03.03.2014 Views

Numerical Methods Course Notes Version 0.1 (UCSD Math 174, Fall ...

Numerical Methods Course Notes Version 0.1 (UCSD Math 174, Fall ...

Numerical Methods Course Notes Version 0.1 (UCSD Math 174, Fall ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

7.2. RICHARDSON EXTRAPOLATION 93<br />

Exercises<br />

(7.1) Derive the approximation<br />

f ′ (x) ≈<br />

4f(x + h) − 3f(x) − f(x − 2h)<br />

6h<br />

using Taylor’s Theorem.<br />

(a) Assuming that f(x) has bounded derivatives, give the accuracy of the above approximation.<br />

Your answer should be something like O ( h ?) .<br />

(b) Let f(x) = x 3 . Approximate f ′ (0) with this approximation, using h = 1 4 .<br />

(7.2) Let f(x) be an analytic function, i.e., one which is infinitely differentiable. Let ψ(h) be the<br />

centered difference approximation to the first derivative:<br />

ψ(h) =<br />

f(x + h) − f(x − h)<br />

2h<br />

(a) Show that ψ(h) = f ′ (x) + h2<br />

3! f ′′′ (x) + h4<br />

5! f (5) (x) + h6<br />

7! f (7) (x) + . . .<br />

(b) Show that<br />

8 (ψ(h) − ψ(h/2))<br />

h 2 = f ′′′ (x) + O ( h 2) .<br />

(7.3) Derive the approximation<br />

f ′ (x) ≈<br />

4f(x + 3h) + 5f(x) − 9f(x − 2h)<br />

30h<br />

using Taylor’s Theorem.<br />

(a) What order approximation is this? (Assume f(x) has bounded derivatives of arbitrary<br />

order.)<br />

(b) Use this formula to approximate f ′ (0), where f(x) = x 4 , and h = <strong>0.1</strong><br />

(7.4) Suppose you want to know quantity Q, and can approximate it with some formula, say φ(h),<br />

which depends on parameter h, and such that φ(h) = Q + a 1 h + a 2 h 2 + a 3 h 3 + a 4 h 4 + . . .<br />

Find some linear combination of φ(h) and φ(−h) which is a O ( h 2) approximation to Q.<br />

(7.5) Assuming that φ(h) = Q + a 2 h 2 + a 4 h 4 + a 6 h 6 . . ., find some combination of φ(h), φ(h/3)<br />

which is a O ( h 4) approximation to Q.<br />

(7.6) Let λ be some number in (0, 1). Assuming that φ(h) = Q + a 2 h 2 + a 4 h 4 + a 6 h 6 . . ., find some<br />

combination of φ(h), φ(λh) which is a O ( h 4) approximation to Q. To make the constant<br />

associated with the h 4 term small in magnitude, what should you do with λ? Is this practical?<br />

Note that the method of Richardson Extrapolation that we considered used the value λ = 1/2.<br />

(7.7) Assuming that φ(h) = Q + a 2 h 2 + a 4 h 4 + a 6 h 6 . . ., find some combination of φ(h), φ(h/4)<br />

which is a O ( h 4) approximation to Q.<br />

(7.8) Suppose you have some great computational approximation to the quantity Q such that<br />

ψ(h) = Q + a 3 h 3 + a 6 h 6 + a 9 h 9 . . . Can you find some combination of ψ(h), ψ(h/2) which is<br />

a O ( h 6) approximation to Q?<br />

(7.9) Complete the following Richardson’s Extrapolation Table, assuming the first column consists<br />

of values D(n, 0) for n = 0, 1, 2:<br />

n\m 0 1 2<br />

0 2<br />

1 1.5 ?<br />

2 1.25 ? ?<br />

(See equation 7.4 if you’ve forgotten the definitions.)

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!