Page 137 - INTRODUCTION TO THE CALCULUS OF VARIATIONS
P. 137

124                                                         Regularity

                       Show that
                                                       0             ∞
                                         u x 1 x 1  ,u x 2 x 2  ∈ C (Ω) ,u x 1 x 2  / ∈ L  (Ω)
                                                                         2
                                                            0
                                                         ∈ C ,while u/∈ C ,infact u is not even
                       which implies that ∆u = u x 1 x 1  + u x 2 x 2
                       in W  2,∞ .
                       Exercise 4.3.4 The present exercise (in the spirit of the preceding one) will
                                                                      1
                       give an example of a function u/∈ W 2,1 ,with ∆u ∈ L .Let
                                                 ©     2             ª
                                             Ω = x ∈ R :0 < |x| < 1/2
                                        u (x)= u (x 1 ,x 2 )= log |log |x|| ,if x ∈ Ω .
                                                        1
                       Show that u/∈ W 2,1  (Ω) while ∆u ∈ L (Ω).

                       4.4    Some general results

                       The generalization of the preceding section to integrands of the form f =
                       f (x, u, ∇u) is a difficult task. We will give here, without proof, a general the-
                       orem and we refer for more results to the literature. The next theorem can be
                       found in Morrey [75] (Theorem 1.10.4).
                                                                                          n
                                               n
                       Theorem 4.11 Let Ω ⊂ R be a bounded open set and f ∈ C  ∞  (Ω × R × R ),
                                                              ¡         ¢
                                                                          and similarly for the
                       f = f (x, u, ξ).Let f x =(f x 1  , ..., f x n  ), f ξ = f ξ 1  , ..., f ξ n
                                                                                         n
                                                                               n
                       higher derivatives. Let f satisfy, for every (x, u, ξ) ∈ Ω × R × R and λ ∈ R ,
                                  ⎧                p                      p
                                               α 1 V − α 2 ≤ f (x, u, ξ) ≤ α 3 V
                                  ⎪
                                  ⎪
                                  ⎪
                                  ⎪
                                  ⎪
                                  ⎪
                                  ⎪                           p−1                  p−2
                                  ⎨  |f ξ | , |f xξ | , |f u | , |f xu | ≤ α 3 V  , |f uξ | , |f uu | ≤ α 3 V
                            (C)
                                  ⎪
                                  ⎪                  n
                                  ⎪                 X
                                  ⎪             2                                  2
                                  ⎪       p−2                                p−2
                                  ⎪    α 4 V  |λ| ≤         (x, u, ξ) λ i λ j ≤ α 5 V  |λ|
                                  ⎪                     f ξ i ξ j
                                  ⎩
                                                    i,j=1
                                                  2
                                            2
                                    2
                       where p ≥ 2, V =1 + u + |ξ| and α i > 0, i =1, ..., 5, are constants.
                          Then any minimizer of
                                     ½       Z                                       ¾
                                                                               1,p
                             (P)   inf I (u)=   f (x, u (x) , ∇u (x)) dx : u ∈ u 0 + W 0  (Ω)
                                               Ω
                       is in C  ∞  (D), for every D ⊂ D ⊂ Ω.
                       Remark 4.12 (i) The last hypothesis in (C) implies a kind of uniform convexity
                       of ξ → f (x, u, ξ); it guarantees the uniform ellipticity of the Euler-Lagrange
                       equation. The example of the preceding section, obviously, satisfies (C).
                          (ii) For the regularity up to the boundary, we refer to the literature.
   132   133   134   135   136   137   138   139   140   141   142