Sei sulla pagina 1di 5

Taylor polynomials: applications and error estimation

MA 123, Fall 2014


Taylor polynomials: applications and error estimation
Taylor polynomials
Recall that given a function f(x), the Taylor polynomial of order n for f(x)
near x = a is dened as follows
P
n
(x) =
n

k=0
f
(k)
(a)
k!
(x a)
k
= f(a) +
f

(a)
1!
(x a) + +
f
(n)
(a)
n!
(x a)
n
Note that the actual degree of P
n
(x) may be smaller than n (for example, if
f
(n)
(a) = 0).
When the function f(x) is nice, we can use the Taylor polynomials of f(x)
to approximate f(x) near x = a and
P
n
(x) f(x) as n .
Taylor polynomials: applications and error estimation
Error bound
Now we would like to estimate the error in approximating f(x) by its Taylor
polynomial P
n
(x) of order n. The error is the dierence
E
n
(x) = f(x) P
n
(x),
so, we want to nd the bound on the magnitude of the error |E
n
(x)|.
The following result, known as the Taylors Inequality or the Lagrange Error
Bound, helps to estimate the error.
Theorem (The Lagrange Error Bound for P
n
(x))
Suppose f(x) and all its derivatives are continuous. Then
|E
n
(x)| = |f(x) P
n
(x)|
M
(n + 1)!
|x a|
n+1
where |f
(n+1)
(x)| M on the interval between a and x.
Taylor polynomials: applications and error estimation
Finding the error bound
Example
What is the maximal error possible in using the approximation
sin x x
x
3
3!
+
x
5
5!
when |x| 0.3?
Solution: Observe that
x
x
3
3!
+
x
5
5!
is the Taylor polynomial P
5
(x) for sin x near x = 0. Hence, we are looking for
the error bound in the approximation sin x P
5
(x).
We have E
5
(x) = sin x P
5
(x) and since | sin
(n+1)
(x)| 1 for every x, by the
Lagrange Error Bound theorem we get
|E
5
(x)|
1
(5 + 1)!
|x|
6+1
Taylor polynomials: applications and error estimation
Approximating by Taylor polynomials
Finally, since |x| 0.3 then
|E
5
(x)|
1
(5 + 1)!
|x|
6+1

|0.3|
5+1
(5 + 1)!
= 0.0000010125
which can serve as an estimate for the maximal possible error.
In fact, the estimate above can be improved since in the case of sin x we have
P
5
(x) = P
6
(x).
Hence, again under the assumption |x| 0.3, by the Lagrange Error Bound
theorem we have
| sin x P
5
(x)| = | sin x P
6
(x)| = |E
6
(x)|
1
(6 + 1)!
|x|
6+1

|0.3|
6+1
(6 + 1)!
= 0.00000004339286
This estimate of the maximal error is better than the previous one.
Taylor polynomials: applications and error estimation

Potrebbero piacerti anche