Example 6.  How good did it get?

Solution 6.

Since all the graphs"look alike", we will need to "trust" that the analytic solution is known to be:

[Graphics:../Images/FiniteDifferenceMod_gr_150.gif]


[Graphics:../Images/FiniteDifferenceMod_gr_151.gif]


[Graphics:../Images/FiniteDifferenceMod_gr_152.gif]

[Graphics:../Images/FiniteDifferenceMod_gr_153.gif]
[Graphics:../Images/FiniteDifferenceMod_gr_154.gif]

[Graphics:../Images/FiniteDifferenceMod_gr_155.gif]

[Graphics:../Images/FiniteDifferenceMod_gr_156.gif]
[Graphics:../Images/FiniteDifferenceMod_gr_157.gif]

[Graphics:../Images/FiniteDifferenceMod_gr_158.gif]

[Graphics:../Images/FiniteDifferenceMod_gr_159.gif]
[Graphics:../Images/FiniteDifferenceMod_gr_160.gif]

[Graphics:../Images/FiniteDifferenceMod_gr_161.gif]



[Graphics:../Images/FiniteDifferenceMod_gr_162.gif]
[Graphics:../Images/FiniteDifferenceMod_gr_163.gif]
[Graphics:../Images/FiniteDifferenceMod_gr_164.gif]


Observe that the error for 50 subintervals is very close to [Graphics:../Images/FiniteDifferenceMod_gr_165.gif] of the error for 25 subintervals.  

[Graphics:../Images/FiniteDifferenceMod_gr_166.gif]

[Graphics:../Images/FiniteDifferenceMod_gr_167.gif]

Since the version of the finite difference method we are using is known to be of order  [Graphics:../Images/FiniteDifferenceMod_gr_168.gif], we expect this to happen.  

When extrapolation is used in a method of order [Graphics:../Images/FiniteDifferenceMod_gr_169.gif], a formula like [Graphics:../Images/FiniteDifferenceMod_gr_170.gif] is used.  

Extrapolation works !

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(c) John H. Mathews 2004