n
(t)| Q|
n
(t) = e
iE
n
t/h
e
iE
n
t/h
n
| Q|
n
=
n
| Q|
n
, (5)
independent of time. Of course, the most general solution to (1) is a linear
combination of specic solutions
|(t) =
n
c
n
e
iE
n
t/h
|
n
, (6)
where the expansion coecients c
n
are determined by the initial state of the
system.
In general, the expectation value
Q(t) (t)| Q|(t) , (7)
1
of an observable Q is time-dependent: From (1) and (2) follows that
d
dt
Q(t) =
i
h
(t)| [Q, H(t)] |(t) . (8)
If Q does not commute with the Hamiltonian, then in general Q(t) is not
conserved.
Example
Find the time-dependence of S(t) for a spin S in a constant magnetic eld
B.
We choose the z axis to be in the direction of B. The Hamiltonian reads
H = B S = BS
z
. (9)
The spin operators obey the usual commutation relations,
[S
x
, S
y
] = iS
z
, [S
y
, S
z
] = iS
x
, [S
z
, S
x
] = iS
y
. (10)
Substituted into (8), this yields
d
dt
S
x
(t) = BS
y
,
d
dt
S
y
(t) = BS
x
,
d
dt
S
z
(t) = 0. (11)
The solution is a vector
S(t) = S
[cos(t + ) x + sin(t + ) y] + S
z, (12)
that precesses around the z-axis with an angular frequency B.
Time-dependent Hamiltonians
As an example of a system with a time dependent Hamiltonian, we consider
a spin-1/2 in a precessing magnetic eld
B(t) = b [cos(t) x cos(t) x] + B z (13)
For a spin-1/2, the spin operator S can be expressed in terms of the standard
Pauli matrices
x
=
0 1
1 0
,
y
=
0 i
i 0
,
z
=
1 0
0 1
, (14)
2
as S = h(
x
x +
y
x +
z
z)/2. The Hamiltonian is
H(t) = B(t) S =
h
2
B be
it
be
it
B
. (15)
Although we will solve the problem exactly and in general, we will be
particularly interested in the situation where B b, i.e. where the time-
dependent part of H(t) is a small perturbation to the time-independent part
H
0
=
hB
2
z
=
hB
2
1 0
0 1
, (16)
We will answer the following question: suppose the system starts out in the
ground state
0
=
1
0
, (17)
of the unperturbed system H
0
at time t = 0, what is the probability to nd
the system in the excited state
1
=
0
1
, (18)
at a later time t? The dierence in energy between these two states is E =
hB, and this energy must be supplied by the time-dependent part of the
Hamiltonian if the transition is to occur.
Our task is to solve the time-dependent Schrodinger equation i hd/dt =
H(t)(t) with the initial condition (0) =
0
. This is most conveniently
done by transforming to two functions u(t) and v(t), dened through
(t) =
u(t)e
it/2
v(t)e
it/2
. (19)
Substitution into the time-dependent Schrodinger equation leads to a dier-
ential equation for u(t) and v(t), namely
d
dt
u(t)
v(t)
=
i
2
Q
u(t)
v(t)
where Q =
B b
b B
. (20)
We note that Q
2
=
2
I where
=
(B )
2
+ (b)
2
. (21)
It is straight-forward to check that the general solution to (20) is
u(t)
v(t)
I cos(t/2) + i
Q
sin(t/2)
u(0)
v(0)
. (22)
3
Figure 1: Left: Probability P(t) to nd the system in the excited state.
Right: Maximum of P(t) as a function of .
Proof: Consider the time-derivative of the operator on the right-hand side,
keeping in mind that I = Q
2
/:
d
dt
I cos(t/2) + i
Q
sin(t/2)
i
Q
2
cos(t/2)
I
2
sin(t/2)
=
i
2
Q
cos(t/2) + i
Q
sin(t/2)
. (23)
The initial condition (t) =
0
corresponds to u(0) = 1 and v(0) = 0, so
that
u(t) = cos(t/2) + i
B
sin(t/2), (24)
v(t) = i
b
sin(t/2). (25)
The probability P(t) to nd the system in the excited state
1
at time t
equals
P(t) = |v(t)|
2
=
(b)
2
(B )
2
+ (b)
2
sin
2
(t/2). (26)
It oscillates with a frequency /2 between 0 and a maximum value
P
=
(b)
2
(B )
2
+ (b)
2
. (27)
These oscillations are known as Rabi oscillations. As a function of , P
) h/2 at a frequency
/2. When the time-dependent perturbation is small (b B) and B,
the oscillation is all the way from h/2 to h/2, at an angular frequency
= b that is much lower than the spin precession frequency B.
5