Sunteți pe pagina 1din 7

Dynamical Systems 2013

Class 2
Department of Electrical Engineering
Eindhoven University of Technology
Siep Weiland
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 1 / 25
Outline
1 Example on population growth
2 Numerical implementations
3 Existence and uniqueness of solutions
4 Potential functions
5 Summary
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 2 / 25
Example
Outline
1 Example on population growth
2 Numerical implementations
3 Existence and uniqueness of solutions
4 Potential functions
5 Summary
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 3 / 25
Example
example -population growth
N(t) number of species at time t. Model

N = rN (1 N/) = f (N)
with growth constant r > 0 and carrying capacity > 0.
2 0 2 4 6 8 10 12 14
6
5
4
3
2
1
0
1
2
3
The function f (N) with r = 1 and = 10
Fixed points at N

1
= 0 and N

2
= .
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 4 / 25
Example
the logistic equation
Model:

N = rN (1 N/)
called the logistic equation or Verhulst model
after Belgian mathematician Pierre-Francois Verhulst (1804-1849)
models population growth with carrying capacity and growth rate r
(also called Malthusian parameter).
growth is negative if N > : mortality rate.
growth is positive if N < : birth rate.
Derivative f

(N) = r (1 2N/) gives
f

(N

1
) = f

(0) > 0 so N

1
= 0 is unstable
f

(N

2
) = f

() < 0 so N

2
= is stable.
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 5 / 25
Example
Pierre-Francois Verhulst (1804-1849)
Belgian mathematician
PhD at Gent University in
1825.
Estimated in 1846 the
upperbound = 9.439.000
for Belgian population.
(Currently: = 11.007.020
and r = 0.0082)
Contributions:
Calculus of variations
Number theory
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 6 / 25
Example
Solutions of logistic equation
Solutions with dierent initial conditions N(0) = N
0
:
0 1 2 3 4 5 6 7 8 9 10
0
5
10
15
time t
s
o
lu
tio
n
x
solutions of logistic equation
0 1 2 3 4 5 6 7 8 9 10
0
5
10
15
time t
s
o
lu
tio
n
x
solutions of logistic equation
r = .8, = 10 r = .2, = 10
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 7 / 25
Example
analytic expression of solution
In fact, one can show that
N(t) =
N
0
N
0
+ ( N
0
) exp(rt)
(1)
Indeed:
separation of variables:

N
N(1 N/)
=

N
N
+

N
N
= r
integrate: log(|N|) log(| N|) = rt + C
0
rewrite: |N|/| N| = C exp(rt) where C = exp(C
0
)
now solve for N to infer (1)
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 8 / 25
Numerical implementations
Outline
1 Example on population growth
2 Numerical implementations
3 Existence and uniqueness of solutions
4 Potential functions
5 Summary
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 9 / 25
Numerical implementations
Matlab implementation to compute solutions
How to solve initial value problem x = f (x), x(0) = x
0
in Matlab?
>> t0=0; te=10;
>> x0=1;
>> r=1; alpha=10;
>> [t,x] = ode45(@logistic,[t0 te],x0,[],r,alpha);
>> plot(t,x);
>> title( solution logistic equation)
Here, logistic is the function
function dxdt = logistic(t,x,r,alpha);
dxdt=r*x*(1-x/alpha);
Mind the arguments!! (t, x) are obligatory. (r , ) are passed to logistic
in last arguments of ode45
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 10 / 25
Numerical implementations
ODE solvers
Matlab has many solvers:
variable time-step
non-sti solvers
ode45
ode23
xed time-step solvers
ode1
ode2
ode3
ode4
variable time-step sti
solvers
ode15s
ode23s
ode23t
ode23tb
. . . You choose. . .
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 11 / 25
Existence and uniqueness of solutions
Outline
1 Example on population growth
2 Numerical implementations
3 Existence and uniqueness of solutions
4 Potential functions
5 Summary
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 12 / 25
Existence and uniqueness of solutions Existence
existence of solutions
Alarming problem 1
Consider ow x = 1 + x
2
with x(0) = 0
Integration yields

dt =

1
1 + x
2
dx
with solution x(t) = tan(t).
Solution exists only for /2 < t < /2 as
x(t) if t /2
Bottom line:
Solutions may not exist for all times!
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 13 / 25
Existence and uniqueness of solutions Uniqueness
uniqueness of solutions
Alarming problem 2
Consider ow x = x
1/3
with x(0) = 0
Integration yields

dt =

x
1/3
dx
or, equivalently, t + C =
3
2
x
2/3
. Hence, solution x(t) =

2
3
t

3/2
.
But x(t) = 0 is obviously also a solution.
So we have two solutions emerging from x(0) = 0!
Bottom line:
Solutions may not be unique!
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 14 / 25
Existence and uniqueness of solutions Existence and uniqueness theorem
existence and uniqueness theorem
But fortunately life has some guarantees:
Theorem
If both f and f

are continuous in an open interval I containing the point
x
0
, then the ow x = f (x) with initial value x(0) = x
0
has a unique
solution x(t) on some time interval (, ) with > 0.
These two smoothness conditions on f guarantee both existence and
uniqueness of solutions on some interval of time around t = 0.
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 15 / 25
Existence and uniqueness of solutions Existence and uniqueness theorem
Remarks on existence theorem
Remarks on existence and uniqueness theorem:
Solutions of x = f (x), x(0) = x
0
are also called ows and sometimes
denoted as x(t; x
0
).
uniqueness means that two solutions with the same initial condition
will be equal on the largest interval of time where both are dened.
Solutions will never cross each other: Impossible that
x(t
0
; x
1
) = x
0
= x(t
0
; x
2
)
for some time instant t
0
when x
1
= x
2
.
(since if they would, there is no unique solution with initial condition
x(t
0
) = x
0
.)
We will mostly assume existence and uniqueness of solutions.
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 16 / 25
Potential functions
Outline
1 Example on population growth
2 Numerical implementations
3 Existence and uniqueness of solutions
4 Potential functions
5 Summary
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 17 / 25
Potential functions
potentials -denition
Denition
A function V : R R is a potential function for the ow x = f (x) if
f =
dV
dx
.
Why interesting?
Let V(t) = V(x(t)) then

V(t) =
dV
dx
dx
dt
=

dV
dx

2
0 for all t
so that V(t) is non-increasing along solutions of the ow
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 18 / 25
Potential functions
potentials and stability
Theorem
If V is a potential for the ow x = f (x) then
V(x(t)) is non-increasing along its solutions.
x

is a local minimum of V if and only if x

is a stable xed point.


x

is a local maximum of V if and only if x

is an unstable xed point.


Hence,
Stability of xed points can be read o from any potential function
Will relate to potential elds and potential energy later
Try to prove this theorem!
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 19 / 25
Potential functions
examples of potential functions
Example
Consider ow x = x x
3
. This has potential
V(x) =

x
0

3
d = x
2
/2 + x
4
/4 + C
Conclude
For any initial condition x
0
, V(x(t)) will be non-increasing
V has local minima at x

= 1.
V has local maximum at x

= 0
In words: the potential of a system is never increasing and decreases
towards a local minimum.
(no philosophical connotation is meant here . . . )
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 20 / 25
Potential functions
examples of potential functions
The potential function V(x) = x
2
/2 + x
4
/4
1.5 1 0.5 0 0.5 1 1.5
0.25
0.2
0.15
0.1
0.05
0
0.05
0.1
0.15
Potential function V
x
V
(x
)
Note: x on horizontal axis.
Local minima correspond to stable xed points, local maxima to unstable
ones.
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 21 / 25
Potential functions
potentials and positive invariance
Denition
A set V is (positive) invariant for x = f (x) if the ow x(t; x
0
) V for all
time t 0 whenever x
0
V.
So, V is invariant if solutions stay in V once they start there.
Example
Some examples of positive invariant sets
if V = {x

} consist of a xed point only


V = [/2, 3/2] is positive invariant for the ow x = sin(x)
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 22 / 25
Potential functions
potentials and positive invariance
Suppose V is potential function for system x = f (x).
Dene sub-level set
V

:= {x
0
| V(x
0
) }
of all initial points with potential at most .
Theorem
The sub-level set V

is positive invariant for x = f (x) for any value of .


So:
x
0
V

= x(t) V

for all t 0
Proof: Potential functions are non-increasing functions along trajectories.
Thus: V(x
0
) implies V(x(t)) V(x(0)) .
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 23 / 25
Summary
Outline
1 Example on population growth
2 Numerical implementations
3 Existence and uniqueness of solutions
4 Potential functions
5 Summary
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 24 / 25
Summary
Summary
We studied logistic equation in detail.
Can compute solutions of nonlinear autonomous systems in Matlab
Existence and uniqueness theorem.
Flows never intersect each other.
Introduced potential functions for autonomous systems
You can now characterize stability of xed point x

by
computing explicit solutions
looking at phase diagrams
determining sign of f

(x

)
determine extremal points of potential functions.
to previous class to next class
Class 2 (TUE) Dynamical Systems 2013 Siep Weiland 25 / 25

S-ar putea să vă placă și