The Free Particle

Gaussian Wave Packets

The Gaussian wave packet initial state is one of the few states for which both the
{|x i} and {|p i} basis representations are simple analytic functions and for which the
time evolution in either representation can be calculated in closed analytic form. It
thus serves as an excellent example to get some intuition about the Schrödinger
equation.

We define the {|x i} representation of the initial state to be

«1/4 2
− x 2
„
1 i p0 x 4σ
ψx (x, t = 0) = hx |ψ(0) i = e~ e x (5.10)
2 π σx2
√
The relation between our σx and Shankar’s ∆x is ∆x = σx 2. As we shall see, we
choose to write in terms of σx because h(∆X )2 i = σx2 .

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 292
 The Free Particle (cont.)

Before doing the time evolution, let’s better understand the initial state. First, the
symmetry of hx |ψ(0) i in x implies hX it=0 = 0, as follows:
Z ∞
hX it=0 = hψ(0) |X |ψ(0) i = dx hψ(0) |X |x ihx |ψ(0) i
−∞
Z ∞
= dx hψ(0) |x i x hx |ψ(0) i
−∞
«1/2 2
Z ∞ „
1 − x 2
2σ
= dx x e x =0 (5.11)
−∞ 2 π σx2

because the integrand is odd.

Second, we can calculate the initial variance h(∆X )2 it=0 :

«1/2 2
Z ∞ „
1 − x 2
h(∆X )2 it=0 = dx x 2 − hX i2t=0 = σx2
` ´ 2σ
e x (5.12)
−∞ 2 π σx2

where we have skipped a few steps that are similar to what we did above for hX it=0
and we did the final step using the Gaussian integral formulae from Shankar and the
fact that hX it=0 = 0.

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 293
 The Free Particle (cont.)

We calculate the {|p i}-basis representation of |ψ(0) i so that calculation of hPit=0

and h(∆P)2 it=0 are easy (by contrast, Shankar Example 4.2.4 does this in the {|x i}
basis):
Z ∞
ψp (p, t = 0) = hp |ψ(0) i = dx hp |x ihx |ψ(0) i
−∞
«1/4 2
1
Z ∞ i
„
1 i − x 2
= √ dx e − ~ px
e~ p0 x
e 4σ x
2π~ −∞ 2 π σx2
(p−p0 )2
!1/4
1 − 2
4 σp
= e (5.13)
2 π σp2

where the √1 comes from the normalization |p i = √1 |k i, where σp ≡ 2 ~σ , and the

~ ~ x
final step is done by completing the square in the argument of the exponential and
R∞ 2 √
using the usual Gaussian integral −∞ du e −u = π. With the above form for the
{|p i}-space representation of |ψ(0) i, the calculation of hPit=0 and h(∆P)2 it=0 are
calculationally equivalent to what we already did for hX it=0 and h(∆X )2 it=0 , yielding

hPit=0 = p0 h(∆P)2 it=0 = σp2 (5.14)

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 294
 The Free Particle (cont.)

We may now calculate |ψ(t) i. Shankar does this only in the {|x i} basis, but we do it
in the {|p i} basis too to illustrate how simple it is in the eigenbasis of H. The result
is of course

(p−p0 )2
!1/4
1 − 2 i p2
ψp (p, t) = hp |ψ(t) i = e 4 σp
e− ~ 2m
t
(5.15)
2 π σp2

That is, each {|p i} picks up a complex exponential factor for its time evolution. It is
immediately clear that hPi and h(∆P)2 i are independent of time. Calculationally, this
occurs because P, and (∆P)2 simplify to multiplication by numbers when acting on
|p i states and the time-evolution complex-exponential factor cancels out because the
two expectation values involve hψ | and |ψ i. Physically, this occurs because the P
operator commutes with H; later, we shall derive a general result about conservation
of expectation values of operators that commute with the Hamiltonian. Either way
one looks at it, one has

hPit = hPit=0 = p0 h(∆P)2 it = h(∆P)2 it=0 = σp2 (5.16)

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 295
 The Free Particle (cont.)

Let’s also calculate the {|x i} representation of |ψ(t) i. Here, we can just use our
propagator formula, Equation 5.6, which tells us
Z ∞
ψx (x, t) = hx |ψ(t) i = dx 0 [U(t)]xx 0 hx 0 |ψ(0) i
−∞
Z ∞
r 0 2
„ «1/4 (x 0 )2
m i m (x−x ) 1 i p0 x 0
−
= dx 0 e~ 2t e~ e 4 σx2

−∞ 2π~i t 2 π σx2
23
«–−1/2 ` p0 ´2
x − t
»q „
i ~t
= 2 π σx2 1 + exp 4− “ m
6 7
”5
2 m σx2 4 σ2 1 + i ~ t x 2 m σx2

i p02
„ « „ «
i
exp p0 x exp − t (5.17)
~ ~ 2m

where we do the integral in the usual fashion, by completing the square and using the
Gaussian definite integral.

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 296
 The Free Particle (cont.)

The probability density in the {|x i} basis is

P(x) = |hx |ψ(t) i|2

2 3
" «2 !#−1/2 ` p0 ´2
x− m t
„
~t
= 2 π σx2
6 7
1+ 4−
exp 6 (5.18)
„ 7
2 m σx2 “ ”2 « 5
t
2 σx2 1 + 2 m σ2
~
x

p0
Because the probability density is symmetric about x = m
t, it is easy to see that

p0 p0
hX it = t = hX it=0 + t (5.19)
m m

i.e., the particle’s effective position moves with speed p0 /m, which is what one
expects for a free particle with initial momentum p0 and mass m.

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 297
 The Free Particle (cont.)
The variance of the position is given by the denominator of the argument of the
Gaussian exponential (one could verify this by calculation of the necessary integral),
" „ «2 # " „ «2 #
2 ~t 2 ~t
h(∆X ) it = σx2 1+ = h(∆X ) it=0 1 + (5.20)
2 m σx2 2 m σx2

The position uncertainty grows with time because of the initial momentum uncertainty
of the particle – one can think of the {|p i} modes with p > p0 as propagating faster
than p0 /m and those with p < p0 propagating more slowly, so the initial wavefunction
spreads
q out over time. In the limit of large time (t  2 m σx2 /~), the uncertainty
h(∆X )2 it grows linearly with time. The “large time” condition can be rewritten in a
more intuitive form:

σx2 σx σx
t  t0 = 2 m =m = (5.21)
~ σp σv

where σv = σp /m is the velocity uncertainty derived from the momentum uncertainty.

So, t0 is just the time needed for the state with typical velocity to move the width of
the initial state. We should have expected this kind of condition because σx and ~ are
the only physical quantities in the problem. Such simple formulae can frequently be
used in quantum mechanics to get quick estimates of such physical phenomena; we
shall make such use in the particle in a box problem.

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 298
 The Free Particle (cont.)

Position-Momentum Uncertainty Relation

Before leaving the free particle, we note an interesting relationship that appeared
along the way. Recall that, because the position and momentum operators do not
commute, [X , P] = i ~, no state is an eigenstate of both. If there is no uncertainty in
one quantity because the system is in an eigenstate of it, then the uncertainty in the
other quantity is in fact infinite. For example, a perfect position eigenstate has a
delta-function position-space representation, but it then, by the alternative
representation of the delta function, Equation 3.146, we see that it is a linear
combination of all position eigenstates with equal weight. The momentum uncertainty
will be infinite. Conversely, if a state is a position eigenstate, then its position-space
representation has equal modulus everywhere and thus the position uncertainty will be
infinite.

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 299
 The Free Particle (cont.)

When we considered the Gaussian wave packet, which is neither an eigenstate of X

nor of P, we found that the t = 0 position and momentum uncertainties were

~2
h(∆X )2 it=0 = σx2 h(∆P)2 it=0 = σp2 = (5.22)
4 σx2

Hence, at t = 0, we have the uncertainty relation

q q
~
h(∆X )2 it=0 h(∆P)2 it=0 = (5.23)
2

We saw that, for t > 0, the position uncertainty grows while the momentum
uncertainty is unchanged, so in general we have
q q
~
h(∆X )2 i h(∆P)2 i ≥ (5.24)
2

We will later make a general proof of this uncertainty relationship between

noncommuting observables.

Section 5.1 Simple One-Dimensional Problems: The Free Particle Page 300
 The Particle in a Box

The Hamiltonian
A “box” consists of a region of vanishing potential energy surrounded by a region of
infinite potential energy:
 L
0 |x| ≤ 2
V (x) = lim L (5.25)
V0 →∞ V0 |x| > 2

It is necessarily to include the limiting procedure so that we can make mathematical

sense of the infinite value of the potential when we write the Hamiltonian. Classically,
such a potential completely confines a particle to the region |x| ≤ L/2. We shall find a
similar result in quantum mechanics, though we need a bit more care in proving it.

The classical Hamiltonian is

p2
H(x, p) = + V (x) (5.26)
2m

Section 5.2 Simple One-Dimensional Problems: The Particle in a Box Page 301
 The Particle in a Box (cont.)
Postulate 2 tells us that the quantum Hamiltonian operator is
P2
H(X , P) = + V (X ) (5.27)
2m

Next, we want to obtain an eigenvalue-eigenvector equation for H. For the free

particle, when V (X ) was not present, it was obvious we should work in the {|p i}
basis because H was diagonal there, and then it was obvious how P acted in that basis
and we could write down the eigenvalues and eigenvectors of H trivially. We cannot
do that here because V (X ) and hence H is not diagonal in the {|p i} basis. Moreover,
regardless of basis, we are faced with the problem of how to interpret V (X ). Our
usual power-series interpretation fails because the expansion is simply not defined for
such a function – its value and derivatives all become infinite for |x| ≥ L/2.

Shankar glosses over this issue and jumps to the final differential equation; thereby
ignoring the confusing part of the problem! We belabor it to make sure it is clear how
to get to the differential equation from H and the postulates. The only sensible way
we have to deal with the above is to write down matrix elements of H in the {|x i}
basis because our Postulate 2 tells us explicitly what the matrix elements of X are in
this basis. Doing that, we have

P2
hx |H(X , P)|x 0 i = hx | |x 0 i + hx |V (X ) |x 0 i (5.28)
2m

Section 5.2 Simple One-Dimensional Problems: The Particle in a Box Page 302
 The Particle in a Box (cont.)

Let’s look at each term separately. For the first term, since it is quadratic in P, let’s
insert completeness to get the P’s separated:

P2 1
Z ∞
hx | |x 0 i = dx 00 hx |P |x 00 ihx 00 |P|x 0 i
2m 2m −∞
Z ∞
~2
» –» –
d d
=− dx 00
δ(x − x 00 ) δ(x 00
− x 0
)
2 m −∞ dx dx 00
2 Z ∞ » –
~ d d
=− dx 00 δ(x − x 00 ) 00
δ(x 00 − x 0 )
2 m dx −∞ dx
~2 d
» –
d
=− δ(x − x 0 )
2 m dx dx
~2 d2
=− δ(x − x 0 ) (5.29)
2m d(x 0 )2

where in last step we used Equation 3.127,

dn dn
δ(x − x 0 ) = δ(x − x 0 )
dx n dx 0 n

Section 5.2 Simple One-Dimensional Problems: The Particle in a Box Page 303
 The Particle in a Box (cont.)

For the second term, we can approach it using a limiting procedure. Suppose V (X )
were not P
so pathological; suppose it has a convergent power series expansion
V (X ) = ∞ k
k=0 Vk X . Then, we would have

∞ ∞
X X ` ´k
hx |V (X ) |x 0 i = Vk hx | X k |x 0 i = Vk x 0 hx |x 0 i
k=0 k=0
∞
X ` ´k
= Vk x 0 δ(x − x 0 ) = δ(x − x 0 ) V (x 0 )
k=0

where we have allowed X to act to the right on |x 0 i. This is not a strict application
of Postulate 2; if one wants to be really rigorous about it, one ought to insert
completeness relations like we did for P 2 . For example, for X 2 we would have
Z ∞ Z ∞
hx |X 2 |x 0 i = dx 00 hx |X |x 00 ihx 00 |X |x 0 i = dx 00 x δ(x − x 00 ) x 00 δ(x 00 − x 0 )
−∞ −∞

= x δ(x − x 0 ) = (x 0 )2 δ(x − x 0 )
2

For X k , we have to insert k − 1 completeness relations and do k − 1 integrals. The

result will be of the same form.

Section 5.2 Simple One-Dimensional Problems: The Particle in a Box Page 304
 The Particle in a Box (cont.)

The key point in the above is that we have figured out how to convert the operator
function V (X ) into a simple numerical function V (x) when V (X ) can be expanded as
a power series. To apply this to our non-analytic V (X ), we could come up with an
analytic approximation that converges to the non-analytic one as we take some limit.
(One could use a sum of tan−1 or tanh functions, for example.) The point is that if
we used the expansion and then took the limit, we would obtain a result identical to
the above. So we write

hx |V (X ) |x 0 i = δ(x − x 0 ) V (x 0 ) (5.30)

With the above results, we have that the matrix elements of H are given by:

~2 d2
» –
hx |H |x 0 i = δ(x − x 0 ) − + V (x 0
) (5.31)
2 m d(x 0 )2

Section 5.2 Simple One-Dimensional Problems: The Particle in a Box Page 305
 The Particle in a Box (cont.)

Thus, for an arbitrary state |f i, we have that (using completeness as usual)

Z ∞
hx |H |f i = dx 0 hx |H |x 0 ihx 0 |f i
−∞
Z ∞
~2 d2
» –
= dx 0 δ(x − x 0 ) − 0 2
+ V (x 0 ) f (x 0 )
−∞ 2 m d(x )
Z ∞
~2 d 2 f (x 0 )
» –
= dx 0 δ(x − x 0 ) − 0 2
+ V (x 0 ) f (x 0 )
−∞ 2 m d(x )
~2 d 2 f (x)
=− + V (x) f (x) (5.32)
2 m dx 2

Section 5.2 Simple One-Dimensional Problems: The Particle in a Box Page 306