Beruflich Dokumente
Kultur Dokumente
breaking it down into a linear combination of the basis vectors such that
f (x) = a1 e1 + + an en
Approaching the subject mathematically, let's say we are given a function space V with basis = {b1 , ... , bn }, and within this space we want to approach, or approximate the analog function f () by
In fact, Fourier was the one to rst come up with a groundbreaking theorem for this type of problem, and in it he proves that any continuous, periodic signal can be approximated as a sum of sines and cosines. So let's use those as our basis vectors and try to solve for the cocients, a1 , ... ., an . With a Fourier series, we are going to try to write a series representation for f (x) on L x L in the form,
f ( x) =
n=0
An cos
nx nx + Bn sin L L n=1
From our background in dierential equations and calculus, we see that a Fourier series is, in some way, a combination of the Fourier sine and Fourier cosine series. For now, it's not critically important to know if the series will actually converge to f (x), but we'll just assume it does. Next, we need to determine the coecients An and Bn . Now, cutting some corners here but bear with me, imagine the graphs of sine and cosine. They are perfectly opposite by inspection. It's a little long, but it's relatively simple to prove that the sets are mutually
nx nx and cos L L n=1 n=1 orthogonal on the interval L x L. The specic formulas for this principle are if n = m = 0 L 2L nx mx cos cos dx = L if n = m = 0 L L L 0 if n = m sin
sin
L
mx nx sin dx = L L
L 0
if n = m if n = m
mx L
So let's start o by multiplying both sides of the series above, f (x), by cos L to L.
f ( x) =
n=0
An cos
nx nx + Bn sin L L n=1
L
mx f (x) cos dx = L L
Bn sin
L n=1
nx mx cos dx L L
Integral signs are just fancy summation symbols. So let's change them around just for shits and giggles.
f (x) cos
L
mx dx = An L n=0
cos
L
nx mx cos dx + Bn L L n=1
sin
L
nx mx cos dx L L
Now we'll take advantage of the fact that the sines and cosines are mutually orthogonal. The integral in the second series will always be zero and in the rst series, the integral will be zero if n = m. This reduces to
f (x) cos
L
mx dx = L
Am (2L) Am (L)
if n = m = 0 if n = m = 0
1 2L
f (x) dx
L
Am
f (x) cos
L
mx dx L
m = 1, 2, 3, ...
mx L
Now, do it all over again, ya varmint! But this time, multiply both sides by sin sides from L to L and interchange the integral and summation to get
mx dx = An f (x) sin L L n=0
L
. Integrate both
sin
L
nx mx sin dx L L
In this case the integral in the rst series will always be zero and the second will be zero if n = m and so we get
f (x) sin
L
mx dx = Bm (L) L
f (x) sin
L
mx dx L
m = 1, 2, 3, ...
With Fourier sine and Fourier cosine series, since the functions are conveniently even, it's quite easy to show a reduced form of the coecients in terms of an integral from 0 to L. However, since this Fourier series, or Fourier transform is arbitrary, we have no idea whether f (x) is even, odd, or neither. So there it is right there! 2