Quantum Mechanics - II

Approximation Methods :
Perturbation Theory - II
(Degenerate Case)
Dipan Kumar Ghosh
UM-DAE Centre for Excellence in Basic Sciences
Kalina, Mumbai 400098
September 29, 2019

1 Introduction
(0)
In the non-degenerate perturbation theory, we had a unique state | ψn i corresponding to
(0)
energy En to which the perturbed system approached at the parameter λ → 0. however,
if the zeroth order eigenstate is degenerate, it means that in addition to the energy, there
is some other observable, say A, which characterizes the states. Our basis states are then
eigenstates of both H0 and the operator Â
(0) (0) (0)
H0 | ψmα i = Em | ψmα i
(0)
where α is a quantum number corresponding to Â. The energy Em is independent of
(0)
the index α. This implies that any linear combination of | ψmα i with α = 1, 2, . . . is an
eigenstate of H0 . Thus, if we start from any unperturbed state and doing a perturbation
expansion reach a perturbed state and then reduce the parameter λ to approach zero, we
may arrive at a totally different linear combination of the basis states. If [V, A] 6= 0, the
zeroth order eigenstates of H = H0 + V are not eigenstates of Â. This lifts the degeneracy
associated with H0 .
For simplicity, let us take the unperturbed state to be two-fold degenerate. Let

H | ψn i = (H0 + V ) | ψn i = En | ψn i (1)

The unperturbed energy levels are given by

(0)
H0 | ψnα i = En(0) | ψnα
(0)
i, where α = 1, 2 (2)
 (0)
The states {| ψnα i} are orthonormalized
(0) (0)
hψnα | ψnβ i = δα,β

In the lowest order of approximation, we can write

(0) (0)
| ψn i = a01 | ψn1 i + a02 | ψn2 i

where a01 and a02 are constants. Substituting this in (1) we have
   
(0) (0) (0) (0)
(H0 + V ) a01 | ψn1 i + a02 | ψn2 i = En a01 | ψn1 i + a02 | ψn2 i
     
(0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)
En a1 | ψn1 i + a2 | ψn2 i + V a1 | ψn1 i + a2 | ψn2 i = En a1 | ψn1 i + a2 | ψn2 i
(3)
(0) (0)
From (3), multiplying both sides by hψn1 | and hψn2 |, we get the following equations for
the coefficients a01 and a02 .
(0) (0) (0) (0)
En(0) a01 + a01 hψn1 | V | ψn1 i + a02 hψn1 | V | ψn2 i = En a01 (4)
(0) (0) (0) (0)
En(0) a02 + a01 hψn2 |V | ψn1 i + a02 hψn2 |V | ψn2 i = En a02 (5)

These are simultaneous equations in the coefficients a01 and a02 and the solutions exist only
if the secular determinant vanishes, i.e.,
(1)
V11 − En V12
(1) =0 (6)
V21 V22 − En

(0) (0) (1) (0)

where, Vαβ = hψnα | V | ψnβ i and En = En − En . The solution of (6) gives the change
in energy in the first order perturbation,
p
En(1) = −(V11 − V22 ) ± (V11 − V22 )2 + 4V12 V21 (7)

The eigen vectors which removes the degeneracy are the eigen vectors that diagonalize
the matrix. These are the states to which the perturbed state approaches in the limit of
λ → 0. The above can easily be extended to the case where the degeneracy is g-fold. In
this case the secular equation is an g × g matrix.

2 Examples
2.1 Linear Stark Effect in Hydrogen Atom
We have seen that because of degeneracy of levels for n > 1, it was not possible to use
the non-degenerate perturbation theory for the excited states of hydrogen atom. Let us
consider n = 2 level of the H-atom. The energy of the state is −1/4 Ry and the state
is 4-fold degenerate (we ignore the spin degeneracy here which cannot be lifted by the
electric field). The perturbation Hamiltonian is
V = −ez | E |= −er cos θ | E |
The unperturbed states have the singlet state l = 0 and the triplet state l = 1 with m =
−1, 0, +1. We need to construct the secular matrix with element hl, m | V | l0 , m0 i. Since
V depends only r and θ and is independent of the azimuthal angle φ, hl, m | V | l0 , m0 i = 0,
except when m = m0 . Further, Since V is of odd parity, it can only connect states of
opposite parity, i.e. it gives non-zero value only between states of l = 0 and l = 1. In this
case the only non vanishing state is h0, 0 | V | 1, 0i and its conjugate. The matrix element
can be calculated using explicit forms of the hydrogen atom wavefunctions ψn,l,m , which
for n = 2, m = 0 are
 
1 1 r
ψ2,0,0 = √ p 3 1 − e−r/2a0
4π 2a0 2a0
r
3 1 r −r/2a0
ψ2,1,0 = p e cos θ
4π 6a30 2a0
The matrix elements are calculated in position representation as follows :
√ Z ∞  Z π
3 1 1 r 4 −r/a0
h2, 0, 0 | r cos θ | 2, 1, 0i = √ 3 2π 1− r e cos2 θ sin θdθ
4π 12a0 2a0 0 2a0 0
Z ∞  Z 1
1 r
= 4
1− r4 e−r/a0 µ2 dµ (where µ = cos θ)
8A0 0 2a0 −1
 
1 2·3·4 1 2·3·4·5
= −
12a40 (1/a0 )5 2a0 (1/a0 )6
a0
= (24 − 60) = −3a0 (8)
12
Using (8) in the 4 × 4 secular equation, we have
| 2, 0, 0i | 2, 1, 0i | 2, 1, 1i | 2, 1, 1̄i
| 2, 0, 0i 0 −3a0 0 0
−eE | 2, 1, 0i −3a0 0 0 0 =0 (9)
| 2, 1, 1i 0 0 0 0
| 2, 1, 1̄i 0 0 0 0
If one looks at the left hand corner of the matrix, it can be seen to be proportional to
Pauli’s σx with eigenvalues ±1. Thus the eigenvalues of the secular equation are
∆E = ±3eEa0 (10)
The corresponding eigen vectors can be readily found to be
1
ψ+ = √ (| 2, 0, 0i+ | 2, 1, 0i) (11)
2
1
ψ− = √ (| 2, 0, 0i− | 2, 1, 0i) (12)
2
Note that the states | 2, 1, 1i and | 2, 1, 1̄i are still degenerate with the unperturbed eigen-
value −1/4 Ry.
Comments :

• The 2s and 2p levels shown in the figure are really degenerate. Due to the spin-orbit
interaction, the 2p level splits into 2p1/2 and 2p3/2 . There is, however, degeneracy
persisting between 2s1/2 and 2p1/2 levels.

• The degeneracy between 2s1/2 and 2p1/2 levels is removed in QED due to Lamb shift.
However, the energy separation introduced by the degenerate perturbation theory
is much larger than the Lamb shift.

1
√ (| 2, 0, 0i− | 2, 1, 0i)
2

| 2, 1, 1i and | 2, 1, 1̄i
four fold

1
√ (| 2, 0, 0i+ | 2, 1, 0i)
2

Figure 1: Stark Effect splitting in Hydrogen Atom

2.2 Two dimensional Harmonic Oscillator with perturbation

Consider a two dimensional oscillator with the Hamiltonian
~2 d2 d2 1
H0 = − ( 2 + 2 ) + mω 2 (x2 + y 2 ) (13)
2m dx dy 2
The energies of the oscillator are given by
1 1
(nx + )~ω + (ny + )~ω
2 2
The first excited level has a two fold degeneracy (nx = 0, ny = 1 and nx = 1, ny = 0)
with an energy of 2~ω. In occupation number representation the states of the system are
denoted by | 0, 1i and | 1, 0i where the first index refers to the x-oscillator and the second
to y-oscillator. Now, a perturbation V = mλω 2 x̂ŷ is applied to the system. The resulting
secular equation can be easily obtained by going over to the representation for x and y
r
~
x̂ = (a + a† )
r 2mω
~
ŷ = (b + b† )
2mω
The perturbation is then given by
~ω
λ (a + a† )(b + b†
2
The non-zero matrix elements of the perturbation are as follows :
~ω
V12 = λ h0, 1 | (a + a† )(b + b† ) | 1, 0i
2
The operator a, a† acts only on the first index while the operators b, b† only on the second.
This gives
~ω
V12 = λ
2
A similar calculation gives V21 = V12 . Thus the energy shifts are given by

~ω
−E (1) λ
2 =0
~ω
λ −E (1)
2
~ω
which gives the energy shifts to be ±λ . The eigen states of the perturbed Hamiltonian
2
| 01i± | 10i
can be seen to be .
2
Let us consider the next excited states which are three fold degenerate. These states
are | 2, 0i, | 1, 1i and | 0, 2i. The unperturbed energy is 3~ω. Let the ladder operators for
the motion in the x and y directions be noted by a(a† ) and b(b† ). We then have

~
V = kxy = k (a + a† )(b + b† )
2mω
where the operators a’s commute with the operator b’s. We need the matrix elements
~
hmx , my | V | nx , ny i = k hmx , my | (a + a† )(b + b† ) | nx , ny i. Since a, b, a† , b† either
2mω
create or destroy, the diagonal matrix elements are all zero. The only non-zero elements
are

h1, 1 | V | 2, 0i, h1, 1 | V | 0, 2i, h2, 0 | V | 1, 1iand h0, 2 | V | 1, 1i

 k~
and each of these matrix elements is equal to √ ≡ δ. The matrix V is then
2mω
 
| 2, 0i | 1, 1i | 0, 2i
| 2, 0i 0 δ 0 
 
V =
| 1, 1i δ 0 δ 

| 0, 2i 0 δ 0

The secular determinant is

E −δ 0
−δ E −δ = 0
0 −δ E
√
which has the solution E = 0, ± 2δ. The corresponding eigenstates may be easily found
1 1 √ 1 √
to be √ (| 2, 0i− | 0, 2i), (| 2, 0i + 2 | 1, 1i+ | 2, 0i) and (| 2, 0i − 2 | 1, 1i+ | 2, 0i).
2 2 2