Page 60 - Complementarity and Variational Inequalities in Electronics
P. 60

A Variational Inequality Theory Chapter | 4 51


                           and taking now the limit as λ →+∞, we get

                                                      ∞ (x) ≤ 	 ∞ (x).                 (4.4)

                           The result follows from (4.3) and (4.4).
                           Remark 14. We have
                                                  epi(  ∞ ) = (epi( )) ∞ .

                                          n
                           Indeed, let x 0 ∈ R be such that α 0 =  (x 0 )< +∞.Wehave
                                                    (x,α) ∈ epi(  ∞ )
                                                          ⇔
                                                        ∞ (x) ≤ α
                                                          ⇔
                                                      (λx + x 0 ) −  (x 0 )
                                            (∀λ> 0) :                  ≤ α
                                                             λ
                                                          ⇔
                                             (∀λ> 0) :  (λx + x 0 ) ≤ α 0 + λα
                                                          ⇔
                                           (∀λ> 0) : λ(x,α) + (x 0 ,α 0 ) ∈ epi( )
                                                          ⇔
                                                   (x,α) ∈ (epi( )) ∞ .

                              In the following result, we show that   ∞ turns out to be proper, convex,
                           lower semicontinuous, and positively homogeneous of order 1.
                                                                           n
                                                  n
                           Proposition 8. If   ∈   0 (R ;R ∪{+∞}), then   ∞ ∈   0 (R ;R ∪{+∞}), and
                                                        n
                                         (∀ α ≥ 0), (∀x ∈ R ) :   ∞ (αx) = α  ∞ (x).
                           Proof. Let x 0 ∈ D( ).Wehave

                                                            1
                                                  ∞ (0) = lim   (x 0 ) = 0,
                                                        λ+∞ λ
                           and thus   ∞ is proper. Let us now prove the convexity of   ∞ .Let λ ∈[0,1]
                                       n
                           and x 1 ,x 2 ∈ R .Wehave
                              ∞ (λx 1 + (1 − λ)x 2 )
                                      1
                              = lim    [ (λx 0 + (1 − λ)x 0 + λtx 1 + (1 − λ)tx 2 )
                                t→+∞ t
                                − λ (x 0 ) − (1 − λ) (x 0 )]
   55   56   57   58   59   60   61   62   63   64   65