Page 81 - Complementarity and Variational Inequalities in Electronics
P. 81

72  Complementarity and Variational Inequalities in Electronics



                           4.2.24 Class of (M, ) ∈ Q0 n
                                                                   n
                           We define by Q0 n the set of (M, ) ∈ R n×n × (R ;R∪{+∞}) such that there
                           exists λ 0 > 0 such that

                                                                        n
                                           (∀ 0 <λ ≤ λ 0 ) : R(λI + M, ) = R ,
                                                         n
                           that is, for any 0 <λ ≤ λ 0 and q ∈ R , problem VI(λI + M,q, ) has at least
                           one solution.
                              It is clear that

                                    PD n ⊂ PD0 n , P n ⊂ P0 n , PS n ⊂ PS0 n , DS n ⊂ DS0 n .
                           We also have the following:
                           Proposition 13. We have

                                                      DS n ⊂ P n
                           and

                                                     DS0 n ⊂ P0 n .
                           Proof. Let us first remark that if (M, ) ∈ DS n ∪ DS0 n , then condition (4.12)
                           (or (4.14)) holds. Indeed, here


                                                    =   K 1 ×K 2 ×···×K n
                           as in (4.21) (or (4.24)) where K 1 ,K 2 ,...,K n are nonempty closed convex
                           cones. Thus

                                         D(  ∞ ) = D( ) ∞ = K 1 × K 2 × ··· × K n .
                           If w ∈ K, then for all i ∈{1,...,n}, w i ∈ K i , and since for all j ∈{1,...,n},
                                                   j
                                                            j
                                                     j
                           j 
= i,0 ∈ K i , we obtain  w,e  e = w j e ∈ K.
                              If (M, ) ∈ DS n and x 
= 0, then
                                                 n              n

                                      Mx,	x =     (Mx) i (	x) i =  	 ii (Mx) i x i > 0.
                                                i=1            i=1
                           Here, for all i ∈{1,...,n}, 	 ii > 0, and thus there necessarily exists α ∈
                           {1,...,n} such that (Mx) α x α > 0.
                              If (M, ) ∈ DS0 n and x 
= 0, then I ={i ∈{1,...,n}: x i 
= 0}
=∅, and

                                              Mx,	x =     	 ii (Mx) i x i ≥ 0.
                                                        i∈I
                           So, there necessarily exists α ∈{1,...,n} such that x α 
= 0 and (Mx) α x α ≥ 0.
   76   77   78   79   80   81   82   83   84   85   86