Example
6. Compare the
Horner method and ordinary method of polynomial evaluation for
Solution 6.
Construct a the polynomial and two version of Horner's method.
Aside. This polynomial
has known roots
.
For illustration we can evaluate
.
![[Graphics:../Images/HornerMod_gr_188.gif]](../Images/HornerMod_gr_188.gif)
We are done.
Aside. Consider the
computational time for vector arguments for
evaluating P(x) and
the nested multiplication form Q(x) and
Horner form H(x). The
latter two take roughly 1/2 the
time to do the evaluations
![[Graphics:../Images/HornerMod_gr_190.gif]](../Images/HornerMod_gr_190.gif)
We are really done.
Aside. The product
form of the polynomial,
is
the "best computational form." We can compare the "error"
with it and the other three forms of this polynomial.
![[Graphics:../Images/HornerMod_gr_193.gif]](../Images/HornerMod_gr_193.gif)
![[Graphics:../Images/HornerMod_gr_195.gif]](../Images/HornerMod_gr_195.gif)
![[Graphics:../Images/HornerMod_gr_197.gif]](../Images/HornerMod_gr_197.gif)
An analysis of the error has been done by G. W. Stewart in the
article:
Error
Analysis of the Algorithm for Shifting the Zeros of a Polynomial by
Synthetic Division
G. W. Stewart
Mathematics of Computation, Vol. 25, No. 113. (Jan., 1971), pp.
135-139, Jstor.
(c) John H. Mathews 2004