1: %% ****** Start of file template.aps ****** %
2: %%
3: %%
4: %% This file is part of the APS files in the REVTeX 4 distribution.
5: %% Version 4.0 of REVTeX, August 2001
6: %%
7: %%
8: %% Copyright (c) 2001 The American Physical Society.
9: %%
10: %% See the REVTeX 4 README file for restrictions and more information.
11: %%
12: %
13: % This is a template for producing manuscripts for use with REVTEX 4.0
14: % Copy this file to another name and then work on that file.
15: % That way, you always have this original template file to use.
16: %
17: % Group addresses by affiliation; use superscriqptaddress for long
18: % author lists, or if there are many overlapping affiliations.
19: % For Phys. Rev. appearance, change preprint to twocolumn.
20: % Choose pra, prb, prc, prd, pre, prl, prstab, or rmp for journal
21: % Add 'draft' option to mark overfull boxes with black boxes
22: % Add 'showpacs' option to make PACS codes appear
23: % Add 'showkeys' option to make keywords appear
24: \documentclass[aps,pre,preprint,groupedaddress,showpacs]{revtex4}
25: %\documentclass[aps,prl,preprint,superscriptaddress]{revtex4}
26: %\documentclass[aps,pre,twocolumn,groupedaddress,floatfix]{revtex4}
27: \usepackage{graphicx}
28: % You should use BibTeX and apsrev.bst for references
29: % Choosing a journal automatically selects the correct APS
30: % BibTeX style file (bst file), so only uncomment the line
31: % below if necessary.
32: \bibliographystyle{apsrev}
33:
34: \begin{document}
35: \newcommand{\ud}{{\mathrm d}}
36: \newcommand{\umod}{\mathrm{mod}}
37: \newcommand{\sech}{\mathrm{sech}}
38: % Use the \preprint command to place your local institutional report
39: % number in the upper righthand corner of the title page in preprint mode.
40: % Multiple \preprint commands are allowed.
41: % Use the 'preprintnumbers' class option to override journal defaults
42: % to display numbers if necessary
43: %\preprint{}
44:
45: %Title of paper
46: \title{Nonlinear Stochastic Resonance with subthreshold rectangular
47: pulses}
48:
49: % repeat the \author .. \affiliation etc. as needed
50: % \email, \thanks, \homepage, \altaffiliation all apply to the current
51: % author. Explanatory text should go in the []'s, actual e-mail
52: % address or url should go in the {}'s for \email and \homepage.
53: % Please use the appropriate macro foreach each type of information
54:
55: % \affiliation command applies to all authors since the last
56: % \affiliation command. The \affiliation command should follow the
57: % other information
58: % \affiliation can be followed by \email, \homepage, \thanks as well.
59: \author{Jes\'us Casado-Pascual}
60: \email[]{jcasado@us.es}
61: \homepage[]{http://numerix.us.es}
62: \author{Jos\'e G\'omez-Ord\'o\~nez}
63: \author{ Manuel Morillo}
64: \affiliation{F\'{\i}sica Te\'orica,
65: Universidad de Sevilla, Apartado de Correos 1065, Sevilla 41080, Spain}
66:
67:
68: \date{\today}
69:
70: \begin{abstract}
71: We analyze the phenomenon of nonlinear stochastic resonance (SR) in
72: noisy bistable systems driven by pulsed time periodic forces. The
73: driving force contains, within each period, two pulses of equal constant
74: amplitude and duration but opposite signs. Each pulse starts every
75: half-period and its duration is varied. For {\em subthreshold}
76: amplitudes, we study the dependence of the output signal-to-noise ratio
77: (SNR) and the SR gain on the noise strength and the relative duration of
78: the pulses. We find that the SR gains can reach values larger than
79: unity, with maximum values showing a nonmonotonic dependence on the
80: duration of the pulses.
81: \end{abstract}
82:
83:
84:
85: % insert suggested PACS numbers in braces on next line
86: \pacs{05.40.-a, 05.10.Gg, 02.50.-r}
87: % insert suggested keywords - APS authors don't need to do this
88: %\keywords{}
89:
90: %\maketitle must follow title, authors, abstract, \pacs, and \keywords
91: \maketitle
92: %\section{Introduction}
93: In recent work \cite{Casgomprl,Casgompre}, we have carried out a
94: detailed analytical and numerical study of the nonlinear response of a
95: noisy bistable system subject to a subthreshold time periodic driving
96: force. The force considered in those works is such that it remains
97: constant within the duration of each half period while switching its
98: sign every half period. We have focused our attention on the analysis
99: of the dependence of the output signal-to-noise ratio and the
100: corresponding SR gain on the noise strength. The analytical study was
101: based on a two-state approximation amenable to exact treatment
102: \cite{Casgomprl}. We showed that, for subthreshold input signals of
103: sufficiently long periods, the phenomenon of SR can be accompanied by
104: SR gains larger than unity. This is a genuine characterization of
105: nonlinearity, as SR gains larger than unity are strictly forbidden
106: within a linear response description \cite{dykluc95,dewbia95,casgom03}.
107: The analytical results were corroborated by numerical simulations
108: \cite{Casgompre}.
109:
110: In the last few years, Gingl and collaborators
111: \cite{ginmak00,ginmak01,ginmak02} have carried out analog simulations of
112: noisy bistable systems subject to driving forces of the type given in
113: Eq.~(\ref{pulse})
114: \begin{equation}
115: \label{pulse}
116: F(t)= \left \{ \begin{array}
117: {r@{\quad:\quad}l}
118: A & 0\le t < t_c \\
119: -A& \frac T2 \le t < \frac T2 + t_c \\
120: 0& \textrm{otherwise}
121: \end{array}
122: \right .
123: \end{equation}
124: It is convenient to introduce the parameter $r=2t_c/T$, measuring the
125: fraction of a period during which this driving force has a nonvanishing
126: value (the parameter $r$ in the present paper corresponds exactly to
127: what Gingl and coworkers term ``duty cyle''). In
128: \cite{ginmak00,ginmak01,ginmak02}, the SNR and the SR gain for
129: subthreshold amplitude input signals with $r \le 0.3$ were
130: studied. These authors find SR gains larger than unity, and, also, that
131: increasing the $r$ value lowers the SNR gain. They rationalize their
132: observations by noting that the input SNR increases as $r$ increases,
133: while the output SNR is less sensitive to the value of $r$. The case
134: studied by us in \cite{Casgomprl,Casgompre} corresponds to the largest
135: possible value of the parameter $r$, namely, $r=1$. It seems therefore
136: of interest to extend our analysis to input signals with $r < 1$ in
137: order to compare with the predictions of Gingl et al.
138:
139: Let us consider a system characterized by a single degree of freedom,
140: $x$, whose dynamics (in dimensionless units) is governed by the
141: Langevin equation
142: \begin{equation}
143: \label{langev} \dot{x}(t)=-U'\left[ x(t),t \right]+\xi(t),
144: \end{equation}
145: where $\xi (t)$ is a Gaussian white noise of zero mean with $\langle
146: \xi(t)\xi(s)\rangle = 2D\delta(t-s)$, and $ -U'(x,t)$ represents the
147: force stemming from the time-dependent, archetype bistable potential
148: \begin{equation}
149: \label{potential}
150: U(x,t)=\frac{x^4}{4}-\frac{x^2}{2}-F(t)x,
151: \end{equation}
152: with $F(t)$ given by Eq.~(\ref{pulse}). The one-time correlation
153: function is defined as
154: \begin{equation}
155: \label{ctau}
156: C(\tau)= \frac{1}{T} \int_{0}^{T} \ud t \, \langle x(t+\tau)
157: x(t)\rangle_{\infty}.
158: \end{equation}
159: It can be written exactly as the sum of two contributions: a coherent
160: part, $C_{coh}(\tau)$, which is periodic in $\tau$ with period $T$, and
161: an incoherent part, $C_{incoh}(\tau)$, which decays to $0$ for large
162: values of $\tau$ and reflects the correlation of the output fluctuations
163: about its average (the noisy part of the output). The coherent
164: part $C_{coh}(\tau)$ is given by \cite{RMP,PRA91}
165: \begin{equation}
166: \label{chtau}
167: C_{coh}(\tau) =\frac{1}{T} \int_{0}^{T} \ud t \,\langle x(t+\tau)
168: \rangle_{\infty} \langle x(t) \rangle_{\infty},
169: \end{equation}
170: and $C_{incoh}(\tau)$ is obtained from the difference of
171: Eq.~(\ref{ctau}) and Eq.~(\ref{chtau}). In the expressions above, the
172: susbcript indicates that the averages are to be evaluated in the limit
173: $t \rightarrow \infty$.
174:
175: According to McNamara and Wiesenfeld \cite{McNWie89}, the output SNR,
176: $R_{out}$, is defined in terms of the Fourier transform of the
177: coherent and incoherent parts of $C(\tau)$ as
178: \begin{equation}
179: \label{snr}
180: R_{out} =\frac {\lim_{\epsilon \rightarrow 0+}
181: \int_{\Omega-\epsilon}^{\Omega+\epsilon} \ud\omega\;
182: \tilde{C}(\omega)}{\tilde{C}_{incoh}(\Omega)},
183: \end{equation}
184: where $\tilde{H}(\omega)$ denotes the Fourier cosine transform of
185: $H(\tau)$, i.e., $\tilde{H}(\omega)=2/\pi \int_0^\infty \ud\tau\,H(\tau)
186: \cos (\omega \tau)$. Note that this definition of the output SNR differs
187: by a factor $2$, stemming from the same contribution at $\omega = -
188: \Omega$, from the definitions used in earlier works \cite
189: {RMP,PRA91}.
190: The periodicity of the coherent part gives rise to delta
191: peaks in the spectrum. Thus, the only contribution to the numerator in
192: Eq.\ (\ref{snr}) stems from the coherent part of the correlation
193: function. The output SNR can then be expressed as
194: \begin{equation}
195: \label{snr1}
196: R_{out}=\frac{Q_u}{Q_l},
197: \end{equation}
198: where
199: \begin{equation}
200: \label{num}
201: Q_u= {\frac 2T} \int_0^T \ud \tau \,C_{coh}(\tau) \,\cos
202: (\Omega \tau),
203: \end{equation}
204: and
205: \begin{equation}
206: \label{den}
207: Q_l=\frac 2\pi \int_0^\infty \ud \tau
208: \,C_{incoh}(\tau) \,\cos (\Omega \tau ).
209: \end{equation}
210: A nonmonotonic behavior of the SNR with the strength of the noise is a
211: signature of the phenomenon of SR.
212:
213: The signal-to-noise ratio for an input signal $F(t)+\xi(t)$ is given by
214: \begin{equation}
215: \label{snrinp}
216: R_{inp}=\frac{\frac 12 (f_1^2+g_1^2)}{ \frac 2\pi D}.
217: \end{equation}
218: where
219: \begin{equation}
220: f_1=\frac{2A}{\pi} \sin \Omega t_c
221: \end{equation}
222: and
223: \begin{equation}
224: g_1=\frac{2A}{\pi} \left ( 1-\cos \Omega t_c \right )
225: \end{equation}
226: It is then clear that for fixed values of $A$ and $D$, $R_{inp}$
227: increases with $r$ as pointed out in Ref. \cite{ginmak02}. Also,
228: for given $A$ and $r$, $R_{inp}$ decreases as $D$ increases.
229:
230: The SR gain, $G$, is defined as the ratio of the SNR of the output over
231: that of the input; namely,
232: \begin{equation}
233: \label{gain}
234: G=\frac {R_{out}}{R_{inp}}.
235: \end{equation}
236: SR gain values larger than 1 have been obtained in driven nondynamical
237: systems \cite{ginmak00}, in stochastic resonators with static
238: nonlinearities driven by square pulses \cite{chapeau}, or in noisy
239: bistable systems driven by superthreshold input sinusoidal frequencies
240: \cite{saga}. The existence of SR gains with values larger than 1
241: indicates a truly nonlinear SR.
242:
243: Although the two-state approximation introduced in Ref. \cite{Casgomprl}
244: can, in principle, be extended to analyze systems
245: driven by input signals with $r < 1$, the
246: analytical expressions obtained are too cumbersome to be of practical
247: value. Thus, in the present work, we rely on the numerical treatment of
248: the Langevin equation, Eq.~(\ref{langev}), following the procedure detailed in
249: Ref.~\cite{casgom03}.
250:
251: In Fig.\ \ref{SNR}, we depict the behavior of $R_{out}$ with the noise
252: strength $D$, for input signals of the type given in Eq.~(\ref{pulse}),
253: with subthreshold amplitude $A=0.35$, fundamental frequency
254: $\Omega=0.0024$ and $r=0.1, 0.4, 0.7, 0.95, 0.98, 1$. For all values of
255: $r$, the characteristic nonmonotonic behavior of the SNR with $D$ is
256: obtained. As $r$ increases, the maximum value of $R_{out}$
257: increases. Namely, the longer the potential remains asymmetric during
258: each half-cycle, the larger the maximum height in the SNR is. Therefore,
259: $R_{out}$ is quite sensitive to the duration of the pulses within each
260: half-cycle.
261:
262: \begin{figure}
263: \includegraphics[width=9cm,height=10cm]{SNRvDrp1-r1.eps}
264: \caption{\label{SNR} The output signal-to-noise ratio, $R_{out}$ {\it
265: vs.} the noise strength, $D$, for $r=0.1$ (circles), $r=0.4$
266: (squares), $r=0.7$ (triangles), $r=0.95$ (filled circles),
267: $r=0.98$ (filled squares) and $r=1.0$ (filled triangles).
268: The input signal has a subthreshold amplitude $A=0.35$ and a
269: fundamental frequency $\Omega=2\pi/T=0.0024$.}
270: \end{figure}
271:
272: In Fig.\ \ref{Glow} the behavior of $G$ with the noise strength $D$ is
273: depicted for the same values of $r$ as in the previous figure. For all
274: values of $r$ considered, there exists a range of noise values such that
275: $G$ is larger than unity. The peak value of the SR gain has a
276: nonmonotonic behavior with $r$. At the lowest value of $r$ considered,
277: the SR gain has a rather large peak value. Then, as $r$ is increased,
278: the peak of the SR gain decreases, in agreement with the observations in
279: \cite{ginmak01,ginmak02}. As the duration of the pulses gets larger so
280: that $r$ is closer to $1$, the tendency of the SR gain maximum reverses
281: and a considerable increase in the maximum is observed.
282:
283: The just mentioned features can be rationalized by noting that the SR
284: gain depends on $R_{out}$ and on $R_{in}$. As indicated above, for fixed
285: values of the noise strength, $D$, and amplitude, $A$, the input
286: $R_{in}$ always increases with $r$. Also, for given values of $A$ and
287: $r$, $R_{in}$ decreases monotonically with $D$. On the other hand, the
288: results depicted in Fig.\ \ref{SNR} indicate that there are two main
289: effects on the location of the maximum of $R_{out}$ as $r$
290: increases. First, as noted before, the maximum height increases as $r$
291: increases. Second, the maxima appear at increasingly large values of $D$
292: as the duration of the pulses increases. This second effect manifests
293: itself clearly for pulses of sufficiently long duration, namely for $r$
294: values larger than $r \approx 0.9$, while it is almost unnoticeable for
295: smaller bvalues of $r$. For low values of $r$ (let us say $r =0.1$),
296: even though the peak of $R_{out}$ is the smallest one appearing in Fig.\
297: \ref{SNR}, the corresponding value of $R_{in}$ is so small (due to the
298: smallness of $r$) that the SR gain reaches the large values depicted in
299: Fig.\ \ref{Glow}. As $r$ increases, the height of the $R_{out}$ maximum
300: also increases, but appearing at an approximately constant value of the
301: noise strength. Thus, the increase of $R_{in}$ with $r$ counterbalances
302: the increase of $R_{out}$ in such a way that the SR gain decreases.
303: Finally, for long duration pulses, the shift to the right of the
304: $R_{out}$ maximum and the large increase in its height are the cause of
305: the increase in the maximum gain observed in Fig.\ \ref{Glow}.
306:
307: In conclusion, we observe that the behavior of the SR gain for pulses of
308: relatively short duration is basically a consequence of the behavior of
309: $R_{in}$, in agreement with the observation of Gingl et
310: al. \cite{ginmak00,ginmak01,ginmak02}. On the other hand, for $r$ close
311: to 1, the behavior of the SR gain is dominated by the output SNR.
312:
313: \begin{figure}
314: \includegraphics[width=9cm,height=10cm]{GvDrp1-r1.eps}
315: \caption{\label{Glow} The SR gain, $G$, {\it
316: vs.} the noise strength, $D$, for the same parameter values as in Fig.\
317: (\ref{SNR}).}
318: \end{figure}
319:
320: \begin{acknowledgments}
321: We acknowledge the support of the Direcci\'on General de
322: Ense\~nanza Superior of Spain (BFM2002-03822) and the Junta de
323: Andaluc\'{\i}a.
324: \end{acknowledgments}
325:
326: \begin{thebibliography}{100}
327: \bibitem{Casgomprl} J. Casado-Pascual, J. G\'omez-Ord\'o\~nez,
328: M. Morillo and P. H\"anggi, Phys. Rev. Letts. {\bf 91}, 210601 (2003).
329:
330: \bibitem{Casgompre} J. Casado-Pascual, J. G\'omez-Ord\'o\~nez,
331: M. Morillo and P. H\"anggi, Phys. Rev. E {\bf 68} 061104 (2003).
332:
333: \bibitem{casgom03} J. Casado-Pascual, C. Denk, J. G\'omez-Ord\'o\~nez,
334: M. Morillo, and P. H\"anggi, Phys. Rev. E {\bf 67}, 036109 (2003).
335:
336: \bibitem{dykluc95} M. I. Dykman, D. G. Luchinsky, R. Mannella,
337: P. V. E. McClintock, N. D. Stein, and N. G. Stocks, Il Nuovo
338: Cimento {\bf 17D}, 660 (1995).
339:
340: \bibitem{dewbia95} M. DeWeese and W. Bialek, Il Nuovo Cimento {\bf
341: 17D}, 733 (1995).
342:
343: \bibitem{ginmak00} K. Loerincz, Z. Gingl, and L.B. Kiss, Phys. Lett A
344: {\bf 224}, 63 (1996).
345:
346: \bibitem{ginmak01} Z. Gingl, P. Makra, and R. Vajtai, Fluct.
347: Noise Lett. {\bf 1}, L181 (2001).
348:
349: \bibitem{ginmak02} P. Makra, Z. Gingl and L. B. Kish, Fluct. Noise
350: Lett. {\bf 2}, L147 (2002).
351:
352: \bibitem{RMP}
353: L. Gammaitoni, P. H\"anggi, P. Jung, and F. Marchesoni, Rev. Mod.
354: Phys. {\bf 70}, 223 (1998).
355:
356: \bibitem{PRA91} P. Jung and P. H\"anggi, Europhys. Lett. {\bf 8}, 505
357: (1989); P. Jung and P. H\"anggi, Phys. Rev. A {\bf 44}, 8032 (1991).
358:
359: \bibitem{McNWie89} B. McNamara and K. Wiesenfeld, Phys. Rev. A {\bf
360: 39}, 4854 (1989).
361:
362: \bibitem{chapeau} F. Chapeau-Blondeau, Physics Letters A, {\bf 232}, 41 (1997).
363:
364: \bibitem{saga} P. H\"anggi, M. Inchiosa, D. Fogliatti and A. D. Bulsara,
365: Phys. Rev. E {\bf 62}, 6155 (2000).
366: \end{thebibliography}
367:
368:
369: \end{document}
370:
371:
372:
373:
374: