Quantum Computers Final Exam
January 6, 2004
Do any 4 of the 6 problems.
1. Fidelity
(a) A single qubit is in an unknown pure state |ψi, selected at ran- dom from an ensemble uniformly distributed over the Bloch sphere (p(θ, φ) = 4π1 sin θ). We take a guess |ϕi that is also selected ran- domly on the Bloch sphere. The fidelity F is defined as
F ≡ |hψ|ϕi|2, F ≡ tr (ρψρϕ) ,
for pure states and density operators respectively. Calculate the expected fidelity hF i of our guess by either equations. (Hint: this problem can be solved by either integrating over the Bloch sphere (expressing the states in polar coordinates) or finding the density operator (the expected state) for both states.)
(b) We perform a measurement of the spin of |ψi on the z-axis, that is, a measurement in the basis {|0i, |1i}, the resulting state is described by the density operator
ρ = |hψ|0i|2|0ih0| + |hψ|1i|2|1ih1|.
Calculate the expected fidelity between the density operators be- fore (ρψ = |ψihψ|) and after measurement.
(c) Briefly explain the meaning of the difference between the two fi- delities calculated above.
2. Consider the two qubit state
|ΦiAB = 1
√2|0iA 1
2|0iB+
√3 2 |1iB
!
+ 1
√2|1iA
√3
2 |0iB+1 2|1iB
!
, (a) Compute the partial traces ρ = tr (|Φi hΦ|) and ρ =
(c) Find the Schmidt decomposition of |ΦiAB. (Hint: match each pair of eigenstates from the two density matrices that have the same eigenvalues, and use the square root of that eigenvalue as its coefficient.)
3. Describe the effect of decoherence using the Bloch sphere parametriza- tion of the density matrix. The effect of environment phase-damping on qubit A is
|0iA|0iE →q1 − p|0iA|0iE +√
p|0iA|1iE,
|1iA|0iE →q1 − p|1iA|0iE +√
p|1iA|2iE,
where the environment (in a 3-d Hilbert space) has standard basis {|0iE, |1iE, |2iE}. The Kraus operators for the evolution of ρA is
M0 =q1 − pI, M1 =√
p 1 0
0 0
!
, M2 =√
p 0 0
0 1
!
, and the qubit A evolves to
ρ0A= M0ρAM0+ M1ρAM1+ M2ρAM2. (a) Calculate ρ0A, using ρA= a b
c d
!
.
(b) What would happen if qubit A is allowed to decohere further?
(compare ρA and ρ0A)
(c) The density matrix of qubit A corresponding to the point ~p ≡ (p1, p2, p3) where |~p| ≤ 1 is
ρA = 1
2(I + ~p · ~σ) = 1
2(I + p1σ1+ p2σ2+ p3σ3) .
What is the point ~p0 corresponding to the evolved density operator ρ0A in terms of p1, p2, p3?
(d) Which point in the Bloch ball would be reached asymptotically after further evolution?
4. To distinguish two non-orthogonal states |ϕ1iA and |ϕ2iA, we could prepare an ancillary system B and apply the unitary transformation U that act as
|ϕ1iA|βiB → |ϕ1iA|β1iB,
|ϕ2iA|βiB → |ϕ2iA|β2iB,
then the state of A can be determined without disturbing the state by measuring system B alone.
(a) Discuss the validity of this procedure. (Hint: Derive the relation between |β1iB and |β2iB from the unitarity of U)
(b) What if |ϕ1iA and |ϕ2iA are orthogonal?
5. Suppose there is a physical system that has two possible states: ρ1 with probability p1 and ρ2 with probability p2; that is, its state can be interpreted as p1ρ1 + p2ρ2. We wish to perform a measurement that can determine its state with minimum error.
Suppose our measurement has two possible outcomes, projections E1 and E2 = I − E1; that is, if we measure ρ, we would get the state ρE1 with probability tr (ρE1), and ρE2 with probability tr (ρE2). We guess that the state is ρ1 when E1 is measured and ρ2 otherwise, then the probability of error is
perror = p1tr (ρ1E2) + p2tr (ρ2E1) .
(a) Show that if the eigenstates and eigenvalues of p2ρ2 − p1ρ1 is |ii and λi, that is,
p2ρ2− p1ρ1 =X
i
λi|iihi|,
where the |ii’s form an orthonormal basis, then the error proba- bility is
perror= p1+X
i
λihi|E1|ii.
(b) Find the non-negative operator E1 that minimizes perror, show that the resulting error probability from this projection is
(perror)optimal = p1+ X λi.
Show that
(perror)optimal = 1 2 −1
2kp2ρ2− p1ρ1ktr. (Hint: tr (p2ρ2− p1ρ1) = p2− p1 =Piλi.)
6. Use any definition of (perror)optimal from the previous problem:
(a) Calculate (perror)optimal for ρ1 = ρ2 = ρ. (Hint: consider the cases where p1 > p2 and p2 > p1.)
(b) Calculate (perror)optimal for ρ1 and ρ2 with supports on orthogonal subspaces. That is ρ1ρ2 = 0 = ρ2ρ1 where 0 is the zero matrix.
(Hint: use the form of trace norm kρktr = tr
h
ρ†ρi
1 2
.)
(c) Briefly explain the reason for the value of (perror)optimal in the pre- vious two situations.
Pauli Matrices
σ1 = 0 1 1 0
!
, σ2 = 0 −i i 0
!
, σ3 = 1 0 0 −1
!
General Qubit State
Vector (Pure)
|ψ(a, b)i = a|0i + b|1i = cosθ
2|0i + eiφsinθ
2|1i = |ψ(θ, φ)i
|0i ≡ | ↑zi, |1i ≡ | ↓zi Density Operator (Pure)
ρ(θ, φ) = |ψ(θ, φ)ihψ(θ, φ)| = 1
2(I+ˆn·~σ) = 1
2(I+sin θ cos φσ1+sin θ sin φσ2+cos θσ3) ˆ
n = (sin θ cos φ, sin θ sin φ, cos θ) Density Operator (General)
ρ(~p) = 1
2(I + ~p · ~σ) = 1
2(I + p1σ1 + p2σ2+ p3σ3)
|~p| ≤ 1
Qubit Measurement
P (↑ˆn) = tr (| ↑nˆih↑nˆ |ρ) = tr
1
2(I + ˆn · ~σ)ρ
Partial Trace
Two General Systems
|ψiAB =X
iµ
aiµ|iiA|µiB
ρA= trB(|ψiAB ABhψ|) = X
ijµ
aiµa∗jµ|iiAAhj|
ρB = trA(|ψiAB ABhψ|) =X
iµν
aiµa∗iν|µiB Bhν|
Two Qubits
|ΨiAB = a|00i + b|01i + c|10i + d|11i
ρA = (|a|2+ |b|2)|0ih0| + (ac∗+ bd∗)|0ih1| + (a∗c + b∗d)|1ih0| + (|c|2+ |d|2)|1ih1|
ρB = (|a|2+ |c|2)|0ih0| + (ab∗+ cd∗)|0ih1| + (a∗b + c∗d)|1ih0| + (|b|2+ |d|2)|1ih1|