CASE STUDY
The following example will show how a variety of these methods can be used
Worked example
The motion of an unforced pendulum subject to viscous damping is modelled by
the second order differential equation. | ![]() | where b, is a constant |
by substituting | ![]() | and | ![]() |
![]() | ![]() | ![]() |
substituting in the original equation gives
the equations become
![]() |
![]() |
These occur when
![]() |
![]() |
Therefore, y=0 and sinx =0
The equilibrium points are
The second step is to classify the equilibrium points of the linearisations by examining the Jacobian matrices.
Thus | ![]() | if n is odd and | ![]() | if n is even |
Tr = -b and det = -18 if n is odd and the equilibrium point of the linearisation is a saddle. By the linearisation theorem the nonlinear system shows similar behaviour close to the equilibrium points and they are therefore unstable.
Tr = -b and det = 18 if n is even. The behaviour of the equilibrium point of the linearisation depends on the value of b. If the equilibrium point is stable
the equilibrium point is unstable
the equilibrium point is a centre
By the linearisation theorem the equilibrium points of the nonlinear system are
stable if
unstable if
If b=0 the system needs further investigation.
To investigate further we look for a conserved quantity. To do this we find a first integral.
Since b=0 the equations for the system become
Since this function exists for all values of (x,y) it follows that
E(x,y)= | ![]() | is a conserved quantity. |
Next we need to see if the equilibria are local maxima or minima of E(x,y)
The equilibria occur at | ![]() | with n even therefore | ![]() | and | ![]() |
Thus the equilibria are stationary points of E(x,y)
Next find the Hessian matrix
Exx=18cosx, Eyy=1, Exy=0
The eigenvalues are 18 and 1 and so the equilibrium point is a maximum
Thus the level curves of E(x,y) are closed curves surrounding the equilibrium points and since it is a conserved system the solutions of the system lie along these curves. Thus the solution curves also form closed curves around the equilibrium points and they are nonlinear centres and neutrally stable.
An alternative approach to this problem is to use the Lyapunov function
L(x,y)= | ![]() |
1) first we need to prove it is a Lyapunov function
provided b
If b=0 then L(x,y) is a conserved quantity ,if b>0 L(x,y) is a Lyapunov function
2)To describe the behaviour of a system we first need to find the equilibrium points
given by
the equilibrium points occur at
We need to find if these are stationary points of L(x,y)
![]() | = 0 | ![]() | = 0 at | ![]() |
Therefore the equilibria are stationary points of L(x,y)
We now need to determine whether they are maxima or minima of L(x,y)
Lxx=18cosx Lyy=1 Lxy=0
The Hessian matrices H are given by
![]() | if n is odd | ![]() | if n is even |
If n is odd tr = -17 det = -18 and the equilibrium point is a saddle of L(x,y). thus if b>0 the solution curves of the nonlinear system cross the curves of L(x,y) and the equilibria are nonlinear saddles and unstable.
If b=o the system is conservative and the solution curves of the nonlinear system must follow the level curves of L(x,y) and the equilibria of the nonlinear system will also be a nonlinear saddle and unstable.
If n is even tr = 19, det = 18 and the equilibrium point is a minimum of L(x,y).The level curves of L(x,y)=C therefore form closed curves surrounding the equilibria. If b>0 the solutions of the nonlinear system cross the curves of L(x,y) and the equilibria are therefore attractors and stable.
If b=0 L(x,y) is a conserved quantity and the system is conservative. Thus the solution curves of the system must follow the level curves of L(x,y) and will form closed curves surrounding the equilibria. Thus the equilibria are nonlinear centres and neutrally stable.