}) allows for a proper description of aging effects.
344:
345: In conclusion, we have shown that the two-time dynamics of the coordinate of a quantum dissipative free particle coupled to a
346: thermal bath (coupling $\sim\omega^\delta$ with $0<\delta<2$) displays extremely rich behaviors. According to the values of
347: $T$ and $\delta$, one may find either a localized regime in which the position can be defined in an absolute way and does not
348: age ($T=0,\,0<\delta<1$), or (possibly anomalously) diffusing ones in which it only makes sense to consider the displacement,
349: which displays aging ($T=0,\,1\leq\delta<2$, or $T$ finite, $0<\delta<2$). The aging regime at finite $T$ is properly
350: described by the large times expression of the classical fluctuation-dissipation ratio, which for $\delta\neq 1$ is a
351: self-similar function of
352: $t_w/\tau$, solely parametrized by $\delta$.
353:
354: %%%%%%%%%%%%%%%%%%%%%%%
355: \begin{figure}
356: \vspace{-2mm}
357: \epsfxsize=\columnwidth
358: \centerline{\epsffile{Figure3.eps}}
359: \vspace{0mm}
360: \caption{\label{Figure3}
361: In full lines, the effective inverse temperature~$\beta_{\rm eff}$, as computed from Eq.~(\ref{Teff}), plotted as a
362: function of $\gamma\tau$ for a bath temperature $k_BT={\hbar\gamma/2\pi}$ and two different values of $\delta$. In dashed
363: lines, the corresponding classical effective inverse temperature $\beta_{\rm eff}^{\rm cl,ag}$, as deduced from
364: Eq.~(\ref{selfsimilar}).}
365: \end{figure}
366: %%%%%%%%%%%%%%%%%%%%%%%
367: \vspace{-9mm}
368: \begin{thebibliography}{}
369:
370: \bibitem{Bouchaud}
371: J.-P. Bouchaud, L.F. Cugliandolo, J. Kurchan and M.~M\'ezard, in {\em Spin glasses and random fields}, edited by A.P. Young
372: editor (World Scientific, Singapore, 1997).
373:
374: \bibitem{Cugliandolo}
375: L.F. Cugliandolo, J. Kurchan and G. Parisi, J. Phys. {\bf 4}, 1641 (1994).
376:
377: \bibitem{Pottier}
378: N. Pottier and A. Mauger, Physica A {\bf 282}, 77 (2000).
379:
380: \bibitem{Hakim}
381: V. Hakim and V. Ambegaokar, Phys. Rev. A {\bf 32}, 423 (1985).
382:
383: \bibitem{Aslangul1}
384: C. Aslangul, N. Pottier and D. Saint-James, J. Stat. Phys. {\bf 40}, 167 (1985).
385:
386: \bibitem{Caldeira}
387: A. Caldeira and A.J. Leggett, Ann. Phys. {\bf 149}, 374 (1983).
388:
389: \bibitem{Leggett}
390: A.J. Leggett, S. Chakravarty, A.T. Dorsey, M.P.A. Fisher, A. Garg and W. Zwerger, Rev. Mod. Phys. {\bf 59}, 1 (1987).
391:
392: \bibitem{Schramm}
393: P. Schramm and H. Grabert, J. Stat. Phys. {\bf 49}, 767 (1987).
394:
395: \bibitem{Grabert}
396: H. Grabert, P. Schramm and G.-L. Ingold, Phys. Rep. {\bf 168}, 115 (1988).
397:
398: \bibitem{Aslangul2}
399: C. Aslangul, N. Pottier and D. Saint-James, J. Physique {\bf 48}, 1871 (1987).
400:
401: \bibitem{Weiss}
402: U. Weiss, {\em Quantum dissipative systems\/}, Second Edition, World Scientific, Singapore, 1999.
403:
404:
405: \end{thebibliography}
406:
407: \end{document}
408:
409: \end
410: