Stochastic resonance in symmetric double well: higher harmonics

The work deals with the phenomenon of Stochastic Resonance in its genuine model, proposed by B. McNamara and K. Wiesenfeld for explanation of long-term climatic changes on Earth. It is shown that in two state model the higher harmonics behave in a non-monotonous way with increase of the noise level,...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Вопросы атомной науки и техники
Datum:2001
1. Verfasser: Zhiglo, A.V.
Format: Artikel
Sprache:Englisch
Veröffentlicht: Національний науковий центр «Харківський фізико-технічний інститут» НАН України 2001
Schlagworte:
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/79900
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Stochastic resonance in symmetric double well: higher harmonics / A.V. Zhiglo // Вопросы атомной науки и техники. — 2001. — № 6. — С. 251-254. — Бібліогр.: 8 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1860247505095098368
author Zhiglo, A.V.
author_facet Zhiglo, A.V.
citation_txt Stochastic resonance in symmetric double well: higher harmonics / A.V. Zhiglo // Вопросы атомной науки и техники. — 2001. — № 6. — С. 251-254. — Бібліогр.: 8 назв. — англ.
collection DSpace DC
container_title Вопросы атомной науки и техники
description The work deals with the phenomenon of Stochastic Resonance in its genuine model, proposed by B. McNamara and K. Wiesenfeld for explanation of long-term climatic changes on Earth. It is shown that in two state model the higher harmonics behave in a non-monotonous way with increase of the noise level, possessing one or more maxima. Explicit formulae for third and fifth harmonic amplitudes and corresponding SNR are obtained. Studied by other authors peculiarities, like dips and sharp peaks in output signal do not occur in two state model, thus they only exist in systems with continuous configuration space.
first_indexed 2025-12-07T18:38:35Z
format Article
fulltext STOCHASTIC RESONANCE IN SYMMETRIC DOUBLE WELL: HIGH- ER HARMONICS A.V. Zhiglo Institute for Theoretical Physics, National Science Center"Kharkov Institute of Physics and Technology", Kharkov, Ukraine The work deals with the phenomenon of Stochastic Resonance in its genuine model, proposed by B. McNamara and K. Wiesenfeld for explanation of long-term climatic changes on Earth. It is shown that in two state model the higher harmonics behave in a non-monotonous way with increase of the noise level, possessing one or more maxima. Explicit formulae for third and fifth harmonic amplitudes and corresponding SNR are obtained. Studied by other au- thors peculiarities, like dips and sharp peaks in output signal do not occur in two state model, thus they only exist in systems with continuous configuration space. PACS: 05.40. +j 1. INTRODUCTION Stochastic resonance (SR) has been studied for about 20 years. The principal manifestation of the phe- nomenon is a strong reaction of different output charac- teristics of the system (like component with initial fre- quency in residual-time distribution, signal/noise ratio etc.) to a weak periodic signal. This reaction grows with increasing of the noise level up to certain extent. The motivations for study of the phenomenon as well as the- oretical model are rather naturally presented in [1,2], early works describing SR (initially put forward for ex- planation of the correlation between glacial periods on Earth with the periodic changing of the Earth orbit ec- centricity). Now SR constitutes an important subfield of non-linear physics. According to usual understanding of stochastic resonance phenomenon as non-monotonous dependence of the output as a function of noise intensi- ty, the majority of investigations (both theoretical and experimental) in the case of monochromatic input ana- lyze the component of output with initial frequency. There are several works where higher harmonics are investigated as well as the first one — using different models in continuous configuration space (most com- mon approaches are linear-response theory, numerical integration of Fokker-Planck equation, matrix continued fraction technique) they study the problems of optimal generation or suppressing of higher harmonics [3,4,5]; the strengths of higher harmonics show various peculi- arities such as extremely sharp peaks and resonance-ab- sorption like dips at certain noise intensities. The de- pendence of non-zero intensity of even harmonics on the potential asymmetry is studied in [6]. Different methods used for study of SR-like phenomena and further numer- ous references can be found in [7]. We will demonstrate some qualitative aspects of the problem in two state model, intending to find out what of studied features of higher harmonics can be observed in it. We follow the notation from the work [1]; we will propose some useful representations of higher harmon- ics intensities — absolute and in relation to noise power densities at corresponding frequencies. Let us consider a Brownian particle in an external smooth potential )(xU with a barrier between two wells, subjected to a strong friction. If the noise intensity D (we consider a white noise that provides a term 2 ( )D tξ in the Langevin equation (3); =′)()( tt ξξ )( tt ′−δ ) is much smaller than the height of the barrier between wells U∆ , the average reciprocal transition time through the barrier is given by Kramers formula: 1 (0) ( ) exp 2 m Ur U U x Dπ ∆ж цўў ўў= −з ч и ш ; 0 and mx are positions of potential maximum and mini- mum. We consider a symmetric potential (with dimen- sionless x ) 42 )( 42 xxxU +−= (1) thus      −= D r 4 1exp 2 1 π . (2) Let us assume that beyond the chaotic force 2D tξ ( ) , a friction and − ∂ ∂U x/ the particle is subjected to an external periodic force A tcosω . For overdamped sys- tem the dependence x t( ) is to be found from  cos ( )x x x A t D t= − + +3 2ω ξ . (3) Under these conditions ( )x t is a periodic function having period 2 /T π ω= (providing 0 1/t t r− ? ); the point is to study the behavior of its different harmonics as functions of D . We mainly study the power spectrum, being obtained as a Fourier transform of autocorrelation function ( ) ( )x t x t τ+ . ( )kSNR D , defined as a ratio of signal power in -spike at kωΩ = to noise power density at PROBLEMS OF ATOMIC SCIENCE AND TECHNOLOGY. 2001, № 6, p. 251-254. 251 this frequency is also investigated; non-monotonous 1( )SNR D dependence is a usual manifestation of SR. Two state model, in which there are only two states for the system to occupy ( mx x= ± ), may be considered a limit of described above dynamical system in which the transition time is much greater than the relaxation time within a well. It is natural to assume minima of ( )U x as mx± ; for the potential (1) 1mx = . In such an approximation the distribution ( , )p x t re- duces to ( )n t± — probabilities of location near mx± (within right or left well) ( )1n n+ −+ = : ( , ) ( ) ( ) ( ) ( )m mp x t n t x x n t x xδ δ+ −= − + + . (4) Their evolution is given by the rate equation: ( ) ( ) dn n W t n W t dt ± ± ±= − +m m , (5) where ( )W t± are normalized probabilities of transition into  state. In the work [1] the following expression was pro- posed: ( ) exp cosmAx W t r t D ω± ж ц= ±з ч и ш (6) with , , r A D introduced above. The solution of (5) (for sign +) 2 cosh cos exp cosm mAx Axdn n r t r t dt D D ω ω+ += − +ж ц ж ц з ч з ч и ш и ш ( ) ( )P t n Q t+− +є (7) is ( ) exp ( ) exp ( ) t t t n t P dt C Q t P d dtτ τ+ − − −Ґ Ґ Ґ ж цж ц ж ц = − + −з чз ч з чз чи ш и ши ш т т т (8) The integrals in (8) cannot be calculated in terms of known functions. In [1] integrals were evaluated with accuracy /mAx Dε =: , considered to be a small pa- rameter. As a result ( )n t+ only contained the first har- monic that was ~ ε (comparing with the constant com- ponent of distribution at t = t0+1/r). Using more precise expansion in small ε it is possi- ble to account for higher harmonics and to calculate their amplitudes with desired accuracy. However, the problem is rather complicated especially if the aim is to obtain kSNR – because in such a way it is inevitable to take into account terms with different frequencies (not only kω ). In the work [8] the authors obtained results for higher harmonics in similar model considering hop- ping between the wells as process discrete in time (i.e. not only configuration space, but also the time scale was treated as discrete). They also studied interesting fea- tures connected with modulation of equilibrium posi- tions ( )mx t ; this resulted in (rather weak) peaks at even harmonics in power density. We will obtain sum representation of amplitudes in power spectrum; these sums: 1) have sense at all ε 2) can be used for evaluating the largest terms (those are 2~ for k kSNRε ) at small ε . 2. POWER SPECTRUM Using exp( cos ) I ( ) cosn n t n t Z e w e w О = е , where In are modified Bessel functions we can obtain 0 20 1 2 sin 2t n n P dt r t n tε ω Ґ = = + ет , (9) where r r r nn n 0 0 2 22= =I ( ) I ( )ε ε ε ω , . (10) So, using (8) (2 0 0 0sin 2 2 2 2 cos( ) n tn t r t r t r t tn t e Ce e dt r e eε ω ε ω− − − + − Ґ е= + т ) ( )2 0sin 2 2( ) ( )n n t r te t Ce V tε ω π −е+ґ є . (11) Here ( / ) ( )V t V tπ ω+ = . For the conditional probability ( )0 0| , :n t x t+ ( ( ) 00 0 0| , mx xn t x t δ+ = ) ( )0 0 0 2 ( ) 1 0 0( ) ( ) ( ) ( ) ( ) m r t t x xn t t V t e V t tπ δ π− − − + й щ= + − +л ы . (12) The amplitudes of different harmonics in output power are connected with the coefficients kG in 0 0 0( ) , ( ) lim 2 ( ) ( ) 1 t m x t x t H t t V t x π → − Ґ = = − 0 0 cos( )k k k G G k tω ψ Ґ = + +є е . (13) The periodicity of ( )H t is obvious; the autocorrel- tion function 0 0( ) ( ) ( , | . )x t x t K t x tτ τ+ = 0 0( ) ( ) ,x t x t x tτ+ 0 0( , | , ) ( , | , )xy dx dy p x t y t p y t x tτ= +тт (14) is periodic as a function of t in the limit 0t → − Ґ . On averaging over random initial phase in external force (or, equivalently, over the period of t, taking 0t t− and τ constant) 0 2 0 0lim ( ) ( ) , ( ) ( )mt x x t x t x t H t H tτ τ− → − Ґ + = + ( )02 2 ( )1 ( ) ( ) r te H t t τ π τ π − ++ − . (15) The first term (see (13)) gives (after averaging) 2 2 0 1 1( ) ( ) cos 2 kt k H t H t G G kτ ω τ Ґ = + = + е . (16) Fourier transform of (16) only contains δ-spikes at kωΩ = : 252 2 2 0 1 2 ( ) ( ) cos 2 i t k S k GG K e k t dtπ δ ω ҐҐ − Ω = − Ґ − Ω + Ω = е т [ ]2 1 ( ) ( ) 2 k k G k kπ δ ω δ ω Ґ = = Ω − + Ω +е . (17) Using the same procedure one can see that the 2nd term in (15) corresponds to noise component of output power since ( )0 02 22( ) 1 ( ) ( ) ( ) r r t te H t e U tτ τπ τ π τ − −+ − = . (18) with ( / ) ( )U t U tπ ω+ = . Fourier transform ( )NK Ω of (18) (i.e. the second term in (15)) is a regular function with Gaussian maxima near 2 (n n N −w=W О positive integers). So, the useful signal is given by (17); in part, if some 0kG = the power spectrum is regular at kw=W (because of the symmetry of the potential all even harmonics vanish: 2 0lG l N= " О ). According to the definition 2 4 ( ) k k N GSNR K k = W . (19) 3. HARMONICS AMPLITUDES EVALUA- TION Let us represent V(t) (see(11)) as a sum: 0 02 2 2 1 ( ) exp( cos ) exp sin 2 t r t r t k k V t e t e dt k te w e w Ґ =- Ґ = ет ( ) ( ) 0 22 2 , 0 I ( ) I ( ) 2 2 k k k k i t n knr t n n n k kn n k e ei r i n kn w e e w + - е е= Ч + + е Х е . (20) Here kn е means a sum over all 1{ }k kn ZҐ Ґ = О – conse- quences of integers (with finite number of non-zero ele- ments). Let us obtain representation of ( )H t as a Fouri- er sum: , ( ) 2 ( ) ( ) 1 I ( ) k k i t n n l H t t V t e ws s p e Ґ =- Ґ = - = е е ( ) ( ) 2 2 0 2I ( )I ( ) 1 2 k k k k k n l n k l k k ri r im e e w - -е -ґ Ч + Х i tH e wssє е . (21) In the last equation we have introduced , nZs s =О 12 ( )k kk k n lҐ =+ +е , thus n is to be calculated from here; 2 k k m kl N s О = - е . Hs are simply connected with :kG cos( ) , 2i kt i kt k k k k k kG k t H e H e G Hω ωω ψ − −+ = + = For 1e= (21) enables us to obtain approximations for Hs : below / ,z r w= { } { } ( ) 0 ( , ) , 2 k kl n k k iB B n l r imw -е =є + !N means !N ( ) 2 2 1, ( / 2) ( / 2) 1 ( ) 2 ! (2 )! ! ! k k k k n ln k kn l k k H z B O r n k k n l s e e e + Ґ = ж цчз чз= +чз ччззи ш е Х ( )2 1, 2 (2 )! ! ! k kk k k k n ln k n l kn l k k z B k k n l e ++ + Ґ = е ж цж ц чзчз ч» ч зз ччч зз чзи ш и ш е Х . (22) It is obvious that the terms containing the lowest power of e correspond to those with equal-signed , kn n and kl . There is a finite number of such terms. Using only them we get for 0s > ( ) ( ) 0, , 0 2 2 2 2 k k k k k k l n n n l n k l n H i r r im s s s e w - і + + = е е ж цчз= чз ччзи ш +е 1 1 (2 )! ! ! k kn l k k k z k k n l +Ґ = ж цчз чґ з чз чзи ш Х . (23) The calculations give for first non-zero amplitudes kG : ( ) ( ) 2 5 3 1 2 2 2 2 1 / 4 22 4 1 2 4 1 zG z O O z z z ee e e e + = + = + + + + 3 2 3 2 2 1/16 3 (4 1)(4 9) z zG z z e + = + + (24) ( ) ( ) ( )( )( ) 2 225 5 2 2 2 64 / 3 1 14 320 4 1 4 9 4 25 z zzG z z z e - + = + + + . Using the terms with the lowest power of e in NK one can obtain from (19) expressions for kSNR applicable for small e : ( )( ) 2 2 2 2 4 0 0 4 1 ( ) 8 k k GSNR r k O r p w e= + + . (25) Thus ( ) 2 2 2 6 1 2 21 2 4 1 zSNR z O z p ew e e ж цчз чз= + +чз ччзз +и ш 2 6 3 2 1/16 72 4 1 zSNR z z p w e + = + (26) ( ) ( ) ( )( ) 2 2210 5 2 13 2 2 64 / 3 1 14 10 2 4 1 4 9 z zzSNR z z pw e - + = Ч + + . The dependence ( )kSNR w has non-monotonous charac- ter. Unless A or e is large kSNR attains its maximum at 1e< where our approximation is valid. So, maximum positions can be calculated from the above expressions. From expansion of I ( )n e one can derive the following structure of the main term of SNR: ( )2 2/ 2 ( )k k kSNR z Q zw e= , where kQ is some rational function, 2 (0) 0kQ № . There- fore, for small z (i.e. large w or low noise, though the last alternative is restricted by the requirement / 1mAx D e<є ) one can evaluate the position of prin- cipal maximum of kSNR : PROBLEMS OF ATOMIC SCIENCE AND TECHNOLOGY. 2001, № 6, p. 251-254. 253 2 max 0 ( ) 0 0 2 k k z dQ z d SNR UD d d ke e = м ьDп пп п= = « =Юн эп пп по ю . Corresponding graphics of kSNR are depicted in Fig. 1–3. Maxima in Fig. 1 are located at predicted values. Figs. 2,3 show deviation from description based on ne- glecting of intricate z-containing factor behavior – they correspond to low w , when maxima are situated in the region of D, where 1e> . Fig. 3 shows anomalous rela- tive heights and position of multiple maxima of the first three non-zero harmonics. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.00 0.05 0.10 0.15 0.20 SNR1,3,5 D Fig. 1. 1,3,5 0.1SNR ω = – regular situation. Crosses correspond to 1SNR , the light curve to 3SNR , and the dark one to 5SNR 0.0 0.2 0.4 0.6 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 D SNR3 Fig. 2. 3 0.018SNR ω = – low frequency, 2 maxima with equal heights 0.0 0.2 0.4 0.6 0.8 0.00 0.05 0.10 0.15 0.20 SNR1,3,5 D Fig. 3. 1,3,5 0.003SNR ω = The main difference from corresponding characteris- tics of the first harmonic is due to more complicated role of z in the last expressions. This makes possible multiple maxima, which are really observed at such w that the "resonant" D corresponds to 1z » (or r w» ). In this case the behavior of ( )z D is necessary to be accounted for. Due to the different locations of maxima of different factors containing z in kG and kSNR one can find spe- cial values of w that provide several maxima of compa- rable heights for these functions, though universally these functions possess only one maximum in the region of lower and higher frequencies. For 1e> one should use greater number of terms in (21). The number of required terms increases rapidly when e becomes greater than 1. So, in order to find out whether there exist other peculiarities in output charac- teristic for considered system, it is more convenient to use numerical calculation of different harmonics of ( )H t . Such numerical integration shows that the correc- tions to (24) in the vicinities of maxima are not signifi- cant. These precise results do not change the non- monotonous character of both curves Gk ( )ε and SNRk ( )ε (or, equivalently G Dk ( ) and SNR Dk ( ) ). The only difference I would like to mention is less rapid van- ishing of both functions at e® Ґ ( )0DЫ ® – these vanish exponentially according to (24,26) and less rapidly following the numerical results. I am glad to acknowledge the participation of Yu.L. Bolotin in both the putting of the problem and fur- ther elaboration of questions needed consideration as well as for his support in search for relevant literature and inspiring discussions. REFERENCES 1.B. McNamara, K. Wiesenfeld. Theory of Stochastic Resonance // Phys. Rev. A. 1989, v. 39, p. 4854-4869. 2.R. Benzi, A. Sutera, A. Vulpiani. The mechanism of stochastic resonance // J. Phys. A: Math.Gen. 1982, v. 34, p. 10-16. 3.L. Gammaitoni, P. Hänggi, P. Jung, F. Marchesoni. Stochastic Resonance // Rev. Mod. Phys. 1998, v. 70, p. 223-287. 4.P. Jung, P. Hänggi. Stochastic Nonlinear Dynamics Modulated by External Periodic Forces // Europhys. Lett. 1989, v. 8 (6), p. 505-510. 5.P. Bartussek, P. Hänggi, P. Jung. Stochastic Reso- nance in Optical Bistable Systems // Phys. Rev. E. 1994, v. 49, №5, p. 3930-3939. 6.P. Jung, P. Talkner. Suppression of Higher Har- monics at Noise Induced Resonances // Phys. Rev. E. 1995, v. 51, №3, p. 2640-2644. 7.A.R. Bulsara, M.E. Inchiosa, L. Gammaitoni. Noise-Controlled Resonance Behavior in Nonlinear Dynamical Systems with Broken Symmetry // Phys. Rev. Lett. 1996, v. 77, №11, p. 2162-2165. 8.V.A. Shneidman, P. Jung, P. Hänggi. Power Spec- trum of a Driven Bistable System // Europhys. Lett. 1994, v. 26, №8, p. 571-576. 254 A.V. Zhiglo Institute for Theoretical Physics, National Science Center"Kharkov Institute of Physics and Technology", Kharkov, Ukraine PACS: 05.40. +j 1. INTRODUCTION 2. POWER SPECTRUM 3. HARMONICS AMPLITUDES EVALUATION REFERENCES
id nasplib_isofts_kiev_ua-123456789-79900
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 1562-6016
language English
last_indexed 2025-12-07T18:38:35Z
publishDate 2001
publisher Національний науковий центр «Харківський фізико-технічний інститут» НАН України
record_format dspace
spelling Zhiglo, A.V.
2015-04-06T16:42:22Z
2015-04-06T16:42:22Z
2001
Stochastic resonance in symmetric double well: higher harmonics / A.V. Zhiglo // Вопросы атомной науки и техники. — 2001. — № 6. — С. 251-254. — Бібліогр.: 8 назв. — англ.
1562-6016
PACS: 05.40. +j
https://nasplib.isofts.kiev.ua/handle/123456789/79900
The work deals with the phenomenon of Stochastic Resonance in its genuine model, proposed by B. McNamara and K. Wiesenfeld for explanation of long-term climatic changes on Earth. It is shown that in two state model the higher harmonics behave in a non-monotonous way with increase of the noise level, possessing one or more maxima. Explicit formulae for third and fifth harmonic amplitudes and corresponding SNR are obtained. Studied by other authors peculiarities, like dips and sharp peaks in output signal do not occur in two state model, thus they only exist in systems with continuous configuration space.
I am glad to acknowledge the participation of Yu.L. Bolotin in both the putting of the problem and further elaboration of questions needed consideration as well as for his support in search for relevant literature and inspiring discussions.
en
Національний науковий центр «Харківський фізико-технічний інститут» НАН України
Вопросы атомной науки и техники
Anomalous diffusion, fractals, and chaos
Stochastic resonance in symmetric double well: higher harmonics
Стохастический резонанс в симметричной двойной яме: высшие гармоники
Article
published earlier
spellingShingle Stochastic resonance in symmetric double well: higher harmonics
Zhiglo, A.V.
Anomalous diffusion, fractals, and chaos
title Stochastic resonance in symmetric double well: higher harmonics
title_alt Стохастический резонанс в симметричной двойной яме: высшие гармоники
title_full Stochastic resonance in symmetric double well: higher harmonics
title_fullStr Stochastic resonance in symmetric double well: higher harmonics
title_full_unstemmed Stochastic resonance in symmetric double well: higher harmonics
title_short Stochastic resonance in symmetric double well: higher harmonics
title_sort stochastic resonance in symmetric double well: higher harmonics
topic Anomalous diffusion, fractals, and chaos
topic_facet Anomalous diffusion, fractals, and chaos
url https://nasplib.isofts.kiev.ua/handle/123456789/79900
work_keys_str_mv AT zhigloav stochasticresonanceinsymmetricdoublewellhigherharmonics
AT zhigloav stohastičeskiirezonansvsimmetričnoidvoinoiâmevysšiegarmoniki