Mechanisms of chaos in the forced NLS equation

  • Slides: 35
Download presentation
Mechanisms of chaos in the forced NLS equation Eli Shlizerman Vered Rom-Kedar Weizmann Institute

Mechanisms of chaos in the forced NLS equation Eli Shlizerman Vered Rom-Kedar Weizmann Institute of Science eli. shlizerman@weizmann. ac. il http: //www. wisdom. weizmann. ac. il/~elis/

The autonomous NLS equation • Boundary • Periodic • Even B(x+L, t) = B(x,

The autonomous NLS equation • Boundary • Periodic • Even B(x+L, t) = B(x, t) B(-x, t) = B(x, t) • Parameters • Wavenumber k = 2π / L • Forcing Frequency Ω 2

Integrals of motion • The “Particle Number”: • The “Energy”: • The “Perturbation”:

Integrals of motion • The “Particle Number”: • The “Energy”: • The “Perturbation”:

The problem Classify instabilities in the NLS equation Time evolution near plane wave

The problem Classify instabilities in the NLS equation Time evolution near plane wave

Solitons • Solitary wave • Permanent shape B (x , t) = g (x)

Solitons • Solitary wave • Permanent shape B (x , t) = g (x) • Traveling wave solution B (x , t) = g (x - vt) • Localized g (r) = 0 r →±∞ • Particle like • Preserved under collisions

Plane wave solution Bh Re(B(0, t)) Bh θ₀ θ₀ Bpw Im(B(0, t)) Heteroclinic Orbits!

Plane wave solution Bh Re(B(0, t)) Bh θ₀ θ₀ Bpw Im(B(0, t)) Heteroclinic Orbits!

Modal equations • Consider two mode Fourier truncation B(x , t) = c (t)

Modal equations • Consider two mode Fourier truncation B(x , t) = c (t) + b (t) cos (kx) • Substitute into the unperturbed eq. : [Bishop, Mc. Laughlin, Ercolani, Forest, Overmann]

General Action-Angle Coordinates • For b≠ 0 , consider the transformation: • Then the

General Action-Angle Coordinates • For b≠ 0 , consider the transformation: • Then the systems is transformed to: • We can study the structure of [Kovacic]

The Hierarchy of Bifurcations Local Stability for I < 2 k 2 Fixed Point

The Hierarchy of Bifurcations Local Stability for I < 2 k 2 Fixed Point x=0 y=0 x=±x 2 y=0 Stable I>0 I > ½k 2 Unstable I > ½ k 2 - Fixed points in (x, y) are circles in 4 dimensional space

Perturbed motion classification near the plane wave • “Standard” dyn. phenomena • Homoclinic Chaos,

Perturbed motion classification near the plane wave • “Standard” dyn. phenomena • Homoclinic Chaos, Elliptic Circles • Special dyn. phenomena • PR, ER, HR I • Close to the integrable motion H 0 Dashed – Unstable Solid – Stable

Analogy between ODE and PDE I ODE y x PDE H 0 Bpw=Plane wave

Analogy between ODE and PDE I ODE y x PDE H 0 Bpw=Plane wave

Analogy between ODE and PDE I ODE y x PDE H 0 +Bh=Homoclinic Solution

Analogy between ODE and PDE I ODE y x PDE H 0 +Bh=Homoclinic Solution

Analogy between ODE and PDE I ODE y x PDE -Bh=Homoclinic Solution H 0

Analogy between ODE and PDE I ODE y x PDE -Bh=Homoclinic Solution H 0

Analogy between ODE and PDE I ODE y x PDE H 0 +Bsol=Soliton (X=L/2)

Analogy between ODE and PDE I ODE y x PDE H 0 +Bsol=Soliton (X=L/2)

Analogy between ODE and PDE I ODE y x PDE +Bsol=Soliton (X=0) H 0

Analogy between ODE and PDE I ODE y x PDE +Bsol=Soliton (X=0) H 0

I Numerical simulations - Surface plot I H 0 H 0

I Numerical simulations - Surface plot I H 0 H 0

B plane plot

B plane plot

EMBD

EMBD

I-γ plot

I-γ plot

Conclusions • Three different types of chaotic behavior and instabilities in Hamiltonian perturbations of

Conclusions • Three different types of chaotic behavior and instabilities in Hamiltonian perturbations of the NLS are described. • The study reveals a new type of behavior near the plane wave solution: Parabolic Resonance. • Possible applications to Bose-Einstein condensate.

Characterization Tool • An input: Bin(x, t) – can we place this solution within

Characterization Tool • An input: Bin(x, t) – can we place this solution within our classification? • Quantitative way for classification (tool/measure) HC - O(ε), HR - O(ε 1/2), PR - O(ε 1/3) • Applying measure to PDE results

The measure: σmax y x Measure: σmax = std( |B 0 j| max)

The measure: σmax y x Measure: σmax = std( |B 0 j| max)

σmax PDF for fixed ε

σmax PDF for fixed ε

σmax dependence on ε

σmax dependence on ε

Future Work • Capturing the system into PR by variation of the forcing •

Future Work • Capturing the system into PR by variation of the forcing • Instabilities in the BEC • Resonant surface waves

Thank you!

Thank you!

Summary • We analyzed the modal equations with the “Hierarchy of Bifurcations” • Established

Summary • We analyzed the modal equations with the “Hierarchy of Bifurcations” • Established the analogy between ODE and PDE • Numerical simulations of instabilities • Characterization tool

I Analogy between ODE and PDE y H 0 x -Bsol=Soliton (X=L/2) Bpw=Plane wave

I Analogy between ODE and PDE y H 0 x -Bsol=Soliton (X=L/2) Bpw=Plane wave -Bh=Homoclinic Solution +Bsol=Soliton (X=0) +Bh=Homoclinic Solution

The Hierarchy of Bifurcations We can construct the EMBD for all fixed points in

The Hierarchy of Bifurcations We can construct the EMBD for all fixed points in the model:

Previous experiments D. Mc. Laughlin, K. Mc. Laughlin, Overmann, Cai

Previous experiments D. Mc. Laughlin, K. Mc. Laughlin, Overmann, Cai

Evenness condition Without evenness: • For small L - the solutions are correlated D.

Evenness condition Without evenness: • For small L - the solutions are correlated D. Mc. Laughlin, K. Mc. Laughlin, Overmann, Cai

Local Stability • Plane wave: B(0, t)= c(t) • Introduce x-dependence of small magnitude

Local Stability • Plane wave: B(0, t)= c(t) • Introduce x-dependence of small magnitude B (x , t) = c(t) + b(x, t) • Plug into the integrable equation and solve the linearized equation. From dispersion relation get instability for: 0 < k 2 < |c|2

Local Stability • But k is discretized by L so kj = 2πj/L for

Local Stability • But k is discretized by L so kj = 2πj/L for j = 0, 1, 2… (j - number of LUMs) • Substitute to 0 < k 2 < |c|2 and get 2πj/L < |c| < 2π(j+1)/L • As we increase the amplitude the number of LUMs grows.

Validity of the model • For plane wave (b=0): • Substituting the condition for

Validity of the model • For plane wave (b=0): • Substituting the condition for |c| for 1 LUM: 2πj/L < |c| < 2π(j+1)/L j=1 • Then the 2 model is plausible for I < 2 k 2

Analogy between ODE and PDE • Constants of motion • The solution

Analogy between ODE and PDE • Constants of motion • The solution