Fourier Series on other Intervals : Notes

from Class
In class on W we investigated Fourier series on intervals other than (-p, p). For functions that are 2
L periodic (where 2 L is the wavelength), we compute Fourier coefficients and the Fourier series
making the modifications derived in class :

f x = +  an cos  n p x  L + bn  bn sin n p x  L

¶ ¶
a0
2 n=1 n=1

where :

 f x dx  f x cos n p x  L dx  f x sin n p x  L dx

1 L 1 L 1 L
a0 = an = bn =
L -L L -L L -L
2
Example 1 : Consider f (x) = x on (-2, 2):
It is important to recall that 2L is the length of the interval, therefore 2 L = 2 - (-2) so that the value
of L to use in the calculations is L = 2.

 x dx  x cos n p x  2 dx  x sin n p x  2 dx
1 2 1 2 1 2
a0 = 2
an = 2
bn = 2
2 -2 2 -2 2 -2

Now, since f (x) is even on (-2, 2) we can employ symmetry to write :

 x dx  x cos n p x  2 dx
2 2 2 2
a0 = 2
an = 2
bn = 0
2 0 2 0

Now, going the lazy route and letting Mathematica do the heavy lifting :
In[330]:= a0  Integratex2 , x, 0, 2

8
Out[330]=
3
In[332]:= an  Integratex2 Cosn  x  2, x, 0, 2

8 2 n  Cosn    2  n2 2  Sinn 

Out[332]=
n3 3

and let' s just be sure :

In[337]:= bn  Integratex ^ 2 Sinn  x  2, x,  2, 2
Out[337]= 0

It is easy to show that the a coefficients reduce to :

-1n
16
an =
n p
2 2

-1n cos n p x  2
so that our Fourier series is :

f x = 
¶
4 16
+
3 p2 n=1 n2
2 fourierseriesonotherintervals.nb

Verifying through Mathematica :

 16  2  Sum 1n Cosn  x  2  n2 , n, 1, 21, x,  6, 6

4
In[339]:= Plot
3
4

Out[339]= 2

-6 -4 -2 2 4 6

Note that the values are consistent with our function (f (2) = 4 ) and the function repeats with a
periodicity of 4.
Now, let' s examine the same function on (0, 4). Since our limits are not symmetric across the
origin, we cannot make use of symmetry arguments. Our integrals become :

 x dx  x cos n p x  2 dx  x sin n p x  2 dx
1 4 1 4 1 4
a0 = 2
an = 2
bn = 2
2 0 2 0 2 0

Even though the limits have changed, L = 2 since the total length of the interval, 2 L = 4.
Computing coefficients :
Cleara0, an, bn, L, f

L  2;
 I know I have stressed we should not use capital letters for variables;
In[340]:=

but I know that "L" is not restricted for Mathematica use, so let's use it. 
fx_ : x2
a0  1  L Integratefx, x, 0, 4
an  1  L Integratefx Cosn  x  L, x, 0, 4
bn  1  L Integratefx Sinn  x  L, x, 0, 4
32
Out[342]=

8 2 n  Cos2 n    1  2 n2 2  Sin2 n 

3

Out[343]=
n3 3
8 1   1  2 n2 2  Cos2 n   2 n  Sin2 n 
Out[344]= 
n3 3

And we get our three outputs. Knowing that sin (2 n p) is always zero for integer values of n, these
complicated expressions reduce quite nicely, and we have :
16 -16
an = bn =
n2 p 2 np
 fourierseriesonotherintervals.nb 3

16 cos n p x  2 -16 sin n p x  2

Our Fourier series then becomes :

f x = + +
¶ ¶
32
6 n=1 p2 n2 n=1
p n2
Plotting three cycles of this function using the first 31 terms of the expansion :
In[345]:= Plot32  6  Sum16  2  Cosn  x  2  n2   16   Sinn  x  2  n, n, 1, 31, x,  4, 8

15

10

Out[345]=

-4 -2 2 4 6 8

And we reproduce our function over three cycles.