Page 43 - INTRODUCTION TO THE CALCULUS OF VARIATIONS
P. 43

30                                                        Preliminaries

                          (i) ⇒ (ii). This follows from Hölder inequality and the fact that u has a weak
                       derivative; indeed
                                ¯               ¯  ¯               ¯
                                Z  b                Z  b
                                ¯               ¯  ¯               ¯
                                ¯        0      ¯  ¯    0          ¯    0
                                ¯  u (x) ϕ (x) dx¯ = ¯  u (x) ϕ (x) dx¯ ≤ ku k L p kϕk L p 0 .
                                 a                   a
                                ¯               ¯  ¯               ¯
                          (ii) ⇒ (i). Let F be a linear functional defined by
                                                  Z  b
                                                           0
                                   F (ϕ)= hF; ϕi =   u (x) ϕ (x) dx, ϕ ∈ C  ∞  (a, b) .  (1.6)
                                                                        0
                                                   a
                       Note that, by (ii), it is continuous over C 0 ∞  (a, b).Since C 0 ∞  (a, b) is dense in
                        p
                         0
                       L (a, b) (notethat we used herethe fact that p 6=1 and hence p 6= ∞), we can
                                                                               0
                       extend it, by continuity (or appealing to Hahn-Banach theorem), to the whole
                         0
                                                                                      0
                                                                                     p
                        p
                       L (a, b); wehavethereforedefined a continuous linear operator F over L (a, b).
                                                                                    p
                       From Riesz theorem (Theorem 1.13) we find that there exists v ∈ L (a, b) so
                       that
                                                  Z
                                                    b
                                                                          0
                                                                         p
                                   F (ϕ)= hF; ϕi =   v (x) ϕ (x) dx, ∀ϕ ∈ L (a, b) .    (1.7)
                                                   a
                       Combining (1.6) and (1.7) we get
                                Z  b                  Z  b
                                                               0
                                   (−v (x)) ϕ (x) dx = −  u (x) ϕ (x) dx, ∀ϕ ∈ C  ∞  (a, b)
                                                                             0
                                 a                     a
                                                         p
                                               0
                       which exactly means that u = −v ∈ L (a, b) and hence u ∈ W  1,p  (a, b).
                          (iii) ⇒ (ii). Let ϕ ∈ C 0 ∞  (a, b) and let ω ⊂ ω ⊂ (a, b) with ω compact and
                                                                            c
                       such that supp ϕ ⊂ ω.Let h ∈ R so that |h| <dist (ω, (a, b) ). Wehavethen,
                       appealing to (iii),
                           ¯                         ¯
                           Z  b                         µZ                     ¶ 1
                           ¯                         ¯                           p
                                                                            p



                           ¯                         ¯       |u (x + h) − u (x)| dx
                           ¯   [u (x + h) − u (x)] ϕ (x) dx¯ ≤                     kϕk L p 0
                           ¯  a                      ¯     ω
                                             ⎧
                                             ⎨ c |h|kϕk L p 0  if 1 <p < ∞
                                           ≤                                            (1.8)
                                             ⎩
                                                c |h|kϕk    if p = ∞ .
                                                       L 1






                       We know, by hypothesis, that ϕ ≡ 0 on (a, a + h)  and (b − h, b)  if h > 0 and we






                       therefore find (letting ϕ ≡ 0 outside (a, b))
                           b                   b+h                    b
                         Z                   Z                      Z
                            u(x + h) ϕ (x) dx =    u(x + h) ϕ (x) dx =  u(x) ϕ   (x − h) dx . (1.9)
                          a                   a+h                    a
   38   39   40   41   42   43   44   45   46   47   48