Homework 2.

3
Quantum Mechanics II

Pavel Elian Morales Medina

April 30th , 2021

Problem 8.4
(a) Prove the following corollary to the variational principle: If hψ|ψgs i = 0, then hHi ≥ Efe ,
where Efe is the energy of the first exited state. Comment: If we can find a trial function
that is orthogonal to the exact ground state, we can get an upper bound on the first exited
state. In general, it’s difficult to be sure that ψ is orthogonal to ψgs , since (presumably) we
don’t know the latter. However, if the potential V (x) is an even function of x, then the
ground state is likewise even, and hence any odd trial function will automatically meet the
condition for the corollary.
Answer:
We will proceed as the proof of the Variational principle, but taking advantage of the given
orthogonality hψ|ψgs i = 0 as well.
Since the eigenfunctions of H form a complete set, we can express ψ as a linear combination
of them
X
ψ= cn ψn , with Hψn = En ψn .
n

Consequently, with the given condition

hψ|ψgs i = 0
⇒ hψgs |ψi = 0
*  +
X
∴ ψgs  cn ψn = 0

 n
X
cn hψgs |ψn i = 0 (1)
n

Additionally, we know that ψ has to be normalized

*  +
X X XX
hψ|ψi = 1 = cm ψm  cn ψ n = c∗m cn hψm |ψn i (2)

m
 n m n

1
However, notice that for m = gs we have a sum of terms that equals zero, by replacing from
equation 1 as follows
X
c∗gs cn hψgs |ψn i = c∗gs (0) = 0
n

Meaning that m = gs does not contributes to the sums in the expression 2 for the normal-
ization, and we can write
XX
hψ|ψi = 1 = c∗m cn hψm |ψn i
m6=gs n

Now, we must recall that the eigenfunctions themselves, have been orthonormalized, so that
hψm |ψn i = δmn . Thus, equation above simplifies, obtaining
X
1= |cm |2 (3)
m6=gs

On the other hand, we can compute the expectation value of H

*  +
X  X XX
hHi = cm ψm H cn ψ n = c∗m En cn hψm |ψn i


m n m n

Once again, we can notice that for m = gs, it provides a sum equal to zero, that does not
contribute to the total result, then using the orthonormalization of the eigenfunctions as
well,
X XX
hHi = c∗gs En cn hψgs |ψn i + c∗m En cn hψm |ψn i
| {z }
|n {z } m6=gs n δm n
0
X
= Em |cm |2
m6=gs

= Efe |cfe |2 + Ese |cse |2 + Ete |cte |2 + ... (4)

Recall that, by definition, the energies of the states are order as:

Egs < Efe ≤ Ese ≤ Ete ≤ ... ≤ Enth e

Where the sub-indices represent the ground state (gs), first exited state (fe), second exited
state (se), up to the n-th exited state (nth e). Hence, the sum
X
Em |cm |2 = Efe |cfe |2 + Ese |cse |2 + Ete |cte |2 + ...
m6=gs
X
≥ Efe |cfe |2 + Efe |cse |2 + Efe |cte |2 + ... = Efe |cm |2
m6=gs

2
 X X
Em |cm |2 ≥ Efe |cm |2 (5)
m6=gs m6=gs

Employing this condition 5 to the expression 4 for hHi previously calculated

X X
hHi = Em |cm |2 ≥ Efe |cm |2
m6=gs m6=gs
| {z }
1

Finally, substituting the sum from equation 3, getting the desired expression to prove.
⇒ hHi ≥ Efe (6)

(b) Find the best bound on the first exited state of the one-dimensional harmonic oscillator
using the trial function
2
ψ (x) = Axe−bx
Answer:
First, we need to verify that the given trial function do is orthogonal to the ground state of
the harmonic oscillator, remember that for this system,
 mω 1/4 mω 2 ~2 d2 1
ψgs (x) = e− 2~ x and Ĥ = − 2
+ mω 2 x2
π~ 2m dx 2
Then, we need to calculate
Z ∞  mω 1/4 mω 2
2
hψ|ψgs i = A∗ xe−bx e− 2~ x dx
−∞ π~
1/4 Z ∞
∗ mω
 mω
xe−(b+ 2~ )x dx
2
=A
π~ −∞

Nevertheless, noticing that the integrand is an even function (product of an even Gaussian
function and an odd x function) integrated over a symmetric interval around zero, its positive
and negative contributions cancel the integral out.
∴ hψ|ψgs i = 0
As desired to verify. With this result we can continue to the computation of the bound for
the first exited state; from the previous sub-exercise we have explicitly
D   E
ψ Ĥ ψ ≥ Efe (7)
 

Where we are able to compute the left side of the inequality, but, before getting into any
calculation we will solve some integrals that will turn out to be handy.
Z ∞ Z ∞  Z ∞  
2 −gx2 −gx2
 d −1 −gx2
xe dx = x xe dx = x e dx
−∞ −∞ −∞ dx 2g
−1 ∞ d  −gx2 
Z ∞
:0
Z  
−gx2 −1 h −gx2 i∞ 1 2
= xe −e dx = xe
 + e−gx dx
2g −∞ dx 2g  −∞ 2g −∞

3
Part of the remaining term, is the well-known Gaussian integral which is equal to
Z ∞ r
−gx2 π
e dx = (8)
−∞ g

Thus, the whole integral equals to

Z ∞ r
2 −gx2 1 π
xe dx = (9)
−∞ 2 g3

Another useful integral is

Z ∞ ∞   ∞
d −1 −gx2
Z   Z
4 −gx2 3 −gx2 3
xe dx = x xe dx = x e dx
−∞ −∞ −∞ dx 2g
−1 ∞ d  3 −gx2 
Z  
2 −gx2
= xe − 3x e dx
2g −∞ dx
Z ∞
−1 h 3 −gx2i
∞:0 3 2
= xe + x2 e−gx dx
2g  −∞ 2g −∞

Substituting the last equation from equation 9 we have

Z ∞ r
4 −gx2 3 π
xe dx = (10)
−∞ 4 g5

Next, we will find an expression for A using the normalization requirement

hψ|ψi = 1
Z ∞
2
⇒ |A|2 x2 e−2bx dx = 1
−∞

Evaluating the integral from the result at equation 9

2 1 π
r
|A| =1
2 (2b)3

Thus
s
2 (2b)3
|A| = 2 (11)
π

4
With this results, now it will be easy computing
D   E Z ∞ 
~2 d2 1

∗ −bx2 2
ψ Ĥ ψ = A xe − + mω x Axe−bx dx
2 2
 
2m dx 2 2
−∞
Z ∞  2

2 −bx2 ~ d 
−bx2 2 −bx2
 1
2 3 −bx2
= |A| xe − e − 2bx e + mω x e dx
−∞ 2m dx 2
Z ∞
~2 
 h i 1 
2 −bx2 −bx2 −bx2 3 −bx2 2 3 −bx2
= |A| xe − −2bxe − 2b 2xe − 2bx e + mω x e dx
−∞ 2m 2
Z ∞
~2  h i 1 
2 2 −2bx2 2 −2bx2 4 −2bx2 2 4 −2bx2
= |A| − −2bx e − 2b 2x e − 2bx e + mω x e dx
−∞ 2m 2
Z ∞
~2   1 
2 2 −2bx2 2 4 −2bx2 2 4 −2bx2
= |A| − −6bx e + 4b x e + mω x e dx
−∞ 2m 2
Z ∞  Z ∞
3b~2 2b2 ~2 1
  
2 2 −2bx2 2 4 −2bx2
= |A| xe dx + − + mω x e dx
−∞ m −∞ m 2

Using the result from integrals at expressions 9 and 10, we substitute

3b~2 2b2 ~2 1
  r    r 
D   E
2 1 π 2 3 π
ψ Ĥ ψ = |A| + − + mω
 
m 2 (2b)3 m 2 4 (2b)5
 r  
3b~2 2b2 ~2 1
   
1 π 3
= |A|2 + − + mω 2
2 2b m m 2 4b

Substituting |A|2 from the equation 11, we obtain

: 1 
s
3  r 
3b~2 2b2 ~2 1
   
D   E (2b) 1 π 2 3
ψ Ĥ ψ = 
2  + −

+ mω
  
π 2 2b m m 2 4b
2 2 2
   
3b~ 2b ~ 1 3
= + − + mω 2
m m 2 4b
2 2 2
3b~ 3b~ 3mω
= − +
m 2m 8b
2 2
3b~ 3mω
⇒ hHi = +
2m 8b
However from expression 7, corollary of the variational principle, we know that this expec-
tation value is greater or equal to the actual value of the energy Efe . Therefore we will
minimize to find which value of b allow us to get as close to the energy of the first exited
state as possible.

d hHi 3~2 3mω 2

= −
db 2m 8b2
d2 hHi 3mω 2
⇒ =
db2 4b3

5
We know that at the minimum the first derivative has to be zero
d hHi
=0
db
3~2 3mω 2
⇒ − 2 =0
2m 8bmin
~2 mω 2
= 2
2m 8bmin
2m2 ω 2
b2min =
8~2
mω
bmin =
2~
Evaluating the second derivative at this point candidate for a minimum
3
d2 hHi  3mω 2 2~
 
= >0
db2 b=bmin 4 mω

Then, by the second-derivative test we verify that hHi has a local and absolute minimum
(given that there are not more critical points) at bmin , which can be found by simply evalu-
ating
2   3mω 2  2~ 
 3~ mω
hHimin = hHi b=bmin = +
2m 2~ 8 mω
3~ω 3~ω
= +
4 4
Giving the result

3~ω
hHimin = (12)
2
Meaning that for this trial function, the equality at the corollary of the variational principle
actually holds at the minimum of hHi, which may seem surprising, but it’s not that much,
since the trial function was actually Gaussian as the eigenfunction of the ground state is as
well.

Problem 8.6
(a) Use the variational principle to prove that first-order perturbation theory always over-
estimates (or any rate never underestimates) the ground state energy.
Answer: Recalling the first-order perturbation theory correction to the energy

En1 = ψn0 H 0 ψn0 H = H0 + H0

  
where

As a result, for the ground state (gs) we have

1

0  0 0 
Egs = ψgs H ψgs (13)

6
In addition, using the variational principle for this Hamiltonian we get

hHi ≥ Egs
H + H 0 ≥ Egs
0



H + hH 0 i ≥ Egs

0

Particularly, for the wave function of the ground state from the solvable Hamiltonian, we
can write explicitly

0  0 0 
0  0 0 
ψgs H ψgs + ψgs H ψgs ≥ Egs
| {z }
1
Egs

Expression above, can be simplified, using the eigenvalue equation for H 0 at the ground state
H 0 ψgs
0 0 0
= Egs ψgs , the normalization properties of the wave function, and by identifying that
the second term on the left side, actually corresponds to the energy correction from equation
13, then it becomes

0 0 0 1
ψgs H ψgs + Egs ≥ Egs
0
: 1

0  0  1
Egs ψ
 
gs ψgs
 + Egs ≥ Egs

0 1
⇒ Egs + Egs ≥ Egs

As can be observed, the left hand side represents the energy estimation for the ground state
up to first-order non-degenerate perturbation theory, which by the variational principle, is
known to be a greater equal value to the energy eigenvalue of the ground state, best scenario
they are equal, i.e. the estimation is never under the actual value.
⇒ First-order non-degenerate perturbation theory always overestimates Egs

(b) In view of (a), you would expect that the second-order correction to the ground state
is always negative. Confirm that this is indeed the case, by examining Equation 7.15.
Answer: Writing down Equation 7.15 for the second-order correction
X |hψ 0 |H 0 |ψ 0 i|2
m n
En2 = 0 − E0
m6=n
E n m

For the ground state, we know n = gs, after that equation above becomes
X |hψ 0 |H 0 |ψgs i|2
2 m
Egs = 0 0
(14)
m6=gs
E gs − Em

7
It is easy to see that the numerator in the previous sum, is always positive due to the
the modulus squared. Whereas, for the denominator, we must remember that the energy
spectrum is ordered,

Egs < E2 < E3 < ...

⇒ Egs < Em with m 6= gs
∴ Egs − Em < 0

Hence, denominator in sum 14 is always negative. As a result, the second-order correction

is the sum of exclusively negative terms, bringing up a correction that is always negative .

Problem 8.10
Suppose we use a minus sign in our trial wave function (Equation 8.38):

ψ = A [ψ0 (r) − ψ0 (r0 )] .

Without doing any new integrals, find F (x) (the analogous to Equation 8.52) for this case,
and construct the graph. Show that there is no evidence of bonding.14 (Since the variational
principle gives an upper bound, this doesn’t prove that bonding cannot occur for such state,
but it certainly doesn’t look promising.)
Answer: We will proceed analogously to the positive sign, starting by finding the normal-
ization constant of the proposed trial function

*1
 
:1
Z Z  Z Z
 
0 2 3
1 = |ψ|2 d3 r = |A|2  ψ 0 (r)
 2 3
d r + ψ 0 (r ) d r − 2 ψ0 (r) ψ0 (r0 ) d3 r
 




Given that ψ0 is normalized, the first to integrals equal one, while the last one can be replaced
by the defined quantity I ≡ hψ0 (r)|ψ0 (r0 )i obtaining

1 = 2 |A|2 [1 − I]
1
⇒ |A|2 =
2 (1 − I)

Where
"  2 #
x2
   
R 1 R
I = e−R/a 1+ + =e −x
1+x+ (15)
a 3 a 3

Secondly, recalling the eigenvalue equation of ψ0 (r) will help us to calculate the expectation
value of H for the trial function,

~2 2 e2 1
 
− ∇ − ψ0 (r) = E1 ψ0 (r)
2m 4π0 r

8
Consequently, for the Hamiltonian of the Hydrogen molecule ion
~2 2 e2
  
1 1
Ĥψ = A − ∇ − + [ψ0 (r) − ψ0 (r0 )]
2m 4π0 r r0
 2  
e 1 1 0
= E1 ψ − A ψ0 (r) − ψ0 (r )
4π0 r0 r
And, for the expectation value
 2         
2 e 1 1 0
hHi = E1 − 2 |A| ψ0 (r) 0 ψ0 (r) − ψ0 (r) ψ0 (r )
   
4π0 r r
Using the direct integral, D and the exchange integral, X defined as
     
1 a  a  −2R/a 1 1 −2x
D ≡ a ψ0 (r) 0 ψ0 (r) = − 1 +
  e = − 1+ e
r R R x x
     
1 0 R
X ≡ a ψ0 (r) ψ0 (r ) = 1 + e−R/a = (1 + x) e−x
r a
Where x ≡ R/a. Allowing us to write hHi in terms of E1 getting
 
(D − X)
hHi = 1 + 2 E1
(1 − I)
Nevertheless, this energy, is only from the electrons, there is also potential energy associated
with the proton-proton repulsion
e2 1 2a 2
Vpp = = − E1 = − E1
4π0 R R x
Combining all this information and the variational principle, we can obtain than the total
energy of the system, in units of −E1 is less than
hHi + Vpp
F− (x) =
−E1
"
#
− x1 + 1 + 1
e−2x + (1 + x) e−x


x 2
= −1 + 2 2 +
1 + x + x3

1− e−x x
 h i
 + (x + 1) e−2x 2 −x  −x x2
2 −1
 + (x + x ) e +1 − e 1+x+ 3
= −1 +  2

1 − e−x 1 + x + x3
 
x
   
−2x 2 x2 −x
2 (x + 1) e + x
 + x − 1 −x
  − 3
e
= −1 +  2

1 − e−x 1 + x + x3

x

" #
2 (x + 1) e−2x + 23 x2 − 1 e−x


⇒ F− (x) = −1 + 2
1 − e−x 1 + x + x3

x

9
To plot this function, we use Wolfram Mathematica 12.1 as graphic utility, and obtain the
following result:

Figure 1: F+ (x) and F− (x) represent a plus and minus sign respectively in their trial function.

As can be seen in the figure, there is no evidence of bonding, given that the energy is always
grater than −1, meaning that it does have more energy than a neutral atom plus a free
proton. However, we have to remember that the variational principle is an upper bound so
it does not guarantee that bonding do not happen, but it this case, it does not give a clue
either.

10