Department of Physics Indian Institute of Technology Madras

PH5842: Adv. QCQI 2025 — Assignment 0 Solutions

Q1: Distance between states on the Bloch sphere

(i) We are given that the density matrices for the pure states |ψ⟩ and |ϕ⟩ can be
written as:
I + ⃗r · ⃗σ I + ⃗s · ⃗σ
ρψ = |ψ⟩⟨ψ| = and ρϕ = |ϕ⟩⟨ϕ| =
2 2
where ⃗r and ⃗s are the respective Bloch vectors and ⃗σ = (σx , σy , σz ) is the vector
of Pauli matrices.
We want to find a relationship between |⟨ψ|ϕ⟩|2 and ⃗r ·⃗s. We can use the fact that:

|⟨ψ|ϕ⟩|2 = ⟨ψ|ϕ⟩⟨ϕ|ψ⟩ = Tr(|ψ⟩⟨ψ||ϕ⟩⟨ϕ|) = Tr(ρψ ρϕ )

Now, substitute the expressions for ρψ and ρϕ :

 
2 I + ⃗r · ⃗σ I + ⃗s · ⃗σ 1
|⟨ψ|ϕ⟩| = Tr = Tr ((I + ⃗r · ⃗σ )(I + ⃗s · ⃗σ ))
2 2 4

Expanding the product:

1
|⟨ψ|ϕ⟩|2 = Tr (I + ⃗r · ⃗σ + ⃗s · ⃗σ + (⃗r · ⃗σ )(⃗s · ⃗σ ))
4
Now, we can use the linearity of the trace:
1
|⟨ψ|ϕ⟩|2 = (Tr(I) + Tr(⃗r · ⃗σ ) + Tr(⃗s · ⃗σ ) + Tr((⃗r · ⃗σ )(⃗s · ⃗σ )))
4
We know the following properties:
• Tr(I) = 2 (for 2x2 identity matrix)
• Tr(⃗r ·⃗σ ) = 0 (trace of any Pauli matrix is zero, and ⃗r ·⃗σ is a linear combination
of Pauli matrices with real coefficients)
• Tr(⃗s · ⃗σ ) = 0 (same as above)
• Tr((⃗r · ⃗σ )(⃗s · ⃗σ )) = 2(⃗r · ⃗s) ...as shown in the appendix
Substituting these values:
1 1 1
|⟨ψ|ϕ⟩|2 = (2 + 0 + 0 + 2(⃗r · ⃗s)) = (2 + 2⃗r · ⃗s) = (1 + ⃗r · ⃗s)
4 4 2
Therefore, we have shown that:
1
|⟨ψ | ϕ⟩|2 = (1 + ⃗r · ⃗s)
2

1
 (ii) First, let’s find the difference between the two density operators:

I + ⃗r · ⃗σ I + ⃗s · ⃗σ (⃗r − ⃗s) · ⃗σ
ρ1 − ρ2 = − =
2 2 2

Now, we need to find |ρ1 −ρ2 |, which is the positive square root of (ρ1 −ρ2 )† (ρ1 −ρ2 ).
Since ρ1 and ρ2 are Hermitian, ρ1 − ρ2 is also Hermitian, so (ρ1 − ρ2 )† = ρ1 − ρ2 .
Therefore:
s 2 r
p (⃗
r − ⃗
s ) · ⃗
σ 1
||ρ1 − ρ2 ||1 = (ρ1 − ρ2 )2 = = ((⃗r − ⃗s) · ⃗σ )2
2 4

Let d⃗ = ⃗r − ⃗s. Then we need to evaluate (d⃗ · ⃗σ )2 . Using the identity (⃗a · ⃗σ )(⃗b · ⃗σ ) =
(⃗a · ⃗b)I + i(⃗a × ⃗b) · ⃗σ , we have:

(d⃗ · ⃗σ )2 = (d⃗ · d)I

⃗ + i(d⃗ × d)
⃗ · ⃗σ = |d|
⃗ 2 I + 0 = |d|
⃗ 2 I = |⃗r − ⃗s|2 I

Thus,
|⃗r − ⃗s| √
r
1 |⃗r − ⃗s|
||ρ1 − ρ2 ||1 = |⃗r − ⃗s|2 I = I= I
4 2 2
Now, we take the trace:
 
|⃗r − ⃗s| |⃗r − ⃗s| |⃗r − ⃗s|
Tr||ρ1 − ρ2 ||1 = Tr I = Tr(I) = (2) = |⃗r − ⃗s|
2 2 2

Finally, we substitute this back into the expression for the trace distance:
1 1
D(ρ1 , ρ2 ) = Tr||ρ1 − ρ2 ||1 = |⃗r − ⃗s|
2 2

Therefore, the trace distance between two qubit density operators is:
1
D(ρ1 , ρ2 ) = |⃗r − ⃗s|
2

Alternative: Refer appendix for proof to ∥(r − s) · σ∥1 = 2|⃗r − ⃗s| result.

Q2: Entanglement measure

(i) First, find the complex conjugate of the state |Ψ⟩:

|Ψ∗ ⟩ = a∗ |00⟩ + b∗ |01⟩ + c∗ |10⟩ + d∗ |11⟩

Next, we need to evaluate σY ⊗ σY . Recall that:

 
0 −i
σY =
i 0

2
 Therefore,
 
    0 0 0 −1
0 −i 0 −i  0 0 1 0
σY ⊗ σY = ⊗ = 
i 0 i 0 0 1 0 0
−1 0 0 0

Now, we need to compute ⟨Ψ|(σY ⊗ σY )|Ψ∗ ⟩. It is easiest to write out the vectors
and matrix explicitly:
⟨Ψ| = a∗ b∗ c∗ d∗


   ∗  ∗
0 0 0 −1 a −d
∗
0 0 1 0   b   c∗ 
(σY ⊗ σY )|Ψ∗ ⟩ = 
    
 0 1 0 0   c∗  =  b∗ 


−1 0 0 0 d∗ −a∗
Then,

−d∗
 

 c∗ 
⟨Ψ|(σY ⊗σY )|Ψ∗ ⟩ = a∗ b∗ c∗ d∗  ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗
 b∗  = −a d +b c +c b −d a = 2(b c −a d )


−a∗

Finally, we take the absolute value:

C(|Ψ⟩) = |⟨Ψ|(σY ⊗ σY )|Ψ∗ ⟩| = |2(b∗ c∗ − a∗ d∗ )| = 2|bc − ad|

Now, let us look at the determinant of the matrix M :

 
a b
M=
c d

det(M ) = ad − bc
| det(M )| = |ad − bc| = |bc − ad|
Comparing this to our expression for C(|Ψ⟩), we see:

C(|Ψ⟩) = 2| det(M )|

Therefore, C(|Ψ⟩) is proportional to the absolute value of the determinant of the

matrix M . The constant of proportionality is 2.

(ii) First, recall the definition of the bipartite state and the matrix M:

|Ψ⟩AB = a|00⟩AB + b|01⟩AB + c|10⟩AB + d|11⟩AB

 
a b
M= .
c d

3
To find the reduced density matrix ρA , we need to trace out subsystem B from the
density matrix ρ = |Ψ⟩⟨Ψ|. Let’s start by writing out ρ:

ρ = |Ψ⟩⟨Ψ| = (a|00⟩ + b|01⟩ + c|10⟩ + d|11⟩)(a∗ ⟨00| + b∗ ⟨01| + c∗ ⟨10| + d∗ ⟨11|)

Expanding this, we get:

ρ =|a|2 |00⟩⟨00| + ab∗ |00⟩⟨01| + ac∗ |00⟩⟨10| + ad∗ |00⟩⟨11|

+ ba∗ |01⟩⟨00| + |b|2 |01⟩⟨01| + bc∗ |01⟩⟨10| + bd∗ |01⟩⟨11|
+ ca∗ |10⟩⟨00| + cb∗ |10⟩⟨01| + |c|2 |10⟩⟨10| + cd∗ |10⟩⟨11|
+ da∗ |11⟩⟨00| + db∗ |11⟩⟨01| + dc∗ |11⟩⟨10| + |d|2 |11⟩⟨11|

Now, we trace out subsystem B. Recall that T r(|0⟩⟨0|) = T r(|1⟩⟨1|) = 1 and

T r(|0⟩⟨1|) = T r(|1⟩⟨0|) = 0. Also note that T rB acts only on the B part of the
tensor product and leaves A untouched:

ρA = TrB [ρ] = (|a|2 +|b|2 )|0⟩⟨0|+(ac∗ +bd∗ )|0⟩⟨1|+(ca∗ +db∗ )|1⟩⟨0|+(|c|2 +|d|2 )|1⟩⟨1|

In matrix form, this is:

|a| + |b|2 ac∗ + bd∗

 2 
ρA =
ca∗ + db∗ |c|2 + |d|2

Now, let’s compute M M † :

   ∗ ∗
a b † a c
M= , M = ∗
c d b d∗

  ∗ ∗  2
|a| + |b|2 ac∗ + bd∗
 
a b
† a c
MM = =
c d b∗ d∗ ca∗ + db∗ |c|2 + |d|2
Therefore, ρA = M M † .
Entanglement and Rank: A pure bipartite state is entangled if it cannot be
written as a product state |ψ⟩A ⊗|ϕ⟩B . The Schmidt decomposition theorem states
that any pure bipartite state can be written as:
X√
|Ψ⟩ = pi |i⟩A |i⟩B
i

The reduced density matrix ρA has eigenvalues equal to pi . The number of non-zero
Schmidt coefficients is called the Schmidt rank of the state. A state is entangled if
its Schmidt rank is greater than 1. The rank of the reduced density matrix ρA is
equal to the Schmidt rank of the state. Therefore, |Ψ⟩AB is entangled if and only
if the reduced state ρA has a rank greater than 1.
Concurrence and Entanglement: From part (i), we know that C(|Ψ⟩) = 2|ad−
bc|. This is twice the absolute value of the determinant of the matrix M. If the

4
 state is separable (not entangled), it can be written in a product form, which
implies that the determinant of M is zero (ad = bc), and therefore C(|Ψ⟩) = 0.
If the state is entangled, the determinant is non-zero (ad ̸= bc), and therefore
C(|Ψ⟩) > 0. Therefore, the concurrence C(|Ψ⟩) is non-zero if and only if the state
|Ψ⟩ is entangled.

Q3: Positive Operator Valued Measures (POVMs)

Let, ⟨ψ1 |ψ2 ⟩ = c =⇒ 0 ≤ c ≤ 1
Completeness: Let’s first check if the operators sum to identity:

M1 + M2 + M3 = R(I − |ψ1 ⟩⟨ψ1 |) + R(I − |ψ2 ⟩⟨ψ2 |) + (1 − 2R)I + R(|ψ1 ⟩⟨ψ1 | + |ψ2 ⟩⟨ψ2 |)

= RI −R|ψ1 ⟩⟨ψ1 |+RI −R|ψ2 ⟩⟨ψ2 |+I −2RI +R|ψ1 ⟩⟨ψ1 |+R|ψ2 ⟩⟨ψ2 | = 2RI −2RI +I = I
So, M1 + M2 + M3 = I.
Positivity of M1 and M2 : In |ψ1 ⟩ and |ψ2 ⟩ basis,

R(1 − c2 ) 0
   
0 0
M1 = , M2 =
0 R(1 − c2 ) 0 0

The eigenvalues of M1 and M2 are 0 and R(1−c2 ) ≥ 0, are non-negative by construction.

Positivity of M3 : In |ψ1 ⟩ and |ψ2 ⟩ basis,

1 − R + Rc2
 
c
M3 =
c 1 − R + Rc2

Eigenvalues of M3 are {1 − c − R + Rc2 , 1 + c − R + Rc2 }. To ensure the least eigenvalue

to be non-negative, we condition
1 1
1 − c − R + Rc2 ≥ 0 =⇒ R ≤ = , Since 0≤c≤1
1+c 1 + ⟨ψ1 |ψ2 ⟩

Q4: Evedropping and disturbance

(i) First, we need to ensure that the POVM operators are positive semi-definite. The
key constraint comes from Eϕ as the other two are manifestly positive. If A is
too large, Eϕ will have negative eigenvalues, invalidating the POVM. To find the
maximum allowed value for A, we can find the minimum eigenvalue of Eϕ and set
it to zero. This is the boundary where the operator remains positive semi-definite.
Let’s find |u⟩⟨u| and |v⟩⟨v|.

cos2 α
   
cos α
cos α sin α
|u⟩⟨u| = cos α sin α =
sin α cos α sin α sin2 α

sin2 α
   
sin α
sin α cos α
|v⟩⟨v| = sin α cos α =
cos α sin α cos α cos2 α

5
Therefore,

cos α + sin2 α
 2   
2 sin α cos α 1 sin 2α
|u⟩⟨u| + |v⟩⟨v| = =
2 sin α cos α sin2 α + cos2 α sin 2α 1

Now, we can write Eϕ as:

     
1 0 1 sin 2α 1 − A A sin 2α
Eϕ = (1 − 2A) +A =
0 1 sin 2α 1 A sin 2α 1 − A

The eigenvalues of this matrix can be found by solving the characteristic equation:

|Eϕ − λI| = 0

1 − A − λ A sin 2α
= (1 − A − λ)2 − (A sin 2α)2 = 0
A sin 2α 1 − A − λ
(1 − A − λ) = ±A sin 2α
λ = 1 − A ± A sin 2α = 1 − A(1 ∓ sin 2α)

So, the eigenvalues are:

λ1 = 1 − A(1 − sin 2α) and λ2 = 1 − A(1 + sin 2α)

For Eϕ to be positive semi-definite, both eigenvalues must be non-negative. There-

fore, the smallest eigenvalue must be non-negative. Since 1 + sin 2α > 1 − sin 2α,
λ2 is the smallest eigenvalue. Thus we require:
1
1 − A(1 + sin 2α) ≥ 0 =⇒ A ≤
1 + sin 2α

Therefore, the maximum value of A is:

1
A=
1 + sin 2α

With this value of A, we can calculate the probabilities of outcomes ū, v̄, and ϕ
when Anita sends |u⟩ or |v⟩ with equal probability.
The probability of outcome ϕ when Anita sends |u⟩ is:
 
1 − A A sin 2α
Pr(ϕ|u) = ⟨u|Eϕ |u⟩ = ⟨u| |u⟩
A sin 2α 1 − A
  

1 − A A sin 2α cos α
= cos α sin α = (1−A)(cos2 α+sin2 α)+2A sin 2α sin α cos α
A sin 2α 1 − A sin α
1 sin2 2α
= 1 − A + A sin2 2α = 1 − +
1 + sin 2α 1 + sin 2α

6
 Similarly, the probability of outcome ϕ when Anita sends |v⟩ is:

1 sin2 2α
Pr(ϕ|v) = ⟨v|Eϕ |v⟩ = 1 − A + A sin2 2α = 1 − +
1 + sin 2α 1 + sin 2α

The average probability of obtaining outcome ϕ is:

1 1 sin2 (2α) sin(2α)

P (ϕ) = Pr(ϕ|u) + Pr(ϕ|v) = 1 − A + A sin2 2α = =
2 2 1 + sin(2α) 1 + sin(2α)

(ii) We are given that Anita sends either |u⟩ or |v⟩ with equal probability, where
   
cos α sin α π
|u⟩ = , |v⟩ = , 0≤α≤ .
sin α cos α 4

Sunil intercepts the qubit and measures it in the {|0⟩, |1⟩} basis. If Sunil gets
outcome 0, he forwards |u⟩ to Bharat. If he gets outcome 1, he forwards |v⟩ to
Bharat. Bharat performs his POVM and gets a conclusive result (ū or v̄). We
want to find the probability that Bharat’s conclusive outcome is different from
the state Anita originally sent.
Case 1: Anita sends |u⟩
• Sunil measures |u⟩ in the {|0⟩, |1⟩} basis. The probability of getting outcome
0 is |⟨0|u⟩|2 = cos2 α. In this case, Sunil sends |u⟩ to Bharat, and Bharat
obtains the correct conclusive outcome v̄.
• The probability of Sunil getting outcome 1 is |⟨1|u⟩|2 = sin2 α. In this case,
Sunil sends |v⟩ to Bharat. For Bharat to have a conclusive outcome and for it
to be the wrong one (ū), he needs to obtain the outcome Eū . The probability
of this happening is A(I − |u⟩⟨u|)|v⟩ = A|⟨v|v⟩ − |⟨u|v⟩|2 = A(1 − sin2 (2α)).
Therefore, when Anita sends |u⟩, the probability of Bharat getting a wrong con-
clusive outcome is:

P (Error | Anita sends |u⟩) = sin2 (α) · A(1 − sin2 (2α)).

Case 2: Anita sends |v⟩

• Sunil measures |v⟩ in the {|0⟩, |1⟩} basis. The probability of getting outcome
1 is |⟨1|v⟩|2 = cos2 α. In this case, Sunil sends |v⟩ to Bharat, and Bharat
obtains the correct conclusive outcome ū.
• The probability of Sunil getting outcome 0 is |⟨0|v⟩|2 = sin2 α. In this case,
Sunil sends |u⟩ to Bharat. For Bharat to have a conclusive outcome and for it
to be the wrong one (v̄), he needs to obtain the outcome Ev̄ . The probability
of this happening is A(I − |v⟩⟨v|)|u⟩ = A|⟨u|u⟩ − |⟨v|u⟩|2 = A(1 − sin2 (2α)).

7
 Therefore, when Anita sends |v⟩, the probability of Bharat getting a wrong con-
clusive outcome is:

P (Error | Anita sends |v⟩) = sin2 α · A(1 − sin2 (2α)).

Total Error Probability

Since Anita sends |u⟩ and |v⟩ with equal probability (1/2), the total probability of
Bharat making an error (getting a conclusive outcome that is wrong) is:
1 1
P (Error) = · P (Error | Anita sends |u⟩) + · P (Error | Anita sends |v⟩)
2 2
1 1
P (Error) = · [sin2 α · A(1 − sin2 (2α))] + · [sin2 α · A(1 − sin2 (2α))]
2 2
P (Error) = A · sin2 α · (1 − sin2 (2α)).

Therefore, the probability of Bharat getting a conclusive outcome that differs from
what Anita sent is A · sin2 α · (1 − sin2 (2α)).

Q5: Schmidt Decomposition

(i) To show that the states |ĩ⟩B are mutually orthogonal, we need to show that ⟨ĩ|k̃⟩ =
0 for i ̸= k.
First, let’s write out the inner product:
 † !
X X X
⟨ĩ|k̃⟩ =  cij |j⟩B  ckl |l⟩B = c∗ij ckl ⟨j|l⟩B
j l j,l

Since {|j⟩B } is an orthonormal basis, ⟨j|l⟩B = δjl . Thus,

X X
⟨ĩ|k̃⟩ = c∗ij ckl δjl = c∗ij ckj
j,l j

Now, consider the reduced density matrix ρA :

  
X X
ρA = TrB [|ψ⟩⟨ψ|] = TrB  cij |i⟩A |j⟩B   c∗kl ⟨k|A ⟨l|B 
i,j k,l

X X X
ρA = cij c∗kl |i⟩A ⟨k|A TrB [|j⟩B ⟨l|B ] = cij c∗kl |i⟩A ⟨k|A δjl = cij c∗kj |i⟩A ⟨k|A
i,j,k,l i,j,k,l i,j,k

8
 We are given that {|i⟩A } is the basis in which ρA is diagonal. This means that
⟨i|ρA |k⟩ = λi δik , where λi are the eigenvalues of ρA . Therefore,
* +
X X X
⟨i|ρA |k⟩ = i ci′ j c∗k′ j |i′ ⟩A ⟨k ′ |A k = ci′ j c∗k′ j ⟨i|i′ ⟩A ⟨k ′ |k⟩A = cij c∗kj = λi δik
i′ ,j,k′ i′ ,j,k′ j

∗
P
Therefore, j cij ckj = λi δik . Taking the complex conjugate, we have:
X
c∗ij ckj = λ∗i δik = λi δik
j

Substituting this back into our expression for ⟨ĩ|k̃⟩:

X
⟨ĩ|k̃⟩ = c∗ij ckj = λi δik
j

This shows that the states |ĩ⟩B are orthogonal for i ̸= k since ⟨ĩ|k̃⟩ = 0 for i ̸= k.
They are not necessarily normalized, however.
Schmidt Decomposition: To find the Schmidt decomposition, we normalize the
states |ĩ⟩B :
√
Let |ĩ⟩B = λi |i⟩B . Note that since {|i⟩A } diagonalizes ρA , therefore, λi are
eigenvalues of ρA . Now, we rewrite the original state |ψ⟩ using the orthonormal
set |i⟩A and orthonormalized set |i⟩B :
X X X X X p Xp
|ψ⟩ = cij |i⟩A |j⟩B = |i⟩A cij |j⟩B = |i⟩A |ĩ⟩B = |i⟩A λi |i⟩B = λi |i⟩A |i⟩B
i,j i j i i i

This is the Schmidt decomposition

√ of the state |ψ⟩, where {|i⟩A } and {|i⟩B } are
the Schmidt bases, and λi are the Schmidt coefficients.

(ii) We will find the Schmidt form of the given states.

|00⟩+|01⟩+|11⟩
State |Ψ1 ⟩AB = √
3
Step 1: Write the state as a matrix We express |Ψ1 ⟩AB in terms of subsystem
A’s basis (|0⟩A , |1⟩A ) and subsystem B’s basis (|0⟩B , |1⟩B ):
1

|Ψ1 ⟩AB = √ |0⟩A (|0⟩B + |1⟩B ) + |1⟩A |1⟩B .
3
This corresponds to the following coefficient matrix MΨ1 :
 √ √ 
1/ 3 1/√3
MΨ1 = .
0 1/ 3

9
Step 2: Perform Singular Value Decomposition (SVD) Performing SVD
on MΨ1 (refer appendix to How to Perform Singular Value Decomposition (SVD)
for a Given Matrix for more details), we write:

MΨ1 = U ΣV † ,

where:
√ √
√
 
5+1 √1− 5  q 
q √ √ 1


2( 5+5) 10−2 5 6 5+3 0
U = q
, Σ= √
,
 
q √ q
1
√2 1
5+5 0 6 3− 5
5+5 10
 √ √ 
√ 5−1 5+1
√ −q √
10−2 5 2( 5+5)
V = q .
 
1
√ q
√2


10 5+5 5+5

√ √
The singular values ( λ1 , λ2 ) are:

1 √ √ 
r  r 
p  p 1
λ1 = 5+3 , λ2 = 3− 5 .
6 6

Step 3: Write the Schmidt form The Schmidt decomposition is:

1 √ √ 
r  r 

′ ′ 1
|Ψ1 ⟩AB = 5 + 3 |0 ⟩A |0̃ ⟩B + 3 − 5 |1′ ⟩A |1̃′ ⟩B ,
6 6
where: - The orthonormal basis for subsystem A and B is:
 ′     ′   
|0 ⟩A |0⟩ |0 ⟩B † |0⟩
=U , =V ,
|1′ ⟩A |1⟩ |1′ ⟩B |1⟩

|00⟩+|11⟩
State |Ψ2 ⟩AB = √
2
Step 1: Write the state in matrix form The given state can be represented
as a 2x2 matrix:  
1 1 0
C=√
2 0 1
Step 2: Perform SVD on the matrix
The SVD of matrix C is given by: C = U ΣV † . In this case, the SVD yields:
   
1 0 1 1 0
U =V = and Σ = √
0 1 2 0 1

Step 3: Interpret the results

The Schmidt decomposition is obtained from the SVD as follows:

10
 X
|ψ⟩ = λi |ui ⟩A |vi ⟩B
i

where λi are the singular values (diagonal elements of Σ), and |ui ⟩A and |vi ⟩B are
the corresponding column vectors of U and V † , respectively.
Therefore, the Schmidt decomposition of the given state is:

1 1
|ψ⟩ = √ |0⟩A |0⟩B + √ |1⟩A |1⟩B
2 2
This decomposition shows that the state has two non-zero Schmidt coefficients,
both equal to √12 , indicating that it is a maximally entangled state.
|00⟩+|01⟩+|10⟩+|11⟩
State |Ψ3 ⟩AB = 2
Step 1: Represent the Quantum State as a Matrix The state |ψ⟩ can be
written in terms of a coefficient matrix A, where each entry corresponds to the
coefficients of |i⟩A |j⟩B . For this state:

1
|ψ⟩ = (|00⟩ + |01⟩ + |10⟩ + |11⟩),
2
The coefficient matrix A is:
 
1/2 1/2
A= .
1/2 1/2

Step 2: Perform SVD on A The SVD of A is: A = U ΣV † ,

where: U and V are unitary matrices, Σ is adiagonal
√ matrix
√ containing the singular
1/√2 −1/√ 2
values. Performing SVD yields: U = V = , Σ = diag(1, 0).
1/ 2 1/ 2
Step 3: Write the Schmidt Decomposition The Schmidt decomposition of
|ψ⟩ is given by:

d−1
X
|ψ⟩ = σi |ui ⟩A |vi ⟩B ,
i=0

where: σi are the singular values (entries of Σ), |ui ⟩A , |vi ⟩B are the orthonormal
basis vectors derived from U, V † .
For this case:
Singular values: σ0 = 1, σ1 = 0, √ √
Basis vectors: |u0 ⟩A√= (|0⟩ + |1⟩)/ 2 = |+⟩, |u1 ⟩A√= (|0⟩ − |1⟩)/ 2 = |−⟩,
|v0 ⟩B = (|0⟩ + |1⟩)/ 2 = |+⟩, |v1 ⟩B = (|0⟩ − |1⟩)/ 2 = |−⟩.
Thus, the Schmidt decomposition is: |ψ⟩ = 1(|u0 ⟩A |v0 ⟩B ) = |+⟩ ⊗ |+⟩.

11
 This shows that the state is separable since there is only one non-zero Schmidt
coefficient.

(iii) We will evaluate the reduced state on subsystem HA , write down its eigenvalues,
and calculate the Shannon entropy for each of the states |Ψi ⟩AB from part (ii).
The reduced density matrices are:
     
(1) 2/3 1/3 (2) 1/2 0 (3) 1/2 1/2
ρA = , ρA = , ρA = .
1/3 1/3 0 1/2 1/2 1/2

Eigenvalues:
(1) √ √
• ρA : Eigenvalues are λ1 = 16 (3 + 5) and λ2 = 16 (3 − 5)
(2)
• ρA : Eigenvalues are λ1 = 1/2 and λ2 = 1/2.
(3)
• ρA : Eigenvalues are λ1 = 1 and λ2 = 0.
Shannon Entropy Calculation: We calculate the Shannon entropy for each
eigenvalue distribution using the formula:
X
H({λi }) = − λi ln(λi ).
i

• State |Ψ1 ⟩:
√  √  1 √ 1 √
     
1 1 
H1 = − 3 − 5 log2 3− 5 − 5 + 3 log2 5+3 ≈ 0.55
6 6 6 6

• State |Ψ2 ⟩:
      
1 1 1 1 1
H2 = − log2 + log2 = − log2 = log2 (2) = 1
2 2 2 2 2

• State |Ψ3 ⟩:
H3 = −λ1 log2 λ1 = − log2 (1) = 0

Entanglement Comparison: The Shannon entropy of the reduced density ma-

trix is a measure of entanglement. Higher entropy indicates greater entanglement.
Therefore:
• |Ψ2 ⟩ > |Ψ1 ⟩ > |Ψ3 ⟩ order of entanglement.
• |Ψ3 ⟩ has the least entanglement; In fact, is actually a product state, and is
not entangled state. |Ψ2 ⟩ has the maximal entanglement; In fact, is actually
a Bell state, and is maximally entangled state.
Alternative method: Refer appendix to Schmidt Decomposition of a bipartite
pure state. Read textbook Nielsen and Chuang chapter 2 section 5 for more details.

12
Appendix:

1 Proof: T r[(⃗r · ⃗σ )(⃗s · ⃗σ )] = 2(⃗r · ⃗s)

we use the properties of the Pauli matrices. Here’s the step-by-step proof:

Step 1: Expand the Dot Products

The dot products ⃗r · ⃗σ and ⃗s · ⃗σ are defined as:
⃗r · ⃗σ = rx σx + ry σy + rz σz , ⃗s · ⃗σ = sx σx + sy σy + sz σz .
Thus, the product is:
X
(⃗r · ⃗σ )(⃗s · ⃗σ ) = ri sj (σi σj ), where i, j = x, y, z
i,j

Step 2: Use the Pauli Matrix Product Rule

The product of two Pauli matrices satisfies the relation: σi σj = δij I + iϵijk σk , where: I
is the 2 × 2 identity matrix, δij is the Kronecker delta, ϵijk is the Levi-Civita symbol.
Substitute this into the expression for the product:
X
(⃗r · ⃗σ )(⃗s · ⃗σ ) = ri sj (δij I + iϵijk σk ).
i,j

Separate the terms:

X X
(⃗r · ⃗σ )(⃗s · ⃗σ ) = ( ri sj δij )I + i( ri sj ϵijk σk ).
i,j i,j,k

Step 3: Simplify Each Term

First Term:
The Kronecker delta simplifies to:
X X
ri sj δij = ri si = ⃗r · ⃗s.
i,j i

Thus, the first term becomes: (⃗r · ⃗s)I.

Second Term:
The Levi-Civita symbol sums to zero because it is antisymmetric and ri sj are symmetric
under exchange of indices. Thus:
X
i( ri sj ϵijk σk ) = 0.
i,j,k

So, the product reduces to: (⃗r · ⃗σ )(⃗s · ⃗σ ) = (⃗r · ⃗s)I.

13
Step 4: Take the Trace
The trace of a scalar multiple of the identity matrix is proportional to the scalar itself.
For a 2 × 2 identity matrix I, we have: Tr[cI] = 2c, where c = (⃗r · ⃗s).
Thus: Tr[(⃗r · ⃗σ )(⃗s · ⃗σ )] = 2(⃗r · ⃗s).

2 Proof, ∥(r − s) · σ∥1 = 2|⃗r − ⃗s|

To simplify trace norm: ∥(r − s) · σ∥1 , where (r − s) · σ = (rx − sx )σx + (ry − sy )σy +
(rz − sz )σz , we proceed as follows:

Step 1: Express the operator

The term (r − s) · σ is a Hermitian operator because it is a linear combination of the
Pauli matrices σx , σy , σz , which are Hermitian. The coefficients rx − sx , ry − sy , rz − sz
are real numbers.

Step 2: Eigenvalues of the operator

The eigenvalues of a Hermitian operator determine its trace norm. For qubit systems,
the eigenvalues of (r − s) · σ can be computed as follows:

1. Compute the magnitude of the Bloch vector difference:

q
|r − s| = (rx − sx )2 + (ry − sy )2 + (rz − sz )2 .

2. The eigenvalues of (r − s) · σ are given by: ±|r − s|.

Step 3: Trace norm

The trace norm of an operator A is the sum of the absolute values of its eigenvalues.
For (r − s) · σ, the eigenvalues are +|r − s| and −|r − s|. Thus, the trace norm is:

∥(r − s) · σ∥1 = | + |r − s|| + | − |r − s|| = 2|r − s|.

Step 4: Simplify the expression

Substitute this result into the trace distance formula:
1 1
D(ρ1 , ρ2 ) = ∥(r − s) · σ∥1 = (2|r − s|) = |r − s|.
2 2

Final Result:
The trace norm simplifies to: ∥(r − s) · σ∥1 = 2|r − s|, and the trace distance between
two qubit density operators is directly proportional to the Euclidean distance between
their Bloch vectors: D(ρ1 , ρ2 ) = |r − s|.

14
3 How to Perform Singular Value Decomposition (SVD) for a
Given Matrix
The SVD of a matrix A is related to the eigenvalue decomposition of A† A and AA† .
Here’s how to proceed:

Step 1:Form the matrices

• Compute A† A (a square n × n matrix).

• Compute AA† (a square m × m matrix).

Step 2: Find eigenvalues

• The eigenvalues of A† A and AA† are the same.

• The singular values of A, denoted as σi , are the square roots of the non-zero
eigenvalues of either A† A or AA† :
p
σi = λ i .

Step 3: Find eigenvectors

• The right-singular vectors (columns of V ) are the eigenvectors of A† A.

• The left-singular vectors (columns of U ) are the eigenvectors of AA† .

Step 4: Construct the SVD Components

Once you have the eigenvalues and eigenvectors:

• Matrix V : Form V = [v1 , v2 , ..., vn ], where each column is a normalized eigenvec-

tor of A† A.

• Matrix U : Form U = [u1 , u2 , ..., um ], where each column is a normalized eigen-

vector of AA† .

• Matrix Σ: Create a diagonal matrix Σ with singular values (σ1 , ..., σr , 0, ..., 0) on
the diagonal, where r = rank(A).

Step 5: Verify Orthogonality

Ensure that:
• Columns of U and V are orthonormal (U † U = Im , V † V = In ).

• Check that the decomposition satisfies:

A = U ΣV † .

15
Performing Singular Value Decomposition (SVD) from the Polar
Decomposition
√
The polar decomposition of a matrix A ∈ Cm×n expresses √ A as: A = U A† A, where:
U is a unitary (or orthogonal in the real case) matrix, A† A is a positive semi-definite
Hermitian matrix. Starting from the polar decomposition:
√ √ √ √
1. Given A = U A† A, decompose A† A = VD DVD† , where D is diagonal and
positive semi-definite.
√ √ √
2. Substitute into A = U A† A = U (VD D DVD† ) = (U VD ) D(VD† ), which gives the
SVD form: √
A = U ′ DVD† ,
where U ′ = U VD .

Alternatively, For a square matrix, the polar decomposition can also be written as:
A = RS, where: R is unitary, S is symmetric positive semi-definite.
The relationship between SVD and polar decomposition can be utilized to derive one
from the other:

1. Start with the SVD of A = UA ΣVA† .

2. Define the unitary matrix R = UA VA† .

3. Define the symmetric positive semi-definite matrix S = VA ΣVA† .

Thus, the polar decomposition becomes:

A = RS = (UA VA† )(VA ΣVA† ) = UA (ΣVA† ).

16