Recitation Class No.

7 in Quantum Mechanics 2 (77605)

Yuval Vinkler
The Racah Institute of Physics, The Hebrew University, Jerusalem, 91904, Israel

November 27, 2013

In class you solved the problem of the hydrogen-like atom
H=

p~2
Ze2

,
2
r

(1)

by writing the appropriate differential eigenproblem and solving it to get

H|n, l, mi = En |n, l, mi,

(2)

with

Z 2 e4
,
(3)
2~2 n2
and the quantum numbers are n = 0, 1, . . ., 0 l (n1) and l m l. This gives a degeneracy of n2
to each energy level. The Hamiltonian is obviously invariant under rotations, and this SO(3) symmetry
is known to generate degeneracy of 2l + 1 in the energies. The fact that the degeneracy of the hydrogen
atom is larger points to some hidden symmetry. In this recitation class we will find that symmetry, and
using it will construct an algebraic solution to the spectrum of the hydrogen atom. This was first done
by Pauli in 1926 and it was put in context of group theory formalism by Bargmann in 1936.
En =

The LRL Vector

We begin by looking at the classic Kepler problem, described by the classic Hamiltonian
H=

p~2
c
,
2 r

(4)

with c = Ze2 for the hydrogen-like problem and c = GM for gravitational problems. The known
classical bound solution depicts an elliptical orbit, which is fixed (it does not precess.) In central-power
problems there is a symmetry for rotations in 3D, which means that the angular momentum is a conserved
~ = ~r p~ is a constant vector, and the orbit is bound to the plane perpendicular to L,
~
quantity. That is, L
as both ~r and p~ reside in that plane at all times. But we would not expect the orbit to be constant and
closed. In fact only for two types of central potentials the solution is a closed, stable orbit: 1/r and r2
(Bertrands theorem). The fact that for Keplers problem the orbit is stable and does not precess hints
to another symmetry and conserved quantity. Indeed, there is a quantity called the Laplace-Runge-Lenz
(LRL) vector, given by
~ = 1 p~ L
~ c ~r,
M
(5)

r
which is conserved.
~ is conserved, let us understand its meaning. It is obvious that M
~ L
~ = 0 so
Before proving that M
~ lies in the plane of the orbit. Furthermore, M
~ points along the major axis of the ellipse defining the
M
orbit. This can be shown by looking at its value at the perihelion, and since it is fixed that is its values
at all times. So the fact that it is conserved relates the fact that the major axis of the orbit is fixed, thus
it does not precess.
1

~ is fixed. We get
Now we will show that indeed M
~
dM
1
~ + 1 p~ L
~ c ~r + c r~
r,
= p~ L
dt

r
r2

(6)

~ = 0 since it is constant. From Hamilton equations

but of course L
H
p~
=
~
p

H
c
p~ =
= 3 ~r,
~r
r

~r

and now
r =

p~ ~r
d 2
1
~r = ~r~r =
.
dt
r
~r2

(7)

(8)

Now we can explicitly write

~
dM
dt

c
c
c
c p~
1 ~
c p~ ~r
p~ L ~r + 2 r~
r = 3 ~r (~r p~)
+
~r =

r
r
r
r r2 r
c
c
c
3 (~
p ~r~r r2 p~) + 3 p~ ~r~r
p~ = 0,
r
r
r

(9)

and indeed it is a conserved quantity.

The Quantum LRL vector

We would like to define a quantum version of the LRL vector, simply writing
~ c ~r,
~ = 1 p~ L
M

(10)

with all classical quantities replaced with operators. However, such an operator wont be hermitian, as
(AB) = B A we get, for example
~ z = (px Ly py Lx ) = Ly px Lx py 6= px Ly py Lx ,
(~
p L)

(11)

~ does not commute. In general

and similarly for the rest of the components. This is because p~ and L
~ = L
~ p~. The solution to this is by noticing that for any operator A, the operator defined by
(~
p L)
A = (A + A )/2 is hermitian by construction. So our quantum LRL vector will be defined as


~ = 1 p~ L
~ L
~ p~ c ~r.
M
2
r

(12)

It is a tedious, but doable task, to prove the following identities

h
i
~ ,H = 0
M
~ M
~ =M
~ L
~ =
L

0

2H  ~ 2
~
M
=
L + ~2 + c2

X
[Mi , Lj ] = i~
ijk Mk
2

[Mi , Mj ]

~ X
2i H
ijk Lk .

(13)

~ accordingly by
Working at subspaces of eigenstates of H, we can rescale M
r
~
~
M
M,
2E

(14)

and then the last relation becomes

[Mi , Mj ] = i~

ijk Lk .

(15)

~, L
~ and themselves define the Lie algebra of SO(4) rotations
The commutation relations between M
in 4-dimensions. This is the true symmetry of the problem with the apparent symmetry of SO(3) only
nested within it.

SO(4) Symmetry

Our first task is constructing the operators for rotations in 4-dimensions. We will denote the axes by
numbers, with the commutator [ri , pj ] = i~ij , and the generators of rotations (i.e. angular momentum
operators) by the plane in which they induce rotations. So the ordinary three angular momentum
operators will be denoted
Lx = ypz zpy

r2 p3 r3 p2 = J23

Ly = zpx xpz

r3 p1 r1 p3 = J31

Lz = xpy ypx

r1 p2 r2 p1 = J12

(16)

and we will need another three operators, given by

J14

= r1 p4 r4 p1 ,

J24

= r2 p4 r4 p2 ,

J34

= r3 p4 r4 p3 ,

(17)

and we claim that Mx = J14 , My = J24 and Mz = J34 . Let us show that this is indeed the case, by
examining some of the commutations relations
[Mx , My ]

[r1 p4 r4 p1 , r2 p4 r4 p2 ] = i~r1 p2 i~r2 p1 = i~J12 = i~Lz ,

[Lx , My ]

[r2 p3 r3 p2 , r2 p4 r4 p2 ] = i~r4 p3 + i~r3 p4 = i~J34 = i~Mz ,

(18)

and the rest of the needed commutation relations are obtained similarly. So we saw that the Lie algebra
induced by rotations in 3-dimensions combined with the transformations generated by the LRL-vector
is identical to the Lie algebra of rotations in 4-dimensions. We would now like to obtain the Casimir
operators and the Cartan subalgebra of this group.
To this end we define a new set of operators
~ =
A
~
B


1 ~
~ ,
L+M
2

1 ~
~
LM
2

(19)

and the following commutation relations hold

[Ai , Aj ]

X
1
~X
1
[Li + Mi , Lj + Mj ] = i
ijk (Lk + Mk + Mk + Lk ) = i~
ijk (Lk + Mk ) =
4
4
2
k
k
X
i~
ijk Ak ,
k

[Bi , Bj ]

X
1
~X
1
[Li Mi , Lj Mj ] = i
ijk (Lk Mk Mk + Lk ) = i~
ijk (Lk Mk ) =
4
4
2
k
k
X
i~
ijk Bk ,
k

[Ai , Bj ]

1
~X
[Li + Mi , Lj Mj ] = i
ijk (Lk Mk + Mk Lk ) = 0.
4
4

(20)

So in this new basis the generators of the SO(4) symmetry were decomposed to two sets of generators,
each having a Lie algebra of SO(3) and each independent of the other, which means that
SO(4) SO(3) SO(3).

(21)

Diagonalizing each SO(3) algebra is immediate, as we now that we can construct a common set of
eigenvectors for the following operators: A2 , B 2 , Az and Bz , denoted by |a, b, ma , mb i with eigenvalues
Az /Bz |a, b, ma , mb i =
2

A /B |a, b, ma , mb i =

~ma/b |a, b, ma , mb i
~2 a(a + 1)/b(b + 1)|a, b, ma , mb i,

(22)

with the spectrum a, b = 0, 1/2, 1, . . . and a/b ma/b a/b. The degeneracy of any Hamiltonian
having the symmetry SO(4) is immediately (2a + 1)(2b + 1). These are general results for SO(4).

Spectrum of the Hydrogen Atom

Using these results to solve the problem of the hydrogen-atom, we note that in our case, because of
~ L
~ = 0, we can write
M
2


1~
~ = 1 M 2 + L2 ,
M +L
A2 =
4
4
2


1
1~
2
~ =
B =
M L
M 2 + L2 ,
(23)
4
4
and A2 = B 2 , which means that a = b. Moreover, from that we may deduce that
M 2 = 4A2 L2 ,

(24)

going back to the original scaling of M , we get that

2E 2
M

(25)

2E
4A2 L2 .

(26)

M2
which means
M2 =
Recalling that
M2 =


2H  ~ 2
L + ~2 + c2 ,

(27)

we equate the two expressions

L2 ~2

c2
= 4A2 L2
2E
4

(28)

and plug in the eigenvalues for A2 which are ~2 a(a + 1) resulting in

E=

1
1
c2
c2
,
=

2
2
2~ 4a(a + 1) + 1
2~ (2a + 1)2

(29)

recalling that a = 0, 1/2, 1, . . . we get that (2a + 1) = 0, 1, 2, . . . which we can denote now simply by n,
and recalling that c = Ze2 , the energies of the hydrogen atom are given by
En =

Z 2 e4
,
2~2 n2

(30)

and the degeneracy is (2a + 1)(2b + 1) = n2 .

~ =A
~ +B
~ is the
We are left with the task of finding the allowed values of angular momentum. Since L
sum of two independent angular momenta, you will see next week in class that the values it can get are
|a b| l a + b,
and here a = b = (n 1)/2 so 0 l (n 1).

(31)