Example 3.  Consider the function  [Graphics:Images/TaylorPolyMod_gr_189.gif].  
3 (b).  Investigate the error in the approximation over the interval [-0.5, 0.5].
Solution 3 (b).

Background. Lagrange form of the Remainder.  The Lagrange form of the error is  [Graphics:../Images/TaylorPolyMod_gr_242.gif]  where  c  is known to exist and lies somewhere between  0  and  x.

[Graphics:../Images/TaylorPolyMod_gr_243.gif]


[Graphics:../Images/TaylorPolyMod_gr_244.gif]
[Graphics:../Images/TaylorPolyMod_gr_245.gif]


First we need to bound the size of the term  [Graphics:../Images/TaylorPolyMod_gr_247.gif]  for values of  c  in the interval  [Graphics:../Images/TaylorPolyMod_gr_248.gif].  This can easily be done graphically, but to do it analytically with derivatives is quite messy.  We choose to look at the following graph to see what is happening.

[Graphics:../Images/TaylorPolyMod_gr_249.gif]

[Graphics:../Images/TaylorPolyMod_gr_250.gif]


How big does  [Graphics:../Images/TaylorPolyMod_gr_252.gif]  get ?  Looking at the graph we can estimate it to be  [Graphics:../Images/TaylorPolyMod_gr_253.gif].  

That's good enough.

How big does the error  [Graphics:../Images/TaylorPolyMod_gr_254.gif]  get ?  Notice that [Graphics:../Images/TaylorPolyMod_gr_255.gif][Graphics:../Images/TaylorPolyMod_gr_256.gif].  
We will use the bound the first portion[Graphics:../Images/TaylorPolyMod_gr_257.gif]  and  then bound the portion  [Graphics:../Images/TaylorPolyMod_gr_258.gif]  over the interval  [-0.5, 0.5]  by evaluating it at  [Graphics:../Images/TaylorPolyMod_gr_259.gif]  

[Graphics:../Images/TaylorPolyMod_gr_260.gif]

[Graphics:../Images/TaylorPolyMod_gr_261.gif]

Now multiply the two numbers together to find the error bound for Lagrange's remainder formula.

    [Graphics:../Images/TaylorPolyMod_gr_262.gif]  

This is a quite a bit larger than the actual maximum error we found. Remember is an "error bound."

[Graphics:../Images/TaylorPolyMod_gr_263.gif]

[Graphics:../Images/TaylorPolyMod_gr_264.gif]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(c) John H. Mathews 2004