Fourier series: relation of coefficients

divB
Messages
85
Reaction score
0
Hi,

The Fourier series can (among others) expressed in terms of sines and cosines with coefficients a_n and b_n and solely by sines using amplitudes A_n and phase \phi_n.

I want to express the latter using a_n and b_n. Using

<br /> a_n = A_n \sin(\phi_n) \\<br /> b_n = A_n \cos(\phi_n)<br />

I quickly found A_n by expressing the arccos and arcsin. For \phi_n I would get

\phi_n = \arccos \frac{a_n}{\sqrt{a_n^2 + b_n^2}}

However, according to the German Wikipedia (http://de.wikipedia.org/wiki/Fourier-Reihe) this seems not so trivial. One option there is (for n \neq 0):

\phi_n = 2 \arctan \frac{b_n}{A_n + a_n}

or using arctan and signum function. What I am missing or is my approach also correct?

Thanks,
divB
 
Physics news on Phys.org
You are starting with the two styles of the Fourier transforms:
$$\sum_n a_n\cos nx + b_n\sin nx = \sum_m A_m\sin(mx+\phi_m)$$
... and you want to relate ##\{a_n,b_n\}## to ##\{A_m,\phi_m\}## ... is that correct?
 
yes, exactly. And I relate them with \sin(a+b) = \sin(a)\cos(b) + \cos(a)\sin(b) to get:

<br /> A_0 = a_0/2\\<br /> a_0 = 2A_0<br />

<br /> a_n = A_n \sin\phi_n \\<br /> b_n = A_n \cos\phi_n<br />

For represeting A_n I get using both equations above:

<br /> A_n = a_n/\sin\phi_n \\<br /> cos\phi_n = b_n/A_n \\<br /> \phi_n = \arccos b_n/A_n = \arcsin \sqrt{1-b_n^2/A_n^2} \\<br /> A_n = \frac{a_n}{\sqrt{1-\frac{b_n^2}{A_n^2}}} \\<br /> \dots \\<br /> A_n = \sqrt{a_n^2 + b_n^2}<br />

I think this is correct, right?
In the same spirit, I can derive

<br /> \phi_n = \arccos\frac{b_n}{A_n} = \arccos\frac{b_n}{\sqrt{a_n^2 + b_n^2}}<br />

... but according to the German Wikipedia this is wrong. Why?
 
That works ...

To put it another way - you can expand each term in the sin-only expansion in terms of the sin-cos expansion ... i.e.

$$A_n\sin(nx+\phi_n)=\sum_m a_m\cos(mx)+b_m\sin(mx)$$

Then you apply the identity.
Since you are expanding an arbitrary sine in terms of sines and cosines, the sum should only have one term in it... (for each value of n on the RHS.)

Leads you to:

##A_n^2=a_n^2+b_n^2##

##\phi_n=\arctan(a_n/b_n)##

... sure. Nicely done.
 

Similar threads

Replies
4
Views
3K
Replies
6
Views
2K
Replies
11
Views
2K
Replies
1
Views
1K
Replies
3
Views
2K
Replies
16
Views
2K
Back
Top