Physics 342 Lecture 5

Schrödinger’s Equation
Lecture 5

Physics 342
Quantum Mechanics I

Wednesday, February 3rd, 2010

Today we discuss Schrödinger’s equation and show that it supports the basic
interpretation of the fundamental object of study in quantum mechanics:
the wave function. The operational procedures of quantum mechanics begin
with this PDE, and its associated boundary and initial conditions. So while
it is easy to state the basic axioms, and even motivate the expressions, the
shift in focus is dramatic.
In classical mechanics, our goal is always to find or describe the salient
features of a curve x(t), the trajectory of a particle under some impressed
force. The relation of force to curve is provided by Newton’s second law:
F = m ẍ(t), and the physically inspired initial (or boundary) conditions.
The job is straightforward, if complicated at times. The interpretation is
simple: The particle moves along the curve.
In quantum mechanics, our initial goal will be to find the probability density
|Ψ(x, t)|2 for a particular system, from which we can calculate expectation
values (the outcomes of experiment). The relation of a classical potential
to the density (or equivalently, Ψ(x, t), the wave function) is provided by
Schrödinger’s equation.
There are a few obvious questions that come up as we exploit the probabilis-
tic interpretation of the wave function. We know that if we have a probabil-
ity density ρ(x, t), we canRcompute the outcome of experiments that rely on
∞
position, since hf (x)i = −∞ f (x) ρ(x, t) dx. What, then, should we make
of experiments that measure velocity (or momentum)? A related question
is why, given that it is ρ(x, t) that we want, does Schrödinger’s equation
govern Ψ(x, t)? Shouldn’t we have a PDE that tells us about the evolution
of ρ(x, t) itself?

1 of 9
5.1. WITHOUT FURTHER ADO Lecture 5

5.1 Without Further Ado

Here it is – for a complex field Ψ(x, t), Schrödinger’s equation says that the
dynamical evolution of Ψ is given by (in one spatial dimension):

∂Ψ ~2 ∂ 2 Ψ
i~ =− + V (x) Ψ (5.1)
∂t 2 m ∂x2
for a potential V (x). This is the quantum mechanical analogue of the clas-
sical Newton’s second law: the dynamical evolution of a particle is given by
its trajectory x(t) governed by mẍ = −∇V .
Our statistical interpretation is provided by viewing Ψ(x, t)∗ Ψ(x, t) as a
density: ρ(x, t) = |Ψ(x, t)|2 so that
Z b Z b
P (a, b) = ρ(x, t) dx = |Ψ(x, t)|2 dx (5.2)
a a

is the probability a particle is found between x = a and x = b. We are talking

about the probability density associated with a particle, that explains the
m in Schrödinger’s equation. The constant ~ tells us this is not a classical
system. If we turn off ~, we get Ψ = 0, which seems reasonable – classically,
the probability density would be given by: ρ(x, t)CM = δ(x − x(t)), so that
with probability 1, the particle is on its classical trajectory x(t). Without
some notion of x(t), there is no probability density at all, so Ψ = 0 is as
good a solution as any. Of course, ~ is doing more than setting units, but
we will see that later on.
No statistical interpretation exists in Schrödinger’s
R ∞ equation alone – for ex-
ample, the equation itself does not demand that −∞ |Ψ(x, t)|2 dx = 1 for all
time. But, it does support the statistical interpretation R ∞in the sense that if
we solved the above, and required that, at time t = 0, −∞ |Ψ(x, 0)|2 dx = 1,
R∞
then (as we’ll show), −∞ |Ψ(x, t)|2 dx = 1 for all time – so once a probability
density, always a probability density.
R∞
We can see this by taking the time derivative of −∞ |Ψ(x, t)|2 dx, and show-
ing that, as long as Ψ(x, t) satisfies (5.1), the time derivative is zero, so that
the normalization remains intact. It is, then, up to us to ensure that Ψ(x, t)
is appropriately normalized at some time. Take:
Z ∞ Z ∞
d ∂
2
|Ψ(x, t)| dx = (Ψ∗ (x, t) Ψ(x, t)) dx, (5.3)
dt −∞ −∞ ∂t

2 of 9
5.1. WITHOUT FURTHER ADO Lecture 5

then we just need the temporal derivative of the norm (squared) of Ψ(x, t):
∂ ∂Ψ∗ ∂Ψ
(Ψ∗ (x, t) Ψ(x, t)) = Ψ + Ψ∗ , (5.4)
∂t ∂t ∂t
which is fine, but this is under an integral w.r.t. x, so we’d like a mechanism
for turning temporal derivatives into spatial ones (preparatory to integra-
tion by parts). Schrödinger’s equation, and its complex conjugate provide
precisely the desired connection:
∂Ψ i ~ ∂2Ψ i
= 2
− V (x) Ψ
∂t 2 m ∂x ~ (5.5)
∂Ψ∗ i ~ ∂ 2 Ψ∗ i
=− 2
+ V (x) Ψ∗ ,
∂t 2 m ∂x ~
and we can write the above as:
 2 ∗ 2

∂ ∗ i~ ∂ Ψ ∗ ∂ Ψ
(Ψ (x, t) Ψ(x, t)) = − Ψ+Ψ . (5.6)
∂t 2m ∂x2 ∂x2
Now, under the integral, we can use integration by parts1 to simplify the
above – consider the first term:
Z ∞ 2 ∗  ∗ ∞ Z ∞
∂ Ψ ∂Ψ ∂Ψ∗ ∂Ψ
2
Ψ dx = Ψ − dx. (5.9)
−∞ ∂x ∂x −∞ −∞ ∂x ∂x

We can get rid of the boundary term by assuming (an additional require-
ment) that Ψ(x, t) −→ 0 as |x| −→ ∞. This is a reasonable assumption,
and a common one when dealing with densities (charge distributions do not
extend to infinity except in artificial cases, for example). Then, going back
to the original time derivative of the density:
Z ∞ Z ∞  2 ∗ 2

d ∗ i~ ∂ Ψ ∗ ∂ Ψ
(Ψ (x, t) Ψ(x, t)) = − Ψ+Ψ dx
dt −∞ 2 m −∞ ∂x2 ∂x2
Z ∞ ∗
∂Ψ ∂Ψ ∂Ψ∗ ∂Ψ

i~ (5.10)
= − dx
2 m −∞ ∂x ∂x ∂x ∂x
= 0,
1
For two functions u(x) and v(x), we have
Z b
d
(u(x) v(x)) dx = u(x) v(x)|ba , (5.7)
a dx

and using the product rule on the left gives:

Z b» –
du(x) dv(x)
v(x) + u(x) dx = u(x) v(x)|ba , (5.8)
a dx dx
from which the “usual” form of integration by parts follows.

3 of 9
5.1. WITHOUT FURTHER ADO Lecture 5

and we have the result: The time derivative of the probability that a particle
exists somewhere spatially is zero, hence we can normalize Ψ(x, t) so that
the probability the particle exists somewhere is = 1, and it will remain 1 for
all time.
Notice that the statement:
Z
d
ρ(x, t) dx = 0 (5.11)
dt
is reminiscent of charge conservation. This is a “probability” conservation
sentiment. By analogy with charge, though, we should be able to find a
“probability current” associated with the conservation. This must be an
object whose x-derivative is equal to the right-hand-side of (5.10) (the top
line, before we integrate by parts) – easy to find in this case. We want:
 2 ∗ 2

d i~ ∂ Ψ ∗ ∂ Ψ
J(x, t) = − − Ψ+Ψ , (5.12)
dx 2m ∂x2 ∂x2

(the minus sign is convention, and we will use it in a moment to make contact
with other conservation laws) and it is pretty clear that:

∂Ψ ∂Ψ∗
 
i~
J(x, t) = − Ψ∗ − Ψ (5.13)
2m ∂x ∂x

has the desired property.

Interesting – then we have another view of the conservation statement – we
would write it, in one dimension, as
Z ∞ Z ∞
d d
ρ(x, t) dx = − J(x, t) dx. (5.14)
dt −∞ −∞ dx

Here, the right-hand side is clearly zero by the fundamental theorem of

calculus, and the requirement that Ψ vanish at spatial infinity. We have
basically rewritten integration by parts, but this rewrite allows us to make
a “differential” continuity statement:
∂ρ ∂J
=− . (5.15)
∂t ∂x
In three dimensions, this would read;
∂ρ i~
= −∇ · J J=− [Ψ∗ ∇Ψ − Ψ ∇Ψ∗ ] . (5.16)
∂t 2m

4 of 9
5.2. EXPECTATION VALUES Lecture 5

Now we see the analogy: In E&M, the rate of change of charge in Ra volume
is determined by the charge flowing
H in through a surface. We have V ρ dτ as
the charge in the volume, and J · da is the surface integral for the current
– then the statement is
Z I
∂ρ d
= −∇ · J −→ ρ dτ = − J · da, (5.17)
∂t dt V ∂V
where we get the integral form by integrating both sides of the differential
one and applying the divergence theorem. The same conservation law holds
for mass flow in a fluid. Evidently, in quantum mechanics, the local proba-
bility of finding a particle in a volume can change over time according to a
probability flux that can be used to account for probability “flowing” into
and out of the volume through its boundary.

5.2 Expectation Values

Let’s take stock: We have a PDE that governs the temporal evolution of a
“wavefunction” Ψ(x, t) given a classical potential V (x), Schrödinger’s equa-
tion. This PDE supports a probabilistic interpretation for the absolute value
(squared) of Ψ(x, t), i.e. “|Ψ|2 is the probability density for finding the par-
ticle at location x at time t.” provided Ψ(x, t) vanishes at spatial infinity.
That gives us a boundary condition for Ψ(x, t), but we are missing the initial
condition – what is the spatial function Ψ(x, t = 0)?R We’ll come back to this,
∞
but one thing is clear, it must have the property: −∞ |Ψ(x, t = 0)|2 dx = 1
for the probabilistic interpretation to hold. We have, so far, a complex
Ψ(x, t) to be found subject to:

∂Ψ ~2 ∂ 2 Ψ
i~ =− + V (x) Ψ
∂t 2 m ∂x2
Ψ(x → ±∞, t) = 0 (5.18)
Z ∞
|Ψ(x, 0)|2 dx = 1.
−∞

If we had the wavefunction, all manner of interesting physical properties

could be calculated. We can start with the simplest – the expectation value
of position: hxi. From the density, we know that
Z ∞ Z ∞
hxi = x ρ(x, t) dx = Ψ∗ x Ψ dx (5.19)
−∞ −∞

5 of 9
5.2. EXPECTATION VALUES Lecture 5

which is reasonable. We have put x in between Ψ and its complex conjugate,

that doesn’t matter here, just notation.
One can define the expectation of “velocity” to be the time-derivative of hxi,
i.e. if we know how the expected position changes in time, we might call that
the “expected” velocity. Since expectation values proceed by integration
over all positions, the notion of time-dependence for x itself is not particu-
larly useful. The objects with time-dependence are expectation values, and
these inherit their time-dependence directly from the wavefunction’s, so we
are sort of forced into viewing particle “velocity” as the time-dependence of
the particle’s expected position. At that point, we have
Z ∞
dhxi ∂
= (ρ(x, t) x) dx
dt −∞ ∂t
Z ∞
∂ρ
= x dx
−∞ ∂t
Z ∞ (5.20)
∂J
=− x dx
−∞ ∂x
Z ∞
= J dx,
−∞
∂ρ
where we have used the local form of conservation: ∂t = − ∂J
∂x , and integra-
tion by parts.
This makes some sense when we think of the current density from E&M,
or the fluid flux density in hydrodynamics – in either case, with appropri-
ate interpretation of ρ, we have J = ρ v – in our present setting, the ρ is
probability density, and the v then represents some sort of velocity with
which theR probability is flowing – the natural expectation value here would
∞
be hvi = −∞ ρ v dτ in one dimension. But that’s what the last line of (5.20)
is. There is a sleight-of-hand going on here, we just associated v with prob-
ability flux, and yet, we’d like to understand the expectation value of v in
terms of a particle’s velocity – don’t take the above argument as anything
more than motivation at this point.
Pressing on, we can write the integral itself in a more illuminating form
using the definition of J from (5.13):
Z ∞ Z ∞     
i~ ∂ ∗ ∗ i~ ∂
hvi = J dx = Ψ Ψ −Ψ Ψ dx
−∞ −∞ 2m ∂x 2m ∂x
Z ∞   (5.21)
∗ i~ ∂
=− Ψ Ψdx
−∞ m ∂x

6 of 9
5.2. EXPECTATION VALUES Lecture 5

where we used integration by parts to combine the two terms. In the end,
we can move the mass over to the right, and define the expectation value of
momentum (of the particle, now):
Z ∞  
~ ∂
hpi = hm vi = Ψ∗ Ψ dx. (5.22)
−∞ i ∂x

Looking at the expectation value of position again,

Z ∞
hxi = Ψ∗ x Ψ dx (5.23)
−∞

we are tempted to define “operators” x and p (whose expectation values are

given by the above) via:

~ ∂
x∼x p∼ . (5.24)
i ∂x
These are operators in the sense (obvious for p) that they act on the wave
function.
This seems like a strange move, and it is a literal translation of what the
expectation values are telling us, so there’s no reason to believe any of it
just yet – we will let the correctness of the predictions of this theory do
the convincing. For now it is the case that with a sensible definition of
expectation value, and the peculiar factoring of probability density provided
by the wave function, we are led to the above association.
With this in hand, we can define any classical quantity’s expectation value
∂
by replacing x ∼ x and p ∼ ~i ∂x , and sandwhiching it between the wave
function and its complex conjugate. For a classical function Q(x, p), we have
Z ∞  
∗ ~ ∂
hQi = Ψ Q x, Ψ dx. (5.25)
−∞ i ∂x
Things are beginning to look suspiciously like hα| Q |αi with an inner prod-
uct defined by integration, which is no accident. What is also no acci-
dent is the fact that for the classical Hamiltonian, a function of x and p:
2
H = 2pm + V (x), our prescription for the expectation value is:
Z ∞
~2 ∂ 2
 
∗
hHi = Ψ − + V (x) Ψ dx, (5.26)
−∞ 2 m ∂x2
and if you cover up the integral and Ψ∗ , what you have is one side of
Schrödinger’s equation.

7 of 9
5.2. EXPECTATION VALUES Lecture 5

Homework

Reading: Griffiths, pp. 12–20.

Problem 5.1

Griffiths 1.4. In part c., you are asked where the particle is most likely to
be found – technically, what we want is the maximum of ρ(x) here – that
is, the location of the particle with the largest ρ(x) dx.

Problem 5.2

Griffiths 1.7. This problem is about (an example of) Ehrenfest’s theorem
– it provides good practice with integration by parts, in addition to the
computational utility of the result.

Problem 5.3

Working from the separated form of Schrödinger’s equation: For Ψ(x, t) =

ψ(x) φ(t), we had
~2 d2 ψ(x)
− + V (x) ψ(x) = E ψ(x)
2 m dx2 (5.27)
iEt
φ(t) = e− ~ .

Consider a physical configuration in which we have V (x) = 0 and require

that ψ(0) = ψ(d) = 0. This represents a particle confined to a one-
dimensional “box” of width d. The wavefunction (and hence the proba-
bility density) is zero outside the interval, and continuity requires that the
wavefunction must then vanish at the endpoints of the interval.

a. Solve the upper equation for ψ(x), with boundary conditions. This
should look familiar by now, and there will be an integer n that indexes
your solutions. Write the “energy” E as a function of n, and form the full
wavefunction: Ψn (x, t). This should have an undetermined overall constant
in it.

b. Suppose we have a wavefunction Ψ(x, t) = Ψn (x, t) – in order to

interpret ρ(x, t) = Ψ∗ (x, t) Ψ(x, t) as a probability density, we must have

8 of 9
5.2. EXPECTATION VALUES Lecture 5

R∞
−∞ ρ(x, t) dx = 1. Fix your overall constant in Ψn (x, t) by normalizing
the wavefunction.

c. Again, using Ψ(x, t) = Ψn (x, t), calculate:

hxi and hpi. (5.28)

9 of 9