Tutorial11 Solution
Tutorial11 Solution
where [A, B] = AB − BA is the commutator of A and B, and the Levi-Civita symbol jk` is defined by
1
(j, k, `) is an even (cyclic) permutation of (1, 2, 3)
jk` = −1 (j, k, `) is an odd permutation of (1, 2, 3)
0 otherwise
Remark: The interpretation as rotation applies to an arbitrary single-qubit gate U (when ignoring global phases),
since it can always be represented as U = eiα R~n (θ) with α ∈ R and a suitable rotation operator R~n (θ).
Solution
(a) We first note that for j = k, the commutator is clearly zero, as is the Levi-Civita symbol.
For j = 1 and k = 2, by an explicit calculation,
0 1 0 −i 0 −i 0 1
[σ1 , σ2 ] = −
1 0 i 0 i 0 1 0
i 0 −i 0 2i 0
= − =
0 −i 0 i 0 −2i
3
X
= 2iσ3 = 2i 12` σ` ,
`=1
and similarly [σ2 , σ3 ] = 2iσ1 and [σ3 , σ1 ] = 2iσ2 (see also exercise 3.1). Finally, we note that an interchange
j ↔ k flips the sign of the commutator, [σj , σk ] = −[σk , σj ], and likewise the sign of jk` by definition. In
summary, we have verified the relation for all possible cases of j, k ∈ {1, 2, 3}.
1 M. A. Nielsen, I. L. Chuang: Quantum Computation and Quantum Information. Cambridge University Press (2010), Exercise 4.6
1
Expanding in terms of individual Pauli matrices leads to:
h i X 3 X3 a2 b3 − a3 b2 σ1
~a · ~σ , ~b · ~σ = aj bk [σj , σk ] = 2i aj bk jk` σ` = 2i a3 b1 − a1 b3 · σ2 = 2i(~a × ~b) · ~σ .
j,k=1 j,k,`=1 a1 b2 − a2 b1 σ3
(b) The statement follows from the fact that different Pauli matrices anti-commute, i.e., σj σk = −σk σj for j 6= k
(see exercise 3.1), and that squaring a Pauli matrix gives the identity:
n o 3
X 3
X
~a · ~σ , ~b · ~σ = aj bk {σj , σk } = aj bk δjk 2I = 2(~a · ~b)I.
j,k=1 j,k=1
(c) In general, applying a unitary matrix U to a quantum state |ψi corresponds to a conjugation of the density
matrix by U :
ρ 7→ U ρU † .
In our case, U = R~n (θ), and U † = R~n (−θ) (inverse rotation).
Inserting the Bloch representation of the density matrix leads to
I 1
R~n (θ)ρR~n (−θ) = + R~n (θ)(~r · ~σ )R~n (−θ)
2 2
I 1 i i
= + cos(θ/2)2 (~r · ~σ ) + cos(θ/2) sin(θ/2) (~r · ~σ )(~n · ~σ ) − cos(θ/2) sin(θ/2) (~n · ~σ )(~r · ~σ )
2 2 2 2
1 2
+ sin(θ/2) (~n · ~σ )(~r · ~σ )(~n · ~σ )
2
I 1 i 1
= + cos(θ/2)2 (~r · ~σ ) + cos(θ/2) sin(θ/2) [~r · ~σ , ~n · ~σ ] + sin(θ/2)2 (~n · ~σ )(~r · ~σ )(~n · ~σ ) .
2 |2 {z } | 2 {z } |2 {z }
1 2 3
To further simplify 2 , we use part (a) together with the identity 2 cos(α) sin(α) = sin(2α) for any α ∈ R:
i 1 1
2 = cos(θ/2) sin(θ/2) 2i(~r × ~n) · ~σ = − sin(θ) (~r × ~n) · ~σ = sin(θ) (~n × ~r) · ~σ .
2 2 2
Also, since ~n is a unit vector, (~n · ~σ )2 = I (see lecture). Inserted into 3 leads to
1
3 = sin(θ/2)2 (~n · ~r)(~n · ~σ ) − sin(θ/2)2 (~r · ~σ )
2
1 1
= (1 − cos(θ)) (~n · ~r)(~n · ~σ ) − sin(θ/2)2 (~r · ~σ ).
2 2
Combining parts 1 , 2 , 3 , and using the identity cos(α)2 − sin(α)2 = cos(2α), we obtain:
I 1
R~n (θ)ρR~n (−θ) = + cos(θ)~r + sin(θ)(~n × ~r) + (1 − cos(θ)) (~n · ~r)~n ·~σ
2 2| {z }
r0
~
The expression for the new Bloch vector ~r0 is exactly Rodrigues’ rotation formula, as required.