Homework 2.

2
Quantum Mechanics II

Pavel Elian Morales Medina

April 22nd , 2021

Problem 7.35
Here is an application of the technique developed in Problem 7.34. Consider the Hamiltonian
   
 0 0 0 a b
H 0 = 0  0  , H 0 = a∗ 0 c 
0 0 0 b∗ c∗ 0

To start, we recall the definition of the PD from Equation (7.100)

 
  

PD = ψa0 ψa0  + ψb0 ψb0  (1)

Then, we notice that by orthogonality of the states

   
  : 1  0 
0  :0  
PD ψa0 = ψa0 ψa0 
ψa0 + ψb ψb ψa0 = ψa0
 
   
  : 0  0 
0  :1  
PD ψb0 = ψa0 ψa0 
ψb0 + ψb ψb ψb0 = ψb0
 
(2)

(a) Find the projection operator PD (it’s a 3 × 3 matrix) that projects onto the subspace
spanned by
   
 0 1  0 0
ψa = 0 and  ψb = 1

0 0

Then construct the matrices H̃ 0 and H̃ 0 .

First, notice that has the matrix H 0 suggests, we are in a vector space of dimension 3, thus
a matrix representation of an operator must have dimension 3 × 3. Then, remember that for
a diagonal matrix D we have the following eigenvalue equation
 
D1,1 0 0
D (êi ) =  0 D2,2 0  êi = Di,i (êi ) with i = 1, 2, 3
0 0 D3,3

1
Where êi are the canonical unit vectors,
     
1 0 0
ê1 = 0
  ; ê2 = 1
 ; ê3 = 0
 (3)
0 0 1
also Di,i are their corresponding eigenvalues and the diagonal entries of the matrix. Observe
that as the definition the given states indicates,
ψa0 = ê1 and ψb0 = ê2 (4)
Next, to obtain the matrix for the operator PD , we will base on D realizing that expressions
in Eq.(2) are actually eigenvalue equations with eigenvalue 1 for each state, therefore, its the
first two diagonal elements have been found as Eq.(3) suggests
 
1 0 0
PD = 0 1 0 
0 0 D3,3
Even tough a diagonal element is still undetermined, we can also find it by realizing that
the projection operator projects into the subspace spanned only by ψa0 and ψb0 , i.e. for the
unit vector ê3 its eigenvalue has to be D3,3 = 0, given that is a vector linearly independent
from both and its projection cannot get out of the 2-dimensional subspace. Consequently
 
1 0 0
PD = 0 1 0
0 0 0

Finally we recall the definitions from equation (7.102):

H̃ 0 = H 0 + PD H 0 PD , H̃ 0 = H 0 − PD H 0 PD (5)
Before substituting we compute
   
1 0 0 0 a b 1 0 0
0
PD H PD = 0  1 0 a∗ 0 c  0 1 0
0 0 0 b∗ c∗ 0 0 0 0
  
1 0 0 0 a 0
PD H 0 PD = 0 1 0 a∗ 0 0
0 0 0 b∗ c∗ 0
 
0 a 0
PD H 0 PD = a∗ 0 0
0 0 0
Substituting into the definitions, first we get
   
 0 0 0 a 0
H̃ 0 = 0  0  + a∗ 0 0
0 0 0 0 0 0

2
  
 a 0
H̃ 0 = a∗  0  (6)
0 0 0
Second, we obtain
   
0 a b 0 a 0
H̃ 0 = a∗ 0 c  − a∗ 0 0
b∗ c∗ 0 0 0 0
 
0 0 b
H̃ 0 =  0 0 c  (7)
b∗ c∗ 0

(b) Solve for the eigenstates of H̃ 0 and verify... i. that its spectrum is non degenerate, ii.
that the non-degenerate eigenstate of H 0
 
 0 0
ψc = 0
1

is also an eigenstate of H̃ 0 with the same eigenvalue.

For calculating the eigenstates of H̃ 0 we need its eigenvalues in first place. To compute them
we’ll use its characteristic polynomial as follows
 
det H̃ 0 − GI3×3 = 0
 
−G a 0
det  a∗ −G 0 =0
0
0 0  −G

Due to the last row has two zero elements, we calculate the determinant by expanding
through the 3rd row.

⇒ +0 + (−1)3+3 [0 − G] ( − G)2 − |a|2 = 0

 

2 0 − 20 G + G2 0 − G2 + 2G2 − G3 + G |a|2 − 0 |a|2 = 0

−G3 + G2 (0 + 2) + G −20 − 2 + |a|2 + 2 0 − 0 |a|2 = 0


(8)

To solve the previous equation, we may be tempted to use Cardano’s expression for the roots
of a cubic polynomial, nevertheless we might guess the first root by noticing that the only
coefficient that can cancel out the term −0 |a|2 is G |a|2 , solving for G we have

−G |a|2 = −0 |a|2

⇒ G = 0

3
Testing our educated guess for G by replacing this value into equation (8)

−03 + 03 + 202 − 202 − 2 0 + 0 |a|2 + 2 0 − 0 |a|2 = 0

0=0

Meaning that G = 0 do is a root, thus we can factor out (G − 0 ) and equation 8 becomes

(G − 0 ) −G2 + 2G − 2 + |a|2 = 0

Now, we simply need to use the quadratic form, for the second degree polynomial in paren-
thesis
q
−2 ± 42 − 4 (−1) −2 + |a|2


1
q
⇒G= =∓   42 + 4
|a|2
42 − 
2 (−1) 2
G± =  ± |a|

Summarizing, the three eigenvalues found are

G = 0 and G± =  ± |a| (9)

Continuing, we proceed to calculate the associated eigenvalues, by using the eigenvalue

equation,
    
 a 0 x1 x1
a∗  0  x2  = G x2 
0 0 0 x3 x3
   
x1 + ax2 x1
a ∗ x1 + x2  = G x2  (10)
0
 x3 x3

For the first case G = 0 :

x1 + ax2 = 0 x1
a ∗ x1 + x2 = 0 x2
0 x3 = 0 x3

The last equation of the system give us x3 = 1, moreover by noticing that the first and
second one do not depend on x3 if we set x1 = x2 = 0 is easy to see that the system is solved.
Obtaining the already normalized eigenstate defined by the problem as |ψc0 i, explicitly
 
 0 0
ψc = 0
1

4
Back to equation (10), now we set G± =  ± |a|:
|a|
x1 + ax2 = (
 ± |a|) x1 ⇒ x2 = ± x1

a
|a|
a ∗ x1 + 
x2 = (
 ± |a|) x2 ⇒ x1 = ± x2
a∗
0 x3 = ( ± |a|) x3 ⇒ x3 = 0
Since the last equation give us x3 = 0 now set x1 = x2 = 0 is not a normalized eigenstate
solution, however noticing that the first two equations are not linearly independent we can
take either of them, so the non-normalized vector is
 
x 1
ψG± = ± |a| x1 
 
a
0
Imposing the normalization condition
 
  x1
1 = ψG± ψG± = x1 ± |a| 0 ± |a|

 
a 1
x a 1
x
0
a∗
1 = x21 + x21
s a
1
x1 =
1 + a∗a
r
a
x1 =
a + a∗
Substituting back into the eigenstate
   
r 1 r 1
  a  |a|  a  p a∗ 
ψG± = ±a = ± a
a + a∗ a + a∗
0 0
Getting the eigenstates
   
p1a∗ 1
r r
  a    a  p a∗ 
ψG+ =
a
 and ψG− = − a (11)
a + a∗ a + a∗
0 0
(i) Noticing that each eigenvalue associated just one eigenvector, and that this eigenvectors
are different, we can verfiy that its spectrum is non-degenerate.

(ii) We have found that |ψc0 i do is an eigenstate of H̃ 0 with eigenvalue 0 . As the problem
request we must verify that this also the eigenvalue for the case of H 0 then
    
 0 0 0 0
0 = 0  = 0 ψc0

0 0
  
H ψc = 0  0    
0 0 0 1 0

Therefore for H 0 the eigenstate |ψc0 i has an eigenvalue 0 as it did for the H̃ 0 .

5
(c) What are the “good states”, and what are their energies to first order in the perturba-
tion?
Since we are dealing with 2-fold degeneracy (|ψa0 i and |ψb0 i have as eigenvalue ), recall that
the first order correction from degenerate perturbation theory is given by
 
1
q
1 2 2
E± = Waa + Wbb ± (Waa − Wbb ) + 4 |Wab |
2
Consequently we need to compute
  
0 a b 1
0 0 0

  

Waa = ψa H ψa = 1 0 0 a∗ 0 c
   0
b∗ c∗ 0 0
 

0
= 1 0 0 a∗ = 0
b∗
  

0 0 0
0 a b 0
Wbb = ψb H ψb = 0 1 0 a∗ 0 c
     1
b∗ c∗ 0 0
 

a
= 0 1 0 0=0
c∗
  

0 0 0
0 a b 0
Wab = ψa H ψb = 1 0 0 a∗ 0 c
     1
b∗ c∗ 0 0
 

a
= 1 0 0 0=a
c∗
Hence, we can substitute and find the first order energy corrections for the degenerate states.
1√
 
1
q
2 ∗
E± = 0 + 0 ± (0 − 0) + 4Wab Wab = ± 4aa∗ = ± |a|
2 2
Meaning that the degenerate eigenvalue is being split by the presence of the perturbation,
thus the energy up to the first correction is

E = E 0 + E 1 =  ± |a| = G±

Noticing that this energy is precisely the definition of the eigenvalue G± previously found;
it comes right after that the “good states” are their eigenvalues, already calculated as well
 
r 1
  a  p a∗ 
ψG± = ± a
a + a∗
0

6
Problem 7.38
Suppose the Hamiltonian H, for a particular quantum system, is a function of some parameter
λ; let En (λ) and ψn (λ) be the eigenvalues and eigenfunctions of H (λ). The Feynman-
Hellmann theorem states that
   
∂En  ∂H 
= ψn   ψn
∂λ ∂λ 
(assuming either that En is non degenerate, or -if degenerate- that the ψns are the “good”
linear combinations of the degerate eigenfunctions).

(a) Prove the Feynman-Hellmann theorem. Hint: Use Equation 7.9.

Using perturbation theory, we set as the unperturbed Hamiltonian, H (λ), the one given by
problem. Writing its eigenvalue equation
H (λ) ψ (λ) = En (λ) ψ (λ)
Or, using perturbation theory notation, we just add a super-index indicating this corresponds
to the solved Hamiltonian.
H 0 (λ) ψ 0 (λ) = En0 (λ) ψ 0 (λ)
Now, let us define a new Hamiltonian, as the original plus a perturbation
∂H 0 ∂H (λ)
Hnew = H 0 + λH 0 with H0 = =
∂λ ∂λ
Hence, from perturbation theory, we compute the first order correction for the perturbed
Hamiltonian Hnew using Eq. (7.9), then
0
   
0  ∂H  0
1

0  0 0  
En = ψn H ψn = ψn 
  ψ
∂λ  n
Also, from notation of perturbation theory we can replace En1 by its respectively coefficient
from the expansion of the energy as function of λ, having
0
     
1 ∂En  ∂H
= ψn0  ψ 0
1! ∂λ ∂λ  n
0
   
∂En 0  ∂H  0

⇒ = ψn  ψ
∂λ ∂λ  n
Into the original notation
   
∂En  ∂H (λ) 
= ψn  ψn
∂λ ∂λ 
Obtaining the statement of the Feynman–Hellmann theorem
   
∂En  ∂H 
= ψn  ψn (12)
∂λ ∂λ 

7
(b) Apply it to the one-dimensional harmonic oscillator, (i) using λ = ω (this yields a
formula for the expectation value of V ), (ii) using λ = ~ (this yields hV i), and (iii) using
λ = m (this yields a relation between hT i and hV i). Compare your answers to Problem 2.12,
and the Virial theorem predictions (Problem 3.37).
For the simple harmonic oscillator we know
 
1 1  2
p̂ + (mωx)2

En = n + ~ω and Ĥ =
2 2m
Where
p̂2 1
T = and V = mω 2 x2
2m 2
(i) For λ = ω we compute
    
∂En ∂ 1 1
= n+ ~ω = n + ~
∂λ ∂ω 2 2
Also,
 
∂H ∂ 1  2 2 1
= p̂ + (mωx) =  (
2mωx)
 (mx) = mωx2
∂λ ∂ω 2m 2m


While
   
∂H 2ω 1 2 2V
= mωx = mω 2 x2
2
=
∂λ 2ω |2 {z } ω ω
V

Thus, using the Feynman–Hellmann theorem

     
1  2V 
⇒ n+ ~ = ψn  ψn (13)
2 ω 

Simplifying the notation, while solving for the expectation value of V

 
1 ~ω
hV i = n + (14)
2 2

(ii) For λ = ~ we compute

    
∂En ∂ 1 1
= n+ ~ω = n + ω
∂λ ∂~ 2 2
Also,
 
∂H ∂ 1  2 2 1 ∂
p̂2


= p̂ + (mωx) =
∂λ ∂~ 2m 2m ∂~
d2 −~ d2
 
1 ∂
= −~2 2 =
2m ∂~ dx m dx2

8
While
~ −~ d2 2
     2
∂H 2 d 1 p̂ 2
= 2
= −~ 2
=
∂λ ~ m dx dx ~m ~m 2
  2
2 p̂ 2T
= =
2m
~ |{z} ~
T

Thus, using the Feynman–Hellmann theorem

     
1  2T 
⇒ n+ ω = ψn  ψn (15)
2 ~

Simplifying the notation, while solving for the expectation value of T

 
1 ~ω
hT i = n+ (16)
2 2

This expectation value verifies the previous result obtained by performing integrations at
problem (2.12). (iii) For λ = m we compute
  
∂En ∂ 1
= n+ ~ω = 0
∂λ ∂m 2

Also,
 
∂H ∂ 1  2 −1  2 1
p̂ + (mωx)2 p̂ + (mωx)2 +  (
 
= 2
= 2mωx)
 ωx
∂λ ∂m 2m 2m 2m

2
p̂2 p̂2 p̂2
 
(
mωx) 2 2 1 1
+ (ωx)2

=− 2 − + (ωx) = − 2
+ (ωx) 1 − = − 2
2m 2m2
 2m 2 2m 2
 
1  p̂2 1 1
= − + mω 2 x2  = (−T + V )

m 2m |2 {z }
|{z} m
T V

Thus, using the Feynman–Hellmann theorem

   
1 
⇒ 0 = ψn  (−T + V )ψn
 (17)
m

Simplifying the notation, while solving for the expectation value of V

0 = h(−T + V )i = − hT i + hV i

⇒ hV i = hT i (18)
Result that agrees and also was been obtained by the Virial theorem.

9
Problem 7.42
The Feynman-Hellmann theorem (Problem 7.38) can be used to determine the expectation
values of 1/r and 1/r2 for hydrogen.37 The effective Hamiltonian for the radial wave functions
is (Equation 4.53)
~2 d2 ~2 ` (` + 1) e2
H=− + − ,
2m dr2 2m r2 4π0 r
and the eigenvalues (expressed in terms of `)38 are (Equation 4.70)
me4
En = .
32π 2 20 ~2 (N + `)2

From the given H we compute

~2 ∂ `2 + ` ~2
 
∂H −2e ∂H
= and = = (2` + 1)
∂e 4π0 r ∂` 2m ∂` r2 2mr2
On the other hand, with the given En we calculate
∂En −m
4e3


= 2
∂e 2
32π 2 0 ~2 (N + `)
Additionally,
∂En −me4 ∂ 2 −me4 1
= (N + `) = (−2)
∂` 2 2 2
32π 0 ~ ∂` 2 2 2
32π 0 ~ (N + `)3
Finally recalling the definitions below from eqns. (4.72) and (4.67) respectively;
1 me2
≡ and n≡N +`
a 4π0 ~2

(a) Use λ = e in the Feynman-Hellmann theorem to obtain h1/ri. Check your result against
Equation 7.56.
Using the Feynman–Hellmann theorem for λ = e, and substituting from the calculations
above.
 
∂En ∂H
=
∂e ∂e
 
−m 3

−2e
⇒ 4e =
32π 2 20 ~2 (N + `)2 4π0 r
 
m (4π0 ) 1
4e3 =


2
2
32π 2 0 ~2 (N + `) (2e) r
me2
 
1
=
4π0 ~2 (N + `)2 r
2
   
me 1 1
2 2 =
4π ~ r
| {z0 } | (N {z
+ `)
}
a−1 n−2

10
From the recalled definitions  
1 1
= 2 (19)
r an
Result that does agree with eq. (7.56).

(b) Use λ = ` to obtain h1/r2 i. Check your answer with Equation 7.57.
Using the Feynman–Hellmann theorem for λ = `, and substituting from the calculations
above.
 
∂En ∂H
=
∂` ∂`
4
 2 
−me 1 ~
( −2) = (2` + 1)
32π 2 20 ~2 (N + `)3 2mr2
me4 (2)
 
2m 1
3 2 =
32π 0 ~ (N + `) ~ (2` + 1)
2 2 2 r2
2m2 e4
 
1
=
16π 2 20 ~4 (N + `)3 (2` + 1) r2
m2 e4
   
2 1
=
16π 2 20 ~4 (N + `)3 (2` + 1) r2
| {z }
a−2

From the recalled definitions

 
1 1
=
` + 12


r2 a2 n3

Result that does agree with eq. (7.57).

11