Page 27 - Complementarity and Variational Inequalities in Electronics
P. 27

The Convex Subdifferential Relation Chapter | 2 17


                                             n
                           Indeed, let U,V ∈ R satisfy the complementarity relation 0 ≤ U ⊥ V ≥ 0.
                           Then (∀h ≥ 0) :
V,h ≥ 0, and since 
V,U = 0, we see that

                                                (∀h ≥ 0) :
V,h − U ≥ 0,

                           meaning that −V ∈ ∂
 R (U). Reciprocally, if −V ∈ ∂
 R (U), then U ≥ 0
                                                                            n
                                                n
                                                +                           +
                           and (∀h ≥ 0) :
V,h − U ≥ 0. Setting h = 2U, we obtain 
V,U ≥ 0. Then,
                           setting h = 0, we get 
V,U  0. Thus V ⊥ U. Moreover, for any H ≥ 0, we
                           may set h = H + U to see that 
V,H ≥ 0. It results in V ≥ 0. Thus U,V ∈ R n
                           satisfy the complementarity relation.
                                        n
                              Aset K ⊂ R is a cone if
                                                   (∀α> 0) : αK ⊂ K.

                           If K is a nonempty closed convex cone, then

                                                        0 ∈ K,
                                                      K + K ⊂ K,

                           and
                                                   (∀α> 0) : αK ⊂ K.

                           Remark 5. The relation K + K ⊂ K is a consequence of the convexity and
                           the cone property. Indeed, if z ∈ K + K, then z = z (1)  + z (2)  with z (1)  ∈ K
                           and z (2)  ∈ K.The set K is a cone and thus 2z (1) ,2z (2)  ∈ K. This results in
                                      1
                               1
                           z = 2z (1)  + 2z (2)  ∈ K since K is convex.
                               2      2
                                                           ∗
                              The dual cone of K is denoted by K and defined by
                                                      n
                                             ∗
                                            K ={w ∈ R :
w,v ≥ 0, ∀v ∈ K}.
                                          ∗
                           We remark that K is the set of vectors w forming an acute angle with all the
                           vectors v of K, that is,
                                                          
w,v
                                           cos(≺ w,v  ) =        ≥ 0,∀v ∈ K.
                                                         ||w||||v||
                                   o
                                          ∗
                           The set K =−K is called the polar cone of K (see Fig. 2.10).
                           Remark 6. If K is a nonempty closed convex cone, then (K ) = K.
                                                                            ∗ ∗
                                                                               n
                           Example 13. Let K = V where V is a real-vector subspace of R . Then
                                                        ∗
                                                             ⊥
                                                       K = V .
   22   23   24   25   26   27   28   29   30   31   32