More Background

    A drawback of Newton's method is the requirement that the Jacobian matrix be computed, which requires the calculation of  [Graphics:Images/BroydenMethodMod_gr_71.gif]  partial derivatives per iteration.  One obvious improvement is to use a finite difference approximation for the Jacobian matrix.  For two dimensions this is

        [Graphics:Images/BroydenMethodMod_gr_72.gif]
where h is small.  Notice that this will require  [Graphics:Images/BroydenMethodMod_gr_73.gif]  function evaluations.  

Exploration

[Graphics:../Images/BroydenMethodMod_gr_74.gif]




[Graphics:../Images/BroydenMethodMod_gr_75.gif]
[Graphics:../Images/BroydenMethodMod_gr_76.gif]
[Graphics:../Images/BroydenMethodMod_gr_77.gif]
[Graphics:../Images/BroydenMethodMod_gr_78.gif]
[Graphics:../Images/BroydenMethodMod_gr_79.gif]
[Graphics:../Images/BroydenMethodMod_gr_80.gif]




[Graphics:../Images/BroydenMethodMod_gr_81.gif]




[Graphics:../Images/BroydenMethodMod_gr_82.gif]
[Graphics:../Images/BroydenMethodMod_gr_83.gif]
[Graphics:../Images/BroydenMethodMod_gr_84.gif]
[Graphics:../Images/BroydenMethodMod_gr_85.gif]
[Graphics:../Images/BroydenMethodMod_gr_86.gif]
[Graphics:../Images/BroydenMethodMod_gr_87.gif]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(c) John H. Mathews 2005