Quantum Information and Computation

The Quantum Circuit Model

鄭皓中
[email protected]
Outline
 𝑥1

What is Computation? 𝑥2 𝑓 𝑥1 , 𝑥2 = 𝑥1 ⊕ 𝑥2

→ 𝑛 𝐴 𝑛 𝐴
• 𝑛 𝑛
𝑥 = 𝑏1 … 𝑏𝑛 0

• 𝒞𝑛

• 𝑓: 0,1 𝑛 → 0,1 𝑚

𝑓(𝑥) 𝑥 ∈ ℤ𝑛2 ≔ 0,1 𝑛 .

• 0
1
•
→
Randomized Classical Computation
•
→ 0…0
• 𝑏1 … 𝑏𝑛 𝑏1 … 𝑏𝑛 0 … 0
𝑏1 … 𝑏𝑛 𝑟1 … 𝑟𝑘 0 … 0 𝑟1 … 𝑟𝑘
•
𝑟1 … 𝑟𝑘
• 𝑓
𝑓 𝑥 2Τ3 𝑥
𝑟1 … 𝑟𝑘
• ≤ 1Τ3
𝑛
𝑛
1Τ3 ≤ 𝜀
→ ≥ 1−𝜀
Quantum Computation Model
• 𝑛 𝑏1 … 𝑏𝑛 0 … 0
𝑏1 … 𝑏𝑛 0 … |0⟩
• 𝒞𝑛

•
𝐶1 , 𝐶2 , … .
•

•
Elementary Quantum Gates
• •

• 𝑍

• 𝜓 𝜙 ↦ 𝜙 |𝜓⟩
• •
Universal Quantum Gate Sets [§4.5, N&C]
• 𝑛≥1 𝑛

𝑉𝑘 2×2

• 2
→ 𝑑×𝑑 𝑈 𝑈 = 𝑉1 ⋯ 𝑉𝑘 𝑘 ≤ 𝑑(𝑑 − 1)/2
→ 𝑛
• 1
• {𝐻, 𝑇} 1 {CNOT, 𝐻, 𝑇}
• {𝐻, CCNOT}
• 1 𝜀
1Τ𝜀
Classical vs. Quantum Computation
• ∀
|𝜓⟩ 𝑈 ∃ 𝑈 −1 = 𝑈 † 𝑈 −1 𝑈 𝜓 = |𝜓⟩
•
•

𝑓: ℤ𝑛2 → ℤ𝑚
2
ሚ ℤ𝑛+𝑚
𝑓: 2 → ℤ 𝑛+𝑚
2 𝑓ሚ 𝑥, 𝑦 ≔ 𝑥, 𝑦 ⊕ 𝑓 𝑥 , ∀𝑥 ∈ ℤ𝑛2 , 𝑦 ∈ ℤ𝑚
2 .

→ 𝑓ሚ 𝑓ሚ 𝑓ሚ 𝑥, 𝑦 = (𝑥, 𝑦) 𝑏 ⊕ 𝑏 = 0 ⋯ 0 ∀𝑏 ∈ ℤ∗2
• 𝑓
𝑥, 𝑦 ↦ |𝑥, 𝑦 ⊕ 𝑓 𝑥 ⟩.
• →
Circuit Implementation of Quantum Algorithms
•

•
Computational Resources
•

𝑛
𝑈
𝑈
The Big-𝑂 Notation for Asymptotic Analysis
∀𝑛 ≥ 𝑁

• 𝑂 𝑓 𝑛 𝑔 ∃𝑐 > 0, 𝑁 ∈ ℕ 𝑔 𝑛 ≤ 𝑐𝑓(𝑛)
→𝑇 𝑛 =𝑂 𝑓 𝑛 𝑇 𝑓
→𝑇 𝑛 = 𝑂 poly 𝑛 𝑇 = 𝑂 𝑛𝑘 𝑘
→𝑇 𝑛 ≠ 𝑂 poly 𝑛 e𝑛 , 2 𝑛 , 𝑛log 𝑛

• Ω(𝑓 𝑛 ) 𝑔 ∃𝑐 > 0, 𝑁 ∈ ℕ 𝑔 𝑛 ≥ 𝑐𝑓(𝑛)

→𝑇 𝑛 =Ω 𝑓 𝑛 𝑇 𝑓
• Θ(𝑓 𝑛 ) 𝑔 𝑂 𝑓 𝑛 Ω(𝑓 𝑛 )
𝑔(𝑛)
• 𝑔(𝑛) = 𝑜 𝑓 𝑛 limsup =0
𝑛→∞ 𝑓(𝑛)
•
Complexity Classes
•
𝑛 → 𝑛
•

•
2Τ3
•

•
Relations – A Glimpse of The Complexity Zoo [Diagram]
𝑛×𝑛
𝑛×𝑛
The Oracle Model (Black Box Promise Problems)
•
𝑛 𝑚
𝑓: 0,1 → 0,1
→

• 𝑓 𝑓
𝑓

• 𝑓
𝑓
•
Examples of The Oracle Model

𝑛
𝑓: 0,1 → 0,1
𝑥 𝑓 𝑥 =1
𝑥

𝑓: ℤ𝑀 → ℤ𝑀
𝑝 𝑓 𝑥+𝑝 =𝑓 𝑥 𝑥
𝑝
The Quantum Oracle
𝑛 𝑚
• 𝑓: 0,1 → 0,1
𝑈𝑓

→
→ 𝑈𝑓
Quantum Parallelism

• 2𝑛 𝑛

𝑓
→

1 1
σ1𝑥1 ,…,𝑥𝑛 =0 𝑥1 ⋯ 𝑥𝑛 = 1
𝐻 ⊗𝑛 0 ⊗𝑛 = ( 0 + |1⟩) ⊗ ⋯ ⊗ ( 0 + |1⟩) = 2𝑛 2𝑛
σ𝑥∈ 0,1 𝑛 |𝑥⟩
2𝑛
Why Using The Query Model?
•

→
•

•
→
The Deutsch–Jozsa Algorithm (1/2)

𝑓: 0,1 𝑛 → 0,1

𝑓
𝑓(𝑥) = 0 1

• 𝒇𝟎 𝒇𝟏 𝒇𝒙 𝒇ഥ𝒙
The Deutsch–Jozsa Algorithm (2/2)
•
2𝑛
2𝑛 Τ2 + 1
→

•
𝑂(𝑛)
𝑈𝑓 : 𝑥 𝑦 = 𝑥 |𝑦 ⊕ 𝑓 𝑥 ⟩
Circuit Diagram of the DJ Algorithm
Recap of Linear Algebras
𝑣1
𝑣2
• 𝒗= ⋮ σ𝑑𝑖=1 𝑣𝑖
𝑣𝑑
1
→ 𝒖= ⋮ 𝒖† 𝒗
1
•
1
⊗𝑛 1
→ + = ⋮
2𝑛
1
•
→ +1 −1
Proof of the DJ Algorithm (1/2)
𝑛 𝑛
|−⟩

𝑈𝑓 : 𝑥 𝑦 ↦ 𝑥 |𝑓 𝑥 ⊕ 𝑦⟩
Proof of the DJ Algorithm (2/2)

𝑓 |𝑓⟩

→ → →
□
Remarks on the DJ Algorithm
•
(2𝑛 + 1) 𝑋 𝑛
•

→
>1−𝜖 𝑂 log 1Τ𝜖 .
→ 𝐾 2Τ2𝐾 < 𝜖
•

•
• 𝐻 ⊗𝑛 𝑥 = 1
σ𝑦∈ 0,1 𝑛 −1 𝑥⋅𝑦 𝑦
2𝑛
𝑥⋅𝑦
Intuition behind the DJ Algorithm (1/2)
•
→
•
→

0−1 0−1 0−1 |−⟩ 𝑋

1 2
↦ 1 𝑋 2
=−1 2

0−1 0−1 0−1 |−⟩

0 2
↦ 0 𝐼 2
=+0 2

⇒ 𝑏 0−1
↦ −1 𝑏 𝑏 0−1
, ∀𝑏 ∈ 0,1
2 2

0−1 0−1
(𝛼0 0 + 𝛼1 |1⟩) 2
↦ (𝛼0 0 − 𝛼1 |1⟩) 2
0 /|1⟩
Intuition behind the DJ Algorithm (2/2)
• ෡𝑓
𝑈 𝑥
෡𝑓
𝑈 𝑥 : 𝑦 ↦ |𝑦 ⊕ 𝑓 𝑥 ⟩
→ |−⟩ ෡𝑓
𝑈 𝑥
−1 𝑓 𝑥

𝑈𝑓 𝑥 |−⟩ ↦ −1 𝑓 𝑥 𝑥 |−⟩, ∀𝑥 ∈ 0,1 𝑛

Concluding Remarks (1/2)
• →
→
→
•

→
• ⊂

•
Concluding Remarks (2/2)
•
•
→

• In principle

•
⊂
Entanglement is necessary for advantage in quantum computation
•
→
• 𝒞𝑛 𝑛 𝐴 1 2
𝒞𝑛 𝑛
→ 𝐴
•
𝛼1 𝛼2 ⋯ |𝛼𝑛 ⟩
2 𝑈 1 2 𝑈 𝛼1 𝛼2 |𝛼3 ⟩ ⋯ |𝛼𝑛 ⟩
4×4 𝑈 𝛼1 𝛼2 𝑎 00 + 𝑏 01 +
𝑐 10 + 𝑑|11⟩ 𝛽1 𝛽2 𝑛
Quantum Interference (1/2)
𝐻 𝐻
• 0 → + → 0

𝐻 𝐻
1
2 0 : 12
1
0
2 1
2 1 : 12
0 1 0
2 0 : 12
1
2 1
−1 1 : −12 |0⟩
2
Quantum Interference (2/2)

𝑈1 𝑈2

𝛽0,0
0 0
𝛽0,1
𝛼0,0
𝛼0,1 1 1 Pr final outcome is 3
2
0 𝛼0,2 = σ𝑗 𝛼0,𝑗 𝛽𝑗,3
2 2
𝛼0,3 𝛽2,3
3 3
𝛽3,3
Why BQP ⊂ PSPACE? [§4.5.5, N&C], [§5, Preskill]
•

• 𝑛 σ𝑥∈ 0,1 𝑛 𝛼𝑥 |𝑥⟩

•

→
{𝐻, 𝑇, CNOT}

• 𝑈1 𝑈2
Pr first qubit is 0
= 00 𝑈2 𝑈1 00 2
+ 01 𝑈2 𝑈1 00 2
Final Remarks
• {𝐻, 𝑇, CNOT}
→ §

•
•

→ 𝐻 𝑋 𝑍

→ { , 𝑇}
References (1/2)
•

•
•
•
•
•
References (2/2)
•

•
•

•
QIP 2024 Tutorial by Robin Kothari