Arnold Sommerfeld Center LudwigMaximiliansUniversität München

Prof. Dr. Viatcheslav Mukhanov Winter term 2019/2020

Exercises on Quantum Mechanics II (TM1/TV)

Solution 2, discussed October 28 - November 1, 2019

Exercise 9 (central tutorial)

(i) Show that if ÂB̂ = 1̂ = Ĉ Â then we have
B̂ = Â−1 = Ĉ.

Remember that the inverse operator Â−1 satises |ψi = Â |χi if and only if |χi = Â−1 |ψi.
(ii) Given an example of operators  and B̂ in Hilbert space for which ÂB̂ = 1̂ holds but for which
B̂ Â 6= 1̂.
 −1
(iii) Let  be an operator such that Â2 = λ1̂ where λ 6= 1 is a complex number. Write  + 1̂ explicitly
in terms of Â.

Solution
(i) To prove that B̂ = Ĉ we can multiply the rst equation by Ĉ from the left:
(1) (2)
Ĉ = Ĉ(ÂB̂) = (Ĉ Â)B̂ = B̂.

To show that B̂ = Ĉ is the inverse operator to Â, we rst assume that |ψi = Â |χi and act on this
with Ĉ and nd |χi = Ĉ |ψi which shows one direction of the equivalence. The other direction can be
shown analogously by acting with B̂ .
(ii) We can consider a rescaled version of ladder operator in the harmonic oscillator. Remember that there
we have an orthonormal basis of states |ni where n = 0, 1, 2, . . . and creation and annihilation operator
which act as √ √
↠|ni = n + 1 |n + 1i , â |ni = n |n − 1i .
Consider now modied ladder operators such that
b̂† |ni = |n + 1i

and (
|n − 1i , n > 0
b̂ |ni =
0, n=0

We can now choose  = b̂ and B̂ = b̂† . Acting on any basis state we nd
ÂB̂ |ni = b̂b̂† |ni = b̂ |n + 1i = |ni

so ÂB̂ = 1̂ but acting on the ground state |0i we have

B̂ Â |0i = b̂† b̂ |0i = 0

so  and B̂ are not inverse of each other.

(iii) Since  squares to something proportional to identity, any function of  that can be written as power
series in  can be reduced to the form
α1̂ + β Â.
Let's try this ansatz for an inverse of Â:
  −1   
1̂ =  + 1̂  + 1̂ =  + 1̂ α1̂ + β  = α1̂ + α + β  + β Â2

The left and right hand side are equal if α and β satisfy
α + β = 0, 1 = α + βλ.

These equations have unique solution

1 1
α= , β=−
1−λ 1−λ
which gives us a candidate for an inverse operator
 −1 1̂ − Â
 + 1̂ = .
1−λ
One can now easily check that this operator is not only right inverse (which is true by our construction)
but also a left inverse.
Alternative solution: we can also use the geometric series to nd the inverse operator. We have
 −1 ∞  k ∞ ∞
1̂ − Â
 + 1̂
X X X
= −Â = λl + (−Â) λm =
m=0
1−λ
k=0 l=0

where in the middle part we split the sum over all integers to a sum over all even integers 2l plus a
sum over all odd integers 2m + 1.

Exercise 10
Consider the operator  = dx .
d

(i) Use the Taylor expansion to nd out how eα acts on wavefunctions. Interpret the result physically.
(ii) How do operators B̂ ≡ sinh(αÂ) and Ĉ ≡ sin(αÂ) act on wavefunctions?

Solution
(i) Acting with eα on a function ψ(x) we have
∞  k
α d
α dx
X αk d
e ψ(x) = e ψ(x) = ψ(x) = ψ(x + α)
k! dx
k=0

so the resulting operator corresponds to a space translation of the wavefunction.

(ii) We have
  eα − e−αÂ
B̂ = sinh α = .
2
By the result of the previous exercise this acts on ψ(x) as
1
B̂ψ(x) = (ψ(x + α) − ψ(x − α)) .
2
Replacing α → iα and dividing by i we nd
i
Ĉψ(x) = (ψ(x − iα) − ψ(x + iα)) .
2
Exercise 11
Find ÂE† forD Â = |ϕi
D E hψ|. Remember that  is dened such that for any two vectors |ψi and |χi we have
†

ψ|Âχ = † ψ|χ .

Solution By denition of the conjugate operator we have

hχ| † |ηi =  |χi , |ηi = hψ|χi |ϕi , |ηi = hχ|ψi hϕ|ηi

Comparing both sides we see that

† = |ψi hϕ| .

Exercise 12
(i) Show that for an orthonormal basis |δj i we have the completeness relation
|δk i δ k  = 1̂.
X


(ii) Consider a product of two operators Ĉ = ÂB̂ . Remember

 that the matrix elements of the operator Â
in the orthonormal basis |δj i were dened as Aj k ≡ δ j  Â |δk i. Show that the components of Ĉ are
given by the usual product of matrices.
(iii) Show that the operator  can be reconstructed from its components via
X
Aj k |δj i δ k 


 =
jk

Solution
(i) Since we have an orthonormal basis, any vector |ψi can be expanded as
X
|ψi = aj |δj i
j

and  X
k 
δ k |ψ =

aj δ |δj = ak .
j

Let us now apply k δ k hδk | to an any vector |ψi

P  

X  X
|δk i δ k |ψ =

ak |δk i = |ψi .
k k

Since this holds for all vectors, we see that k δ k hδk | = 1̂.
P  

(ii) We have X
 X
(Ĉ)j l = δ j  ÂB̂ |δl i = δ j  Â |δk i δ k  B̂ |δl i = Aj k B k l




k k
which is the usual rule for matrix multiplication. In the middle equation we inserted the completeness
relation.
(iii) Plugging in the expression for components
X

 X
δ j  Â |δk i |δj i δ k  = |δj i δ j  Â |δk i δ k  = Â.




j,k j,k

Exercise 13
It was shown in the lecture that the matrix elements of the conjugate operator are

j † h
 i∗
δ  Ĉ |δl i = δ l  Ĉ |δj i ,

i.e. the matrix of components is complex conjugate transpose. Use this to show that (ÂB̂)† = B̂ † † .
Solution Let us calculate

j  † h
 i∗ X h
 i∗ X h
 i∗ h
 i∗
δ  ÂB̂ |δl i = δ l  ÂB̂ |δj i = δ l  Â |δk i δ k  B̂ |δj i = δ l  Â |δk i δ k  B̂ |δj i



k k
X h
 i h
 i X

δ k  † |δl i δ j  B̂ † |δk i = δ j  B̂ † |δk i δ k  † |δl i


=
k k

= δ j  B̂ † † |δl i .



Exercise 14 (central tutorial)

 E D E
Consider a change of orthonormal basis |δi → δ̃ in the Hilbert space described by U j k ≡ δ̃ j |δk .

(i) Show that U j k are components of unitary matrix, i.e. (U † )j k = (U k j )∗ = (U −1 )j k .

(ii) Show that the components of bra vectors in the old and in the new basis are related by
X
ψ̃k = ψj (U † )j k
j

(iii) Show that the matrix elements of operators transform as

X
Ãjm = U j k Ak l (U † )l m .
kl

Solution
(i) We have D E D E∗
δ j |δ̃k = δ̃ k |δj = (U k j )∗ = (U † )j k .

To see that this is an inverse of U , calculate

XD ED E D E
δ̃ j |δl δ l |δ̃k = δ̃ j |δ̃k = δkj = (1̂)jk .
X
(U U † )j k = U j l (U † )l k =
l l

Multiplication in opposite order is analogous.

(ii) Expressing the bra vector in two bases,
X D  X
 X D ED 
ψ̃k δ̃ k  = hψ| = ψj δ j  = ψj δ j |δ̃k δ̃ k 
 
k j j,k

Comparing both sides we see that

X D E X
ψ̃k = ψj δ j |δ̃k = ψj (U † )j k
j j

(iii) The matrix elements of  in the new basis are

D   E XD E
 D E X
Ãjm = δ̃ j  Â δ̃m = δ̃ j |δk δ k  Â |δl i δ l |δ̃m = U j k Ak l (U † )l m .
 
k,l k,l

Exercise 15
Consider a Hermitian operator  and a unitary operator Û .
(i) Show that the trace of the operator  is independent of the choice of the basis. What property of the
trace follows from the hermiticity of Â?
(ii) How are spectra of  and of Û ÂÛ † related?
Solution
(i) Let us calculate the trace of  in the new basis
X D   E X D E
 D E X
X

δ̃ j  Â δ̃j = δ̃ j |δk δ k  Â |δl i δ l |δ̃j = δ l |δk δ k  Â |δl i = δ k  Â |δk i .


j j,k,l k,l k

In particular we can choose a basis which diagonalizes Â. In this basis the eigenvalues are real (because
 is Hermitian) so also their sum, i.e. the trace is real.

(ii) Recall that λ ∈ C is in the spectrum of  if there exists a vector |ψi such that
 |ψi = λ |ψi .

Acting on this equation with Û from the left, we nd that the vector |ηi ≡ Û |ψi satises the equation
Û ÂÛ † |ηi = Û ÂÛ † Û |ψi = λÛ |ψi = λ |ηi ,

i.e. is an eigenvector of Û ÂÛ † with the same eigenvalue λ. Since Û is unitary (and so in particular
invertible), this provides a one-to-one correspondence between eigenvectors of  and Û ÂÛ † .

Exercise 16
Consider a linear
  operator acting on a Hilbert space such that it maps one orthonormal basis into another
one, Û |δj i = δj0 . How can you write this operator in terms of basis vectors? Find its hermitian conjugate.

Solution
(i) If we consider action of |δk0 i δ k  on basis vectors,
P

k
X
|δk0 i δ k |δj = δj0

  

which is exactly how Û acts.

(ii) By the denition of conjugate operator, we have

j  †
0j 
δ  Û = δ 

which is the same as the action of k |δk i δ 0k ,

P


j X
|δk i δ 0k  = δ 0j  .



δ 
k

Exercise 17 (central tutorial)

The position operator x̂ is hermitian. The momenum operator satises the commutation relation
[x̂, p̂] = i~1̂.

Does this imply that p̂ is a hermitian operator? Can there exist nite dimensional matrices x̂ and p̂ which
satisfy these commutation relations?

Solution
(i) The commutation relations don't force p̂ to be hermitian. If we had a hermitian solution (like the
standard p̂ = −i~ ∂x
∂
) we can add to it any complex multiple of x̂ without spoiling the commutation
relations. Now we have
†
(p̂ + αx̂) = p̂† + α∗ x̂† = p̂ + α∗ x̂
which agrees with p̂ + αx̂ only if α is real. So taking α with non-vanishing imaginary part is a counter-
example to the question.
(ii) There cannot exist any nite-dimensional matrices x̂ and p̂ which would satisfy the canonical com-
mutation relations. The reason for it is that the trace of any commutator vanishes by cyclicity of the
trace, h i    
Tr Â, B̂ = Tr ÂB̂ − B̂ Â = Tr ÂB̂ − ÂB̂ = 0

while the trace of the identity matrix on the right-hand side is the dimension of the vector space. We
thus nd an equation
0 = Tr [x̂, p̂] = i~ Tr 1̂ = i~n
which is a contradiction with nite dimensionality of the vector space.

Exercise 18 (central tutorial)

Check by direct calculation that
Z h i
00 00
dq 00 X q q00 P q q0 − P q q00 X q q0 = i~δqq0 = i~δ(q − q 0 )

where the matrix elements of X̂ are X q q0 ≡ hq| X̂ |q 0 i and similarly for P̂ .

Solution We can rewrite the formula in bra-ket notation as

Z h i
dq 00 hq| X̂ |q 00 i hq 00 | P̂ |q 0 i − hq| P̂ |q 00 i hq 00 | X̂ |q 0 i = i~ hq|q 0 i = i~δ(q − q 0 )

Now we have the basic identication between vectors and wave functions
ψ(q) = hq|ψi

and so
hq| X̂ |ψi = qψ(q), hq| P̂ |ψi = −i~∂q ψ(q).
Choosing |ψi = |q i so that ψ(q) = hq|q i = δ(q − q ) we nd the matrix elements
0 0 0

hq| X̂ |q 0 i = qδ(q − q 0 ), hq| P̂ |q 0 i = −i~∂q δ(q − q 0 ) = −i~δ 0 (q − q 0 ).

We can now use this in the equation above

Z h i
I= dq 00 hq| X̂ |q 00 i hq 00 | P̂ |q 0 i − hq| P̂ |q 00 i hq 00 | X̂ |q 0 i
Z
= dq 00 [qδ(q − q 00 )(−i~)δ 0 (q 00 − q 0 ) − (−i~)δ 0 (q − q 00 )q 00 δ(q 00 − q 0 )]

= q(−i~)δ 0 (q − q 0 ) − (−i~)δ 0 (q − q 0 )q 0 = −i~(q − q 0 )δ 0 (q − q 0 )

= i~δ(q − q 0 ).

We used the relation xδ 0 (x) = −δ(x) which follows by acting on test function φ(x),
Z Z Z Z
d
xδ 0 (x)φ(x)dx = − δ(x) (xφ(x))dx = − δ(x) [φ(x) + xφ0 (x)] dx = − δ(x)φ(x)dx
dx