Page 20 - INTRODUCTION TO THE CALCULUS OF VARIATIONS
P. 20

Presentation of the content of the monograph                        7

                Furthermore, equality holds if and only if A is a disk (i.e., ∂A is a circle).
                   We can rewrite it into our formalism (here n =1 and N =2) by parametriz-
                ing the curve
                                ∂A = {u (x)= (u 1 (x) ,u 2 (x)) : x ∈ [a, b]}
                and setting
                                           Z
                                             b q
                                                      02
                                                 02
                        L (∂A)= L (u)=          u + u dx
                                                 1    2
                                            a
                                             Z  b                 Z  b
                                            1
                        M (A)= M (u)=           (u 1 u − u 2 u ) dx =  u 1 u dx .
                                                                        0
                                                           0
                                                    0
                                                                        2
                                                           1
                                                    2
                                            2  a                   a
                The problem is then to show that
                                                                     √
                            (P)  inf {L (u): M (u)= 1; u (a)= u (b)} =2 π.
                                                                     n
                   The problem can then be generalized to open sets A ⊂ R with sufficiently
                regular boundary, ∂A,and it readsas
                                           n   n          n−1
                                    [L (∂A)] − n ω n [M (A)]  ≥ 0
                                                        n
                where ω n is the measure of the unit ball of R , M (A) stands for the measure
                of A and L (∂A) for the (n − 1) measure of ∂A.Moreover, if A is sufficiently
                regular (for example, convex), there is equality if and only if A is a ball.
                0.3    Presentation of the content of the mono-
                       graph
                To deal with problems of the type considered in the previous section, there are,
                roughly speaking, two ways of proceeding: the classical and the direct meth-
                ods. Before describing a little more precisely these two methods, it might be
                                                                 N
                enlightening to first discuss minimization problems in R .
                            N
                   Let X ⊂ R , F : X → R and
                                       (P)  inf {F (x): x ∈ X} .
                   The first method consists, if F is continuously differentiable, in finding solu-
                tions x ∈ X of
                                            0
                                          F (x)= 0,x ∈ X.
                Then, by analyzing the behavior of the higher derivatives of F, we determine if x
                is a minimum (global or local), a maximum (global or local) or just a stationary
                point.
   15   16   17   18   19   20   21   22   23   24   25