Page 111 - Complementarity and Variational Inequalities in Electronics
P. 111

102  Complementarity and Variational Inequalities in Electronics


                             and

                                              (∀e ∈ ker(M),e 
= 0) : q,e 
= 0,       (4.71)
                             then problem VI(M,q, ) has at most one solution.
                           e) If the function   is strictly convex, then problem VI(M,q, ) has at most
                             one solution.

                                                             T
                           Proof. Let us first remark that ker(M + M ) = ker(M) and that part a) and the
                           first relation in part c) are direct consequences of parts a) and c) in Corollary 7.
                           Using the equivalence of problem VI(M,q, ) with problem (4.66),wealsoget
                                1                          1
                                  Mu 1 ,u 1  + q,u 1  +  (u 1 ) =  Mu 2 ,u 2  + q,u 2  +  (u 2 ),
                                2                          2

                           from which we deduce the second relation in part c). We have indeed M(u 1 −
                           u 2 ) = 0, and thus

                             Mu 1 ,u 1  −ÐMu 2 ,u 2  = Mu 1 ,u 1  −ÐMu 2 ,u 1  + Mu 2 ,u 1  −ÐMu 2 ,u 2
                                               = M(u 1 − u 2 ),u 1  + Mu 2 ,u 1 − u 2
                                               = M(u 1 − u 2 ),u 1  + u 2 ,M(u 1 − u 2 ) = 0.

                           It is easy to check that part d) is a direct consequence of part c). It suffices to
                           set e = u 2 − u1 and to see that e = 0. Suppose on the contrary that e 
= 0. Then
                           from part c) we get

                                              −Ðq,e =  (u 1 + e) −  (u 1 ),
                           and then using (4.70), we get  q,e = 0 and a contradiction to (4.71).
                              Moreover, let x 0 ∈ D( ). Then, for all v ∈ ker(M),wehave

                                                  T
                                                M x 0 ,v = x 0 ,Mv = 0.
                           Therefore, condition (4.63) in this case is equivalent to condition (4.67).Partb)
                           is then a direct consequences of part b) in Corollary 7.
                                                                         1
                              Finally, if   is strictly convex, then the function x  à   Mx,x + q,x +
                                                                         2
                            (x) also is strictly convex, and thus problem (4.66) has at most one solu-
                           tion. The result in part e) follows since problem (4.66) is equivalent to problem
                           VI(M,q, ).
                           Example 53. Let
                                                     ⎛             ⎞
                                                        1   −10
                                                 M = ⎝ −1    1   0 ⎠
                                                     ⎜
                                                                   ⎟
                                                        0    0   1
   106   107   108   109   110   111   112   113   114   115   116