next up previous contents
Next: The Shooting Method, Nonlinear Up: BOUNDARY VALUE PROBLEMS (BVP) Previous: BOUNDARY VALUE PROBLEMS (BVP)   Contents


The Shooting Method

Linear Case:

Take

(75) $\displaystyle Y''=p(x)Y'+q(x)r \quad x_0\le x\le b \quad Y(x_0)=\alpha,  Y'(x_0)=0$

and

(76) $\displaystyle Y''=p(x)Y'+q(x)Y\quad x_0\le x\le b\quad Y(x_0)=0,  Y'(x_0)=1.$

If $p, q, r$ continuous and $q > 0$ on $[x_0, b]$ then the Lipschitz condition exists for cast as a system $\Rightarrow$ both (75) and (76) have unique solutions.

Take $Y_1(x)$ solution of (75) and $Y_2(x)$ solution of (76)

(77) $\displaystyle \Rightarrow Y(x)=Y_1(x)+\frac{\beta-Y_1(b)}{Y_2(b)}Y_2(x),$

(provided $Y_2(b)\ne 0.$ can be checked to be unique solution of

(78) $\displaystyle BVP\equiv\left\{\begin{array}{ll} Y''=p(x)Y'+q(x)Y+r(x) & x_0\le x\le b Y(x_0)=\alpha Y(b)=\beta \end{array}\right.$

Remark: Note that if $Y_2$ is solution of $Y''=p(x)Y'+q(x)Y$ and $Y_2(x_0)=Y_2(b)=0\Rightarrow Y_2=0$.

Summary:

So the shooting-method strategy amounts to the following: Replace (78) by 2 IVP (75) and (76). Use appropriate method to solve (75) and (76) and piece solution as per (77). Figure (13) shows graphically the construction of the solution $Y(x)$ in terms of $Y_1(x)$ and $Y_2(x)$.

Figure 13: Graphical construction of the solution.
\includegraphics[height=3in]{shoot.eps}

ALGORIHM (from Burden and Faires $p$ 582)

  1. Set $h=(b-x_0)/N$
    $u_{1,0}=\alpha\\
u_{2,0}=\alpha\\
v_{1,0}=0\\
v_{2,0}=1$

  2. for $i=0\ldots N-1$

      $\displaystyle \left[\begin{array}{l}
\mbox {Use ERK4 (or some other suitable IV...
...solve for}\\
u_{1, i+1}, u_{2, i+1}, v_{1, i+1}, v_{2, i+1}
\end{array}\right.$    


  $\displaystyle w_{1,0}$ $\displaystyle =$ $\displaystyle \alpha$
  $\displaystyle w_{2,0}$ $\displaystyle =$ $\displaystyle \frac{\beta-u_{1,N}}{v_{1,N}}$


      output $\displaystyle (x_0; w_{1,0} w_{2,0})$
      here $\displaystyle w_{1,0} $$\displaystyle \mbox { is an approximation to } Y(x_0) \mbox { and }
w_{2,0} \mbox{ an approximation to } Y'(x_0)$


  for $\displaystyle i$ $\displaystyle =$ $\displaystyle 1, \cdots N$
  $\displaystyle x$ $\displaystyle =$ $\displaystyle x_0+ih$
  $\displaystyle w_1$ $\displaystyle =$ $\displaystyle u_{1,i}+w_{2,0}v_{1,i}$
  $\displaystyle w_2$ $\displaystyle =$ $\displaystyle u_{2,i}+w_{2,0}v_{2,i}$
  $\displaystyle \mbox {output }(x, w1, w2)$    
      $\displaystyle w_2$ is an approximation to $\displaystyle Y'(x_i)$
      and $\displaystyle w_1$    is an approximation to$\displaystyle Y(x_i)$

END

$\Box$


next up previous contents
Next: The Shooting Method, Nonlinear Up: BOUNDARY VALUE PROBLEMS (BVP) Previous: BOUNDARY VALUE PROBLEMS (BVP)   Contents
Juan Restrepo 2003-05-02