Page 166 - Essentials of applied mathematics for scientists and engineers
P. 166

book   Mobk070    March 22, 2007  11:7








                     156  ESSENTIALS OF APPLIED MATHEMATICS FOR SCIENTISTS AND ENGINEERS
                       and the equation is said to be singular as ε → 0. If we set xε = u we find an equation for u as

                                                          2
                                                         u + u − ε = 0                             (10.16)
                       With ε identically zero, u = 0or −1. Assuming that u may be approximated by a series like
                       (10.5) we find that
                                                        2      2              3
                                        (−a 1 − 1)ε + a − a 2 ε + (2a 1 a 2 − a 3 )ε + ··· = 0     (10.17)
                                                       1
                                                            a 1 =−1

                                                            a 2 = 1                                (10.18)
                                                            a 3 =−2

                       so that
                                                         1
                                                                       2
                                                   x =− − 1 + ε − 2ε + ···                         (10.19)
                                                         ε
                       The three-term approximation of the negative root is therefore x =−10.92, within 0.03% of
                       the exact solution.
                            As a third algebraic example consider
                                                         2
                                                        x − 2εx − ε = 0                            (10.20)
                       This at first seems like a harmless problem that appears at first glance to be amenable to a regular
                                                      2
                       perturbation expansion since the x term is not lost when ε → 0. We proceed optimistically
                       by taking

                                                                   2     3
                                                 x = x 0 + a 1 ε + a 2 ε + a 3 ε + ···             (10.21)
                       Substituting into (10.20) we find

                                                                                2
                                                                2
                                       2


                                      x + (2x 0 a 1 − 2x 0 − 1)ε + a + 2x 0 a 2 − 2a 1 ε + ··· = 0  (10.22)
                                       0
                                                                1
                       from which we find
                                                      x 0 = 0
                                                      2x 0 a 1 − 2x 0 − 1 = 0                      (10.23)
                                                        2
                                                      a + 2x 0 a 2 − 2a 1 = 0
                                                       1
                       From the second of these we conclude that either 0 =−1 or that there is something wrong.
                       That is, (10.21) is not an appropriate expansion in this case.
                            Note that (10.20) tells us that as ε → 0, x → 0. Moreover, in writing (10.21) we have
                       essentially assumed that ε → 0 in such a manner that  x  → constant. Let us suppose instead
                                                                        ε
   161   162   163   164   165   166   167   168   169   170   171