Page 89 - Complementarity and Variational Inequalities in Electronics
P. 89

80  Complementarity and Variational Inequalities in Electronics


                           Indeed, if 0 <t ≤ 1, then the result follows from (4.13) since z ∈ D( ) ∞ ,z 
= 0,
                           whereas if t = 0, then the result is trivial since z 
= 0. We know that

                                            (1 − t)z + tMz,h ≥ 0,∀h ∈ D(  ∞ ).

                                                                      j
                                                                        j
                           Let j ∈{1,...,n}.Using (4.13),wemay set h = z,e  e to get
                                                      2
                                                (1 − t)z + t(Mz) j z j ≥ 0.
                                                      j
                           This last relation holds for all j ∈{1,...,n}, and since

                                                  (1 − t)z + tMz,z ≤ 0,

                           we get a contradiction:

                                            2                      2
                                   0 ≥ (1 − t)z + t(Mz) k z k +  (1 − t)z + t(Mz) j z j > 0.
                                            k                      j
                                                          j
=k
                              c) We prove that PS n ⊂ AC n .Let (M, ) ∈ PS n , t ∈[0,1], and z ∈ B((1 −
                           t)I + tM, ). We claim that z = 0. Suppose on the contrary that z 
= 0. Here
                                   ∗
                           (D(  ∞ )) ={0}, and the second relation in (4.27) yields
                                                  (1 − t)z + tMz = 0.

                           If t = 0, then z = 0, a contradiction. If 0 <t ≤ 1, then
                                                          (1 − t)
                                                   Mz =−        z,
                                                             t
                                 ∗
                           so that ν =− (1−t)  ≤ 0 is a real eigenvalue of M, a contradiction to (4.17).
                                        t
                              Let us now give some additional properties that will be used later.
                                                 n
                           Proposition 20. Let   : R → R be a proper convex lower semicontinuous
                           function with closed domain, and let M ∈ R n×n .If

                                              (M, ) ∈ PD0 n ∪ P0 n ∪ PS0 n ,

                           then
                                              (∀λ> 0) : (λI + M, ) ∈ AC n .

                           Proof. Let λ> 0.
                              a) If (M, ) ∈ PD0 n , then condition (4.11) on M entails that λI + M satis-
                           fies condition (4.10), and the result is a consequence of Proposition 19.
   84   85   86   87   88   89   90   91   92   93   94