Derivation of the free particle propagator
In general the kernel of the Schrödinger integral equation, A(x, t; y, 0) is
Z
ψ(x, t) = dyA(x, t; y, 0)ψ(y, 0)dy . (1)
Suppose that have solved the time independent Schrödinger equation:
ĤφE (x) = EφE (x) , (2)
i.e., we know the eigenfunctions and eigenvalues of the Hamiltonian Ĥ.
Then, for a discrete spectrum we know that:
X
ψ(x, t) = cEn φEn (x)e−iEn t/~ , (3)
n
with Z
cEn = dyφ∗En (y)ψ(y, 0) . (4)
Substituting Eq. (4) into Eq. (9) we obtain
Z X
ψ(x, t) = dy φ∗En (y)φEn (x)e−iEn t/~ψ(y, 0) , (5)
n
which compared with Eq. (1), gives
X
A(x, t; y, 0) = φ∗En (y)φEn (x)e−iEn t/~ . (6)
n
Note that we have used the fact that the energy eigenfunctions are prop-
erly orthonormalized in writing the expansion of Eq. (9). The continuous
spectrum case is analogous. We have
Z
ψ(x, t) = dEc(E)φE (x)e−iEx/~ , (7)
with Z
c(E) = dxφ∗E (y)ψ(y, 0) (8)
Substituting Eq. (8) into Eq. (7) we obtain
Z
ψ(x, t) = dydEφ∗E (y)φE (x)e−iEt/~ψ(y, 0) , (9)
1
which compared with Eq. (1), gives
Z
A(x, t; y, 0) = dEφ∗E (y)φE (x)e−iEt/~ . (10)
Even in this case we have impicitly assumed that the energy eigenfunctions
were properly normalized.
Let us now consider the free particle case. The eigenfunctions of the
Hamiltonian are also eigenfunctions of the momentum
p̂φp (x) = pφp (x) (11)
p2
Ĥφp (x) = φp (x) , (12)
2m
p2 ~2 k 2
with E = 2m = 2m and
φp (x) = Aeipx/~ (13)
where A is the normalization factor and −∞ < p < +∞. As we have
mentioned a few times, the energy spectrum has a two-fold degeneracy:
√
for each energy eigenvalue E, two eigenfunctions exist with p = ± 2mE.
Now, we have argued that we can properly normalize this state also in the
continuum, using the defintion of the delta function:
Z +∞
1
δ(x − y) = dueiu(x−y) . (14)
2π −∞
It is important, however, to notice that the normalization factor will be dif-
ferent in the case we want properly normalized eigenfunctions of the energy
or the momentum. As we have seen, the following expression
1
φp (x) = √ eipx/~ (15)
2π~
gives the correctly normalized momentum eigenfunctions:
Z
dxφ∗p (x)φp′ (x) = δ(p − p′ ) (16)
By using the property of the δ(x) functions one can check that
m 1 1 √
4
φE (x) = √ e±i 2mEx/~ (17)
2E 2π~
gives the correctly normalized momentum eigenfunctions:
Z
dxφ∗E (x)φE ′ (x) = δ(E − E ′ ) (18)
2
We have now all the ingredients to calculate the free-particle propagator.
Z +∞
A(x, t; y, 0) = dEφ∗E (x)φE (y)e−iEt/~ (19)
0
Z +∞ m 1 1 √
2
= dE e±i 2mE(x−y)/~e−iEt/~ , (20)
0 2E 2π~
We can now make a substitution:
√ √
2mE 2m 1
k= ⇒ dk = √ dE , (21)
~ ~ 2 E
which gives
+∞
1
Z
2 t/2m
A(x, t; y, 0) = dkeik(x−y) e−i~k . (22)
2π −∞
Note that this is exactly the result which we would have obtained, had we
(more cleverly) started from the expression of the propagator in terms of
the eigenfunction of the momentum (k = p/~)
Z
2
A(x, t; y, 0) = dpφ∗p (x)φp (y)e−ip t/2m~ . (23)
Using the formula (which can be obtained by completing the square in
the exponent)
Z +∞ r
−ak 2 +bk π b2
dke = e 4a (24)
−∞ a
we finally obtain:
r
m − m(x−y)2
A(x, t; y, 0) = e 2i~t . (25)
2πi~t
