Page 107 - INTRODUCTION TO THE CALCULUS OF VARIATIONS
P. 107

94                                                      Direct methods

                       and from (H3), we find that there exists γ > 0 so that
                                                            1
                                                     p    p                      n
                                |f (x, u, ξ)| ≤ γ (1 + |u| + |ξ| ) , ∀ (x, u, ξ) ∈ Ω × R × R .  (3.6)
                                             1
                       In particular we deduce that

                                             |I (u)| < ∞, ∀u ∈ W 1,p  (Ω) .

                          Step 2 (Derivative of I). We now prove that for every u, ϕ ∈ W  1,p  (Ω) and
                       every   ∈ R we have
                                               Z
                             I (u +  ϕ) − I (u)
                          lim                =    [f u (x, u, ∇u) ϕ + hf ξ (x, u, ∇u); ∇ϕi] dx .  (3.7)
                           →0
                                                Ω
                       We let
                                     g (x,  )= f (x, u (x)+  ϕ (x) , ∇u (x)+  ∇ϕ (x))
                       so that                           Z
                                              I (u +  ϕ)=   g (x,  ) dx .
                                                          Ω
                                  1
                                                                                       1
                       Since f ∈ C we have, for almost every x ∈ Ω,that   → g (x,  ) is C and
                       therefore there exists θ ∈ [− | | , | |], θ = θ (x), such that
                                             g (x,  ) − g (x, 0) = g   (x, θ)

                       where
                          g   (x, θ)= f u (x, u + θϕ, ∇u + θ∇ϕ) ϕ + hf ξ (x, u + θϕ, ∇u + θ∇ϕ); ∇ϕi .

                       The hypothesis (H3) implies then that we can find γ > 0 so that, for every
                                                                       2
                       θ ∈ [−1, 1],
                       ¯               ¯
                       ¯ g (x,  ) − g (x, 0) ¯               p     p      p       p
                       ¯               ¯  = |g   (x, θ)| ≤ γ (1 + |u| + |ϕ| + |∇u| + |∇ϕ| ) ≡ G (x) .
                                                     2

                       ¯               ¯
                                                                 1
                       Note that since u, ϕ ∈ W  1,p  (Ω),wehave G ∈ L (Ω).
                          We now observe that, since u, ϕ ∈ W  1,p  (Ω), we have from (3.6) that the
                                                                     1
                       functions x → g (x, 0) and x → g (x,  ) are both in L (Ω).
                          Summarizing the results we have that
                                      g (x,  ) − g (x, 0)   1
                                                       ∈  L (Ω) ,

                                    ¯               ¯
                                    ¯ g (x,  ) − g (x, 0) ¯               1
                                    ¯               ¯  ≤  G (x) ,with G ∈ L (Ω)

                                    ¯               ¯
                                      g (x,  ) − g (x, 0)
                                                      → g   (x, 0) a.e. in Ω .
   102   103   104   105   106   107   108   109   110   111   112