Nonlinear Dynamics Nonlinear Dynamics
Moreno Marzolla
Dip. di Informatica—Scienza e Ingegneria (DISI) Università di Bologna
http://www.moreno.marzolla.name/
Complex Systems 2
Introduction: Dynamics of Simple Maps
Complex Systems 4
Dynamical systems
● A dynamical system can be informally defined as any system where some fixed rule describes the time
dependence of the position and velocity of a point in geometric space
● Examples
– Planet orbiting around a star
– Oscillating pendulum
– Chemical reaction
– Cellular automata (more about these later)
Fixed points
● There are many different types of motion
● For example, a moving object may reach a fixed point
Fixed point (e.g., a pendulum coming to a complete stop due to friction).
Limit cycle (the system state eventually repeats itself; e.g., planet orbiting around a star)
Quasiperiodi orbit (the system is periodic, but its state does not precisely repeat; e.g.,
multiple planets orbiting a star with non-resonant orbits)
Complex Systems 6
Chaos
● For a long time it was believed that every dynamical system had either a fixed point, a periodic orbit or a quasiperiodic orbit
● Now, we understand that there are plenty of examples of systems that do not fall in any of the above classes
– Turbulence in water or air
– Wobble of planets following complicate orbits
– Weather pattern
– Electric activity of the brain
– Double rod pendulum
Example
Complex Systems 8
Logistic map
● Simple model of population growth
– when the population size is small, the population will increase at a rate proportional to the current population.
– the growth rate will decrease at a rate proportional to the value obtained by taking the theoretical "carrying capacity"
of the environment less the current population
● Note: the book uses the slightly different formulation
we will not use this: we adopt the standard formulation at the top of this slide, as commonly used
x
n+1= r x
n( 1− x
n) , r ∈[0, 4] , x
n∈[ 0,1]
x
n+1= 4 r x
n( 1−x
n)
Logistic map
● The equation is fully deterministic: apparently, nothing surprising can happen there
x
n+1= f ( x
n)= r x
n( 1− x
n)
0 1 / 2 1 x
y r / 4
Complex Systems 10
Fixed points
● We want to study the steady-state dynamics of the logistic map, for every value of r
– We start from a given x0 and iterate the recurrence xn+1 = f(xn) = rxn( 1 – xn )
● What are the fixed points of f?
– Those values x for which f(x) = x
Logistic map
● There are two fixed points: x = 0 and x = (r - 1) / r
– The second one is valid only if r ≥ 1
x
n+1= f ( x
n)= r x
n( 1− x
n)
0 1 / 2 1 x
y r / 4
y = x
(r-1) / r
Complex Systems 12
Iterates of the logistic map (r = 2.8)
Iterates of the logistic map (r = 3.2)
Complex Systems 14
Iterates of the logistic map (r = 3.52)
Iterates of the logistic map (r = 4)
Complex Systems 16
Classification of fixed points
● Let xF be a fixed point for function f
– If f '(xF) = 0 super-stable
– If f '(xF) < 1 attracting and stable
– If f '(xF) = 1 neutral
– If f '(xF) > 1 repelling and unstable
The logistic map
● r ≤ 1
– Iterations eventually converge to the fixed point 0 (stable)
● 1 < r ≤ 3
– Unstable fixed point 0
– Stable fixed point (1 - r) / r
● r > 3
– Chaotic behavior, period-doubling bifurcations
Complex Systems 18
For those mathematically inclined
Robert L. Devaney, An Introduction to Chaotic Dynamical Systems, 2nd edition, Westview Press, 2003, ISBN 978-0813340852
Complex Systems 20
Higher Dimensions
Higher dimensions
● Chaotic behavior can be observed by iterating some simple maps in higher dimensions
● Example: Arnold's cat map
Γ :( x , y)→ ( ( 2x+ y) mod 1,( x+ y) mod 1 )
Complex Systems 22
Arnold's cat map
● This (and similar) map is usually shown to illustrate the Poincaré recurrence theorem
– Certain systems will, after a sufficiently long but finite time, return to a state very close to the initial state
● Iteration of the cat map eventually produces the initial image
Baker's map
Complex Systems 24
Baker's map
Iterating the baker's map
Complex Systems 26
Invariant image
Strange attractors
● Strange attractors are attractors with a fractal structure
● Let us consider the Hénon map (introduced by the French astronomer Michel Hénon) that maps two points (xt, yt) into a new pair of points (xt+1, yt+1), as follows:
where a, b are two constants
● Again, this is a fully deterministic map
x
t +1= a− x
t2+ b y
ty
t +1= x
tComplex Systems 28
The Hénon map for a=1.29, b=0.3
t xt
(Only xt is shown, random initial values)
The Hénon map for a=1.29, b=0.3
t xt
(Only xt is shown, random initial values)
Complex Systems 30
The Hénon map for a=1.29, b=0.3
t xt
(Only xt is shown, random initial values)
x = 0.838486... is an unstable attractor
The Hénon attractor viewed at different scales
(plots of y
tversus x
t)
Complex Systems 32
Hénon map attractor
● The Hénon map attractor is made of those points that map into the attractor
– In other words, the attractor is invariant in the Hénon map
The Hénon map attractor can be computed by warping a square according to the Hénon map
Producer-Consumer Dynamics
Complex Systems 34
Predator-Prey Model
Volterra and Lotka
● F = small fish population
● S = shark population
● a = reproduction rate of small fish
● b = number of small fish that a shark can eat
● c = amount of energy that a small fish supplies to a shark
– If c is large, cSF will be large, meaning that the shark population increases
● d = death rate of sharks
dF
dt = F (a−bS ) dS
dt = S (cF −d )
Population Dynamics
Fixed point at F = d/c, S = a/b
Complex Systems 36
Generalization
● The Volterra-Lotka model can be extended to an arbitrary number n of species
where xi represents the i-th species and Aij represents the effect that species j has on species i
dx
idt = x
i∑
j=1 n
A
ij( 1−x
j)
Example
Sample Models → Biology → Wolf Sheep Predation