Limit Cycle : Nonlinear Phenomenon
A limit cycle is an isolated closed trajectory
Neighboring trajectories are not closed; they spiral either toward or
away from the limit cycle
Self-sustained Oscillations
Example:
ṙ = r (1 − r2 ) = f (r); θ̇ = 1
where r ≥ 0
Radial and angular dynamics uncoupled
Fixed points:
I r∗ = 0 : Unstable
I r∗ = 1 : Stable
Van der Pol Oscillator
ẍ + µ (x2 − 1) ẋ + x = 0
where µ ≥ 0 is a parameter
Nonlinear Damping
I |x| > 1 : Damping
I |x| < 1 : Pumping
Stable Limit Cycles
Ruling Out Closed Orbits
I Closed orbits are impossible in “gradient systems” which can
be written in the form ẋ = −∇V , where V (x) is a
continuously differentiable single-valued scalar function (a
“potential function”)
I Consider a system ẋ = f (x) with a fixed point x∗ .
Suppose that we can find a Lyapunov function, i.e., a
continuously differentiable, real-valued function V (x) with the
following properties:
I V is positive definite, i.e. V (x) > 0 for all x 6= x∗ and
V (x∗ ) = 0
dV ∗
dt < 0 for all x 6= x
I
∗
I Then x is globally asymptotically stable: for all initial
conditions, x(t) → x∗ as t → ∞
I So the system has no closed orbits
Dulac’s Criterion
I Let ẋ = f (x) be a continuously differentiable vector field
defined on a simply connected subset R of the plane
I If there exists a continuously differentiable, real-valued
function g(x) such that ∇.(g ẋ) has one sign throughout R,
then there are no closed orbits lying entirely in R
Poincaré Bendixson Theorem
I R is a closed, bounded subset of the plane
I ẋ = f (x) is a continuously differentiable vector field on an
open set containing R
I R does not contain any fixed points
I There exists a trajectory C that is confined in R, i.e. it starts
in R and stays in R for all future time
I Then either C is a closed orbit, or it spirals toward a closed
orbit as t → ∞
I So R contains a closed orbit
Trapping Region R : a closed connected set such that the vector
field points inward everywhere on the boundary of R
Then all trajectories in R are confined
Glycolysis
I Fundamental biochemical process by which living cells obtain
energy by breaking down sugar
I In yeast cells, as well as in yeast or muscle extracts, glycolysis
can proceed in an oscillatory fashion, with the concentrations
of various intermediates changing cyclically with time period
of several minutes
A simple model of these oscillations in dimensionless form:
ẋ = −x + ay + x2 y
ẏ = b − ay − x2 y
Here x and y are the concentrations of adenosine diphosphate
(ADP) and fructose-6-phosphate (F6P), and a, b > 0 are
parameters
Nullclines
ẋ = 0 on curve y = x/(a + x2 ) — vertical arrows
ẏ = 0 on curve y = b/(a + x2 ) — horizontal arrows
Trapping Region
ẋ − (−ẏ) = −x + ay + x2 y + (b − ay − x2 y) = b − x
dy ẏ
So if x > b then −ẏ > ẋ and dx = ẋ is more negative than −1
Since all arrows along the boundary are inwards, the marked region
is a trapping region
However, we still cannot conclude that there is a closed orbit inside
the trapping region, as there is a fixed point in the region at the
intersection of the nullclines
If the fixed point is a repeller, then we can prove the existence of a
closed orbit in the modified punctured region
Punctured Region removing the Fixed Point
ẋ = f (x, y) = −x + ay + x2 y
ẏ = g(x, y) = b − ay − x2 y
Jacobian:
!
∂f ∂f
a + x2
∂x ∂y −1 + 2xy
A= ∂g ∂g =
∂x ∂y
−2xy −(a + x2 )
Fixed Points: x∗ = b; y∗ = b
a+b2
4 +(2a−1)b2 +(a+a2 )
∆ = a + b2 > 0, τ = −b a+b2
For (x∗ , y ∗ ) to be a repeller : τ >0
1 √
b2 = (1 − 2a ± 1 − 8a)
2
Stable Limit Cycle
Weakly Nonlinear Oscillator
ẍ + x + h(x, ẋ) = 0
where 0 ≤ << 1 and h(x, ẋ) is a smooth function
For example:
I Duffing Oscillator:
ẍ + x + x3 = 0
I van der Pol oscillator with very small values of µ:
ẍ + x + ẋ(x2 − 1) = 0
= 0.1; Initial condition close to (0, 0)
Slowly winding spiral goes to approximately circular limit cycle
Relaxation Oscillation
Shape of Periodic Orbits
For instance, reconsider the Van der Pol oscillator:
ẍ + µ (x2 − 1) ẋ + x = 0
When µ >> 1 : strongly nonlinear limit
Extremely slow buildup followed by a sudden discharge :
Relaxation Oscillations
Found in stick-slip processes, firing of neuronal cells, etc...
d 1
ẍ + µ (x2 − 1) ẋ = (ẋ + µ[ x3 − x])
dt 3
Let
1
F (x) = x3 − x, w = ẋ + µF (x)
3
Then the van der Pol system gives:
ẇ = ẍ + µẋ(x2 − 1) = −x
Implying
ẇ = −x, ẋ = w − µF (x)
Further change of variable: y = w/µ gives
ẋ = µ[y − F (x)]
1
ẏ = − x
µ
Separation of timescales
1 3
Cubic nullcline where ẋ = 0 : y = F (x) = x −x
I Suppose that the initial condition is not too close to the cubic
nullcline, i.e. y − F (x) ∼ O(1), then ẋ = µ[y − F (x)]
implies that |ẋ| ∼ O(µ) >> 1 whereas |ẏ| ∼ O( µ1 ) << 1
I So the velocity is large in the horizontal direction and small in
the vertical direction, so trajectories move practically
horizontally
I If the initial condition is above the nullne, then y − F (x) > 0
and so ẋ > 0: trajectory moves toward the nullcline
I As it approaches the null-cline, y − F (x) gets smaller.
I When y − F (x) ∼ O(µ−2 ), then ẋ and ẏ become comparable
∼ O(µ−1 )
I Crosses cubic nullcline vertically
van der Pol oscillator with µ = 10
