Heuristically (ignoring problems of interchange of integrals that don’t make
sense), the Fourier inversion formula can be written as follows
ψ(q) =
1
√
2 π
∫+∞
−∞
eikqψ ̃(k)dk
=
1
√
2 π
∫+∞
−∞
eikq
(
1
√
2 π
∫+∞
−∞
e−ikq
′
ψ(q′)dq′
)
dk
=
1
2 π
∫+∞
−∞
(∫+∞
−∞
eik(q−q
′)
ψ(q′)dk
)
dq′
=
∫+∞
−∞
δ(q−q′)ψ(q′)dq′
Physicists interpret the above calculation as justifying the formula
δ(q−q′) =
1
2 π
∫+∞
−∞
eik(q−q
′)
dk
and then go on to consider the eigenvectors
|k〉=
1
√
2 π
eikq
of the momentum operator as satisfying a replacement for the Fourier series
orthonormality relation (equation 11.1), with theδ-function replacing theδnm:
〈k′|k〉=
∫+∞
−∞
(
1
√
2 π
eik′q
)(
1
√
2 π
eikq
)
dq=
1
2 π
∫+∞
−∞
ei(k−k
′)q
dq=δ(k−k′)
11.4 Linear transformations and distributions
The definition of distributions as linear functionals on the vector spaceS(R)
means that for any linear transformationAacting onS(R), we can get a linear
transformation onS′(R) as the transpose ofA(see equation 4.1), which takes
Tto
AtT:f∈S(R)→(AtT)[f] =T[Af]∈C
This gives a definition of the Fourier transform onS′(R) as
FtT[f]≡T[Ff]
and one can show that, as forS(R) andL^2 (R), the Fourier transform provides
an isomorphism ofS′(R) with itself. Identifying functionsψwith distributions