Example 5.  Consider the function  [Graphics:Images/TaylorPolyMod_gr_348.gif].  
5 (a).  Find the terms up to  [Graphics:Images/TaylorPolyMod_gr_349.gif]  in the Maclaurin series for  f[x].
Solution 5 (a).

5 (a).  Find the terms up to  [Graphics:../Images/TaylorPolyMod_gr_352.gif]  in the Maclaurin series for  f[x].

[Graphics:../Images/TaylorPolyMod_gr_353.gif]


[Graphics:../Images/TaylorPolyMod_gr_354.gif]

Remark.  If you just find the "Series" it will include a "Big [Graphics:../Images/TaylorPolyMod_gr_355.gif]"  term, which cannot be used in either evaluations or graphing, we eliminate it with the command "Normal."  The "Big [Graphics:../Images/TaylorPolyMod_gr_356.gif]  term lets us know the power of x in the "remainder.

[Graphics:../Images/TaylorPolyMod_gr_357.gif]

[Graphics:../Images/TaylorPolyMod_gr_358.gif]

[Graphics:../Images/TaylorPolyMod_gr_359.gif]

[Graphics:../Images/TaylorPolyMod_gr_360.gif]

Aside.  Of course the "full" Maclaurin series has an infinite number of terms.  Mathematica is capable of finding sums of infinite series.  

The following is just for fun !

There are two ways to get infinite sums, the old way and the new way.

[Graphics:../Images/TaylorPolyMod_gr_361.gif]

[Graphics:../Images/TaylorPolyMod_gr_362.gif]

The "symbolic" command is:

[Graphics:../Images/TaylorPolyMod_gr_363.gif]

[Graphics:../Images/TaylorPolyMod_gr_364.gif]

Now graph f[x] and the Maclaurin polynomial s[x] over the interval [-3, 3].  

[Graphics:../Images/TaylorPolyMod_gr_365.gif]

[Graphics:../Images/TaylorPolyMod_gr_366.gif]

[Graphics:../Images/TaylorPolyMod_gr_367.gif]

The above approximation is not too good near the endpoints, this is why we want to graph f[x] and the Maclaurin polynomial s[x] over the interval [-2, 2].  

[Graphics:../Images/TaylorPolyMod_gr_368.gif]

[Graphics:../Images/TaylorPolyMod_gr_369.gif]

[Graphics:../Images/TaylorPolyMod_gr_370.gif]

Background for part 5 (b).  Use the fact that the series is "alternating" to investigate the error for the Maclaurin polynomial of degree n = 20 over the interval  [-2.0, 2.0].  

Now look closely at the "error" when the series is used to approximate the function.  How close are were the two curves in part (a) ?

[Graphics:../Images/TaylorPolyMod_gr_371.gif]

[Graphics:../Images/TaylorPolyMod_gr_372.gif]

[Graphics:../Images/TaylorPolyMod_gr_373.gif]

This series is "very nice" because it is alternating, and for that reason the error bound is the magnitude of the "next non-zero term" in the series.

[Graphics:../Images/TaylorPolyMod_gr_374.gif]

[Graphics:../Images/TaylorPolyMod_gr_375.gif]

[Graphics:../Images/TaylorPolyMod_gr_376.gif]

[Graphics:../Images/TaylorPolyMod_gr_377.gif]

The error bound for the entire interval  [-2.0, 2.0] is

[Graphics:../Images/TaylorPolyMod_gr_378.gif]

[Graphics:../Images/TaylorPolyMod_gr_379.gif]

[Graphics:../Images/TaylorPolyMod_gr_380.gif]

[Graphics:../Images/TaylorPolyMod_gr_381.gif]

This estimate is "conservative" and is a little larger than the "actual" maximum error which occurred at   [Graphics:../Images/TaylorPolyMod_gr_382.gif]  

[Graphics:../Images/TaylorPolyMod_gr_383.gif]

[Graphics:../Images/TaylorPolyMod_gr_384.gif]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(c) John H. Mathews 2004