Page 122 - INTRODUCTION TO THE CALCULUS OF VARIATIONS
P. 122

Relaxation theory                                                 109

                   We conclude this section with two examples.

                Example 3.30 Let us return to Bolza example (Example 3.10). We here have
                n =1,
                                                      ¡  2  ¢ 2  4
                                  f (x, u, ξ)= f (u, ξ)= ξ − 1  + u
                                        1
                              ½       Z                                 ¾
                                                                 1,4
                                                  0
                      (P)  inf I (u)=     f (u (x) ,u (x)) dx : u ∈ W 0  (0, 1)  = m.
                                       0
                We have already shown that m =0 and that (P) has no solution. An elementary
                computation (cf. Example 1.53 (ii)) shows that
                                             ⎧
                                             ⎨ f (u, ξ)  if |ξ| ≥ 1
                                   f  ∗∗  (u, ξ)=
                                                  u     if |ξ| < 1 .
                                             ⎩     4
                Therefore u ≡ 0 is a solution of

                                      1
                            ½       Z                                  ¾
                   ¡ ¢                                          1,4
                                                  0
                    P    inf I (u)=    f  ∗∗  (u (x) ,u (x)) dx : u ∈ W  (0, 1)  = m =0.
                                                                0
                                     0
                                   1,4
                The sequence u ν ∈ W 0  (ν ≥ 2 being an integer) constructed in Example 3.10
                satisfies the conclusions of the theorem, i.e.
                            u ν   u in W  1,4  and I (u ν ) → I (u)= 0,as ν →∞ .
                                       2
                Example 3.31 Let Ω ⊂ R be aboundedopen setwithLipschitz boundary. Let
                         ¡   2  ¢
                      1,4
                u 0 ∈ W   Ω; R  be such that
                                        ZZ
                                            det ∇u 0 (x) dx 6=0 .
                                          Ω
                                            2
                Let, for ξ ∈ R 2×2 , f (ξ)= (det ξ) ,
                             ½       ZZ                                   ¾
                                                                 1,4  ¡  2 ¢
                     (P)  inf I (u)=     f (∇u (x)) dx : u ∈ u 0 + W  Ω; R  = m
                                                                 0
                                        Ω
                            ½       ZZ                                    ¾
                                                                 1,4  ¡  2  ¢
                   (P)   inf I (u)=     f  ∗∗  (∇u (x)) dx : u ∈ u 0 + W  Ω; R  = m.
                                                                 0
                                      Ω
                We will show that Theorem 3.28 is false, by proving that m> m. Indeed it is
                easy to prove (cf. Exercise 3.6.3) that f  ∗∗  (ξ) ≡ 0, which therefore implies that
                m =0. Let us show that m> 0. Indeed by Jensen inequality (cf. Theorem 1.51)
                                          1,4  ¡  2  ¢
                we have, for every u ∈ u 0 + W  Ω; R ,
                                          0
                       ZZ                         µ        ZZ             ¶ 2
                                     2                1
                           (det ∇u (x)) dx ≥ meas Ω            det ∇u (x) dx  .
                         Ω                          meas Ω   Ω
   117   118   119   120   121   122   123   124   125   126   127