(x k)
(1)
Reminders
Fourier Coefcients
Let f be a T -periodic function, we have : = 2 T ck = 1 T
f (x) =
kZ
ck eikx with
f (t)eikt dt
0
The ck are called the Fourier coefcients of f . This coefcient can be rewritten as an integral over any interval of length T . In particular, we will use :
T /2
1 ck = T
T /2
f (t)eikt dt
(2)
Proof. Let g(t) = f (t)eikt . It is a T -periodic function since we have : g(t + T ) = f (t + T )eik(t+T ) = f (t)eikt eikT = f (t)eikt eik2 = g(t)
=1
Equation (2) is said due to the fact that the integral of a T -periodic function is constant over any interval of length T as can be seen from :
c+T 0 T c+T
g(t)dt =
c c
g(t)dt +
0
g(t)dt +
T
g(t)dt
t=u+T
=
c
+ g(t)dt
g(t)dt +
0
g(u)du =
g(t)dt
0
Fourier Transform
The Fourier transform F () of a real-valued function f (x) is dened by :
F () =
f (x)eix dx
(3)
1 f (x) = 2
F ()eix d
(4)
When two functions are related by the Fourier transform, we note : f (x) F () We have the symmetry property : if f (x) F () then F (x) 2f () and the linearity property : if f (x) F () f (x) F ()
FT FT FT FT FT
(5)
FT
(6)
(7)
(8)
Impulsion train
Lets consider it(x) = pZ (x pT ) a train of T -spaced impulsions and lets compute its Fourier transform. We rst rewrite f using its Fourier coefcients : it(x) =
kZ
ck eikx
1 ck = T
T /2
1 it(t)eikt dt = T
T /2
(t pT )eikt dt
T /2 pZ
1 = T
(t pT )eikt dt
pZT /2
Since the function t (t pT ) is null over the interval [T /2, T /2] for p = 0, we are left with only one term in the sumation :
T /2
1 ck = T
T /2
(t)e
ikt
1 dt = T
(t)eikt dt =
=ei0 =1
1 T
by (7)
(9)
Applying the symmetry property (5) to the Fourier transform of a Dirac (8) we nd : eik(x) 2((x) k) eikx 2(x k) Applying linearity (6) to expression (9) we nally get the equality (1).
FT FT