CS259Q Problem Set 2: Solutions

Problem 2. Quantum No-Signaling Principle

(a) The probability of Bob observing 0 is ⟨ψ| (1 ⊗ |0⟩⟨0|) |ψ⟩ = |α|2 + |γ|2 . The probability of Bob observing
1 is ⟨ψ| (1 ⊗ |1⟩⟨1|) |ψ⟩ = |β|2 + |δ|2 .
γ|0⟩+δ|1⟩
(b) If Alice measures her qubit and obtains 1, Bob’s post measurement state is √ 2 2
. Now, Bob will
|γ| +|δ|
|γ|2 |δ|2
measure 0 w.p. |γ|2 +|δ|2 and 1 with probability |γ|2 +|δ|2 . In general, this is different than in part (a).

(c) We already have calculated Bob’s conditional measurement distribution for Alice measuring 1. If Alice
2
|β|2
obtains 0 upon measuring her qubit, Bob will measure 0 w.p. |α||α|
2 +|β|2 and 1 with probability |α|2 +|β|2 .

We also have that Alice measures 0 w.p. |α|2 + |β|2 and 1 w.p. |γ|2 + |δ|2 . It follows that Bob measures
0 w.p. |α|2 + |γ|2 and 1 is |β|2 + |δ|2 . These are the same measurement statistics as in part (a). We
may interpret this to mean that Alice cannot “signal” or communicate information to Bob simply by
measuring her state.

(d) Suppose Alice applies the unitary U to her state. The probability of Bob observing 0 is

⟨ψ| (U † U ⊗ |0⟩⟨0|) |ψ⟩ = |α|2 + |γ|2

and the probability of Bob observing 1 is

⟨ψ| (U † U ⊗ |1⟩⟨1|) |ψ⟩ = |β|2 + |δ|2 .

These measurement statistics are the same as in (a). Once again, Alice cannot communicate informa-
tion to Bob, by applying a unitary to her state.

1
Problem 3. The GHZ Game

1. A deterministic strategy means that Alice returns A0 whenever she gets a question of 0, and A1 when
she gets a question of 1. Similarly, we have the answers B0,1 and C0,1 . Let us assume such a scenario
can always win; this gives us the following system:
A0 + B0 + C0 ≡ 0 mod 2 (1)
A0 + B1 + C1 ≡ 1 mod 2 (2)
A1 + B0 + C1 ≡ 1 mod 2 (3)
A1 + B1 + C0 ≡ 1 mod 2 . (4)
Let us add all four of these equations, which gives us:
2(A0 + A1 + B0 + B1 + C0 + C1 ) ≡ 1 mod 2 . (5)
This is a contradiction, because as all answers are integers, the left-hand side must be even, while the
right-hand side is odd. Therefore, we cannot solve all four equations simultaneously. The best we can
do is solve three out of the four, meaning a 3/4 win probability — for example, by letting Alice always
answer A0 = A1 = 1, Bob always answer B0 = B1 = 0, and Charlie always answer C0 = C1 = 0.
2. We can just consider two question scenarios: the ⃗q = (0, 0, 0) scenario (which happens with probability
1/4), and only one of the other three since they are permutation-symmetric, for example as if ⃗q =
(0, 1, 1) happened with 3/4 probability.
(i) When they receive questions (0, 0, 0), they keep the state |Ψ⟩ unchanged and measure. They get
outcomes (0, 0, 0), (0, 1, 1), (1, 0, 1), or (1, 1, 0) with one quarter probability each, and return these
as their answers. All of them sum up to zero modulo 2, so all of them win the game in this case
with probability 1.
(ii) When they receive questions (0, 1, 1), then two of them apply U (θ) locally, giving the state:
1
(I ⊗ U (θ) ⊗ U (θ)) |Ψ⟩ = (|0⟩ U (θ) |0⟩ U (θ) |0⟩ − |0⟩ U (θ) |1⟩ U (θ) |1⟩ − (6)
2
− |1⟩ U (θ) |0⟩ U (θ) |1⟩ − |1⟩ U (θ) |1⟩ U (θ) |0⟩) (7)
1
= [|0⟩ (cos θ |0⟩ + sin θ |1⟩)(cos θ |0⟩ + sin θ |1⟩)− (8)
2
− |0⟩ (− sin θ |0⟩ + cos θ |1⟩)(− sin θ |0⟩ + cos θ |1⟩)− (9)
− |1⟩ (cos θ |0⟩ + sin θ |1⟩)(− sin θ |0⟩ + cos θ |1⟩)− (10)
− |1⟩ (− sin θ |0⟩ + cos θ |1⟩)(cos θ + sin θ |1⟩)] (11)
1
= [cos 2θ |000⟩ + sin 2θ |001⟩ + sin 2θ |010⟩ − cos 2θ |011⟩ + (12)
2
+ sin 2θ |100⟩ − cos 2θ |101⟩ − cos 2θ |110⟩ − sin 2θ |111⟩] (13)
Then the measurement outcomes resulting in a win are |001⟩, |010⟩, |100⟩, and |111⟩. The sum of
measurement probabilities for these outcomes are:
(0,1,1)sin2 (2θ) sin2 (2θ) sin2 (2θ) sin2 (2θ)
Pwin = + + + = sin2 (2θ) . (14)
4 4 4 4
Then the overall win probability is:
1 3
Pwin =+ sin2 (2θ) , (15)
4 4
which can be made to be 1 for the choice θ = π/4, which means they can always win the game — this
clearly beats the classical optimum in (a).