Lecture 4

Chapter 4: Lyapunov Stability

Eugenio Schuster
[email protected]

Mechanical Engineering and Mechanics

Lehigh University

Lecture 4 – p. 1/86
 Autonomous Systems
Consider the autonomous system

ẋ = f (x) (1)

where f : D → Rn is a locally Lipschitz map from a domain

D ⊂ Rn into Rn . Suppose x̄ = 0 ∈ D is an equilibrium point
of (1).

Our goal is to characterize and study stability of the

equilibrium x̄ = 0 (no loss of generality).

Lecture 4 – p. 2/86
 Stability
Definition 4.1: The equilibrium point x = 0 of (1) is

stable if, for each ǫ > 0, there is δ = δ(ǫ) > 0 such that

kx(0)k < δ ⇒ kx(t)k < ǫ, ∀t ≥ 0

unstable if not stable

asymptotically stable if it is stable and δ can be chosen
such that
kx(0)k < δ ⇒ lim x(t) = 0
t→∞

The ǫ − δ requirement for stability takes a challenge-answer

form.

Lecture 4 – p. 3/86
 Stability
“Stability is a property of the equilibrium, not of the system"
Stability of the equilibrium is equivalent to stability of the
system only when there exists only one equilibrium (e.g.,
linear systems). In this case stability ≡ global stability.
The equilibrium point x = 0 of (1) is
attractive if there is δ > 0 such that

kx(0)k < δ ⇒ lim x(t) = 0

t→∞

Example: Attractive but unstable

asymptotically stable (a.s.) if it is stable and attractive.
globally asymptotically stable (g.a.s.) if a.s. ∀x(0) ∈ Rn .

Lecture 4 – p. 4/86
 Derivative along the trajectory
Let V : D → R be a continuously differentiable
Definition:
function defined in a domain D ∈ Rn that contains the
origin. The derivative of V along the trajectory (solution) of
(1), denoted by V̇ (x) is given by
n
X ∂V
V̇ (x) = ẋi
∂xi
i=1
n
X ∂V
= fi (x)
∂xi
i=1
∂V ∂V ∂V
= [ , ,..., ][f1 (x), f2 (x), . . . , fn (x)]T
∂x1 ∂x2 ∂xn
∂V
= f (x)
∂x
Lecture 4 – p. 5/86
 Lyapunov Stability Theorem
Theorem 4.1:Let x = 0 be an equilibrium for (1) and D ∈ Rn
be a domain containing x = 0. Let V : D → R be a
continuously differentiable function, such that

V (0) = 0 and V (x) > 0 in D − {0} (2)

V̇ (x) ≤ 0 in D (3)

Then, x = 0 is stable. Moreover, if

V̇ (x) < 0 in D − {0} (4)

then x = 0 is asymptotically stable.

Lecture 4 – p. 6/86
 Lyapunov Stability Theorem
Proof:

Figure 1: Geometric representation of sets.

Lecture 4 – p. 7/86
 Lyapunov Stability Theorem
Lyapunov function candidate

V (0) = 0 and V (x) > 0 in D − {0}

Lyapunov function

V (0) = 0 and V (x) > 0 in D − {0}

V̇ (x) ≤ 0 in D

Lecture 4 – p. 8/86
 Lyapunov Stability Theorem
Lyapunov surface (level surface, level set)

{x|V (x) = c}

Figure 2: Level surfaces of a Lyapunov function.

Lecture 4 – p. 9/86
 Lyapunov Stability Theorem
Positive definite

V (0) = 0, V (x) > 0, ∀x 6= 0

Positive semidefinite

V (0) = 0, V (x) ≥ 0, ∀x 6= 0
V (x) is negative (semi)definite if −V (x) is positive
(semi)definite
Lyapunov Theorem
V pdf + V̇ nsdf → stable
V pdf + V̇ ndf → asymptotically stable

Lecture 4 – p. 10/86
 Lyapunov Stability Theorem
Example 4.4: Consider the pendulum equation with friction

ẋ1 = x2
g   
k
ẋ2 = − sin x1 − x2
l m

Conclusion Lyapunov’s Stability Theorem’s conditions are

only sufficient.

Lecture 4 – p. 11/86
 Region of Attraction
When the origin x = 0 is asymptotically stable, we are often
interested in determining how far from the origin the
trajectory can be and still converge to the origin as t → ∞.
This gives rise to the definition of region of attraction (also
called region of asymptotically stability, domain of
attraction, or basin).
Definition: Let φ(t, x) be the solution of (1) that starts at initial
state x at time t = 0. The, the region of attraction is defined
as the set of all points x such that limt→∞ φ(t, x) = 0
Question: Under what conditions will the region of attraction
be the whole space Rn ? In other words, for any initial state
x, under what conditions the trajectory φ(t, x) approaches
the origin as t → ∞, no matter how large kxk is. If an a.s.
equilibrium point at the origin has this property, it is said to
be globally asymptotically stable (g.a.s.).
Lecture 4 – p. 12/86
Global Lyapunov Stability Theorem
Let x = 0 be an equilibrium for (1). Let
Theorem 4.2:
V : Rn → R be a continuously differentiable function, such
that
V (0) = 0 and V (x) > 0, ∀x 6= 0 (5)

kxk → ∞ ⇒ V (x) → ∞ (6)

V̇ (x) < 0, ∀x 6= 0 (7)

Then, x = 0 is globally asymptotically stable and is the

unique equilibrium point.

Lecture 4 – p. 13/86
 Chetaev’s Instability Theorem
Theorem 4.3: Let x = 0 be an equilibrium for (1). Let
V : D → R be a continuously differentiable function, such
that V (0) = 0 and V (x0 ) > 0 for some x0 with arbitrarily
small kx0 k. Define a set

U = {x ∈ Br |V (x) > 0}

where
Br = {x ∈ Rn |kxk ≤ r}.
Suppose that V̇ (x) > 0 in U . Then x = 0 is unstable.

Crucial Condition: V̇ must be positive in the entire set where

V > 0.

Lecture 4 – p. 14/86
 Chetaev’s Instability Theorem
Proof:

Figure 3: Set U for V (x) > 0.

Lecture 4 – p. 15/86
 Chetaev’s Instability Theorem
Example 4.7: Consider the second order system

ẋ1 = x1 + g1 (x)
ẋ2 = −x2 + g2 (x)

where g1 () and g2 () are locally Lipschitz functions that

satisfy the inequalities

|g1 (x)| ≤ kkxk2 , |g2 (x)| ≤ kkxk2

Use the Lyapunov function candidate V (x) = 12 (x21 − x22 ) and

Chetaev’s theorem to show that the origin is unstable.

Lecture 4 – p. 16/86
 Invariance Principle
Example 4.4: Consider the pendulum equation with friction

ẋ1 = x2
g   
k
ẋ2 = − sin x1 − x2
l m

We consider the Lyapunov function candidate

x22
g   
k
V (x) = − (1 − cos x1 ) + ⇒ V̇ (x) = − x22
l 2 m

The energy Lyapunov function fails to satisfy the asymptotic

stability condition of Theorem 4.1.
But can V̇ (x) = 0 be maintained at x 6= 0?

Lecture 4 – p. 17/86
 Invariance Principle
Idea: If we can find a Lypunov function in a domain
containing the origin whose derivative along the trajectories
of the system is negative semidefinite, and if we can
establish that no trajectory can stay identicaly at points
where V̇ (x) = 0 except at the origin, then the origin is
asymptotically stable (LaSalle’s Invariance Principle).

Lecture 4 – p. 18/86
 Invariance Principle
Let x(t) be a solution of the autonomous system ẋ = f (x).
Definition: The point p is a positive limit point of x(t) if exists
a sequence {tn } with tn → ∞ as n → ∞ such that x(t) → p
as n → ∞.
Definition: The set of all positive limit points of x(t) is called
the positive limit set of x(t).
Definition: A set M is (positively) invariant w.r.t. ẋ = f (x) if
x(0) ∈ M ⇒ x(t) ∈ M for all t ∈ R (t ∈ R+ ).
Definition: x(t) approaches M as t → ∞ if for each ǫ > 0
exists T > 0 such that dist(x(t), M )(= inf x∈M kp − xk) < ǫ for
all t > T .
x(t) → M as t → ∞ ⇒ limt→∞ dist(x(t), M ) = 0
x(t) → M as t → ∞ NOT ⇒ ∃ limt→∞ x(t)

Lecture 4 – p. 19/86
 Invariance Principle
Lemma 4.1: If x(t), solution of ẋ = f (x), is bounded for all
t ≥ 0, then it has a nonempty positive limite set L+ which is
compact and invariant. Moreover,

x(t) → L+ as t → ∞

Lecture 4 – p. 20/86
 Invariance Principle
Theorem 4.4 (LaSalle’s Theorem): Let Ω ⊂ D be a compact set
that is positively invariant w.r.t. ẋ = f (x). Let V : D → R be
a continuously differentiable function such that V̇ (x) ≤ 0 in
Ω. Let E be the set of all points in Ω where V̇ (x) = 0. Let M
be the largest invariant set in E . Then, every solution
starting in Ω approaches M as t → ∞.

Unlike Lyapunov’s theorem, LaSalle’s theorem does

Note:
NOT require the function V (x) to be positive definite.

Note: When we are interested in showing that x(t) → 0 as

t → ∞, we need to establish that the largest invariant set in
E is the origin.

Lecture 4 – p. 21/86
 Invariance Principle
Proof:

Lecture 4 – p. 22/86
 Invariance Principle
Corollary 4.1 (4.2):Let x = 0 be an equilibrium point of
ẋ = f (x). Let V : D(Rn ) → R be a continuously
differentiable (radially unbounded, positive definite) function
on a domain D containing the origin (on Rn ), such that
V̇ (x) ≤ 0 in D (in Rn ). Let S = {x ∈ D(Rn )|V̇ (x) = 0} and
suppose that no solution can stay identically in S , other
than the trivial solution. Then, the origin is (globally)
asymptotically stable.

Lecture 4 – p. 23/86
 Invariance Principle
LaSalle’s theorem
relaxes the negative definiteness requirement for V̇ (x)
of Lyapunov’s theorem
does not require V (x) to be positive definite
gives an estimate of the region of attraction Ω which is
not necessarily a level set of V (x), i.e.,
Ωc = {x ∈ Rn |V (x) ≤ c}
applies not only to equilibrium points but also to
equilibrium sets

Lecture 4 – p. 24/86
 Invariance Principle
Example:

ẋ = −|x|x + (1 − |x|)xy
1
ẏ = − (1 − |x|)x2
8

Lecture 4 – p. 25/86
 Linear Systems and Linearization
The linear time-invariant system

ẋ = Ax

has an equilibrium point at the origin.

if det(A) 6= 0 the equilibrium point is isolated
if det(A) = 0 the system has an equilibrium subspace
(nontrivial null space of A)
Note: A linear system CANNOT have multiple equilibrium
points
For a given initial state x(0), the solution of the system is

x = eAt x(0)

Lecture 4 – p. 26/86
 Linear Systems and Linearization
For all A, there exists a nonsingular transformation P such
that  
J1
−1
P AP = J = 
 . . .


Jr
where Ji is a Jordan block of order mi associated with
eigenvalue λi of A, i.e.,
 
λi 1 0 ... ... 0
 0 λi 1 0 . . . 0
 

 . ... ..
 ..

.

Ji =  .
 
 .. ... 
0 
 ..
 
...
 . 1 


0 ... ... ... 0 λi m ×m

i i

Lecture 4 – p. 27/86
 Linear Systems and Linearization
Then,
r X
X mi
eAt = P eJt P −1 = tk−1 e(λi t) Rik
i=1 k=1

where mi is the order of the Jordan block Ji . If an n × n

matrix A has a repeated eigenvalue λi of algebraic
multiplicity qi , the the Jordan blocks associated with λi have
order one if and only if rank(A − λi I) = n − qi .

Lecture 4 – p. 28/86
 Linear Systems and Linearization
Theorem 4.5: The equilibrium point x = 0 of ẋ = Ax is
1. stable ⇔ Reλi ≤ 0 and for every eigenvalue with
Reλi = 0 and algebraic multiplicity qi ≥ 2,
rank(A − λi I) = n − qi , where n is the dimension of x
2. (globally) asymptotically stable ⇔ Reλi < 0 for all i
Proof:
If Reλi > 0, e(λi t) → ∞ as t → ∞. Moreover, an
eigenvalue on the imaginary axis (Reλi = 0) could give
rise to unbounded terms if order of associated Jordan
block is higher than one (rank(A − λi I) 6= n − qi ) due to
the term tk−1 .
If Reλi < 0, e(λi t) → 0 as t → ∞, i.e., e(At) → 0 as t → ∞
(a.s.). Since x(t) depends linearly on x(0), a.s. of origin
is global.
Lecture 4 – p. 29/86
 Linear Systems and Linearization
When all eigenvalues of A satisfy Reλi < 0, A is called a
Hurwitz matrix or a stability matrix. The origin of ẋ = Ax is
a.s. if and only if A is Hurwitz.
Lyapunov function candidate

V (x) = xT P x, P > 0, P = PT

with derivative along the trajectories of ẋ = Ax

V̇ (x) = xT (P A + AT P )x = −xT Qx

Lyapunov equation

P A + AT P = −Q, Q > 0, Q = QT

Lecture 4 – p. 30/86
 Linear Systems and Linearization
Theorem 4.6: A matrix A is a stability matrix or Hurwitz
matrix, that is, Reλi < 0 for all eigenvalues of A, if and only
if for any given positive definite symmetric matrix Q there
exists a positive definite symmetric matrix P that satisfies
the Lyapunov equation. Moreover, if A is stability matrix or
Hurwitz matrix, the P is the unique solution of the Lyapunov
equation.

Lecture 4 – p. 31/86
 Linear Systems and Linearization
Proof:

Lecture 4 – p. 32/86
 Linear Systems and Linearization
What if Q = C T C is positive semidefinite?

Theorem: A matrix A is a stability matrix or Hurwitz matrix,

that is, Reλi < 0 for all eigenvalues of A, if and only if for a
given positive semidefinite symmetric matrix Q = C T C there
exists a positive definite symmetric matrix P that satisfies
the Lyapunov equation where the pair (A, C) is observalbe.
Moreover, if A is stability matrix or Hurwitz matrix, the P is
the unique solution of the Lyapunov equation.

Lecture 4 – p. 33/86
 Linear Systems and Linearization
Proof:

Lecture 4 – p. 34/86
 Linear Systems and Linearization
Theorem 4.7 (Lyapunov’s First Method or Lyapunov’s Indirect
Method): Let x = 0 be an equilibrium point for the nonlinear
system
ẋ = f (x)
where f : D → Rn is continuously differentiable and D is a
neighborhood of the origin. Let

∂f 
A= (x)
∂x x=0

1. The origin is asymptotically stable if Reλi < 0 for all i

2. The origin is unstable if Reλi > 0 for some i
Note:If A has eigenvalues on imaginary axis and no
eigenvalues with Reλi > 0 we CANNOT make conclusions.
Lecture 4 – p. 35/86
 Linear Systems and Linearization
Proof:

Lecture 4 – p. 36/86
 Nonautonomous Systems
Consider the nonautonomous system

ẋ = f (t, x) (8)

where f : [0, ∞] × D → Rn is piecewise continuous in t and

locally Lipschitz in x on [0, ∞] × D, and D ⊂ Rn is a domain
that contains the origin x = 0. The origin is an equilibrium
point of (8) at t = 0 if
f (t, 0) = 0, ∀t ≥ 0

NOTE: An equilibrium point at the origin could be a

translation of a nonzero equilibrium point, or more generally,
a translation of a nonzero solution (trajectory) of the system.

Lecture 4 – p. 37/86
 Nonautonomous Systems
Suppose ȳ(τ ) is a solution of the system
dy
= g(τ, y)
dτ
defined for all τ ≥ a. The change of variables
x = y − ȳ(τ ), t=τ −a

transforms the system into the form

˙ + a) , f (t, x)
˙ ) = g(t + a, x + ȳ(t + a)) − ȳ(t
ẋ = g(τ, y) − ȳ(τ

Since
˙ + a) = g(t + a, ȳ(t + a)),
ȳ(t ∀t ≥ 0
the origin x = 0 is an equilibrium point of the transformed
system at t = 0.
Lecture 4 – p. 38/86
 Nonautonomous Systems
By examining the stability behavior of the origin as an
equilibrium point for the transformed system, we
determine the stability behavior of the solution ȳ(τ ) of
the original system.

Stability of trajectories ≡ Stability of equilibria of nonautonomous systems

If ȳ(τ ) is not constant, the transformed system will be

non-autonomous even when the original system
autonomous, i.e., even when g(τ, y) = g(y).
The stability behavior of solutions in the sense of
Lyapunov can be done only in the context of studying
the stability behavior of the equilibria of
non-autonomous systems.

Lecture 4 – p. 39/86
 Stability
While the solutions of autonomous systems depend only on
t − t0 , the solutions of nonautonomous systems may
depend on both t and t0 . We then refine definitions.
Definition 4.4:The equilibrium point x = 0 of (8) is
stable if, for each ǫ > 0, there is δ = δ(ǫ, t0 ) > 0 such that

kx(0)k < δ ⇒ kx(t)k < ǫ, ∀t ≥ t0 ≥ 0 (9)

uniformly stable (u.s.) if δ is independent of t0

unstable if not stable
attractive if there is a positive constant c = c(t0 ) such
that
x(t) → 0 as t → ∞, for all kx(t0 )k < c
uniformly attractive (u.a.) if c is independent of t0

Lecture 4 – p. 40/86
 Stability
asymptotically stable (a.s) if stable and attractive
uniformly asymptotically stable (u.a.s) if u.s. and u.a.
globally uniformly asymptotically stable (g.u.a.s) if u.a.s.
and c arbitrarily large

The equilibrium point x = 0 of (8) is

Definition 4.5:
exponentially stable if there exist positive constants c, k ,
and λ such that

kx(t)k ≤ kkx(t0 )ke−λ(t−t0 ) , ∀ kx(t0 )k < c (10)

and globally exponentially stable if (10) is satisfied for any

initial state x(t0 ).

Lecture 4 – p. 41/86
 Comparison Functions
The solutions of nonautonomous systems depend on both t
and t0 . We refine the stability definitions, so that they hold
uniformly in the initial time t0 , using special comparison
functions.
Definition 4.2: A continuous function α : [0, a) → [0, ∞) is said
to belong to class K if it is strictly increasing and α(0) = 0. It
is said to belong to class K∞ if a = ∞ and α(r) → ∞ as
r → ∞.
Definition 4.3: A continuous function
β : [0, a) × [0, ∞) → [0, ∞) is said to belong to class KL if for
each fixed s, the mapping β(r, s) belongs to class K with
respect to r and, for each fixed r, the mapping β(r, s) is
decreasing with respect to s and β(r, s) → 0 as s → ∞.

Lecture 4 – p. 42/86
 Comparison Functions
Examples:

α(r) = rc ∈ K if c > 0
α(r) = rc ∈ K∞ if c > 0
β(r, s) = rc e−s ∈ KL if c > 0

NOTE: See more in book (Example 4.16)

Lecture 4 – p. 43/86
 Comparison Functions
The next lemma states properties of class K and class KL
functions
Lemma 4.2: Let α1 and α2 be class K functions on [0, a), α3
and α4 be class K∞ functions, and β be a class KL
function. Denote the inverse of αi by αi−1 . Then,
α1−1 is defined on [0, α1 (a)) and belongs to class K
α3−1 is defined on [0, ∞) and belongs to class K∞
α1 ◦ α2 belongs to class K
α3 ◦ α4 belongs to class K∞
σ(r, s) = α1 (β(α2 (r), s)) belongs to class KL

Lecture 4 – p. 44/86
 Comparison Functions
The next lemmas show how class K and class KL enter
into Lyapunov analysis
Lemma 4.3: Let V : D → R be a continuous positive definite
function defined on a domain D ⊂ Rn that contains the
origin. Let Br ⊂ D for some r > 0. Then, there exist class K
functions α1 and α2 , defined on [0, r), such that

α1 (kxk) ≤ V (x) ≤ α2 (kxk)

for all x ∈ Br . If D = Rn , the functions α1 and α2 will be

defined on [0, ∞) and the foregoing inequality will hold for
all x ∈ Rn . Moreover, if V (x) is radially unbounded, then α1
and α2 can be chosen to belong to class K∞ .

Lecture 4 – p. 45/86
 Comparison Functions
For a quadratic positive definite function V (x) = xT P x,
Lemma 4.3 follows from the inequality

λmin (P )kxk22 ≤ xT P x ≤ λmax (P )kxk22

Lemma 4.4: Consider the scalar autonomous differential

equation
ẏ = −α(y), y(t0 ) = y0
where α is a locally Lipschitz class K function defined on
[0, a). For all 0 ≤ y0 < a, this equation has a unique solution
y(t) defined for all t ≥ t0 . Moreover,

y(t) = σ(y0 , t − t0 )

where σ is a class KL defined on [0, a) × [0, ∞).

Lecture 4 – p. 46/86
 Stability
Lemma 4.5: The equilibrium point x = 0 of (8) is

uniformly stable (u.s.) if and only if there exist a class K

function α and a positive constant c, independent of t0 ,
such that

kx(t)k ≤ α(kx(t0 )k), ∀ t ≥ t0 ≥ 0, ∀ kx(t0 )k < c (11)

uniformly asymptotically stable (u.a.s) if and only if

there exist a class KL function β and a positive
constant c, independent of t0 , such that

kx(t)k ≤ β(kx(t0 )k, t − t0 ), ∀ t ≥ t0 ≥ 0, ∀ kx(t0 )k < c (12)

globally uniformly asymptotically stable (g.u.a.s.) if and

only if inequality (12) is satisfied for any initial state x(t0 )
Lecture 4 – p. 47/86
 Stability
Theorem 4.8: Let x = 0 be an equilibrium point for (8) and
D ⊂ Rn be a domain containing the origin. Let
V : [0, ∞) × D → R be a continuous differentiable function
such that
W1 (x) ≤ V (t, x) ≤ W2 (x) (13)

∂V ∂V
+ f (t, x) ≤ 0 (14)
∂t ∂x
for all t ≥ 0 and for all x ∈ D, where W1 (x) and W2 (x) are
continuous positive definite functions on D. Then, x = 0 is
uniformly stable (u.s.).

Lecture 4 – p. 48/86
 Stability
Proof:

Figure 4: Geometric representation of sets.

Lecture 4 – p. 49/86
 Stability
Theorem 4.9: Let x = 0 be an equilibrium point for (8) and
D ⊂ Rn be a domain containing the origin. Let
V : [0, ∞) × D → R be a continuous differentiable function
such that
W1 (x) ≤ V (t, x) ≤ W2 (x) (15)

∂V ∂V
+ f (t, x) ≤ −W3 (x) (16)
∂t ∂x
for all t ≥ 0 and for all x ∈ D, where W1 (x), W2 (x) and W3 (x)
are continuous positive definite functions on D. Then, x = 0
is uniformly asymptotically stable (u.a.s.).

Lecture 4 – p. 50/86
 Stability
Theorem 4.9 (cont’): Moreover, if r and c are chosen such
that Br = {kxk ≤ r} ⊂ D and c < minkxk=r W1 (x), then every
trajectory starting in {x ∈ Br |W2 (x) ≤ c} satisfies

kx(t)k ≤ β(kx(t0 )k, t − t0 ), ∀ t ≥ t0 ≥ 0

for some class KL function β . Finally, if D = Rn and W1 (x)

is radially unbounded, then x = 0 is globally uniformly
asymptotically stable (g.u.a.s).

Lecture 4 – p. 51/86
 Stability
Proof:

Lecture 4 – p. 52/86
 Stability
Theorem 4.10: Let x = 0 be an equilibrium point for (8) and
D ⊂ Rn be a domain containing the origin. Let
V : [0, ∞) × D → R be a continuous differentiable function
such that
k1 kxka ≤ V (t, x) ≤ k2 kxka (17)

∂V ∂V
+ f (t, x) ≤ −k3 kxka (18)
∂t ∂x
for all t ≥ 0 and for all x ∈ D, where k1 , k2 and k3 are
positive constants. Then, x = 0 is exponentially stable
(e.s.). Moreover, if the assumptions hold globally, then x = 0
is globally exponentially stable (e.s.).

Lecture 4 – p. 53/86
 Stability
Proof:

Lecture 4 – p. 54/86
 LTV Systems and Linearization
The stability behavior of the origin as an equilibrium point
for the linear time varying (LTV) system

ẋ = A(t)x (19)

can be characterized in terms of the state transition matrix

of the system. The solution of (19) is given by

x(t) = Φ(t, t0 )x(t0 )

where Φ(t, t0 ) is the state transition matrix.

Lecture 4 – p. 55/86
 LTV Systems and Linearization
Theorem 4.11: The equilibrium point x = 0 of (19) is (globaly)
uniformly asymptotically stable if and only if the state
transition matrix satisfies the inequality

kΦ(t, t0 )k ≤ ke−λ(t−t0 ) , ∀t ≥ t0 ≥ 0

for some positive constants k and λ.

Note: Impractical. It needs to compute Φ(t, t0 )!
For linear systems with isolated equilibrium, stability results
are automatically globally. Moreover,
asymptotical stability ⇐⇒ exponential stability

Lecture 4 – p. 56/86
 LTV Systems and Linearization
For LTV systems, uniform asymptotic stability cannot be
characterized by the location of the eigenvalues of the
matrix A as it is done for LTI systems.
Example 4.22:

Lecture 4 – p. 57/86
 LTV Systems and Linearization
Theorem 4.12: Let x = 0 be the exponentially stable
equilibrium point of (19). Suppose A(t) is continuous and
bounded. Let Q(t) be a continuous, bounded, positive
definite, symmetric matrix. Then, there is a continuously
differentiable, bounded, positive definite, symmetric matrix
P (t) that satisfies

−Ṗ (t) = P (t)A(t) + AT (t)P (t) + Q(t)

Hence,
V (t, x) = xT P (t)x
is a Lyapunov function for the system that satisfies the
conditions of Theorem 4.10.
Note: It plays the role of Theorem 4.6 for LTV systems.
Lecture 4 – p. 58/86
 LTV Systems and Linearization
Proof: Take
Z ∞
P (t) = ΦT (τ, t)Q(τ )Φ(τ, t)dτ
t

as gramian.

Lecture 4 – p. 59/86
 LTV Systems and Linearization
Theorem 4.13:Let x = 0 be and equilibrium point for the
nonlinear system
ẋ = f (t, x)
where f : [0, ∞) × D → Rn is continuously differentiable,
D = {x ∈ Rn |kxk2 < r} and the Jacobian matrix [∂f /∂x] is
bounded and Lipschitz on D, uniformly in t. Let

∂f 
A(t) = (t, x)
∂x x=0

Then, the origin is an exponentially stable equilibrium point

for the nonlinear system if it is an exponentially stable
equilibrium point for the linear system

ẋ = A(t)x
Lecture 4 – p. 60/86
 LTV Systems and Linearization
Proof: Apply Theorem 4.12.

Lecture 4 – p. 61/86
 Converse Theorems
Lyapunov Theorems:

V exists ⇒ x = 0 u.a.s or e.s. (20)

Converse Lyapunov Theorems:

V exists ⇐ x = 0 u.a.s or e.s. (21)

Note: They just guarantee existence of Lyapunov function

V but do NOT tell us how to obtain that Lyapunov function!

Lecture 4 – p. 62/86
 Converse Theorems
Theorem 4.14:Let x = 0 be an equilibrium point for the
nonlinear system
ẋ = f (t, x) (22)

where f : [0, ∞] × D → Rn is continuously differentiable,

D = {x ∈ Rn |kxk < r}, and the Jacobian matrix [∂f /∂x] is
bounded on D, uniformly in t. Let k , λ, and r0 be positive
constants with r0 < r/k . Let D0 = {x ∈ Rn |kxk < r0 }.
Assume that the trajectories of the system satisfy

kx(t)k ≤ kkx(t0 )ke−λ(t−t0 ) , ∀ kx(t0 )k ∈ D0 , ∀t ≥ t0 ≥ 0

(23)
Note: This is the definition of exponential stability!

Lecture 4 – p. 63/86
 Converse Theorems
Then, there is a function V : [0, ∞) × D0 → R that satisfies
the inequalities

c1 kxk2 ≤ V (t, x) ≤ c2 kxk2 (24)

∂V ∂V
+ f (t, x) ≤ −c3 kxk2 (25)
∂t ∂x

∂V
≤ c4 kxk (26)
∂x

for some positive constants c1 , c2 , c3 , and c4 . Moreover, if

r = ∞ and the origin is globally exponentially stable, then
V (t, x) is defined and satisfies the aforementioned
inequalities on Rn . Furthermore, if the system is
autonomous, V can be chosen independent of t.
Lecture 4 – p. 64/86
 Converse Theorems
Theorem 4.16:Let x = 0 be an equilibrium point for the
nonlinear system
ẋ = f (t, x) (27)

where f : [0, ∞] × D → Rn is continuously differentiable,

D = {x ∈ Rn |kxk < r}, and the Jacobian matrix [∂f /∂x] is
bounded on D, uniformly in t. Let β be a class KL function
and r0 a positive constant such that β(r0 , 0) < r. Let
D0 = {x ∈ Rn |kxk < r0 }. Assume that the trajectories of the
system satisfy

kx(t)k ≤ β(kx(t0 )k, t − t0 ), ∀ t ≥ t0 ≥ 0, ∀ kx(t0 )k < c (28)

Note: This is the definition of uniform asymptotic stability!

Lecture 4 – p. 65/86
 Converse Theorems
Then, there is a function V : [0, ∞) × D0 → R that satisfies
the inequalities

α1 (kxk) ≤ V (t, x) ≤ α2 (kxk) (29)

∂V ∂V
+ f (t, x) ≤ −α3 (kxk) (30)
∂t ∂x

∂V
≤ α4 (kxk) (31)
∂x

for some positive constants α1 , α2 , α3 , and α4 are class K

functions defined on [0, r0 ]. If the system is autonomous, V
can be chosen independent of t.

Lecture 4 – p. 66/86
 Converse Theorems
Theorem 4.15:Let x = 0 be an equilibrium point for the
nonlinear system
ẋ = f (t, x) (32)

where f : [0, ∞] × D → Rn is continuously differentiable,

D = {x ∈ Rn |kxk < r}, and the Jacobian matrix [∂f /∂x] is
bounded on D, uniformly in t. Let

∂f 
A(t) = (t, x) (33)
∂x x=0

Then, x = 0 is an exponentially stable equilibrium point for

the nonlinear system if and only if it is an exponentially
stable equilibrium point for the linear system
ẋ = A(t)x (34)

Lecture 4 – p. 67/86
 Converse Theorems
Let x = 0 be an equilibrium point for the
Corollary 4.3:
nonlinear system
ẋ = f (t, x) (35)

where f : [0, ∞] × D → Rn is continuously differentiable in

some neighborhood D of the origin. Let

∂f 
A(t) = (t, x) (36)
∂x x=0

Then, x = 0 is an exponentially stable equilibrium point for

the nonlinear system if and only if A is Hurwitz.

Lecture 4 – p. 68/86
 Boundedness
Lyapunov analysis can be used to show boundedness of
the solution of the state equation, even when there is no
equilibrium point at the origin.
Example: ẋ = −x + δ sin(t), x(t0 ) = a, a > δ > 0

Lecture 4 – p. 69/86
 Boundedness
Definition 4.6: The solutions of ẋ = f (t, x) are
uniformly bounded if there exists a positive constant c,
independent of t0 ≥ 0, and for every a ∈ (0, c), there is
β = β(a) > 0, independent of t0 such that

kx(t0 )k ≤ a ⇒ kx(t)k ≤ β, ∀t ≥ t0

uniformly ultimately bounded with ultimate bound b if

there exists a positive constants b and c, independent of
t0 ≥ 0, and for every a ∈ (0, c), there is T = T (a, b) ≥ 0,
independent of t0 such that

kx(t0 )k ≤ a ⇒ kx(t)k ≤ b, ∀t ≥ t0 + T

globally uniformly bounded (or ultimately bounded) if

previous conditions hold for any arbitrarily large a.
Lecture 4 – p. 70/86
 Boundedness
The following Lyapunov-like theorem gives sufficient
conditions for uniform/ultimate boundedness.
Theorem 4.18: Let D ⊂ Rn be a domain that contains the
origin and V : [0, ∞) × D → R be a continuous differentiable
function such that

α1 (kxk) ≤ V (t, x) ≤ α2 (kxk)

∂V ∂V
+ f (t, x) ≤ −W3 (x), ∀ kxk ≥ µ > 0
∂t ∂x
for all t ≥ 0 and x ∈ D, where α1 , α2 are class K functions
and W3 (x) is a continuous positive definite function. Take
r > 0 such that Br ⊂ D and suppose that µ < α2−1 (α1 (r)).

Lecture 4 – p. 71/86
 Boundedness
Then, there exists a class KL function β and for every initial
state x(t0 ), satisfying kx(t0 )k ≤ α2−1 (α1 (r)), there is T ≥ 0
(dependent on x(t0 ) and µ) such that the solution of
ẋ = f (t, x) satisfies

kx(t)k ≤ β(kx(t0 )k, t − t0 ), ∀t0 ≤ t ≤ t0 + T (37)

kx(t)k ≤ α−1 (α2 (µ)), ∀t0 ≥ t0 + T (38)

Moreover, if D = Rn and α1 belongs to class K∞ , then the

inequalities above hold for any initial state x(t0 ), with no
restriction on how large µ is.

Lecture 4 – p. 72/86
 Boundedness
Proof:

Lecture 4 – p. 73/86
 Input-To-State Stability
Consider the system

ẋ = f (t, x, u) (39)

where f : [0, ∞] × Rn × Rm → Rn is piecewise continuous in

t and locally Lipschitz in x and u. The input u(t) is a
piecewise continuous function of t. Suppose the unforced
system
ẋ = f (t, x, 0) (40)

has a globally uniformly asymptotically stable equilibrium at

the origin x = 0.
What can we say about the behavior of the system (39) in the
presence of a bounded input u(t)?

Lecture 4 – p. 74/86
 Input-To-State Stability
Example: ẋ = Ax + Bu

Example: ẋ = −3x + (1 + 2x2 )u

Lecture 4 – p. 75/86
 Input-To-State Stability
Suppose we have a Lyapunov function V (t, x) for the
unforced system (40) and let us calculate V̇ in presence of
u. Due to tthe boundedness of u, it is possible in some
cases to show that V̇ is negative outside a ball of radius µ,
where µ depends on sup kuk. This is expected when
f (t, x, u) is Lipschitz, i.e.,

kf (t, x, u) − f (t, x, 0)k ≤ Lkuk

Showing that V̇ is negative outside a ball of radius µ

enables us to apply Theorem 4.18, which states that kx(t)k
is bounded by a class KL function over [t0 , t0 + T ] (37) and
by a class K function over [t0 + T, ∞) (38). Consequently,

kx(t)k ≤ β(kx(t0 )k, t − t0 ) + α−1 (α2 (µ))

Lecture 4 – p. 76/86
 Input-To-State Stability
Definition 4.7:The system (39) is said to be input-to-state
stable (ISS) if there exist a class KL function β and a class
K function γ such that for any initial state x(t0 ) and any
bounded input u(t), the solution x(t) exists for all t ≥ t0 and
satisfies
 
kx(t)k ≤ β(kx(t0 )k, t − t0 ) + γ sup ku(τ )k (41)
t0 ≤τ ≤t

Note: The origin of the unforced system (40) is globally

uniformly asymptotically stable.

Lecture 4 – p. 77/86
 Input-To-State Stability
The following Lyapunov-like theorem gives a sufficient
condition for ISS.
Theorem 4.19: Let V : [0, ∞) × Rn → R be a continuous
differentiable function such that

α1 (kxk) ≤ V (t, x) ≤ α2 (kxk)

∂V ∂V
+ f (t, x) ≤ −W3 (x), ∀ kxk ≥ ρ(kuk) > 0
∂t ∂x
for all (t, x, u) ∈ [0, ∞) × Rn × Rm , where α1 , α2 are class K∞
functions, ρ is a class K function, and W3 (x) is a continuous
positive definite functions on Rn . Then, the system (39) is
input-to-state stable with γ = α1−1 ◦ α2 ◦ ρ.
The function V is called an ISS Lyapunov function.

Lecture 4 – p. 78/86
 Input-To-State Stability
Proof: Direct application of Theorem 4.18 with µ = ρ(kuk).

Lecture 4 – p. 79/86
 Input-To-State Stability
The following lemma is a direct consequence of the
Converse Lyapunov theorem for g.e.s. (Theorem 4.14)
Lemma 4.6: Suppose f (t, x, u is continuosly differentiable
and globally Lipschitz in (x, u), uniformly in t. If the unforced
system (40) has a globally exponentially stable equilibrium
point at the origin x = 0, then the system (39) is ISS.

Lecture 4 – p. 80/86
 Input-To-State Stability
Converse also holds!
Theorem: For the system

ẋ = f (x, u)

the following properties are equivalent

the system is ISS,
there exists a smooth ISS-Lyapunov function (i.e.,
satisfies conditions of Theorem 4.19)
there exists a smooth positive definite radially
unbounded function V and class K∞ functions ρ1 and ρ2
such that the following dissipativity inequality is satisfied
∂V
f (t, x) ≤ −ρ1 (kxk) + ρ2 (kuk) (42)
∂x
Lecture 4 – p. 81/86
 Input-To-State Stability
Young’s Inequality is key to apply this theorem
ǫ p 1 1 1
x y ≤ kxk + q kykq ,
T p
∀ ǫ > 0, ∀ + = 1
p ǫ q p q

Example (4.25): ẋ = −x3 + u

Example (4.26): ẋ = −x − 2x3 + (1 + x2 )u2

Lecture 4 – p. 82/86
 Input-To-State Stability
An interesting application of input-to-state stability arises in
the stability analysis of the cascade system

ẋ1 = f1 (t, x1 , x2 ) (43)

ẋ2 = f2 (t, x2 ) (44)

where f1 : [0, ∞] × Rn1 × Rn2 → Rn1 and

f2 : [0, ∞] × Rn2 → Rn2 are piecewise continuous in t and
locally Lipschitz in x = [xT1 xT2 ]T . Suppose

ẋ1 = f1 (t, x1 , 0), ẋ2 = f2 (t, x2 )

have both g.u.a.s. equilibria at their respective origins.

Under what condition will the origin x = 0 of the cascade system
possess the same property?

Lecture 4 – p. 83/86
 Input-To-State Stability
Lemma 4.7: Under the stated assumptions, if the system
(43), with x2 as input, is input-to-state stable and the origin
of (44) is globally uniformly asymptotically stable, then the
origin of the cascade system (43)-(44) is globally uniformly
asymptotically stable.

Lecture 4 – p. 84/86
 Input-To-State Stability
Proof:

Lecture 4 – p. 85/86
 Input-To-State Stability
Suppose that in the system

ẋ1 = f1 (t, x1 , x2 , u) (45)

ẋ2 = f2 (t, x2 , u) (46)

the x1 -system is ISS with respect to x2 and u, and the

x2 -system is ISS with respect to u. Then, the cascade
system (45)-(46) is ISS with respect to u.

Lecture 4 – p. 86/86