DOC PREVIEW
MIT 18 03 - Fourier series II

This preview shows page 1-2 out of 5 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

18.03 Class 21, March 29 Fourier series II [1] Review [2] Square wave [3] Piecewise continuity [4] Tricks [1] Recall from before break: A function f(t) is periodic of period 2L if f(t+2L) = f(t) . Theorem: Any decent periodic function f(t) of period 2pi has can be written in exactly one way as a *Fourier series*: f(t) = a_0/2 + a_1 cos(t) + a_2 cos(2t) + ... + b_1 sin(t) + b_2 sin(2t) + ... If the need arises, the "Fourier coefficients" can be computed as integrals: a_n = (1/pi) integral_{-pi}^{pi} f(t) cos(nt) dt , n geq 0 b_n = (1/pi) integral_{-pi}^{pi} f(t) sin(nt) dt , n > 0 [2] Squarewave: A basic example is given by the "standard squarewave," which I denote by sq(t) : it has period 2pi and sq(t) = 1 for 0 < t < pi = -1 for -pi < t < 0 = 0 for t = 0 , t = pi This is a standard building block for all sorts of "on/off" periodic signals. It's odd, so a_n = integral_{-¥pi}^pi odd . even dt = 0 for all n . If f(t) is an odd function of period 2pi, we can simplify the integral for bn a little bit. The integrand f(t) sin(nt) is even, so the integral is twice the integral from 0 to pi: bn = (2/pi) integral_0^pi f(t) sin(nt) dt Similarly, if f(t) is even then an = (2/pi) integral_0^pi f(t) cos(nt) dt In our case this is particularly convenient, since sq(t) itself needs different definitions depending on the sign of t. We have: bn = (2/pi) integral_0^pi sin(nt) dt = (2/pi) [ - cos(nt) / n ]_0^pi = (2/pi n) [ - cos(n pi) - (-1) ]= (2/pi n) [ 1 - cos(n pi) ] This depends upon n : n cos(n pi) 1 - cos(n pi) 1 -1 2 2 1 0 3 -1 2 and so on. Thus: bn = 0 for n even = 4pi/n for n odd and sq(t) = (4/pi) [ sin(t) + (1/3) sin(3t) + (1/5) sin(5t) + ... ] This is the Fourier series for the standard squarewave. I used the Mathlet FourierCoefficients to illustrate this. Actually, I built up the function (pi/4) sq(t) = sin(t) + (1/3) sin(3t) + (1/5) sin(5t) + .... (**) and observed the fit. [3] What is "decent"? This is quite amazing: the entire function is recovered from a *discrete* sequence of slider settings. They record the strength of the harmonics above the fundamental tone. The sequence of Fourier coefficients is a "transform" of the function, one which only applies (in this form at least) to periodic functions. We'll see another example of a transform later, the Laplace transform. Let's be more precise about decency. First, a function is *piecewise continuous* if it is broken into continuous segments and such that at each point t = a of discontinuity, f(a-) = lim_{t ---> a from below} f(t) and f(a+) = lim_{t ---> a from above} f(t) exist. They exist at points t = a where f(t) is continuous, too, and there they are equal. So f(t) = 1/t is NOT piecewise continuous, but sq(t) is . A function is "decent" if it is piecewise continuous and is such that at each point of discontinuity, t = a , the value at a is the average of the left and right limits: f(a) = (1/2) (f(a+) + f(a-)) So the square wave is decent, and any continuous function is decent. Addendum to the theorem:At points of discontinuity, the Fourier series can't make up its mind, so it converges to the average of f(a+) and f(a-) . For example, evaluate the Fourier series for sq(t) at t = pi/2: sin( pi/2) = +1 sin(3pi/2) = -1 sin(5pi/2) = +1 ... so 1 = (4/pi) (1 - 1/3 + 1/5 - 1/7 + ...) or 1 - 1/3 + 1/5 - 1/7 + ... = pi/4 Did you know this? It's due to Newton and Leibnitz. [4] Tricks: Any way to get an expression (*) will give the same answer! Example [trig id]: cos(t - pi/4) . How to write it like (*) ? Well, there's a trig identity we can use: a cos(t) + b sin(t) = A cos(t - phi) if (a,b) has polar coord's (A,phi) a = A cos(phi), b = A sin(phi) : For us, A = 1, phi = pi/4, so a = b = 1/sqrt(2) and cos(t - pi/4) = (1/sqrt(2)) cos(t) + (1/sqrt(2) sin(t) . That's it: that's the Fourier series. This means a1 = b1 = sqrt(2) and all the others are zero. Example [linear combinations]: 1 + 2 sq(t) = 1 + (8/pi) ( sin(t) + (1/3) sin(3t) + ... ) Example [shifts]: f(t) = sq(t + pi/2) = (1/2) (4/pi) (sin(t + pi/2) + (1/3) sin(3(t + pi/2)) + ... ) sin(theta + pi/2) = cos(theta) , sin(theta - pi/2) = - cos(theta) so f(t) = (4/pi) (cos(t) - (1/3) cos(3t) + (1/5) cos(5t) - ... ) Example [Stretching]: What about functions of other periods? Suppose g(x) has period 2L . Building blocks: cos(n(pi/L)x) and sin(n(pi/L)x) are periodic of period 2L .Then the Fourier series for g(x) is: g(x) = a_0/2 + a_1 cos((pi/L) x) + a_2 cos((2pi/L) x) + ... + b_2 sin((pi/L) x) + b_2 sin((2pi/L) x) + ... Example: sq((pi/2) x) has period 4 , not 2pi : L = 2. But we can still write (using the *substitution* t = (pi/2) x ) : sq(2pi x) = (4/pi) ( sin((pi/2)x) + (1/3) sin(3(pi/2) x) + ... ) There are integral formulas as well. [Slide]MIT OpenCourseWare http://ocw.mit.edu 18.03 Differential Equations�� Spring 2010 For information about citing these materials or our Terms of Use, visit:


View Full Document

MIT 18 03 - Fourier series II

Documents in this Course
Exam II

Exam II

7 pages

Exam 3

Exam 3

8 pages

Load more
Download Fourier series II
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Fourier series II and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Fourier series II 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?