1 Harmonic Oscillator in Two Dimensions

The harmonic oscillator in two dimensions is an instructive problem for sev-

eral reasons. First, it illustrates how certain kinds of problems with more
than one degree of freedom may be handled. Second, it can be solved in both
Cartesian and cylindrical coordinates under certain circumstances. Third, it
introduces the concept of angular momentum in simplied form. This serves
a preparation for considering angular momentum in three dimensions, which
is a much more complicated topic.

1.1 Solution in Cartesian Coordinates

In Cartesian coordinates x and y, the scaled Hamiltonian has the form
H = (p2 + p2 + x2 + 2y2)=2;
x y
(1)
where we have retained the option of giving the x and y spring constants
dierent values by setting the constant  to a value other than unity.
In position representation the time-independent Schrodinger equation is
!
@ 2 + @ 2 + (x2 + 2y2) = 2E ; (2)
@x2 @y2
where E is the energy and (x; y) is the wavefunction. Solution of this
equation by separation of variables is possible. Setting (x; y) = X (x)Y (y)
in the above equation and dividing by , we nd
! !
1 d2X + x2 + 1 d2Y + 2 y2 = 2E: (3)
X dx2 Y dy2
Since the rst major term is a function only of x and the second is a function
only of y, these terms individually must be equal to constants which we dene
to be 2E and 2E . This results in two equations,
x y

d2X + x2X = 2E X (4)

dx2 x

and
d2Y + 2y2Y = 2E Y; (5)
dy2 y

1
where it is clear that E + E = E .
x y

We recognize equations (4) and (5) as individual harmonic oscillator equa-

tions with respective eigenvalues
E = 1=2; 3=2; 5=2; : : :
x E = =2; 3=2; 5=2; : : : :
y (6)
The total energy is
E = i + j + (1 + )=2;
ij i = 0; 1; 2; : : : ; j = 0; 1; 2; : : : ; (7)
and the associated eigenvectors for the full problem are
ij (x; y) = X (x)Y (y);
i j (8)
where X (x) is the ith eigenvector of equation (4) and Y (y) is the j th eigen-
i j

vector of equation (5).

Table 1 shows the rst 15 energies for the two-dimensional harmonic
oscillator as computed from equation (7) and sorted into order of increasing
energy. Notice that degeneracy, i. e., repeated energy values, occurs for
 = 1, but not for  = 0:7. The existence of degeneracy in the energy means
that some other operator must commute with the Hamiltonian. The nature
of this operator is no mystery | the angular momentum of the system is
conserved when  = 1, but not for other values of . Thus, it seems plausible
that the repeated energy values in table 1 represent states (or superpositions
of states) with denite values of the angular momentum.

1.2 Solution in Polar Coordinates

Angular momentum is best understood in polar coordinates for two dimen-
sional problems. We let r2 = x2 + y2 and  = arctan(y=x), and transform
equation (2) with  = 1 to polar coordinates:
@2 1 @ 1 @ 2 + r 2 = 2E : (9)
@r2 r @r r2 @2
In polar coordinates the kinetic energy may be written (for unit mass)
K = (p2 + L2=r2 )=2;
r
(10)

2
  = 0:7 =1
i j energy i j energy
0 0 0.85 0 0 1.00
0 1 1.55 1 0 2.00
1 0 1.85 0 1 2.00
0 2 2.25 1 1 3.00
1 1 2.55 2 0 3.00
2 0 2.85 0 2 3.00
0 3 2.95 1 2 4.00
1 2 3.25 2 1 4.00
2 1 3.55 0 3 4.00
0 4 3.65 3 0 4.00
3 0 3.85 2 2 5.00
1 3 3.95 0 4 5.00
2 2 4.25 3 1 5.00
0 5 4.35 4 0 5.00
3 1 4.55 1 3 5.00

Table 1: Energies for the two-dimensional harmonic oscillator in terms of

the quantum numbers i and j . The rst three columns show the results for
 = 0:7 while the last three columns show  = 1.

3
where p is the radial component of the momentum and L is the angular mo-
r

mentum about the axis normal to the x-y plane. Comparison with equation
(9) suggests that
p2 = @r @2 1 @ (11)
r 2 r @r
and
L2 = @ @2 : (12)
2
One might naively expect the radial momentum squared operator to take the
form p2 = @ 2 ( )=@r2, but the complexities of polar coordinates introduce
r

the additional term. Note that angular momentum has the same units as
Planck's constant, so angular momentum is dimensionless in a system of
units in which h = 1.
Equation (9) may be solved by separation of variables. Setting (r; ) =
R(r)(), substituting into the above equation, and dividing by , we nd
1 d2R 1 dR 1 d2 + r2 = 2E: (13)
R dr2 rR dr r2 d2
The combination  1@ 2=@2 must be equal to a constant, which we call
m2 for reasons that will soon be evident. The variables R and  therefore
obey the separate equations
1 d2R 1 dR + m2 + r2 = 2E (14)
R dr2 rR dr r2
and
d2 + m2 = 0: (15)
d2
The  equation has obvious solutions  = exp(im). Since we need to
have these solutions single valued in , the parameter m must take on integer
values only. The reason for choosing a negative separation constant m2 is
now clear; it results in periodic solutions to the  equation that make this
single valuedness possible.
Solution of the R equation is more dicult. It turns out that equation (14)
is actually quite similar to the one-dimensional harmonic oscillator equation,
in that it has solutions of the form R(r) = P (r) exp( r2=2) where P (r) is a
polynomial. The function P (r) obeys
d2P + (1=r 2r) dP (m2=r2 + 2 2E )P = 0: (16)
dr2 dr
4
 4
P = r^3 P = r^3 - r/2 P = r^3 - r/2 P = r^3

3
P = r^2 P = r^2 - 1 P = r^2

2
P=r P=r
E
1
P=1

0
-3 -2 -1 0 1 2 3
m

Figure 1: Eigenstates of the two-dimensional harmonic oscillator.

5
The rst four energy levels may be obtained by assuming
P (r) = r3 + ar2 + br + c: (17)
Solutions of this form only exist when E and m take on various integer values.
Figure 1 show the actual values of E and m that solve equation (16), along
with the corresponding polynomials.
From these few low order solutions the pattern is clear. One solution
exists for E = 1, two solutions for E = 2, etc. This is in agreement with
the results shown in table 1. The allowed values of angular momentum for
each energy are E + 1; E + 3; : : : ; E 1, which means that there are E
dierent allowed angular momentum values for each energy.

1.3 Comparison of Solutions

Comparison of the solutions to the two-dimensional harmonic oscillator in
Cartesian and polar coordinates is enlightening. Let us rst determine whether
these solutions are indeed the same, or at least consistent.
For the lowest energy state it was shown in the section on the ono-
dimensional harmonic oscillator that the unnormalized eigenfunction is simply
exp( x2=2). Thus, the two-dimensional Cartesian solution is exp( r2=2),
which is in agreement with the lowest energy solution in polar coordinates.
For the rst excited state, table 1 shows that we either have i = 1 and
j = 0 or i = 0 and j = 1. Since the rst excited eigenfunction of the one-
dimensional harmonic oscillator is x exp( x2=2), the full eigenfunctions rst
excited state of the two-dimensional oscillator from the Cartesian coordin-
ate system analysis are 1 0(x; y) = r cos() exp( r2=2) and 0 1(x; y) =
; ;

r sin() exp( r2=2). However, from the analysis in polar coordinates, the
rst excited state eigenfunctions are 1 1(r; ) = r exp( r2=2 i) and
;

1 1(r; ) = r exp( r =2+ i), where the subscripts of (r; ) indicate respect-
2
;

ively the energy and angular momentum quantum numbers. Comparison of

these two sets of solutions shows that
1; 1 (r; ) = 1;0 (x; y) i 0;1 (x; y) (18)
and
1;1 (r; ) = 1;0 (x; y) + i (x; y):
0;1 (19)

6
Thus, the Cartesian coordinate eigenfunctions represent standing wave solu-
tions in , while the polar coordinate eigenfunctions represent travelling wave
solutions. Similar arguments can be made about solutions of higher order.
The above equations show that travelling waves can be represented in terms
of standing waves in quantum mechanics. Since these equations are easily
inverted, the reverse is true as well.
The angular momentum equation (15) by itself allows the angular mo-
mentum to take on any integer value. However, constraints imposed by the
radial equation (14) further limit the possible values of angular momentum.
Why is this? Classically, motion in an attractive axially symmetric potential
has a maximum possible angular momentum for a given total energy. The
limitation jmj  E in the quantum mechanical two-dimensional oscillator is
thus no surprise.
A more interesting restriction is that when E = 2 there is no m = 0 state,
when E = 3 there are no states with m = 1, etc. The answer to this lies in
the parity operator P , which in two dimensions has the following properties:
P (x; y) = ( x; y): (20)
Thus, for instance, from the Cartesian coordinate system analysis, the rst
excited state has negative parity since P 1 0(x; y) = 1 0(x; y). However, all
; ;

m = 0 states must have positive parity from the polar coordinate analysis,
since they are axially symmetric. Thus, a rst excited state with m = 0
cannot exist. Similar arguments can be made about missing m values in
higher order states. In particular, states with even values of the energy
quantum number n must have even values of m and vice versa.

1.4 General Comments

The dierence between the  = 0:7 and the  = 1 cases is a fundamental
one that is not limited to quantum mechanics. Figure 2 shows classical
particle trajectories for these two cases in x-y space. Note that for  = 1
the particle makes a repeated trajectory which is just a circle. (Depending
on the initial conditions, other trajectories are possible, but they are still
repeated.) For  = 0:7 the trajectory is much more complex, and though
it will eventually repeat, the period for repetition being much longer than
the harmonic oscillator frequencies in either the x or y direction. In general,

7
 lambda = 1 lambda = 0.7
1 1

0 0
y y

-1 -1
-1 0 1 -1 0 1
x x

Figure 2: Trajectory of classical two-dimensional harmonic oscillator in x-y

plane for two dierent values of .

the trajectory will repeat if  is a rational number, i. e., a ratio of integers,

with the repetition period being longer as the integers become bigger. The
trajectory does not repeat if  is an irrational number. Analogously, there
is occasional energy degeneracy in the quantum mechanical problem if  is
rational, but no degeneracy at all if  is irrational. This follows directly from
equation (7).
The two-dimensional harmonic oscillator is a prototype for problems with
more than one degree of freedom in quantum mechanics. By chance we can
solve this particular problem exactly in Cartesian coordinates for all values of
. However, we can easily solve it in polar coordinates only for  = 1. This
we can do because the Hamiltonian has many degenerate eigenvalues when
 = 1, which means that there exists another operator, namely the angular
momentum operator, which commutes with the Hamiltonian. In this case the
computation of the eigenvalues and the eigenvectors in the two-dimensional
problem reduces to the computation of eigenvalues and eigenvectors in two
separate one-dimensional problems. The computational eort is much less
than is needed to solve the two-dimensional problem directly.
Generally speaking, problems in quantum mechanics with more than one
degree of freedom break down into two types, those that factor into a set
8
of smaller problems, as in the  = 1 case above, and those that don't. Let
us call the former type of problem an easy problem and the latter type a
dicult problem. Dicult problems that dier only slightly from an easy
problem are often solved approximately by perturbation theory. It is not an
exaggeration to characterize progress in quantum mechanics as consisting of
exact solutions to easy problems complemented by approximate perturbation
solutions to dicult problems.