Consider the system x = f(x) where f : D — R” continously
differentiable map. Suppose X = 0 is in D, and is an equilibrium
point. Following the principles of Linearization, it is possible to
represent the system as
x = Ax + g(x)
where A = [Of/@x], (Jacobian evaluated at origin), OF /Ox is
continous and ame — 0 as ||x|| > 0 (g(x) > 0 as x + 0).
Let matrix A be Hurwitz. Then by converse Lyapunov argument,
for any positive definite symmetric matrix Q, the solution P of
the Lyapunov equation PA+ A’ P = —@ is positive definite.Consider candidate Lyapunov function V(x) = x" Px for the
nonlinear system. V(x) is a quadratic Lyapunov function
candidate and is positive definite. The derivative of V(x) along
the trajectories of the system is given by:
V(x) = x7 PF(x) + £7 (x)Px
= x! PIAx + g(x)] + [x Al +91 (x)]Px
= x!(PA+A™P)x + 2x! P(x)
—xT Qx + 2x7 Pg(x)
The first term on the right hand side is negative definite, while the
second term is (in general) indefinite. The function g(x) satisfies
IgGl
+ 0, as ||x|| 3 0
Ix]Therefore, for any y > 0, there exists r > 0 such that
lig)Il < vlad, Vial] Xmin(Q)||x||? where Amin denotes the minimum
eigenvalue of a matrix. Amin(Q) is real and positive since Q is
symmetric and positive definite. Thus,
V(x) < —Pamin(@) = 24||PIl]|ixil2, Vill