Page 176 - Complementarity and Variational Inequalities in Electronics
P. 176

The Nonregular Dynamical System Chapter | 5 167


                              x(τ i ;t 0 ,x 0 ) → x . Thus x ∈  (x 0 ). On the other hand, we get the con-
                                            ∗
                                                    ∗
                              tradiction d(x∗, (x 0 )) ≥ ε.
                            v) The set of stationary solutions S(F,ϕ) is invariant. Indeed, if x 0 ∈ S(F,ϕ),
                              then (∀t ≥ t 0 ) : x(τ;t 0 ,x 0 ) = x 0 , and thus γ(x 0 ) ={x 0 }⊂ S(F,ϕ).

                              Thanks to Theorem 12, we can prove that the set  (x 0 ) is invariant by using
                           standard topological arguments (see e.g. [80]).
                           Theorem 17. Suppose that the assumptions of Theorem 11 hold together with
                           conditions (5.36), (5.42), and (5.43). Suppose also that D(∂ϕ) is closed. Let
                           x 0 ∈ D(∂ϕ). The set  (x 0 ) is invariant.

                           Proof. Let z ∈  (x 0 ). There exists {τ i }⊂[t 0 ,+∞) such that τ i →+∞ and
                           x(τ i ;t 0 ,x 0 ) → z.Let τ ≥ t 0 .Wehave z ∈  (x 0 ) ⊂ D(∂ϕ) = D(∂ϕ), and we
                           may use Theorem 12 to obtain

                                           x(τ;t 0 ,z) = lim x(τ;t 0 ,x(τ i ;t 0 ,x 0 )).
                                                      i→∞
                           Then noting from uniqueness property of the solution that

                                         x(τ;t 0 ,x(τ i ;t 0 ,x 0 )) = x(τ − t 0 + τ i ;t 0 ,x 0 ),

                           we get
                                           x(τ;t 0 ,z) = lim x(τ − t 0 + τ i ;t 0 ,x 0 ).
                                                      i→∞
                           Thus setting w i = τ − t 0 + τ i , we see that w i ≥ t 0 , w i →+∞ and
                           x(w i ;t 0 ,x 0 ) → x(τ;t 0 ,z), so that x(τ;t 0 ,z) ∈  (x 0 ).
                              Our goal is now to prove an extension of the LaSalle invariance principle
                           applicable to the first-order evolution variational inequality
                                       dx
                                      
  (t;t 0 ,x 0 ) + F(x(t;t 0 ,x 0 )),v − x(t;t 0 ,x 0 )
                                       dt
                                                                      n
                                        + ϕ(v) − ϕ(x(t;t 0 ,x 0 )) ≥ 0, ∀v ∈ R , a.e. t ≥ t 0 .
                           Lemma 4. Suppose that the assumptions of Theorem 11 hold together with
                                                                                      n
                           conditions (5.36), (5.42), and (5.43). Let   be a compact subset of R .We
                                                       n
                                                    1
                           assume that there exists V ∈ C (R ;R) such that
                              (∀x ∈ D(∂ϕ) ∩  ) :
F(x),∇V(x) + ϕ(x) − ϕ(x −@ V(x)) ≥ 0.  (5.49)

                           Let x 0 ∈ D(∂ϕ).If γ(x 0 ) ⊂  , then there exists a constant k ∈ R such that
                                                 (∀x ∈  (x 0 )) : V(x) = k.
   171   172   173   174   175   176   177   178   179   180   181