Scribd
Upload
11 views16 pages

Axiomatic Taylor Polynomials

Axiomatic Polynomila s

Uploaded by

kartikeyas0310
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
11 views16 pages

Axiomatic Taylor Polynomials

Axiomatic Polynomila s

Uploaded by

kartikeyas0310
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 16

Fundamental Question:

By comparing graphs of transcendental


functions, can we develop polynomials
that approximate those functions?

Answer:
We will axiomatically develop polynomials
to approximate cosine, sine and the
exponential function (base e).
What does axiomatically mean?
What is an axiom?

It's the ship


from Wall­E

The Structure of Mathematics

theorems

axioms

defined terms

undefined terms
Using Desmos to Approximate f(x) = cos x with polynomials

Under sufficient zoom, y = 1 (blue) is a good


approximation of f(x) = cos x (red)
Cosine is (1) an even function
(2) really wavy

Polynomials can be even functions and they


are more wavy the higher the degree

y = cos x (red) and


2 4 6 8 10 12
x
y=1­ + ­ + ­ x x x x x (green)
+
2! 4! 6! 8! 10! 12!

What if we extended this forever and had a


polynomial of infinite degree?
Using Desmos to Approximate f(x) = sin x with polynomials

y=x

y = sin x

3
y=x­ x
3!
3 5
y=x­ x + x
3! 5!

3 5 7
y=x­ x x
+ ­ x
3! 5! 7!

Just like Calculus, the more


you have, the better it gets.
Using Desmos to Approximate f(x) = ex with polynomials
The ex
function
(blue) and
its power
series
represen­
tations (red)

.gif file created by Oleg Alexandrov using MATLAB, August 2007


We axiomatically accept:
2 3 4 5
x x x x
x
e = 1+ x + + + + + ...
2! 3! 4! 5!
3 5 7 9
sin x = x - x + x x + x ...
-
3! 5! 7! 9! -
2 4 6 8
cos x = 1 - x + x x + x ...
-
2! 4! 6! 8! -

Writing Them With Summation...


∞
xn
ex = Σ
n=0
n!
(recall that 0! = 1)

∞ 2n+1 ∞ 2n
x x
sin x = Σ
n=0
n
(­1) .
(2n+1)!
cos x = Σ
n=0
(­1)n.
(2n)!
Working with Taylor Polynomials
Examples
2 3
(0.25x) (0.25x)
1. Let f(x) = e0.25x and T(x) = 1 + 0.25x + +
2! 3!
What is f(0.75)? Is T(0.75) a reasonable approximation of
f(0.75)? Why or why not?

f(0.75) ≈ 1.206230 and T(0.75) ≈ 1.206177. If f(0.75) is


our "true" value and T(0.75) is our "observed" value, then
our percent error is 0.00443%. That's pretty good.

2. sin(0.4) ≈ 0.389418. The one term approximation


sin x ≈ x is okay here, and the two term approximation
3
sin x ≈ x - x is better still. How many terms of
3!
x3 x5 x7 x9 x11 ...
sin x ≈ x - +
3! 5! - 7! + 9! - 11! +

would we need to use to have an approximation of


sin(0.4) that is accurate to the eighth decimal place?
sin(0.4) ≈ 0.38941834
sin(0.4) ≈ 0.4 is not accurate to any decimal places.
(0.4)3
sin(0.4) ≈ 0.4 ­ = 0.389333 is accurate to 3 places
3!
(0.4)3 (0.4)5
sin(0.4) ≈ 0.4 ­ + = 0.38941866 is accurate
3! 5! to 6 places
(0.4)3 (0.4)5 (0.4)7
sin(0.4) ≈ 0.4 ­ + ­ = 0.38941834
3! 5! 7!
is accurate to 8 places, so we needed 4 terms.

Yours: The sixth degree polynomial approximation


T(x) = 1 + 2x ­ 9 x2 ­ 4 x3 + 27
8 x 4
+ 4 x5 ­ 81 x6
2 3 15 80
is used to estimate the values of f(x) = sin(2x) + cos(3x).

By how much is T(0.6) in error when approximating


f(0.6)?

f(0.6) = 0.704837, but T(0.6) = 0.702897,


so in error by 0.00275268
Yours: To 3 decimals, what is the greatest value a such that
1
T(x) = x + 1 x2 + 1 x3 + 1 x4 + x5
5 50 750 15000

approximates f(x) = xe0.2x to within 0.01 for all x in the


interval [0, a]?

1 2 1 3
T(x) = x + x + x + 1 x4 + 1 x5
5 50 750 15000
f(x) = xe0.2x

a = 3.854
Yours: Given: f(x) = cos( 1 x) and
2
2 4 6
Tn(x) = 1 ­ 2 x x ­ x ... n
x2n
+ + + (­1) 2n
2 2! 24 4! 26 6! 2 (2n)!

find the minimum value of n so that Tn(0.5)


approximates f(0.5) to within 0.00001.

Answer
f(0.5) = 0.968912

T0(x) = 1 T0(0.5) = 1 Not within 0.00001 of 0.96812

2
T1(x) = 1 ­ 2 x T1(0.5) = 0.96875 Almost, but still
2 2!
not within
0.00001 of
0.096812. It is
0.000162 off.
2 4
T2(x) = 1 ­ 2 x x
+
2 2! 24 4!

T2(0.5) = 0.968913 This is within 0.0001 of


0.968912 and in fact, it is only
0.0000003387 off, so n = 2 is
already a great approximation.

Axiomatic Homework
1. f(π/3) = ­1.38967 and T8(π/3) = ­1.37639, so the
error is 0.01328 (error is always positive) and the
approximation is an overestimate as f(π/3) < T8(π/3).
2. cos(1.75) = ­0.17824605565
T0(1.75) = 1
T1(1.75) = ­0.53125
T2(1.75) = ­0.1404622396
T3(1.75) = ­0.1803551568
T4(1.75) = ­0.1781735129
T5(1.75) = ­0.178247749 so need n ≥ 5

3. T4(x) approximates f(x) with less than 0.01 error


for up to xmax = 1.085782
2 4 6 8
x ...
4. T(x) = ( x
1- + x
-
x
3! 5! 7! 9!
+ - )
2 8
( x x4
x6
x
- 1- 2! + 4! - 6! + 8! -... )
1 1 1 1 x4 1 1 x6 ...
= (2! - 3!) x2 + (5! - 4! ) + (6! - 7! ) +
1 1 1 1 x4 1 1 x6 ...
= (2! - 3!) x2 - (4! - 5! ) + (6! - 7! ) -
5. Yes, direct substitution works.
6. The graph:

­1 + cos x
lim x =0
x 0

2 4 6 8
x x x x ...
­1 + 1 - + - + -
2! 4! 6! 8!
6. T(x) = x
x2 x4
x6
x8
- 2! + 4! - 6! + 8! - ...
T(x) =
x
x x3 x5
x7
...
T(x) = - + - + -
2! 4! 6! 8!

There is no longer "division by 0" and we can


evaluate T(0) directly: T(0) = 0.

You might also like