Example 2.  Use the value  [Graphics:Images/NewtonImprovedMod_gr_165.gif]  and compare Methods A,B and C for finding the triple root  [Graphics:Images/NewtonImprovedMod_gr_166.gif]  of the equation  [Graphics:Images/NewtonImprovedMod_gr_167.gif].

Solution 2.

[Graphics:../Images/NewtonImprovedMod_gr_168.gif]

[Graphics:../Images/NewtonImprovedMod_gr_169.gif]

 

 

First, we will compute the iterations for each method, and afterward a table comparing the methods is given.  For the method C, all the iterations in the linear search are included.  

Using formula (2), the standard Newton-Raphson method.  

[Graphics:../Images/NewtonImprovedMod_gr_170.gif]

[Graphics:../Images/NewtonImprovedMod_gr_171.gif]

[Graphics:../Images/NewtonImprovedMod_gr_172.gif]

 

 

Using formula (9) Method A, the accelerated Newton-Raphson method with  m=3.  

[Graphics:../Images/NewtonImprovedMod_gr_173.gif]

[Graphics:../Images/NewtonImprovedMod_gr_174.gif]

 

 

Using formula (11) Method B, the modified Newton-Raphson method.

[Graphics:../Images/NewtonImprovedMod_gr_175.gif]

[Graphics:../Images/NewtonImprovedMod_gr_176.gif]

 

 

Using formula (13) Method C, the adaptive Newton-Raphson method.

The details for obtaining  [Graphics:../Images/NewtonImprovedMod_gr_177.gif] are:

[Graphics:../Images/NewtonImprovedMod_gr_178.gif]

[Graphics:../Images/NewtonImprovedMod_gr_179.gif]

 

 

Since  [Graphics:../Images/NewtonImprovedMod_gr_180.gif]  we reject  [Graphics:../Images/NewtonImprovedMod_gr_181.gif]  and set  [Graphics:../Images/NewtonImprovedMod_gr_182.gif]  and continue the iteration using formula (13).  The subroutine makes all these choices automatically.
        

[Graphics:../Images/NewtonImprovedMod_gr_183.gif]

  

[Graphics:../Images/NewtonImprovedMod_gr_184.gif]

[Graphics:../Images/NewtonImprovedMod_gr_185.gif]

 

 

Caveat.  Another shortcoming of Method A, the accelerated Newton-Raphson method is the wrong choice for  m.  Suppose that one chose  [Graphics:../Images/NewtonImprovedMod_gr_186.gif]  instead of  [Graphics:../Images/NewtonImprovedMod_gr_187.gif]?  Let's see what happens!

[Graphics:../Images/NewtonImprovedMod_gr_188.gif]

[Graphics:../Images/NewtonImprovedMod_gr_189.gif]

[Graphics:../Images/NewtonImprovedMod_gr_190.gif]

 

If the wrong choice for  m  is made then convergence will be linear (or slow).  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(c) John H. Mathews 2004