Université M.

Mammeri de Tizi-Ouzou Master 1 Modélisation Mathématique

Faculté des Sciences
Département Mathématiques Juin 2016
Responsable du module: Mme Khellas

Final Exam. Stabilité des systèmes dynamiques

Exercise 1 (03pts). Show that the origin of the system

(1) x_ = (1 + sin2 x)x

is globally exponentially stable.

Solution 1. Recall that the origin is globally exponentially stable if there exist two constants k; > 0 such
that
kx(t)k k kx (0)k exp ( t) ; for all t 0 and x (0) 2 Rn
Here n = 1 and the origin is the unique equilibrium point of (1). An integral solution of (1) is
Z t
x(t) = x (0) exp (1 + sin2 x (s))ds
0

and therefore
jx(t)j jx (0)j exp ( t)
the last estimation holds for any x (0) 2 R; thus the origin is globally exponentially stable with k = = 1:

Exercise 2 (02pts). Use Lyapunov’s indirect method for the following system to investigate whether the origin
is stable, asymptotically, or unstable: 8
< x_ 1 = 2x1 + x31
x_ 2 = x2 + x21
:
x_ 3 = x3

Solution 2. Lyapunov’s indirect method asserts that the nonlinear system is (asymptotically) stable if its
linearized system is too. The linearized system near the origin is given by
0 1 0 10 1
x_ 1 2 0 0 x1
@ x_ 2 A = @ 0 1 0 A @ x2 A
x_ 3 0 0 1 x3

The eigenvalues of the Jacobian matrix at the origin are all strictly negative, thus the nonlinear system is
asymptotically stable.

Exercise 3 (05pts). Consider the system

x_ 1 = x2 ; x_ 2 = g(x1 ) (x1 + x2 )

where g is locally Lipschitz that satis…es g(y) 1 for all y 2 R:

1. Verify that the system has a unique equilibrium point at the origin.

1
 2. Show that
Zx1
V (x) = yg(y)dy + x1 x2 + x22
0
2
is positive de…nite for all x 2 R and radially unbounded, and use it to show that the equilibrium point
x = 0 is globally asymptotically stable.

Solution 3. 1. Obvious.

2. We have
Zx1 Zx1
1
yg(y)dy ydy = x21
2
0 0

Therefore
1 2 1 1 1 1
V (x) x1 + x1 x2 + x22 = xT x = xT Qx
2 2 1 2 2
The matrix Q of the quadratic form is positive de…nite. Hence, V is positive de…nite for all x; and
1 2
radially unbounded (thanks to Rayleigh inequality :V (x) max (Q) kxk2 ):
2
The derivative of V along the trajectories of the system is given by

V_ (x) = x1 g(x1 )x_ 1 + x1 x_ 2 + x2 x_ 1 + 2x2 x_ 2

= x1 x2 g(x1 ) x1 g(x1 ) (x1 + x2 ) + x2 x2 2x2 g(x1 ) (x1 + x2 )
= g(x1 ) x1 x2 x21 x1 x2 2x1 x2 2x22 + x22
= g(x1 )xT Qx + x22

Now, since g(x1 ) 1 for all x1 2 R and xT Qx 0; we have

V_ (x) x21 + 2x1 x2 + 2x22 + x22

= x21 + 2x1 x2 + x22
= (x1 + x2 )2 0

This shows that V_ (x) is semi-positive de…nite. We need to show that the origin is the only point for
which V_ vanishes. Indeed,

V_ (x) = 0 ) x1 + x2 = 0 ) x1 (t) + x2 (t) 0 ) x_ 1 (t) + x_ 2 (t) 0

) x2 (t) 0 ) x1 (t) 0

Since V is radially unbounded and all the assumptions hold globally, we conclude that the origin is
globally asymptotically stable.

Exercise 4 (Krasovskii’s Method). [06pts]

Consider the system x_ = f (x) with f (0) = 0: Assume that f (x) is continuously di¤erentiable and the
@f
Jacobian satis…es
@x
T
@f @f
P (x) + (x) P I; 8x 2 Rn ; where P = P T > 0:
@x @x

2
 R1
1. Using the representation f (x) = 0
f ( x) xd ; show that
xT P f (x) + f T (x)P x xT x; 8x 2 Rn

2. Show that V (x) = f T (x)P f (x) is positive de…nite1 for all x 2 Rn and radially unbounded.
3. Show that the origin is globally asymptotically stable.
Z1 " T
#
@f @f
Solution 4. 1. According to the hypothesis, we have P ( x) + ( x) P d I; hence
@x @x
0
Z 1 Z 1 T
T T T @f @f
x P f (x) + f (x)P x = x P ( x) xd + xT ( x) d Px
0 @x 0 @x
Z Z !
1 1 T
@f @f
= xT P ( x) xd + xT ( x) d P x
0 @x 0 @x
Z " # !
1 T
@f @f
= xT P ( x) + ( x) P d x xT x
0 @x @x

2. It is clear that V (x) = f T (x)P f (x) is positive semide…nite. To show that it is positive de…nite, we need
2
to show that V (x) = 0 i¤ x = 0: Since P > 0; using the fact that V (x) min (P ) kf (x)k2 ; we need to
show that f (x) = 0 i¤ (or implies) x = 0; that is the origin is the only equilibrium point of the system.
Indeed, assume to the contrary, that there is z 6= 0 such that f (z) = 0: Then
zT z z T P f (z) + f T (z)P z = 0
which is a contradiction. Hence, the origin is the unique equilibrium point.
3. Let us show that V is radially unbounded. Using the estimation V (x) min (P ) kf (x)k22 ; for all x 2 Rn ;
it is enough to prove that
lim kf (x)k = +1:
kxk2 !1

xT P f (x) 1
Assume that kf (x)k M as kxk2 ! 1: According to 1: we have ; : for all x 2 Rn :
kxk22 2
xT P f (x) kxk2 kP k2 kf (x)k2 kP k2
On the other hand, M ! 0; as kxk2 ! 1; which constitutes a
kxk22 kxk2 2
kxk2
contradiction. This shows that lim V (x) = +1 and V is radially unbounded.
kxk2 !1

4. We have V (0) = 0, V (x) > 0 for all x 6= 0: and V is radially unbounded. It remains to show that
V_ (x) < 0 for all x 6= 0: Indeed, the derivative of along the trajectories of the system is
" #
T
@f @f
V_ (x) = f T (x) P (x) + (x) P f (x)
@x @x
kf (x)k22
since f (x) = 0 i¤ x = 0; we deduce that V_ (x) < 0 for all x 6= 0: Thus the origin is globally asymptotically
stable.
Exercise 5 (05pts). Describe what you have learned and understood through my course?
1
Hint. Show that f (x) = 0 i¤ x = 0: