MIT OpenCourseWare http://ocw.mit.edu 18.085 Computational Science and Engineering IFall 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.� 18.085 Prof. Strang Quiz 3 Decemb e r 3, 2003 Your name is: Grading 1 2 3 Total Thank you for taking 18.085, I hope you enjoyed it. 1) (35 pts.) Suppose the 2π-p eriodic f(x) is a half-length square wave: f(x) = ⎧⎪⎪⎪⎨ ⎪⎪⎪⎩ 1 for 0 < x < π/2 for − π/2 < x < 0−1 0 elsewhere in [−π, π] (a) Find the Fourier cosine and sine coefficients ak and bk of f(x). � π(b) Compute −π (f(x))2 dx as a number and also as an infinite series using 2the ak and b2 k . x(c) DRAW A GRAPH of the integral I(x) = 0 f(t) dt from −π to π. What are the Fourier coefficients Ak and Bk of I(x)? df(d) DRAW A GRAPH of the derivative D(x) = dx from −π to π. What are the Fourier coefficients of D(x)? (e) If you convolve D(x) ∗ I(x) why do you get the same answer as f(x) ∗ f(x)? Not required to find that answer, just explain D ∗ I = f ∗ f.xxx 2xxx 32) (33 pts.) (a) Compute directly the convolution f ∗f (cyclic convolution with N = 6) when f = (0, 0, 0, 1, 0, 0). [You could connect vectors (f0, . . . , f5) with polynomials f0 + f1w + · · · + f5w5 if you want to.] (b) What is the Discrete Fourier Transform c = (c0, c1, c2, c3, c4, c5) of the vector f = (0, 0, 0, 1, 0, 0)? Still N = 6. (c) Compute f ∗ f another way, by using c in “transform space” and then transforming back. 4xxx 53) (32 pts.) On page 310 the Fourier integral transform of the one-sided decaying pulse f(x) = e−ax (for x ≥ 0 only) is computed for −∞ < k < ∞ as 1 f�(k) = . a + ik (a) Suppose this one-sided pulse is shifted to start at x = L: fL(x) = e−a(x−L) for x ≥ L, fL(x) = 0 for x < L . Find the Fourier integral transform f�L(k). (b) Draw a rough graph of the difference D(x) = F (x) − FL(x) and find its transform �D(k). NOW LET a 0.→ What is the limit of D(x) as a 0?→ What is the limit of �D(k) as a 0?→ (c) The function fL(x) is smooth except for a at x = L, so the decay rate of f�L(k) is . The convolution C(x) = fL(x) ∗ fL(x) has transform �C(k) = with decay rate . Then in x-space this convolution C(x) has a at the point x = . 6xxx
View Full Document
Unlocking...