1: %
2: % PRL manuscript, updated by SF
3: %
4: %\documentclass[aps,preprint,nofootinbib,showpacs]{revtex4}%
5: %\documentclass[aps,article,showpacs,byrevtex,howkeys]{revtex4}%
6: \documentclass[showpacs,twocolumn,floatfix,prl]{revtex4}
7: \usepackage{graphicx}
8: \usepackage{amsfonts}
9: \usepackage{amsmath}
10: \usepackage{amssymb}
11: \usepackage{graphicx}%
12: %
13: \newcommand{\mathbold}[1]{\mbox{{\boldmath $#1$}}}
14: \newcommand{\beq}{\begin{equation}}
15: \newcommand{\eeq}{\end{equation}}
16: \newcommand{\beqn}{\begin{eqnarray}}
17: \newcommand{\eeqn}{\end{eqnarray}}
18: \newcommand{\bearr}{\begin{array}}
19: \newcommand{\enarr}{\end{array}}
20: \newcommand{\derp}[2]{\frac{\partial{#1}}{\partial{#2}}}
21: \newcommand{\toref}[1]{\mbox{(\ref{#1})}}
22: \newcommand{\ket}[1]{|#1\rangle}
23: \newcommand{\bra}[1]{\langle#1|}
24: \newcommand{\bracket}[2]{\langle#1|#2\rangle}
25: \newcommand{\comm}[2]{\left[#1,#2\right]}
26: \newcommand{\anticomm}[2]{\left\{#1,#2\right\}}
27: \newcommand{\dcomm}[3]{\left[#1,#2,#3\right]}
28: \newcommand{\danticomm}[3]{\left\{#1,#2,#3\right\}}
29: \newcommand{\expval}[3]{\langle#1|#2|#3\rangle}
30: \newcommand{\redexpval}[3]{\langle#1||#2||#3\rangle}
31: \newcommand{\media}[1]{\langle#1\rangle}
32: \newcommand{\eps}{\varepsilon}
33:
34: \begin{document}
35:
36:
37: \title{Vortex and translational currents due to broken time-space symmetries}
38:
39: \author{S. Denisov$^1$, Y. Zolotaryuk$^{2}$, S. Flach$^{3}$,
40: and O. Yevtushenko$^4$}
41:
42: \affiliation{$^1$ Institut f\"ur Physik, Universit\"at Augsburg,
43: Universit\"atsstr.1, D-86135 Augsburg, Germany}
44: \affiliation{$^2$ Bogolyubov Institute for Theoretical Physics,
45: National Academy of Sciences of Ukraine,
46: 03680 Kiev, Ukraine}
47: \affiliation{$^3$Max-Planck-Institute for the Physics of Complex Systems,
48: N\"othnitzer Str. 38, D-01187 Dresden, Germany}
49: \affiliation{$^4$ Physics Department, Arnold Sommerfeld Center for Theoretical Physics,
50: Ludwig-Maximilians-Universit{\"a}t M{\"u}nchen, D-80333 M{\"u}nchen, Germany}
51: \date{\today}
52: \begin{abstract}
53: We consider the classical dynamics of a particle in a
54: $d=2,3$-dimensional space-periodic potential under the influence of
55: time-periodic external fields with zero mean. We perform a general
56: time-space symmetry analysis and identify conditions, when the
57: particle will generate a nonzero averaged translational and vortex
58: currents. We perform computational studies of the equations of
59: motion and of corresponding Fokker-Planck equations, which confirm
60: the symmetry predictions. We address the experimentally important
61: issue of current control. Cold atoms in optical potentials and
62: magnetic traps are among possible candidates to observe these
63: findings experimentally.
64:
65: \end{abstract}
66: \pacs{05.45.-a, 05.60.Cd, 05.40.-a}
67: \maketitle
68:
69: %
70: %======================================================================
71: % Intro
72: %======================================================================
73: %
74:
75: The idea of directed motion under the action of an external
76: fluctuating field of zero mean goes back to Smoluchowski and Feynman
77: \cite{Smol}. It has been intensively studied in the past decades
78: again \cite{magna}. It is believed to be connected with the
79: functioning of molecular motors, and can be applied to transport
80: phenomena which range from mechanical engines to an electron gas
81: (see \cite{revs,Reim} and references therein).
82:
83: The separation of the fluctuating fields into an uncorrelated white
84: noise term and a time-periodic field was used to perform a symmetry
85: analysis of the most simple case - a point-like particle moving in
86: a one-dimensional periodic potential \cite{Flach1}. It allowed to
87: systematically choose space and time dependencies of potentials and
88: ac fields such, that a nonzero dc current is generated. Various
89: studies of the dynamical mechanisms of rectification have been
90: reported (e.g. \cite{Den}). Among many experimental reports, we
91: mention the successful testing of the above symmetry analysis using
92: cold atoms in one-dimensional optical potentials \cite{ren}. By use
93: of more laser beams, experimentalists can already fabricate two- and
94: three-dimensional optical potentials, with different symmetries and
95: shapes \cite{2D3D}, with the aim of even more controlled stirring of
96: cold atoms in these setups.
97:
98: A particle which is moving in a $d=2,3$-dimensional periodic
99: potential may contribute to a directed current along a certain
100: direction. At the same time, the particle can perform vortex motion
101: (which is not possible in a one-dimensional setting) generating a
102: nonzero average of the angular momentum. Directed translational
103: currents are supported by unbounded trajectories while vortex
104: currents may be localized in a finite volume. The question is then,
105: how can we control a type of the directed motion? To answer this
106: question, one has to find conditions for an appearance of either
107: purely vertex or purely translational currents. This is the main
108: goal of the present work.
109:
110:
111: In this letter we perform a symmetry analysis of particle motion
112: in a $d=2,3$-dimensional periodic potential, under the influence of external ac fields.
113: We identify the symmetries which ensure that either directed currents, or vortex currents,
114: are strictly zero. Breaking these symmetries one by one allows to control
115: the particle motion, to generate either directed, or vortex, currents, or both simultaneously.
116:
117:
118:
119: We consider the dynamics of a classical particle (e.g. an atom of a cold dilute
120: atom gas, loaded onto a proper optical lattice) exposed to an
121: external potential field:
122: %------------------------------1---------------------------------------
123: \begin{equation}
124: \mu \ddot{{\bf r}} + \gamma \dot{\bf r} = {\bf g}({\bf r},t)+
125: \boldsymbol{\xi}(t) ~,
126: ~~{\bf g}({\bf r},t)=-\boldsymbol{\nabla} U({\bf r},t) ~.\\
127: \label{1sys}
128: \end{equation}
129: %----------------------------------------------------------------------
130: Here ${\bf r}=\{x,y,z\}$ is the coordinate vector of the particle,
131: the parameter $\gamma \ge 0$ characterizes the dissipation strength,
132: and $\mu \ge 0$ defines the strength of the inertial term
133: \cite{dissip2}. The force ${\bf g}({\bf r},t)=\{g_\alpha({\bf
134: r},t)\}$, $\alpha=x,y,z$, is time- and space-periodic:
135: %----------------------------------------------------------------------
136: \begin{equation}
137: {\bf g}({\bf r},t)={\bf g}({\bf r},t+T)={\bf g}({\bf
138: r}+\mathbf{L},t). \label{sp_per}
139: \end{equation}
140: %----------------------------------------------------------------------
141: The absence of a dc bias implies
142: %----------------------------------------------------------------------
143: \begin{equation}
144: \langle {\bf g} ({\bf r} ,t) \rangle_{\mathbf{L},T}\equiv
145: \int_{0}^T\int_{\mathbf{L}} {\bf g} ({\bf r} ,t) \, dt \, dx dy dz=0
146: \label{aver}
147: \end{equation}
148: %----------------------------------------------------------------------
149: where the spatial integration extends over one unit cell.
150:
151: The fluctuating force is modeled by a $\delta$-correlated Gaussian white
152: noise, $\boldsymbol{\xi}(t)=\{\xi_{x},\xi_{y}, \xi_{x}\}$,
153: $\langle \xi_{\alpha}(t) \xi_{\beta}(t') \rangle = 2\gamma D \delta
154: (t-t') \delta_{\alpha \beta}$ ($\alpha,\beta=x,y,z$).
155: Here $D$ is the noise strength. The statistical description of the
156: system (\ref{1sys}) is provided by the Fokker-Planck
157: equations (FPE) \cite{risken}:
158: %----------------------------------------------------------------------
159: \begin{eqnarray}
160: \nonumber
161: \dot{P}(\bf{r},\mathbf{v},t)&=&\{-\nabla_{\mathbf{r}}\cdot \mathbf{v}+
162: \nabla_{\mathbf{v}}
163: \cdot [\gamma
164: \mathbf{v}-\mathbf{g}(\mathbf{r},t)]+\\
165: &+&\gamma D \triangle_{\mathbf{v}}\}P(\bf{r},\mathbf{v},t),~\mu=1,
166: ~\mathbf{v} = \dot {\bf r},
167: \label{fpe_under}\\
168: \gamma \dot{P}(\bf{r},t)&=-&[\nabla_{\mathbf{r}}\cdot
169: \mathbf{g}(\mathbf{r},t)+ D
170: \triangle_{\mathbf{r}}]P(\bf{r},t),~\mu=0 \label{fpe_over}~.
171: \end{eqnarray}
172: %----------------------------------------------------------------------
173: Each of the linear equations
174: (\ref{fpe_under}-\ref{fpe_over}) has a unique attractor solution,
175: $\hat{P}$
176: which is space and time periodic \cite{risken}.
177:
178:
179: \emph{Directed transport.} Let us consider the dc component of the directed current
180: ${\bf j}(t) = \mathbf{v} = \dot {\bf r}$
181: in terms of the attractor $\hat{P}$:
182: %----------------------------------------------------------------------
183: \begin{eqnarray}
184: &&{\mathbf{J}}=\langle \mathbf{\mathbf{v}}\cdot
185: \hat{P}(\mathbf{r},\mathbf{v},t)\rangle_{T,\mathbf{L}},~~\mu=1~,
186: \label{cur_under}\\
187: &&{\mathbf{J}}=\frac{1}{\gamma}\langle {\bf g}({\bf r},t) \cdot
188: \hat{P}(\mathbf{r},t)\rangle_{T,
189: \mathbf{L}},~~\mu=0~.\label{cur_over}
190: \end{eqnarray}
191: %----------------------------------------------------------------------
192: The strategy is now to identify symmetry operations which invert the
193: sign of $\mathbf{j}$, and, at the same time, leave Eq. (\ref{1sys})
194: invariant. If such symmetries exist, the dc current $\mathbf{J}$
195: will strictly vanish. Sign changes of the current can be obtained by
196: either inverting the spatial coordinates, or time (simultaneously
197: allowing for shifts in the other variables). Below we list all
198: operations together with the requirements the force ${\bf g}$ and
199: the control parameters have to fulfill:
200: %----------------------------------------------------------------------
201: \begin{eqnarray}
202: {\widehat S}_1: {\bf r} \to - {\bf r} + {\bf r'},
203: ~t \to t+{\bf \tau};\; ~~~ {\widehat S}_1 ({\bf g}) \to -{\bf g}\;,~~~~~~~\label{S1}\\
204: {\widehat S}_2: {\bf r} \to {\bf r} + {\bf \Lambda},~t \to
205: -t+t';~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\nonumber
206: \\
207: ~~~~~{\widehat S}_2({\bf g}) \to {\bf g}~ (\mbox{if }\gamma=0); ~~~
208: {\widehat S}_2({\bf g}) \to -{\bf g}~ (\mbox{if }\mu=0).~~~~
209: \label{S2}
210: \end{eqnarray}
211: %----------------------------------------------------------------------
212: Here $t'$ and ${\bf r'}$ depend on the particular shape of ${\bf
213: g}({\bf r},t)$. The system must be invariant under a spatial
214: translation by the vector $2 {\bf \Lambda}$ in space and $2{\bf
215: \tau}$ in time, respectively. The vector ${\bf \Lambda}$ is
216: therefore given by ${\bf \Lambda}=\sum_{\alpha} n_{\alpha} {\bf
217: L}_\alpha/2$, $n_\alpha=0,1$, while $\tau = 0,T/2$. By a proper
218: choice of ${\bf g}$ all relevant symmetries can be broken, and one
219: can then expect the appearance of a non-zero dc current $\mathbf{J}$
220: \cite{Supersym}.
221:
222: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
223: %
224: % Fig. 1a,b
225: %______________________________________________________________________
226: \begin{figure}[t]
227: \includegraphics[width=7cm,height=4.4cm,angle=0]{f1a.eps}
228: \includegraphics[width=7.5cm,height=4.4cm,angle=0]{f1b.eps}%{Potential_new.eps}
229: \caption{(color online) (a) Dependence of the current components, $
230: J_{x}$(solid line) and $J_{y}$(dashed line), on $\theta$ for
231: (\ref{1sys}),(\ref{potent1})-(\ref{force1}), with $D=1$,
232: $E_{x}^{(1)}=-E_{x}^{(2)}=2$, $E_{y}^{(1)}=-E_{y}^{(2)}=2.5$. Data
233: are for the overdamped ($ \mu = 0, \gamma = 1 $, thick lines) and
234: underdamped ($ \mu = 1, \gamma = 0.1 $, thin lines) cases,
235: respectively; (b) The time evolution of the mean particle position,
236: $\bar{\mathbf{r}}(t) = \int {\mathbf r} P({\mathbf r}, {\mathbf v},
237: t) d{\mathbf r} d{\mathbf v}$, for $\theta=0$. The trajectories are
238: superimposed on the contour plot of the potential (\ref{potent1}).
239: Curve (\textit{i}) corresponds to $E_x^{(1)}=3$,
240: $E_x^{(2)}=E_y^{(1)}=0$, $E_y^{(2)}=3.5$ and curve (\textit{ii}) -
241: to the parameters of panel (a). The other parameters are: $ D = \mu
242: = 1, \gamma = 0.1 $; (c) The phase lag $\theta^{(0)}_x$ as a
243: function of the dissipation strength $\gamma$. } \label{fig:conv1}
244: \end{figure}
245: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
246:
247: To be more precise, we consider the case of a particle moving in a
248: two-dimensional periodic potential and being driven by an external
249: ac field: ${\bf g}({\bf r},t)=-\boldsymbol{\nabla}V({\bf r})
250: +\mathbf{E}(t)\equiv \mathbf{f}({\bf r}) +\mathbf{E}(t)$. The
251: symmetry ${\widehat S}_1$ holds if the potential force is
252: \textit{anti-symmetric},
253: $\mathbf{f}(-\mathbf{r}+\mathbf{r}')=-\mathbf{f}(\mathbf{r})$,
254: and the driving function is \textit{shift-symmetric},
255: $\mathbf{E}(t+T/2)=-\mathbf{E}(t)$. The symmetry ${\widehat S}_2$
256: holds at the Hamiltonian limit, $\gamma=0$, if the driving force is
257: \textit{symmetric}, $\mathbf{E}(t+t')=\mathbf{E}(-t)$. Finally, the
258: symmetry ${\widehat S}_2$ holds at the overdamped limit, $\mu=0$ ,
259: if the potential force is shift-symmetric,
260: $\mathbf{f}(\mathbf{r}+\mathbf{\Lambda})=-\mathbf{f}(\mathbf{r})$
261: and the driving force is anti-symmetric,
262: $\mathbf{E}(t+t')=-\mathbf{E}(-t)$.
263:
264: In order to break the above symmetries, we choose
265: %----------------------------10-20-------------------------------------
266: \begin{eqnarray}
267: &&V({\bf r})=V(x,y)=\cos(x)[1+\cos(2 y)]~, \label{potent1}\\
268: &&E_{x,y}(t)=E_{x,y}^{(1)}\sin t +E_{x,y}^{(2)}\sin (2 t +\theta)~.
269: \label{force1}
270: \end{eqnarray}
271: %----------------------------------------------------------------------
272: The potential (\ref{potent1}) is shift-symmetric,
273: $\mathbf{\Lambda}=\{\pm\pi,0\}$. The symmetry ${\widehat S}_1$
274: is broken since $E$ is not shift-symmetric. Therefore in general we
275: expect $\textbf{J} \neq 0$.
276:
277: In Fig.\ref{fig:conv1} we show the computational evaluation of
278: equations (\ref{fpe_under},\ref{fpe_over}) \cite{numeric}. We
279: confirm the presence of a nonzero dc current. Applying operations
280: ${\widehat S}_1$ and $\theta \to \theta+\pi$ we conclude
281: $\textbf{J}(\theta+\pi)=-\textbf{J}(\theta)$, which allows for an
282: easy inversion of the current direction, as also confirmed by the
283: data in Fig.\ref{fig:conv1}(a). In the overdamped limit $\mu=0$,
284: ${\widehat S}_2$ is restored for $\theta=0,\pm\pi$, and therefore
285: $\textbf{J}(-\theta)=-\textbf{J}(\theta)$ (thick lines in
286: Fig.\ref{fig:conv1}(a)). Upon approaching the Hamiltonian limit,
287: $\gamma \rightarrow 0$, the points where ${\bf J}=0$ shift from
288: $\theta=0,\pi$ to $\theta=\pm \pi/2$ where the symmetry ${\widehat
289: S}_2$ is restored (thin lines in Fig.\ref{fig:conv1}(a)). In the
290: underdamped regime, the dc-current can be approximated as $J_\alpha
291: \propto J_\alpha^{(0)} \sin[\theta-\theta_\alpha^{(0)}(\gamma)],
292: ~\alpha=x,y$. The phase lag is equal to $\theta^{(0)}_{x,y}= \pi/2$
293: and $\theta^{(0)}_{x,y}= 0$ in the Hamiltonian and overdamped
294: limits, respectively (Fig.\ref{fig:conv1}(c)) \cite{Renzoni_dissip}.
295:
296: Even more control over the current direction is possible, by
297: imposing the symmetry conditions (\ref{S1}-\ref{S2}) on each
298: component $g_{\alpha}(\mathbf{r},t)$ independently. For
299: (\ref{potent1}) and (\ref{force1}) with $E_x^{(2)}=E_y^{(1)}=0$,
300: $\theta=0$ the symmetry transformation $ {\widehat S}_c~:~~ x \to -
301: x,~~~ y \to y,~~~t \to t+\pi~$ implies that the current along the
302: $x$-direction is absent, $J_x=0$, and directed transport is
303: happening along the $y$-axis [see Fig.1b, curve (i)]. We may
304: conclude, that the symmetry analysis turns out to be a powerful tool
305: of predicting and controlling directed currents of particles which
306: move in $d=2,3$-dimensional potentials under the influence of
307: external ac fields. Note that dynamical mechanisms of current
308: rectification of a two-dimensional deterministic tilting ratchet
309: were discussed in Ref. \cite{guano_apes}.
310:
311:
312: %
313: %=======================================================================
314: % VORTICITY
315: %=======================================================================
316: %
317: {\it Vorticity.} At variance to the one-dimensional case,
318: particles in two and three dimensions can perform vortex motion, thereby
319: generating ring currents, or nonzero angular momentum.
320: First of all we note that the particle dynamics is not confined to one
321: spatial unit cell of the periodic potential $U(\mathbf{r},t)$. Even in the
322: case when a directed current is zero due to the above symmetries, $\mathbf{J}=0$, the particle
323: can perform unbiased diffusion in coordinate space. In order to distinguish between
324: directed transport
325: and spatial diffusion on one side, and rotational currents on the other side, we
326: use the angular velocity
327: \cite{k91}
328: %----------------------------------------------------------------------
329: \begin{equation}
330: {\bf \Omega}(t)= [\dot {\bf r}(t) \times \ddot {\bf r}(t)]/\dot {\bf r}^2(t),
331: ~~{\bf
332: J}_{\Omega}=\langle {\bf \Omega}(t) \rangle_t \label{angular2}~,
333: \end{equation}
334: %----------------------------------------------------------------------
335: as a measure for the particle rotation, where
336: $\langle...\rangle_{t}=\lim_{t\rightarrow
337: \infty}\frac{1}{t}\int_{0}^{t}... dt'$. ${\bf \Omega}(t)$ is
338: invariant under translations in space and time. It describes the
339: speed of rotation with which the velocity vector ${\bf {\dot r}}$
340: (the tangential vector to the trajectory ${\bf r}(t)$) encompasses
341: the origin.
342:
343: Using the above strategy, we search for symmetry operations which
344: leave the equations of motion invariant, but do change the sign of
345: the angular velocity. If such symmetries exist, rotational currents
346: strictly vanish on average. The sign of $\mathbf{\Omega}$ can be
347: inverted by either \textit{(i)} time inversion $t\rightarrow -t$
348: together with an optional space inversion $\mathbf{r} \rightarrow
349: \pm\mathbf{r}$, or \textit{(ii)} the permutation of any two
350: variables, e.g. ${\widehat {\cal P}_{xy}}: \{x,y,z \} \to \{y,x,z
351: \}$. That leads to the following possible symmetry transformations:
352: %----------------------------------------------------------------------
353: \begin{eqnarray}
354: {\widehat R}_1~: {\bf r} \to {\widehat {\cal P}}{\bf r} +
355: {\bf r'},~~t \to t+\tau\;;\;~~
356: {\widehat R}_1(\mathbf{g}) \to \mathbf{g}\;,\;~~~~\label{R1}\\
357: {\widehat R}_2: {\bf r} \to {\bf \pm r} + {\bf \Lambda},~~t \to
358: -t+t',\; ~~~~~~~~~~~~~~~~~~~~~~~\nonumber \\
359: ~~{\widehat R}_2({\bf g}) \to {\bf g} ~(\mbox{if }\gamma=0);
360: ~~{\widehat R}_2({\bf g}) \to -{\bf g} ~(\mbox{if
361: }\mu=0).~~~\label{R2}
362: \end{eqnarray}
363: %----------------------------------------------------------------------
364: Here ${\widehat {\cal P}}$ stands for any of the following
365: operations: ${\widehat {\cal P}_{xy}}$, ${\widehat {\cal P}_{yz}}$
366: or ${\widehat {\cal P}_{zx}}$ and $t'$ and ${\bf r'}$ again depend
367: on the particular shape of $ {\bf g}( {\bf r}, t) $.
368:
369: To be concrete, we will again consider a particle moving in a
370: two-dimensional periodic potential and being driven by an external
371: ac field: ${\bf g}({\bf r},t)=-\boldsymbol{\nabla}V({\bf r})
372: +\mathbf{E}(t)\equiv \mathbf{f}({\bf r}) +\mathbf{E}(t)$. For $d=2$
373: there is an additional transformation due to a mirror reflection at
374: any axis, $\hat{\Sigma}_x:\{x,y\}\rightarrow \{x,-y\}$ or
375: $\hat{\Sigma}_y:\{x,y\}\rightarrow \{-x,y\}$,
376: %----------------------------------------------------------------------
377: \begin{equation}
378: {\widehat R}_3~:~~ {\bf r} \to \hat{\Sigma} {\bf r}~,~t \to
379: t+T/2;~~~ {\widehat R}_3(\mathbf{g}) \to \mathbf{g}\ ~~ \label{R4}
380: \end{equation}
381: %----------------------------------------------------------------------
382: Symmetry ${\widehat R}_1$ can be satisfied for the {\it
383: Hamiltonian}, {\it underdamped} and {\it overdamped} cases if
384: ${\widehat {\cal P}}{\bf f}({\widehat {\cal P}}{\bf r}+{\bf r'})
385: ={\bf f}({\bf r})$ and ${\widehat {\cal P}}{\bf E}(t+t')={\bf
386: E}(t)$. The symmetry ${\widehat R}_{2}$ apply for the same cases as
387: their counterparts ${\widehat S}_{2}$ if there is no space
388: inversion. In the presence of space inversion they can be satisfied
389: both in the Hamiltonian [if ${\bf f}(-{\bf r})=-{\bf f}({\bf r})$
390: and ${\bf E}(t+t')=-{\bf E}(-t)$] and {overdamped} [if ${\bf
391: f}(-{\bf r})={\bf f}({\bf r})$ and ${\bf E}(t+t')={\bf E}(-t)$]
392: limits. The symmetry ${\widehat R}_3$ is relevant if $ \hat\Sigma_x
393: $ can be applied: $ f_x(x,-y)=f_x(x,-y)$, $f_y(x,-y)=-f_y(x,-y)$,
394: $E_x(t+T/2) =E_x(t)$ and $E_y(t+T/2)=-E_y(t)$. Similar conditions
395: can be found for $ \hat\Sigma_y $.
396:
397: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
398: %
399: % Fig. 2
400: %______________________________________________________________________
401: \begin{figure}[t]
402: \includegraphics[width=8cm,height=4.7cm,angle=0]{f2.eps}
403: \caption{(color online) (a) Dependence $J_{\Omega}(\theta)$,
404: Eq.(\ref{angular2}), for (\ref{1sys}),
405: (\ref{potent3})-(\ref{force3}), with $\mu=1$, $D=0.5$,
406: $E_x^{(1)}=0.4$, $E_y^{(1)}=0.8$ and $\gamma=0.2$ (solid line),
407: $\gamma=0.05$ (dashed line), and $\gamma=2$ (dashed-dotted line).
408: Insets: the trajectory (left insert) and the corresponding attractor
409: solution, $\bar{\mathbf{r}}(t)$, (right inset) for the case
410: $\gamma=0.2$ and $\theta=\pi/2$. We have used $N=10^{5}$ independent
411: stochastic realizations to perform the noise averaging; (b) The
412: phase lag $\theta_0$ as a function of the dissipation strength
413: $\gamma$.} \label{fig:conv3}
414: \end{figure}
415: %@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
416:
417: We performed numerical integrations of the equation of motion (\ref{1sys})
418: with the following potential
419: and driving force:
420: %----------------------------10-20-------------------------------------
421: \begin{eqnarray}
422: &&V(x,y)=[-3\left (\cos x+\cos y\right )+ \cos x \cos y]/2,~~ \label{potent3}\\
423: &&E_x(t)=E_x^{(1)}\cos t~,~~ E_y(t)=E_y^{(1)}\cos(t+\theta)~.
424: \label{force3}
425: \end{eqnarray}
426: %----------------------------------------------------------------------
427: Averaging was performed over $N=10^5$ different stochastic
428: realizations \cite{problems}. Fig.\ref{fig:conv3} shows the
429: dependence of the rotational current (\ref{angular2}) on the
430: relative phase $\theta$. The system is invariant under the
431: transformation ${\widehat S}_1$ (\ref{S1}), therefore the directed
432: current ${\bf J}=0$. However, for the underdamped case, $\gamma \neq
433: 0$, all the relevant symmetries (\ref{R1}-\ref{R4}) are violated,
434: and the resulting rotational current (\ref{angular2}) is nonzero,
435: and depends on the phase $\theta$ (Fig.2a). Note that symmetry
436: ${\widehat R}_2$ is restored when $\theta=0,\pm \pi$, thus the
437: current disappears in the Hamiltonian and overdamped limits for
438: these values of the phase. The left upper inset in
439: Fig.\ref{fig:conv3} shows the actual trajectory of a given
440: realization, confirming that the particle is acquiring an average
441: nonzero angular momentum, while not leaving a small finite volume
442: due to slow diffusion and absence of directed currents.
443:
444: The exact overdamped limit, $\mu=0$, is singular for the definition
445: (\ref{angular2}) since the velocity of a particle, $ {\bf \dot
446: r\textrm{(t)}}$, is a nowhere differentiable function. The
447: overdamped limit can be approached by increasing $\gamma$ at a fixed
448: $ \mu = 1 $. Alternatively, one may remove the restriction on $ \mu
449: $ allowing for an infinitesimal value $ 0 < \mu \ll 1 $ at a fixed
450: dissipation strength $ \gamma = 1 $. Both parameter choices equally
451: regularize (\ref{angular2}). Numerical simulations for the former
452: way of regularization show that if $ \gamma/\mu \geq 5$ then the
453: rotational current completely reflects the symmetries corresponding
454: to the overdamped case (see dependence $\theta_0(\gamma)$ in
455: Fig.\ref{fig:conv3}(b)).
456:
457: Let us discuss the relation of our
458: results to the case of multidimensional stochastic tilting ratchets
459: under the influence of a colored noise studied previously \cite{ghosh}. Since
460: equivalent (in a statistical
461: sense) stochastic processes, $\boldsymbol{\xi}(t)$, have
462: been used as driving forces, the symmetry ${\widehat R}_1$
463: (\ref{R1}) can be violated only by an asymmetric potential.
464: But all potentials considered in Refs. \cite{ghosh} are invariant
465: under the permutation transformation ${\widehat {\cal P}}$. As a
466: consequence, vortex structures for a \textit{local velocity field}
467: presented in Refs. \cite{ghosh} are completely symmetric
468: (clockwise vortices are mapped into counterclockwise ones by
469: ${\widehat {\cal P}}$) and, therefore, the average rotation
470: for any trajectory equals zero.
471:
472: The phase space dimension is five for $d=2$ and seven for $d=3$.
473: Therefore, in the Hamiltonian limit ($\gamma=0$), Arnold diffusion
474: \cite{arnold} takes place. The particle dynamics is no longer
475: confined within chaotic layers of finite width. That leads to
476: unbounded, possibly extremely slow, diffusion in the momentum
477: subspace via a stochastic web \cite{arnold}. Therefore a direct
478: numerical integration of the equations of motion may lead to
479: incorrect conclusions.
480:
481: There is rich variety of physical systems, where multidimensional
482: ratchets can be observed: cold atoms in two- and three-dimensional
483: potentials (optical guiding) \cite{hagman}, colloidal particles on
484: magnetic bubble lattices \cite{tierno}, ferrofluids
485: \cite{emrj03prl}, and vortices in superconducting films with pinning
486: sites \cite{souza}. Our methods of directed current control can be
487: used for an enhancement of the particle separation in the laser
488: beams of complex geometry \cite{Dholakia}.
489:
490:
491: To conclude, we formulated space-time symmetries for the absence of
492: both directed currents, and rotational currents, for particles
493: moving in spatially periodic potentials, under the influence of
494: external ac fields. Proper choices of these potentials and fields
495: allow to break the above symmetries, and therefore to generate and
496: control directed and rotational vortex currents in an independent
497: way. Numerical studies supplement the symmetry analysis and confirm
498: the conclusions.
499:
500: {\it Acknowledgments.} This work has been partially
501: supported by the
502: DFG-grant HA1517/31-1 (S. F. and S. D.), National Academy of
503: Sciences of Ukraine through the special program for young scientists
504: (Y.Z.). O.Y. acknowledges support from SFB-TR-12.
505:
506:
507: \begin{thebibliography}{10}
508:
509: \bibitem{Smol}
510: M. von Smoluchowski, Phys.~Zeitschrift \textbf{XIII}, 1069 (1912);
511: R. P. Feynman, R. B. Leighton, and M. Sands,
512: \textit{The Feyman Lectures on Physics}, 2nd edn. (Addison Wesley,
513: Reading, MA, 1963), vol.1, chap. 46.
514:
515: \bibitem{magna} M. O. Magnasco, Phys. Rev. Lett. \textbf{71}, 1477 (1993);
516: P. H\"{a}nggi and R. Bartussek, Lect. Notes. Phys.
517: \textbf{476}, 294 (1996).
518:
519: \bibitem{revs} F. J\"{u}licher, A. Ajdari, and J. Prost, Rev. Mod. Phys.
520: \textbf{69}, 1269 (1997);
521: R. D. Astumian and P. H\"anggi, Physics Today
522: \textbf{55} (11), 33 (2002).
523:
524: \bibitem{Reim}
525: P. Reimann, Phys. Rep. \textbf{361}, 57 (2002).
526:
527: \bibitem{Flach1} S.~Flach, O.~Yevtushenko, and Y.~Zolotaryuk, Phys. Rev. Lett.
528: \textbf{84}, 2358 (2000); O. Yevtushenko, S. Flach, Y. Zolotaryuk,
529: and A. Ovchinnikov, Europhys. Lett. {\bf 54}, 141 (2001);
530: P. Reimann, Phys. Rev. Lett. \textbf{86}, 4992 (2001).
531:
532: \bibitem{Den}
533: S. Denisov \textit{et al.}, Phys. Rev. E \textbf{66}, 041104 (2002);
534: H. Schantz et al, Phys. Rev. Lett. {\bf 87}, 070601 (2001).
535:
536: \bibitem{ren} M. Schiavoni {\it et al.},
537: Phys. Rev. Lett. {\bf 90}, 094101 (2003); R. Gommers, S. Denisov,
538: and F. Renzoni, {\it ibid.} \textbf{96}, 240604 (2006).
539:
540: \bibitem{2D3D} M. Greiner {\it et al.},
541: Phys. Rev. Lett. \textbf{87}, 160405 (2001); L. Santos {\it et al.},
542: \textit{ibid} \textbf{93}, 030601 (2004).
543:
544: \bibitem{dissip2} The underdamped regime is given by $\mu=1$, $\gamma > 0$,
545: and the Hamiltonian and overdamped limits by $\gamma=0$ and $\mu=0$, correspondingly.
546:
547: \bibitem{control} E. Dupont {\it et al.}, Phys. Rev. Lett. \textbf{74}, 3596 (1995);
548: J. Gong and P. Brumer, Annu. Rev. Phys. Chem. \textbf{56}, 1 (2005).
549:
550: \bibitem{risken} H. Risken, \textit{The Fokker-Planck equation}
551: (Springer-Verlag, London, 1996).
552:
553: \bibitem{Supersym} Note that the symmetry ${\widehat S}_2$
554: is not present in the corresponding overdamped FP equation
555: (\ref{fpe_over}). Indeed, the transformation (\ref{S2}) involves
556: time inversion, thus it maps stable manifolds (attractors) into
557: unstable ones (repellers). In the presence of a heat bath, an
558: attractor and its symmetry-related image, a repeller, acquire
559: different statistical weights. However, it was shown that the
560: symmetry ${\widehat S}_2$ {\sl is} a property of the stationary
561: \textit{solution} of the corresponding FPE \cite{Den}.
562:
563: \bibitem{numeric} We have solved the FPE in the underdamped case,
564: Eq.(\ref{fpe_under}), using a split-step method \cite{spectral}.
565: FPE in the overdamped limit, Eq.(\ref{fpe_over}), has been solved
566: using the Fourier expansion over spatial variables \cite{Den}.
567:
568: \bibitem{Renzoni_dissip} For a phase lag $\theta^{(0)}$ dependence on
569: the dissipation strength for 1d cold atom ratchets see
570: R. Gommers, S. Bergamini, and F. Renzoni,
571: Phys. Rev. Lett. \textbf{95}, 073003 (2005).
572:
573:
574: \bibitem{spectral} M. D. Feit, J. A. Fleck, Jr., and A. Steiger, J.
575: of Comput. Phys. \textbf{47}, 412 (1982).
576:
577: \bibitem{guano_apes}
578: R. Guantes and S. Miret-Artes, Phys. Rev. E {\bf 67}, 046212 (11)
579: (2003); S. Sengupta, R. Guantes, S. Miret-Artes, and P. H\"anggi,
580: Physica A {\bf 338},
581: 406 (2004).
582:
583: \bibitem{k91}
584: E. Kreyszig, {\em Differential Geometry} (Dover Publications Inc.,
585: New York,
586: 1991).
587:
588: \bibitem{problems}
589: The Fokker-Planck equations are not useful here, since the angular
590: velocity involves the acceleration. Therefore one needs to evaluate
591: averages of products of dynamical variables and white noise terms.
592:
593: \bibitem{ghosh}
594: A.~W. Ghosh and S.~V. Khare, Phys. Rev. Lett. {\bf 84}, 5243 (2000);
595: A.~W. Ghosh and S.~V. Khare, Phys. Rev. E {\bf 67}, 056110 (2003).
596:
597: \bibitem{arnold}
598: B. Chirikov, Phys. Rep. \textbf{52}, 1 (1979).
599:
600: \bibitem{hagman} H. Hagman {\it et al.}, Europhys. Lett. {\bf 81}, 33001 (2008)
601:
602: \bibitem{tierno}P. Tierno, T. H. Johansen, and T. M. Fischer, Phys. Rev. Lett.
603: \textbf{99}, 038303 (2007).
604:
605: \bibitem{emrj03prl}
606: A. Engel et al, Phys. Rev. Lett.
607: {\bf 91} 060602 (2003);
608: A. Engel and P. Reimann, Phys. Rev. E {\bf 70} 051107 (2004).
609:
610: \bibitem{souza} J. Van de Vondel {\it et al.}, Phys. Rev. Lett.
611: \textbf{94}, 057003 (2005); C. C. de Souza Silva {\it et al.},
612: Phys. Rev. Lett.
613: \textbf{98}, 117005 (2007).
614:
615: \bibitem{Dholakia} G. Milne {\it et al.}, Optics Express \textbf{15}, 13972 (2007)
616:
617:
618:
619: \end{thebibliography}
620:
621:
622: \end{document}
623: