1 / 24

Yo Horikawa Kagawa University Japan

Exponential Transient Oscillations and Standing Pulses in Rings of Coupled Symmetric Bistable Maps. Yo Horikawa Kagawa University Japan. 1. 1. Background Exponential transients Initial states → Transient states → Asymptotic states ↑

galvin
Télécharger la présentation

Yo Horikawa Kagawa University Japan

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Exponential Transient Oscillations and Standing Pulses in Rings of Coupled Symmetric Bistable Maps Yo Horikawa Kagawa University Japan 1

  2. 1. Background Exponential transients Initial states → Transient states → Asymptotic states ↑ Duration (Life time) of transients increases exponentially with system size. T ∝ exp(N) T: duration of transients N: system size

  3. 1. Background Systems never reach their asymptotic states in a practical time. → Transient states play important roles. Two kinds of exponential transients I. Metastable dynamics in reaction-diffusion systems (Kawasaki and Ohta 1982) in ring neural networks (Horikawa and Kitajima 2008) II. Transient chaos in coupled map lattices (Crutchfield and Kaneko 1988) in neural networks (Bastolla and Parisi 1998) 3

  4. l 1. Background Examples of exponential transients 1. Bistable reaction-diffusion equation Transient kink, pulse patterns → Spatially homogeneous states: u = ±1 4

  5. 2 1 N 3 4 5 6 7 8 l 1. Background Examples of exponential transients 2. Ring neural network Transient traveling waves and oscillations → Spatially homogeneous states 5

  6. 2 1 N 3 4 5 6 7 8 1. Background Examples of exponential transients 3. Bistable ring of directly coupled maps Traveling waves → Spatially homogeneous states l 6

  7. 1. Background 1. Bistable reaction-diffusion equation 2. Ring neural network 3. Ring of directly coupled maps Symmetric bistability Common kinematics dl/dt~ –exp(–l) l: width of patterns Purpose of this study Whether exponential transients exist in lattices of coupled circle maps. 7

  8. 2 1 N 3 4 5 6 7 8 2. Unidirectionally coupled maps Ring of unidirectionally coupled bistable symmetric circle maps n:index of sites, N:the number of sites t: discrete time xn(t): state of nth site at time t ε: coupling strength Bistable steady states: xn = ±1/2 (1 ≤ n ≤ N) (1a), (2) 8

  9. 2. Unidirectionally coupled maps Random initial states: xn(0) ~ N(0, 0.12) → Traveling pulse waves (xn(t): 1/2⇄ –1/2) Fig. 1(a). Transient pulse waves (ε= 0.2, K = 0.5, N = 20) simulation 9

  10. 2. Unidirectionally coupled maps Random initial states: xn(0) ~ N(0, 0.12) → Traveling pulse waves (xn(t): 1/2⇄ –1/2) Fig. 1(b). Transient pulse waves (ε= 0.8, K = 0.5, N = 20) simulation 10

  11. lh lh 2. Unidirectionally coupled maps N: even (N = 2M) Initial states: → Unstable symmetric pulse wave → Saddle manifold in the state space Stable in the subspace: xn = -xN/2+n (1≤ n ≤ N/2) (3) 11

  12. 2 1 3 4 N 5 6 7 8 3. Bidirectionally coupled maps Ring of bidirectionally coupled bistable symmetric circle maps n:index of sites, N:the number of sites t: discrete time xn(t): state of nth site at time t ε: coupling strength Bistable steady states: xn = ±1/2 (1 ≤ n ≤ N) (1b), (2) 12

  13. 3. Bidirectionally coupled maps Random initial states: xn(0) ~ N(0, 0.12) → Standing pulses (xn(t): 1/2⇄ –1/2) Fig. 1(c). Standing pulse (ε= 0.5, K = 0.1, N = 40) simulation 13

  14. 3. Bidirectionally coupled maps N: even (N = 2M) Initial states: → Unstable symmetric standing pulses → Saddles in the state space Stable in the subspace: xn = -xN/2+n (1≤ n ≤ N/2) (3) lh lh 14

  15. v1 v2 N – l l n1 n2 x1 xN 4. Changes in pulse width Locations of pulse fronts: n1, n2 Speeds of pulse fronts: v1 = Δn1/Δt, v2 = Δn2/Δt Changes in pulse width l → Difference between the speeds of two pulse fronts dl/dt = Δ(n1 – n2)/Δt = v2 – v1 15

  16. v1 v2 N – l l n1 n2 x1 xN 4. Changes in pulse width Changes in pulse width: l ~ exponentially small with pulse width: l and N – l (5) α= 2.375, β = 1.304 in unidirectionally coupled maps α = 0.651, β = 0.487 in bidirectionally coupled maps 16

  17. 4. Changes in pulse width Changes in pulse width: l Initial pulse width: l(0) = l0 < N/2 → l(T) = 0 → T(l0; N): Duration of pulses with initial pulse width l0 (5) (6) (7) 17

  18. 4. Changes in pulse width Simple forms by letting N → ∞ T(l0)~exp(l0) ・・・ Duration increases exponentially with initial pulse width (8) 18

  19. l0 5. Duration of transient pulses 1. Duration of asymmetric pulses: T(l0) T(l0)~exp(l0) Fig. 4. Duration T vs initial pulse width l0 in unidirectionally coupled maps (ε = 0.2, K = 0.5, N = 21) Fig. 7. Duration T vs initial pulse width l0in bidirectionally coupled maps 19

  20. 5. Duration of transient pulses 2. Randomly generated pulses Random initial states: xn(0) ~ N(0, 0.12) → Pulses with initial pulse width obeying the uniform distribution: l0 ~ U(0, N/2) Distribution h(T) of duration T of these pulses (9) (10) 20

  21. 5. Duration of transient pulses 2. Distribution h(T) of duration T of randomly generated pulses Cut-off: Tc = exp(αN/2)/(αβ) ≈ 3×106 (N = 20) Prob{T > Tc} ≈ 4exp(-2)/(αN) ≈ 0.357/N ≈ 0.018 (N = 20) (11) (12) Fig. 5. Distribution of duration of random traveling pulses in unidirectionally coupled maps (ε = 0.2, K = 0.5, N = 20) 21

  22. 5. Duration of tansient pulses 2. Distribution h(T) of duration T of randomly generated pulses Cut-off: Tc = exp(αN/2)/(αβ) ≈ 1.1×106 (N = 40) Prob{T > Tc} ≈ 4exp(-2)/(αN) ≈ 0.832/N ≈ 0.021 (N = 40) (11) (12) Fig. 8. Distribution of duration of random standing pulses in bidirectionally coupled maps (ε = 0.5, K = 0.1, N = 40) 22

  23. 6. Conclusion ・Rings of unidirectionally and bidirectionally coupled maps → Transient traveling pulses and standing pulses ・Duration T of transient pulses increases exponentially with initial pulse width l0. T ∝ exp(l0) ・Duration T of transient pulses generated under random initial conditions is distributed in a power law form. h(T) ~ 1/T 23

  24. 2-2. Duration of pulse waves 2. Duration of pulse waves occurring from random initial states (14) Mean: m(T)~exp(N) SD:σ(T) ~exp(N) Coefficient of variation: CV(T) > 1 Fig. 6. Mean, SD and CV of duration of random pulse waves vs N 24

More Related