Page 189 - INTRODUCTION TO THE CALCULUS OF VARIATIONS
P. 189

176                                            Solutions to the Exercises

                          (ii) We find, if x 6=0,that

                                                    x i
                                              0
                                           = f (|x|)   ⇒ |∇u (x)| = |f (|x|)| .
                                                                     0
                                       u x i
                                                   |x|
                       Assume, for a moment, that we already proved that u is weakly differentiable in
                       B R ,then
                                                        Z  R
                                              p             n−1      p
                                                                  0
                                          k∇uk  p = σ n−1  r    |f (r)| dr ,
                                              L
                                                         0
                       which is the claim.
                                                  , asabove, isindeedthe weak derivative (with
                          Let us now show that u x i
                       respect to x i )of u. Wehavetoprove that,for every ϕ ∈ C ∞  (B R ),
                                                                           0
                                             Z             Z
                                                 uϕ dx = −     ϕu x i dx .              (7.6)
                                                   x i
                                              B R            B R
                       Let  > 0 be sufficiently small and observe that (recall that ϕ =0 on ∂B R )
                        Z               Z              Z
                            uϕ dx =           uϕ dx +     uϕ dx
                              x i                x i         x i
                         B R             B R \B         B
                                          Z              Z             Z
                                                                x i
                                    = −         ϕu x i dx −  uϕ   dσ +     uϕ dx
                                                                |x|          x i
                                           B R \B         ∂B            B
                                          Z           Z           Z           Z
                                                                                      x i
                                    = −      ϕu x i dx +  ϕu x i dx +  uϕ dx −     uϕ   dσ .
                                                                        x i
                                                                                      |x|
                                           B R         B           B            ∂B
                                                                 1
                                             and uϕ  are both in L (B R ), we deduce (this follows
                       Since the elements ϕu x i
                                                   x i
                       from Hölder inequality if p> 1 or from standard properties of integrals if p ≥ 1,
                       see Exercise 1.3.7) that
                                            Z               Z
                                         lim    ϕu x i dx =lim  uϕ dx =0 .
                                          →0              →0      x i
                                             B               B
                       Moreover, by hypothesis, we have the claim (i.e. (7.6)), since
                                 ¯Z           ¯
                                 ¯       x i  ¯              n−1
                                 ¯    uϕ    dσ ≤ σ n−1 kϕk L ∞    f ( ) → 0, as   → 0 .
                                              ¯
                                 ¯       |x|  ¯
                                   ∂B
                          (iii) 1) The first example follows at once and gives
                                   ψ ∈ L p  ⇔ sp < n and ψ ∈ W 1,p  ⇔ (s +1) p< n .
                          2) We find, for every 0 <s< 1/2 and p ≥ 1,that
                            Z  1/2               Z  1/2                       2s−1
                                       sp              −1     2(s−1)     |log 2|
                                 r |log r| dr < ∞,    r   |log r|   dr =          < ∞ .
                             0                    0                       1 − 2s
   184   185   186   187   188   189   190   191   192   193   194