Page 156 - Complementarity and Variational Inequalities in Electronics
P. 156

The Nonregular Dynamical System Chapter | 5 147


                                         dx       −1              −1
                                       R    ∈ RAR   Rx − RB∂ (CR    Rx) + RDu
                                         dt
                                                          ⇔
                                        dz       −1     −1  2       −1
                                           ∈ RAR   z − R  R B∂ (CR    z) + RDu
                                        dt
                                                          ⇔
                                        dz       −1     −1  T      −1
                                           ∈ RAR   z − R  C ∂ (CR    z) + RDu.
                                        dt
                           We set

                                                    n
                                              (∀z ∈ R ) :  (z) =  (CR −1 z).          (5.11)
                           Then
                                                         n
                                             D( ) ={z ∈ R : CR −1 z ∈ D( )},
                           and with Assumption (G1),wehave

                                                                T
                                                n
                                          (∀z ∈ R ) : ∂ (z) = R −1 C ∂ (CR −1 z).
                                                           n
                           This allows us to consider, for x 0 ∈ R , problem Q(Rx 0 ,RAR −1 ,RDu, ):
                                                      n
                           Find a function z :[0,+∞[ → R ; t  → z(t) such that:
                                                       0         n
                                                  z ∈ C ([0,+∞[;R ),                  (5.12)
                                                  dz   1          n
                                                    ∈ L (0,+∞;R ),                    (5.13)
                                                       loc
                                                  dt
                                                      z(0) = Rx 0 ,                   (5.14)
                                   dz         −1
                                     (t) ∈ RAR  z(t) + RDu(t) − ∂ (z(t)), a.e. t ≥ 0.  (5.15)
                                   dt
                           Note that this last differential inclusion is equivalent to the variational inequality

                                         dz         −1
                                        
  (t) − RAR  z(t) − RDu(t),v − z(t)
                                         dt
                                                                   n
                                           +  (v) −  (z(t)) ≥ 0,∀v ∈ R , a.e. t ≥ 0.
                           Proposition 23. Suppose that assumptions (G 1 ) and (G 2 ) are satisfied. If
                           (x,y L ) is a solution of problem P(x 0 ,A,B,C,D,u, ), then z = Rx is a so-
                           lution of problem Q(Rx 0 ,RAR −1 ,RDu, ).
                              Reciprocally, if z is solution of problem Q(Rx 0 ,RAR −1 ,RDu, ), then
                           there exists a function y L such that (R −1 z,y L ) is a solution of problem
                           P(x 0 ,A,B,C,D,u, ).
   151   152   153   154   155   156   157   158   159   160   161