Page 51 - Mathematical Techniques of Fractional Order Systems
P. 51

Nonlinear Fractional Order Boundary-Value Problems Chapter | 2  41

                          m      r 1      r 2      r σ22
                    σ
                 m
             5. D ðϕ Þ 5  P  u m2r 1  P  u r 2 2r 1  P  u r 2 2r 3 ...  P  u r σ22 2r σ21 r σ21 ;
                                                             u
                         r 1 50  r 2 50  r 3 50   r σ21 50
             where m $ 2 and σ $ 2 are positive integer numbers. The above properties
             are proved by Liao (2009). To illustrate the procedures of the HAM, let N [u
             (t)] 5 0 denote a nonlinear differential equation. The HAM is based on a
             kind of continuous mapping u(t)-ϕ(t, p), constructed from the so-called the
             zero-order deformation equation as follows
                           ð1 2 pÞL½ϕðt; pÞ 2 u 0 ðtފ 5 HðtÞh pN½ϕðt; pފ;  ð2:9Þ

             where L is an auxiliary linear operator, u 0 (t) is an initial guess, H(t) is the
             auxiliary function, and h is the nonzero control parameter. When p 5 0, the
             zero-order deformation Eq. (2.9) becomes:
                                    L½ϕðt; 0Þ 2 u 0 ðtފ 5 0;          ð2:10Þ
             which gives ϕ(t,0) 5 u 0 (t). And when p 5 1, since H(t)6¼0 and h6¼0
             Eq. (2.9) is equivalent to:
                                       N ϕðt; 1ފ 5 0;                 ð2:11Þ
                                        ½
             which is exactly the same as the original equation N [u(t)] 5 0, provided ϕ(t,
             1) 5 u(t). So, as the homotopy parameter p increases from 0 to 1, the solution
             ϕ(t, p) of the zero-order deformation Eq. (2.9) varies from the initial guess u 0 (t)
             to the exact solution u(t) of the nonlinear differential equation N [u(t)]5 0. By
             using Taylor theorem, ϕ(t, p) expands in a power series of p as follows
                                                N
                                               X
                                                       m
                                ϕðt; pÞ 5 ϕðt; 0Þ 1  u m ðtÞp ;        ð2:12Þ
                                               m51
                          m
             where u m (t) 5 D ϕ(t, p). From ϕ(t,0) 5 u 0 (t) then Eq. (2.12) becomes:
                                               N
                                              X        m
                                 ϕðt; pÞ 5 u 0 ðtÞ 1  u m ðtÞp :       ð2:13Þ
                                              m51
                When choosing the linear operator, the initial guess, the auxiliary func-
             tion, the control parameter, and from ϕ(t,1) 5 u(t) then the power series
             ϕ(t, p) (2.13) converges at p 5 1, and the so-called homotopy series solution
             is given by:
                                               N
                                              X
                                   uðtÞ 5 u 0 ðtÞ 1  u m ðtÞ:          ð2:14Þ
                                              m51
             which satisfies the original equation N [u(t)] 5 0, as proved by Liao (2009).
             The terms u m (t) are unknown functions and determined by the so-called
             high-order deformation equations. Differentiating the zero-order deformation
             Eq. (2.9) m times with respective to the homotopy parameter p and then
             dividing it by m! and finally setting p 5 0, then the term u m (t) is given by
                                                   m
                                           21
                          u m ðtÞ 5 χ u m21 1 hL fHðtÞD ðpN½ϕðt; pފÞg;  ð2:15Þ
                                  m
   46   47   48   49   50   51   52   53   54   55   56