Robustness of noise-induced synchronization Ngy 16 thng 4 nm 2008 - - PowerPoint PPT Presentation

robustness of noise induced synchronization
SMART_READER_LITE
LIVE PREVIEW

Robustness of noise-induced synchronization Ngy 16 thng 4 nm 2008 - - PowerPoint PPT Presentation

Robustness of noise-induced synchronization Ngy 16 thng 4 nm 2008 () Noise-induced synchronization Ngy 16 thng 4 nm 2008 1 / 25 Outline Problem statement 1 Modelling noise: Stochastic Differential Equations 2 The proof


slide-1
SLIDE 1

Robustness of noise-induced synchronization

Ngày 16 tháng 4 năm 2008

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 1 / 25

slide-2
SLIDE 2

Outline

1

Problem statement

2

Modelling “noise”: Stochastic Differential Equations

3

The proof

4

Limitations of the analysis

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 2 / 25

slide-3
SLIDE 3

Outline

1

Problem statement

2

Modelling “noise”: Stochastic Differential Equations

3

The proof

4

Limitations of the analysis

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 3 / 25

slide-4
SLIDE 4

Mainen & Sejnowski experiment

[Mainen & Sejnowski, 1995]

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 4 / 25

slide-5
SLIDE 5

Synchronization interpretation

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 5 / 25

slide-6
SLIDE 6

Outline

1

Problem statement

2

Modelling “noise”: Stochastic Differential Equations

3

The proof

4

Limitations of the analysis

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 6 / 25

slide-7
SLIDE 7

Random walk and Wiener process

Random walk (discrete-time): xt+∆t = xt + ξt∆t where (ξt)t∈N are Gaussian and mutually independent If one is interested in very rapidly varying perturbations, ∆t has to be very small Wiener process (or Brownian motion) (continuous-time): limit of the random walk when ∆t → 0

−10 −5 5 10 15 2 4 6 8 10

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 7 / 25

slide-8
SLIDE 8

Wiener process and “white noise”

Problem: a Wiener process is not differentiable (why?), thus it is not the solution of any ordinary differential equation Define formally ξt (“white noise”) = “derivative” of the Wiener process Formally: W (t) − W (0) = t

0 ξtdt or dW /dt = ξt or dW = ξtdt

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 8 / 25

slide-9
SLIDE 9

Stochastic Differential Equations

We consider processes driven by “white noise” We would like to write (but it’s not correct, because ξ does not exist) ˙ x = f(x) + g(x)ξ In integral form, it may be more correct x(t) − x(0) = t f(x)dt + t g(x)dW where the last term is a Stieltjes integral against W (which does exist) The integral form can also be written in differential form dx = f(x)dt + g(x)dW

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 9 / 25

slide-10
SLIDE 10

Definition of Itô and Stratonovich integrals

For deterministic function α, the Stieltjes integral (which generalizes Riemann integrals) against α is defined as T β(t)dα = lim

N→∞ N−1

  • 1

β(ti) [α(ti+1) − α(ti)] Thus one can define, by analogy T β(t)dW = lim

N→∞ N−1

  • 1

β(ti) [W (ti+1) − W (ti)] which is the Itô integral But one can also define T β(t)dW = lim

N→∞ N−1

  • 1

β ti + ti+1 2

  • [W (ti+1) − W (ti)]

which is the Stratonovich integral

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 10 / 25

slide-11
SLIDE 11

Independance properties

The two above definitions lead to the same result in the deterministic case (probably, C 1 is required) But there are differences in the stochastic case: Since β(ti) (present) is independent of W (ti+1) − W (ti) (future), one has, for Itô integrals E (β(ti)[W (ti+1) − W (ti)]) = E(β(ti))E(W (ti+1) − W (ti)) = 0 leading to E T β(t)dW

  • = 0

This explains Teramae claim “In the Itô formulation, [. . . ], the correlation between φ and ξ vanishes”

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 11 / 25

slide-12
SLIDE 12

Variable transformation

The variable transformation (“changement de variable” in French) formula is also different for Itô and Stratonovich integrals Consider the function y(x). In the deterministic case, one has, for instance dy dt = ∂y ∂x · dx dt

  • r

dy = ∂y ∂x dx The same rule is valid for Stratonovich integrals (Teramae’s “conventional variable transformation”): if dx = f (x)dt + g(x)dW then dy = ∂y ∂x (f (x)dt + g(x)dW )

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 12 / 25

slide-13
SLIDE 13

Itô’s formula for variable transformation

Consider the Iô SDE dx = f (x)dt + g(x)dW Then for a function y(x), one has (Itô’s formula) dy = ∂y ∂x f (x) + 1 2 ∂2y ∂x2 g(x)2

  • dt + ∂y

∂x g(x)dW This will explain Teramae’s “the disappeared correlation is compensated by the new extra drift term Z ′DZ”

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 13 / 25

slide-14
SLIDE 14

Outline

1

Problem statement

2

Modelling “noise”: Stochastic Differential Equations

3

The proof

4

Limitations of the analysis

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 14 / 25

slide-15
SLIDE 15

Phase reduction

Consider the system ˙ x = f(x) which has a limit-cycle. We would like to find a phase variable φ(x) such that: dφ dt = ω ω = constant Example: a mobile travelling on a circle with constant velocity

θ x

General case (using the chain rule): dφ dt = ∂φ ∂x (x) · dx dt = ∂φ ∂x (x) · f(x) = ω One then has to solve the above PDE to find φ

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 15 / 25

slide-16
SLIDE 16

Phase reduction(continued)

Consider now a small perturbation ξ ˙ x = f(x) + ξ Then the equation on the phase becomes dφ dt = ∂φ ∂x (x) · f(x) + ∂φ ∂x (x) · ξ = ω + ∂φ ∂x (x) · ξ This can be converted into a φ-only equation using some approximations dφ dt = ω + Z(φ)ξ This was equation (2) in [Teramae & Tanaka, 2004]

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 16 / 25

slide-17
SLIDE 17

Stratonovich to Itô switch

Actually, the authors could have done everything in Itô! Let us compute the phase equation obtained previously but using now Itô’s formula (with D = 1

2g2)

dφ = ∂φ ∂x f(x) + D ∂2φ ∂x2

  • dt + ∂φ

∂x dW As above, let Z(φ) = ∂φ

∂x . Then

∂2φ ∂x2 = ∂ ∂xZ(φ) = ∂Z ∂φ ∂φ ∂x = Z ′(φ)Z(φ) Thus dφ = (ω + Z ′(φ)DZ(φ))dt + Z(φ)dW which is equation (3) (after formal division by dt)

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 17 / 25

slide-18
SLIDE 18

Linearization

Consider ˙ φ1 = f (φ1) and ˙ φ2 = f (φ2) Then (using the Taylor expansion assuming φ1 − φ2 very small) ˙ φ1 − ˙ φ2 = f (φ1) − f (φ2) = f (φ1) − (f (φ1) + (φ2 − φ1)f ′(φ1)) = f ′(φ1)(φ1 − φ2) This explains equation (4) if we set ψ = φ1 − φ2

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 18 / 25

slide-19
SLIDE 19

Lyapunov exponent

Consider two infinitesimally close trajectories. The Lyapunov exponent λ verifies (intuitively) δφ(t) ≃ eλtδφ0

If λ > 0, then nearby trajectories diverge = instability If λ < 0, then nearby the trajectories converge = stability

(Remark: if a system is contracting, then λ < 0)

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 19 / 25

slide-20
SLIDE 20

Lyapunov exponent (continued)

Let us manipulate the above expression: eλt ≃ δφ(t) δφ0 λt = ln δφ(t) δφ0

  • λ = 1

t ln δφ(t) δφ0

  • Actually, the Lyapunov exponent is defined as (because we are

interested in long-time behaviour) λ = lim

t→∞

1 t ln δφ(t) δφ0

  • ()

Noise-induced synchronization Ngày 16 tháng 4 năm 2008 20 / 25

slide-21
SLIDE 21

Ergodic hypothesis

Consider a stochastic process x(ω, t) Any physicist knows that (ergodic hypothesis): ∀ω, t lim

T→∞

1 T T x(ω, t′)dt′ =

x(ω′, t)dω′ = E (x(t))

Time average Ensemble average

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 21 / 25

slide-22
SLIDE 22

Ergodic hypothesis (continued)

Remark now that y in equation (5) is defined as y = ln(ψ) = ln(φ1 − φ2) = ln(δφ) Remark that T ˙ y = y(T) − y(0) = ln(δφ(T)) − ln(δφ(0)) = ln δφ(T) δφ(0) thus 1 T lim

T→∞

T ˙ y = 1 T lim

T→∞ ln δφ(T)

δφ(0) = λ By the ergodic hypothesis, we then have λ = E (˙ y) which explains the first line in equation (6).

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 22 / 25

slide-23
SLIDE 23

Probability density

Let P(φ, t) denotes the time-dependent probability density of the random variable φ ∈ [0, 2π] P is constant intuitively means that φ has equal probability of being anywhere in [0, 2π]

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 23 / 25

slide-24
SLIDE 24

Outline

1

Problem statement

2

Modelling “noise”: Stochastic Differential Equations

3

The proof

4

Limitations of the analysis

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 24 / 25

slide-25
SLIDE 25

Some limitations

A lot of unproven statements (ergodicity, uniform distribution of φ in steady state,. . . ). Perhaps those statements are evident for physicists! There is a mistake in the computation of the phase equation, as pointed out by [Yoshimura & Arai, 2007] (Thank you, Francis!). However, this mistake does not alter the result. The analysis is only valid when Z is continuously differentiable up to the second-order, which is not verified for e.g. resetting neuron models (Integrate and Fire, Izhikevich,...)

() Noise-induced synchronization Ngày 16 tháng 4 năm 2008 25 / 25