1: %\documentstyle[aps,prl,multicol,psfig,preprint]{revtex}\def\tc{no}
2: \documentstyle[aps,prl,multicol,psfig]{revtex}\def\tc{yy}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \def\inodot{\char'020}
5: \def\<{\langle}
6: \def\>{\rangle}
7: \def\({\left (}
8: \def\){\right )}
9: \def\[{\left [}
10: \def\]{\right ]}
11: \def\d{{\rm d}}
12: \def\i{{\rm i}}
13: \def\e{{\rm e}}
14: \def\Tr{{\rm tr}}
15: \def\fr{f_{\rm reg}}
16: %\newcommand{\no}[1]{}
17: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
18: \begin{document}
19: \title{Classical and Quantum Hamiltonian Ratchets}
20: \author{Holger Schanz,$^{1}$ Marc-Felix Otto,$^{1}$ Roland Ketzmerick,$^{1}$
21: and Thomas Dittrich$^{2}$}
22: \address{$^{1}$Max-Planck-Institut f\"ur Str\"omungsforschung und Institut
23: f{\"u}r Nichtlineare Dynamik\\ der Universit{\"a}t G{\"o}ttingen,
24: Bunsenstra\ss e 10, 37073 G\"ottingen, Germany}
25: \address{$^{2}$Departamento de F\'\inodot sica, Universidad Nacional,
26: Santaf\'e de Bogot\'a, Colombia}
27: \date{June 10, 2001}
28: \maketitle
29: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
30: \begin{abstract}
31: We explain the mechanism leading to directed chaotic transport in Hamiltonian
32: systems with spatial and temporal periodicity. We show that a mixed phase
33: space comprising both regular and chaotic motion is required and derive a
34: classical sum rule which allows to predict the chaotic transport velocity from
35: properties of regular phase-space components. Transport in quantum
36: Hamiltonian ratchets arises by the same mechanism as long as uncertainty
37: allows to resolve the classical phase-space structure. We derive a quantum
38: sum rule analogous to the classical one, based on the relation between quantum
39: transport and band structure.
40: \end{abstract}
41: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
42: \pacs{05.60.-k, 05.45.Mt}
43: \if\tc\begin{multicols}{2}\fi
44: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
45: Stimulated by the biological task of explaining the functioning of molecular
46: motors, the study of ratchets \cite{Feynman} has widened to a general
47: exploration of ``self-organized'' transport, i.e., transport without external
48: bias, in nonlinear systems \cite{JAP97}. Along with this process, there has
49: been a tendency to reduce the models under investigation from realistic
50: biophysical machinery to the minimalist systems customary in nonlinear
51: dynamics. External noise, for example, which originally served to account for
52: the fluctuating environment of molecular motors, has been replaced by
53: deterministic chaos. This required to include inertia terms in the equations
54: of motion, thus leaving the regime of overdamped dynamics and leading to
55: deterministic inertia ratchets with dissipation \cite{JKH96,Mat00}. It is then
56: a consequent but radical step to abandon friction altogether. Indeed,
57: transport in \emph{Hamiltonian ratchets} was observed numerically if all
58: symmetries were broken that generate to each trajectory a countermoving
59: partner \cite{FYZ00,GH00}.
60:
61: As a parallel development, the desire to realize ratchets in artificial,
62: nanostructured electronic systems, required to consider quantum effects
63: \cite{RGH97,GH00}. \emph{Quantum Hamiltonian ratchets}, however, have been
64: studied only in the framework of one-band systems where no transport occurs
65: \cite{GH00}.
66:
67: In this paper we explain how a Hamiltonian ratchet works. We rely on methods
68: which---although well established in studies of deterministic dynamics---have
69: never before been applied to ratchets. We derive a classical and an analogous
70: quantum sum rule for transport allowing the following conclusions: (i)
71: Directed transport is a property associated with individual invariant sets of
72: the dynamics. A necessary condition for non-zero transport is a mixed phase
73: space with coexisting regular and chaotic regions. (ii) Transport in {\em
74: chaotic} regions can be described quantitatively by using topological and
75: further properties of adjacent {\em regular} regions only. (iii) Quantum
76: transport persists for all times and approaches the classical transport when
77: $\hbar$ is small compared to the major invariant sets of the classical phase
78: space.
79:
80: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
81: \def\figI{
82: \begin{figure}[tb]
83: \centerline{\psfig{figure=fig1.eps,width=\fw}} \caption{\label{figI} (a)
84: Spatial distribution $P(x)$ of a continuously driven system
85: \protect\cite{FYZ-Parameters} after $20,000$ time periods showing the
86: directed transport in a Hamiltonian ratchet. Initially, $10^{4}$
87: trajectories were started at random on the line $p=0$, $x\in [0,1)$ in the
88: chaotic sea. (b) Poincar\'e section $p$ vs $x$ of a unit cell at integer
89: times showing the main chaotic sea, the upper and lower limiting KAM-tori,
90: and the major embedded regular islands. (c,d) As (a,b), but for the kicked
91: Hamiltonian (\protect\ref{KickH}) showing a much more pronounced directed
92: transport.}
93: \end{figure}
94: }
95: \if\tc{\def\fw{78mm}\figI}\fi
96: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
97:
98: We consider a Hamiltonian of the form $H(x,p,t)=T(p)+V(x,t)$, where $T(p)$ is
99: the kinetic energy. The force $-V'$ is periodic in space and time,
100: $V'(x+1,t)=V'(x,t+1)=V'(x,t)$, and has zero mean $\int_{0}^{1}{\rm
101: d}t\,\int_{0}^{1}{\rm d}x\,V'(x,t)=0$. Usually directed transport is
102: demonstrated by following selected trajectories over very long times
103: \cite{FYZ00,GH00} or an ensemble of trajectories which generates spatial
104: distributions as shown in Fig.~\ref{figI}a,c. While this is easily implemented
105: numerically, it gives no clue about the origin of the transport (but see
106: Ref.~\cite{DF01}). Instead, we shall exploit the periodicity of the dynamics
107: with respect to space and time and analyze transport in terms of the invariant
108: sets of phase space, \emph{reduced to the spatio-temporal unit cell} $x,t\in
109: [0,1)$. For any finite invariant set $M$ we define ballistic transport as
110: phase-space volume times average velocity expressed as
111: \begin{equation}\label{transport}
112: \tau_{M}=
113: \int_{0}^{1}{\rm d}t\,
114: \int_{0}^{1}{{\rm d}x}\,
115: \int_{-\infty}^{+\infty}{\rm d}p\,
116: \chi_{M}(x,p,t)\,{\partial H\over\partial p}\,,
117: \end{equation}
118: where $\chi_{M}$ is the characteristic function of $M$. Transport is additive
119: for the union of two or more disjunct invariant sets, i.e., for
120: $M=\bigcup_{i} M_{i}$, with $M_i \cap M_j = \emptyset$ for all $i \neq j$, we
121: have
122: \begin{equation}\label{SumRule}
123: \tau_{M}=\sum_{i}\tau_{M_{i}}\,.
124: \end{equation}
125: This sum rule for Hamiltonian transport has far-reaching consequences to be
126: discussed in the following.
127:
128: For a generic Hamiltonian system, phase space is mixed and comprises an
129: infinite number of minimal invariant sets of different types. For the sake of
130: definiteness we will restrict the following discussion to the most interesting
131: case of a chaotic region containing embedded regular islands
132: (Fig.~\ref{figI}b). In any of these invariant sets the time-averaged velocity
133: $v_i$ is the same for almost all initial conditions (assuming ergodicity for
134: chaotic components). Hence, for a chaotic region, $\tau_{\rm ch}=A_{\rm
135: ch}\,v_{\rm ch}$ with $A_{\rm ch}$ denoting its area in a stroboscopic
136: Poincar\'{e} section. For an embedded island we have $\tau_i=A_i\,v_i$ where
137: $A_i$ includes the areas of the narrow chaotic layers inside the island and of
138: the infinite hierarchy of island chains surrounding it because all these
139: invariant sets share the same mean velocity. This velocity $v_{i}$ is
140: identical to the rational winding number $w_{i}=x_{i}/t_{i}$ of the stable
141: fixed point at the center of the island. In extended phase space, this
142: corresponds to a shift of the island by $x_{i}$ spatial after $t_{i}$ time
143: periods. Typically, the chaotic set is bounded from above and below by two
144: non-contractible KAM-tori $p_{a,b}(x,t)$ enclosing the spatial unit
145: cell. Treating the phase-space region in between as the global invariant set
146: $M$ appearing on the l.h.s.\ of Eq.~(\ref{SumRule}), its transport $\tau_{M}$
147: is obtained from Eq.~(\ref{transport}) as $\<T\>_{{a}}-\<T\>_{{b}}$ with
148: $\<T\>_{{a,b}}=\int_{0}^{1}{\rm d}t\,\int_{0}^{1}{\rm d}x\,T(p_{a,b}(x,t))$
149: denoting averages of the kinetic energy over the tori. Using the sum rule
150: (\ref{SumRule}) we can now express transport of the chaotic region in terms of
151: its adjoining regular components (KAM-tori and islands) as
152: \begin{equation}\label{ChTr}
153: A_{\rm ch}v_{\rm ch}=\<T\>_{{a}}-\<T\>_{{b}}-\sum_{i}A_i\,w_{i}\,.
154: \end{equation}
155: This is our main result on classical transport in Hamiltonian ratchets. Not
156: only does Eq.~(\ref{ChTr}) provide an efficient method to determine the
157: chaotic drift velocity, it also expresses the simple principle generating
158: directed ballistic motion: Decomposing phase space into different invariant
159: sets, these will in general have average velocities different from each other
160: and also different from zero but related by the sum rule
161: (\ref{SumRule}). Therefore a necessary condition for directed chaotic
162: transport in Hamiltonian ratchets is \emph{a mixed phase space}.
163:
164: L\'evy flights \cite{Theo} are a characteristic feature of chaotic motion in a
165: generic mixed phase space and indeed they were observed in Hamiltonian
166: ratchets \cite{FYZ00,GH00}. They reflect the slow exchange between subsets of
167: a chaotic region, separated by leaky barriers \cite{Theo}. As these subsets
168: are not invariant, their contributions to Eq.~(\ref{SumRule}) are contained in
169: the contribution of the chaotic invariant set. L\'evy flights lead to
170: power-law tails in spatial distributions. For example, the asymmetric shapes
171: of the peaks visible in Figs.~\ref{figI}a,c can be attributed to these tails.
172: Notably, in Fig.~\ref{figI}c one clearly sees a mean transport to the right
173: although the same data shows no indication of a power-law tail in this
174: direction. We stress that the sum rule allows to predict the mean velocity of
175: chaotic trajectories without any reference to such details of the chaotic
176: dynamics. This suggests that L\'evy flights are not a necessary element of the
177: mechanism of chaotic transport in Hamiltonian ratchets.
178:
179: In Ref.~\cite{FYZ00} it was shown that a necessary condition for directed
180: transport is the breaking of all symmetries which to each trajectory generate
181: a countermoving partner. For a chaotic set invariant under such a
182: symmetry, this is in agreement with Eq.~(\ref{ChTr}) because then the r.h.s.\
183: vanishes identically. However, chaotic sets can also occur as
184: symmetry-related pairs transporting in opposite directions. Moreover, if
185: phase space cannot be decomposed into invariant subsets, e.g., for an
186: ergodic system, there cannot be transport even with all symmetries broken.
187:
188: Up to now we have only considered transport of invariant sets of the unit
189: cell. For an arbitrary initial distribution transport is determined by
190: projection onto these invariant sets \cite{D+00}. Therefore, the location of
191: an initial distribution \emph{within an invariant set} is irrelevant. This
192: applies also to the location within the temporal unit cell, i.e., to the
193: question of phase dependence discussed in \cite{YFR00}. In particular, in case
194: that the plane $p=0$, $0\le x,t\le 1$ is completely within the chaotic
195: invariant set, any initial condition restricted to this plane will result in
196: the same average transport. We now understand how a Hamiltonian ratchet makes
197: particles initially at rest ($p=0$) move with a predetermined mean velocity
198: as, e.g., in Fig.~1a.
199:
200: We have checked Eq.~(\ref{ChTr}) numerically for a continuously driven system
201: \cite{FYZ-Parameters}. We determined the areas $A_{i}$ and winding numbers
202: $w_{i}$ for the regular islands shown in the Poincar\'e section of
203: Fig.~\ref{figI}b as well as $\<T_{a}\>$ and $\<T_{b}\>$ for the limiting
204: KAM-tori, yielding $v_{\rm ch}=0.092\pm 0.011$. The error estimate includes
205: the uncertainty in the location of the bounding KAM-tori and the contribution
206: from neglected small islands. The result is in agreement with the value
207: $v_{\rm ch}=0.082\pm 0.002$ determined with much more computational effort
208: from the spatial distribution of $10^{4}$ trajectories, started with $p=0$
209: (Fig.~\ref{figI}a).
210: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
211:
212: As a minimal model for directed chaotic transport in Hamiltonian ratchets,
213: we propose a kicked Hamiltonian
214: \begin{equation}\label{KickH}
215: H(x,p,t)=T(p)+V(x)\,\sum_{n}\delta(t-n).
216: \end{equation}
217: It reduces the dynamics to a map for position and momentum $x_{n+1} =
218: x_n+T'(p_{n})$, $p_{n+1} = p_n-V'(x_{n+1})$, just after the kick. As an
219: example we take a symmetric potential $V(x)=(x\,{\rm mod}\,1\,-1/2)^{2}/2$ and an
220: asymmetric kinetic energy $T(p)=|p|+3\sin(2\pi p)/(4\pi^{2})$.
221: We consider the dynamics on a cylinder with transport along the $x$-axis and
222: $p\in [-1/2,+1/2)$ being a periodic variable. Fig.~\ref{figI}d shows the
223: Poincar\'e section for one unit cell. There are only two major invariant
224: sets---a chaotic sea and a regular island centered around a periodic orbit
225: with winding number $w_{\rm reg}=-1$. According to Eq.~(\ref{transport}),
226: transport of the full phase space vanishes identically because of the periodic
227: momentum variable. Applying the sum rule (\ref{SumRule}) the contributions to
228: transport from the two invariant sets cancel exactly,
229: \begin{equation}\label{krsum}
230: A_{\rm ch}\,v_{\rm ch}+A_{\rm reg}\,w_{\rm reg}=0\,.
231: \end{equation}
232: We find the transport velocity of the chaotic component as $v_{\rm ch}=f_{\rm
233: reg}/(1-f_{\rm reg})$, where $f_{\rm reg}=A_{\rm reg}/(A_{\rm reg}+A_{\rm
234: ch})$ denotes the relative area of the regular island. From Fig.~\ref{figI}d,
235: $f_{\rm reg}=0.117\pm 0.001$, thus $v_{\rm ch}=0.133\pm 0.001$ in agreement
236: with $v_{\rm ch}=0.1344\pm 0.0003$ from the spatial distribution of
237: Fig.~\ref{figI}c.
238: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
239: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
240: \def\figII{
241: \begin{figure}
242: \centerline{ \psfig{figure=fig2.eps,width=\fw}\hspace*{3mm} }
243: \caption{\label{figII} Mean position vs time for a wavepacket in system
244: (\protect\ref{KickH}) initialized as the momentum eigenstate with $p=0$ for
245: various values of $h$ (full lines). For decreasing $h$ the classical
246: prediction $v_{\rm ch}\,t$ (dashed line) is approached.}
247: \end{figure}
248: }
249: %\if\tc{\def\fw{79mm}\figII}\fi
250: \if\tc{\def\fw{60mm}\figII}\fi
251: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
252: In order to extend our concept of directed transport in Hamiltonian systems to
253: quantum ratchets, we first demonstrate by a numerical example that quantum
254: Hamiltonian ratchets can work. Fig.~\ref{figII} shows that the average
255: velocity of a wavepacket initialized in the chaotic sea varies between $0$ for
256: large and the classical value $v_{\rm ch}$ for small values of $\hbar$. We
257: explain this behavior in the following.
258:
259: In analogy with our approach to classical transport, we consider the
260: invariants of the quantum dynamics, the stationary states of the
261: time-evolution operator over one period, i.e., $\hat U=\e^{-\i \hat
262: V/\hbar}\e^{-\i \hat T/\hbar}$ for the kicked Hamiltonian
263: Eq.~(\ref{KickH}). They satisfy $\hat U|\phi_{\alpha,k}\>=\exp(-2\pi
264: \i\epsilon_{\alpha}(k))\,|\phi_{\alpha,k}\>$, with the quasienergy
265: $\epsilon_{\alpha}(k)\in[0,1)$ \cite{Sam73}. Similarly, spatial periodicity
266: implies $\phi_{\alpha,k}(x+1,t)=\exp(2\pi\i k)\,\phi_{\alpha,k}(x,t)$ with
267: quasimomentum $k\in[0,1)$ where $h$ is chosen rational for systems periodic in
268: $p$. Quantum transport is related to the expectation values in the stationary
269: states $\overline v_{\alpha,k}\equiv\langle\langle{\phi_{\alpha,k}|\hat
270: v|\phi_{\alpha,k}}\rangle\rangle$ of the velocity operator $\hat v=\hat
271: T'(\hat p)$, where $\langle\langle\cdot\rangle\rangle=\int_{0}^{1}{\rm
272: d}x\int_{0}^{1}{\rm d}t(\cdot)$. Using a generalization of the Hellman-Feynman
273: theorem to time-periodic systems \cite{Sam73}, we express velocities by
274: \emph{band slopes} as
275: \begin{eqnarray}\label{hell}
276: \overline v_{\alpha,k}={\rm d}\epsilon_{\alpha}(k)/{\rm d}k\,.
277: \end{eqnarray}
278: This allows to discuss quantum transport in terms of spectral properties.
279: Examples for quasienergy band spectra are shown in Fig.~\ref{figIII} together
280: with the corresponding velocity distributions. The semiclassical regime is
281: characterized by the existence of two different types of bands and
282: corresponding eigenstates \cite{KMG94,K+00}: Bands pertaining to regular
283: states appear as straight lines in the spectrum, while the chaotic bands show
284: oscillations and wide avoided crossings among themselves. Associating the
285: terms chaotic and regular with the bands is supported by the Husimi
286: representations of the corresponding eigenfunctions (insets in
287: Fig.~\ref{figIII}a). The new aspect introduced into this picture by directed
288: chaotic transport is the \emph{overall slope of the chaotic bands}.
289: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
290: \def\figIII{
291: \begin{figure}
292: \centerline{\psfig{figure=fig3.eps,width=\fw}}
293: \caption{\label{figIII}
294: Distribution of quantum velocities $\overline v$ obtained according to
295: Eq.~(\protect\ref{hell}) for (a) $h=1/128$ and (b) $h=1/4$. The right insets
296: show band spectra for $h=1/32$ and $h=1/4$, respectively. The smaller insets
297: are Husimi representations of characteristic wave functions together with
298: the border of the classical regular island. In (a) the regular and the
299: chaotic wave functions can be associated with the two peaks of $P(\overline
300: v)$ centered around $v_{\rm reg}=-1$ and $v_{\rm ch}$.}
301: \end{figure}
302: }
303: %\if\tc{\def\fw{79mm}\figIII}\fi
304: \if\tc{\def\fw{85mm}\figIII}\fi
305: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
306:
307: Only on a coarse quasienergy scale, the two sets of bands appear to cross. On
308: a sufficiently fine scale, all crossings are avoided. Consequently the actual
309: bands change their character between regular and chaotic at each of the narrow
310: crossings and have {\em no} overall slope. Switching from the latter
311: (``adiabatic'') to the former (``diabatic'') viewpoint is a well-controlled
312: procedure \cite{KMG94}. Formally, this behavior of the bands can be described
313: in terms of their winding number (average slope) with respect to the periodic
314: $(\epsilon,k)$-space: In the adiabatic as well as in the diabatic case, all
315: quasienergy bands must close after an integer number of periods in the
316: $\epsilon$ and $k$ directions, so that their winding number $\overline w$ must
317: be rational. Clearly, in the adiabatic case, all winding numbers are
318: zero. Going from the adiabatic to the diabatic case amounts to a mere
319: reconnection of bands at the crossings, preserving the sum of winding
320: numbers. Thus it must be zero also in the diabatic representation,
321: \begin{equation}\label{qsum}
322: \sum_\alpha \overline w_\alpha^{\rm (ch)} + \sum_\alpha \overline
323: w_\alpha^{\rm (reg)}=0\,.
324: \end{equation}
325: This is the quantum-mechanical analogue of the classical sum rule
326: (\ref{krsum}). Because of the localization of the regular states on tori
327: inside the regular island, the winding number of the regular bands in
328: $(\epsilon,k)$-space is in the semiclassical limit identical to the winding
329: number in $(x,t)$-space of the central periodic orbit, i.e., $\overline
330: w_\alpha^{\rm (reg)}=w_{\rm reg}$. Moreover, in this limit, the fractions of
331: regular and chaotic bands correspond to the relative phase-space volumes
332: $f_{\rm reg}$ and $1-f_{\rm reg}$, respectively. We therefore obtain from
333: Eq.~(\ref{qsum}) the mean slope of the chaotic bands, $\overline w^{\rm
334: (ch)}=f_{\rm reg}/(1-f_{\rm reg})=v_{\rm ch}$, as the classical drift
335: velocity. This is confirmed in Fig.~\ref{figIII}a.
336:
337: The asymptotic quantum transport velocity for a given initial wavepacket
338: $|\psi\rangle$ is an average of band slopes weighted with the overlaps
339: $|\langle\psi|\phi_{\alpha,k}\rangle|^2$.
340: We can now explain our observations in Fig.~\ref{figII}: For $\hbar\ll A_{\rm
341: reg}$, an initial wavepacket prepared in the chaotic region of a single unit
342: cell of the extended system is a superposition of chaotic eigenfunctions from
343: the entire band spectrum. Consequently, its drift velocity is given by the
344: mean slope of the chaotic bands and thus by the classical value $v_{\rm
345: ch}$. In contrast, for $\hbar\gg A_{\rm reg}$, there are no states restricted
346: to the regular or the chaotic set and hence quantum transport does not
347: correspond to classical transport in this regime.
348: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
349: \def\figIV{
350: \begin{figure}
351: \centerline{\psfig{figure=fig4.eps,width=\fw}} \caption{\label{figIV} Form
352: factor $K(x,t)$ for $h=1/128$ (thick line) at (a) $t=30$ and (b) $t=111$, the
353: Heisenberg time of the chaotic component. The two distinct peaks centered
354: around $v_{\rm reg}$ and $v_{\rm ch}$ are the fingerprints of directed
355: transport. The time-dependent width of the chaotic peak follows the classical
356: distribution $P^{\rm (ch)}(x,t)$ of chaotic trajectories (thin line).}
357: \end{figure}
358: }
359: %\if\tc{\def\fw{79mm}\figIV}\fi
360: \if\tc{\def\fw{60mm}\figIV}\fi
361: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
362:
363: Our analysis based on the winding numbers can be applied to predict the mean
364: quantum transport in the semiclassical regime from the classical value. The
365: band spectra, however, encode more detailed information about quantum
366: transport. It can be extracted by a double Fourier transform $\epsilon \to t$,
367: $k \to x$ (discrete position in units of the spatial period) and subsequent
368: squaring of the spectral density, translating two-point correlations in the
369: bands into the entire time evolution of the spatial distribution on the scale
370: of the spatial period. A formal definition of the resulting \emph{generalized
371: form factor} $K(x,t)$ and further details are found in \cite{D+98}. A
372: semiclassical theory for the form factor in Hamiltonian ratchets, which will
373: be published elsewhere, requires to account for the simultaneous presence of
374: regular and chaotic regions in phase space. It relates the form factor
375: $K(x,t)$ to the respective contributions of these invariant sets to the
376: classical spatio-temporal distribution $P(x,t)$ (Fig.~4).
377:
378: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
379: We benefitted from discussions with S.~Flach, P.~H{\"a}nggi, M.~Holthaus and
380: O.~Yevtushenko. MFO acknowledges financial support from the Volks\-wa\-gen
381: foundation. TD thanks for the hospitality enjoyed during stays at the MPI
382: f\"ur Physik komplexer Systeme, Dresden, and the MPI f\"ur
383: Str\"omungsforschung, G\"ottingen, generously financed by the MPG.
384: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
385: \begin{thebibliography}{10}
386:
387: \bibitem{Feynman} R.~P.~Feynman, R.~B.~Leighton, and M.~Sands,
388: {\em The Feynman Lectures on Physics}, Addison-Wesley, Reading, MA (1966),
389: vol.~1, chap.~46.
390:
391: \bibitem{JAP97}
392: F.~J{\"u}licher, A.~Ajdari, and J.~Prost,
393: Rev.~Mod.~Phys. {\bf 69}, 1269 (1997).
394:
395: \bibitem{JKH96}
396: R.~Bartussek, P.~H{\"a}nggi, and J.~G. Kissner,
397: {Europhys.~Lett.}, {\bf 28}, 459 (1994).
398: P.~Jung, J.~G. Kissner, and P.~H{\"a}nggi,
399: Phys.~Rev.~Lett. {\bf 76}, 3436 (1996).
400:
401: \bibitem{Mat00}
402: J.~L. Mateos, Phys.~Rev.~Lett. {\bf 84}, 258 (2000).
403:
404: \bibitem{FYZ00}
405: S.~Flach, O.~Yevtushenko, and Y.~Zolotaryuk,
406: Phys.~Rev.~Lett. {\bf 84}, 2358 (2000).
407:
408: \bibitem{GH00}
409: I.~Goychuk and P.~H{\"a}nggi in J. Freund and T. P{\"o}schel (eds.)
410: {\em Lecture Notes on Physics}, Vol.~{\bf 557},
411: pp. 7-20. Springer, Berlin (2000).
412:
413: \bibitem{RGH97}
414: P.~Reimann, M.~Grifoni, and P.~H{\"a}nggi,
415: Phys.~Rev.~Lett. {\bf 79}, 10 (1997){\bf ;}
416: P.~Reimann and P.~H{\"a}nggi,
417: Chaos {\bf 8}, 629 (1998).
418:
419: \bibitem{FYZ-Parameters} $H=p^2/2+V(x,t)$ with
420: $\partial_{x}V(x,t)=(2\pi/\omega^2)\,[\cos(2\pi x)+0.6\cos(4\pi x+0.4)-2.3\sin(2\pi
421: t)-1.38\sin(4\pi t+0.7)]$ and $\omega=2.4$ corresponding to the parameter set
422: (3) of Fig.~1 in Ref.~\protect\cite{FYZ00} where we have scaled the spatial
423: and temporal period to 1.
424:
425: \bibitem{DF01} S.~Denisov and S.~Flach, preprint nlin.CD/0104006.
426:
427: \bibitem{Theo} T.~Geisel in G.~M.~Zaslavsky (ed.) {\em L\'evy Flights and Related Phenomena in
428: Physics.} Springer, Berlin (1995).
429:
430: \bibitem{D+00} T.~Dittrich, R.~Ketzmerick, M.~F.~Otto, and H.~Schanz,
431: Ann.~Phys. (Leipzig) {\bf 9}, 755 (2000).
432:
433: \bibitem{YFR00} O.~Yevtushenko, S.~Flach, and K.~Richter,
434: Phys.~Rev.~E {\bf 61}, 7215 (2000).
435:
436: \bibitem{Sam73}
437: H.~Sambe,
438: Phys.~Rev.~A {\bf 7}, 2203 (1973).
439:
440: \bibitem{KMG94}
441: A.~R. Kolovsky, S.~Miyazaki, and R.~Graham,
442: Phys.~Rev.~E {\bf 49}, 70 (1994){\bf ;}
443: S.~Miyazaki and A.~R. Kolovsky,
444: Phys.~Rev.~E {\bf 50}, 910 (1994).
445:
446: \bibitem{K+00}
447: In this work we neglect the existence of hierarchical states, but see
448: R.~Ketzmerick, L.~Hufnagel, F.~Steinbach, and M.~Weiss,
449: Phys.~Rev.~Lett. {\bf 85}, 1214 (2000).
450:
451: \bibitem{D+98}
452: T.~Dittrich, B.~Mehlig, H.~Schanz, and U.~Smilansky,
453: Phys.~Rev.~E {\bf 57}, 359 (1998).
454:
455: \end{thebibliography}
456:
457: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
458: \if\tc\else
459: \def\fw{150mm}\figI\newpage
460: \def\fw{150mm}\figII\newpage
461: \def\fw{150mm}\figIII\newpage
462: \def\fw{150mm}\figIV
463: \fi
464: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
465: \if\tc \end{multicols}\fi
466: \end{document}
467: