Chapter 2

Quantum Foundation

§ 2.1 Formalism of Quantum mechanics

1. Basic concept: wave-particle duality

State to describe systems

Prediction evolution

(physics is about) measurement / observables

wave particle
interference Indivisible, discrete
Features superposition principle quantization
duality
The double-slit experiment

How to describe wave-particle duality?

Math Tool: linear algebra (languages of Q. M.)

Superposition discrete eigenvalues of matrix (quantization)

2. States in Q. M. (See front)

Chapter 2

Review of formalism of Q. M.

§ 2.1 States, evolution and measurements

states to describe system

Prediction evolution

measurement, observable

superposition (wave)

wave-particle duality linear algebra

quantization (particle)

1. States in Q. M.

Assumption 1: state ⇔ a vector in Hilbert space

Vector: |𝜓⟩ inner product ⟨𝜓1 |𝜓2 ⟩ ← a complex No.

Conjugate (adjoint): ⟨𝜓| norm ⟨𝜓|𝜓⟩ > 0 a positive No.

Hilbert space: a vector space with inner product structure.

2. Observables in Q. M.

Assumption 2: Observable ⇔ a Hermitian operator in Hilbert space

Operator: a map from state to state |𝜓⟩ → 𝐴|𝜓⟩

𝐴

operator new vector

 𝑏1 𝑎1
Matrix is special operator ⋮ = 𝑀 ⋮
𝑏𝑛 𝑎𝑛

New vector 𝑏𝑖 = ∑𝑗 𝑀𝑖𝑗 𝑎𝑗

Hermitian operator 𝐴 = 𝐴†
∗
𝐴† is defined as ⟨𝜓1 |𝐴|𝜓⟩ ≡ ⟨𝜓2 |𝐴† |𝜓1 ⟩ or (𝐴† )𝑚𝑛 = 𝐴∗𝑛𝑚

Spectral decomposition of a Hermitian operator

𝐴 = ∑ 𝜆𝑖 |𝜆𝑖 ⟩⟨𝜆𝑖 |
𝑖

eigenvalue eigenstates of A

which satisfy eigenfunction: 𝐴|𝜆𝑖 ⟩ = 𝜆𝑖 |𝜆𝑖 ⟩

completeness: ∑𝑖 |𝜆𝑖 ⟩⟨𝜆𝑖 | = 𝐼̂ (identity)

orthonormal: ⟨𝜆𝑖 |𝜆𝑗 ⟩ = 𝛿𝑖𝑗

⇒ expansion: |𝜓⟩ = ∑𝑖 𝑎𝑖 |𝜆𝑖 ⟩ 𝑎𝑖 = ⟨𝜆𝑖 |𝜓⟩

Classical Quantum
Coordinate 𝑥 𝑥̂
Momentum 𝑝 𝑝̂
Energy 𝑝2 𝑝̂ 2
𝐸(𝑥, 𝑝) = + 𝑉(𝑥) ̂ (𝑥̂, 𝑝̂ ) =
𝐻 + 𝑉(𝑥̂) (𝐻𝑎𝑚𝑖𝑙𝑡𝑜𝑛𝑖𝑎𝑛)
2𝑚 2𝑚
3. Evolution in Q. M.

|𝜓(𝑡 = 𝑡0 )⟩ 𝑘𝑛𝑜𝑤𝑛 → |𝜓(𝑡 > 𝑡0 )⟩ ? ?

Assumption 3: 𝑆𝑐ℎ𝑟𝑜̈ 𝑑𝑖𝑛𝑔𝑒𝑟 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛

𝜕|𝜓⟩
𝑖ℏ = 𝐻|𝜓⟩
𝜕𝑡

State derivative new state |𝜓(𝑡)⟩ for any 𝑡 > 𝑡0

|𝜓(𝑡0 )⟩

̂ is t-independent, formal solution:

If 𝐻
 ̂ (𝑡−𝑡0 )
𝑖𝐻
|𝜓(𝑡)⟩ = 𝑒 − ℏ |𝜓(𝑡0 )⟩ ≡ 𝑈(𝑡, 𝑡0 )|𝜓(𝑡0 )⟩

evolution operator
̂ (𝑡−𝑡0 )
𝑖𝐻 ̂ (𝑡−𝑡0 )
𝑖𝐻
Unitary of U: 𝑈 † 𝑈 = 𝑒 ℏ 𝑒− ℏ = 𝐼̂ (identity)

Evolution in Q. M. is unitary.

Norm-preserving (⟨𝜓(𝑡)|𝜓(𝑡)⟩ = ⟨𝜓0 |𝑈 † 𝑈|𝜓0 ⟩ = ⟨𝜓0 |𝜓0 ⟩ )

4. Measurement in Q. M.

Assumption 4: when we measure an observable A under initial state |𝜓⟩, the final

state is collapsed to an eigenstates |𝜆𝑖 ⟩ of A with probability 𝑝𝑖 = |⟨𝜆𝑖 |𝜓⟩|2

(Born statistical interpretation)

(If |𝜓⟩ = ∑𝑖 𝑎𝑖 |𝜆𝑖 ⟩, 𝑝𝑖 = |𝑎𝑖 |2 )

Define a projector (operator) 𝑃𝑖 ≡ |𝜆𝑖 ⟩⟨𝜆𝑖 |

Property: 𝑝𝑖 ≥ 0, 𝑃𝑖2 = 𝑃𝑖 , ∑𝑖 𝑃𝑖 = 𝐼̂

5. Example application: quantum no-cloning theorem

particle 𝐴 𝐵
Initial state |𝜓⟩ |0⟩
Copy (must be unitary) |𝜓⟩ |𝜓⟩
If that is possible, exist unitary operator U, so that

𝑈(|𝜓⟩𝐴 |0⟩𝐵 ) = |𝜓⟩𝐴 |𝜓⟩𝐵 (𝑓𝑜𝑟 𝑎𝑟𝑏𝑖𝑡𝑟𝑦 |𝜓⟩)

Take two particular |𝜓1 ⟩, |𝜓2 ⟩, with ⟨𝜓1 |𝜓2 ⟩ ≠ 0, 1

Then 𝑈|𝜓1 ⟩𝐴 |0⟩𝐵 = |𝜓1 ⟩𝐴 |𝜓1 ⟩𝐵 (1)

𝑈|𝜓2 ⟩𝐴 |0⟩𝐵 = |𝜓2 ⟩𝐴 |𝜓2 ⟩𝐵 (2)

Inner product of (1) and (2)

⟨𝜓1 |𝜓2 ⟩ = ⟨𝜓1 |𝜓2 ⟩2 ⇒ ⟨𝜓1 |𝜓2 ⟩ = 0, 1 ⟹ 𝑐𝑜𝑛𝑡𝑟𝑎𝑑𝑖𝑐𝑡𝑖𝑜𝑛

So it is impossible to clone a quantum state. ⇒ secret of quantum message

 State after measurement

|𝜓𝑓 ⟩ = 𝑃𝑖 |𝜓⟩ = |𝜆𝑖 ⟩⟨𝜆𝑖 |𝜓⟩ ∝ |𝜆𝑖 ⟩

Probability: 𝑝𝑖 = ⟨𝜓𝑓 |𝜓𝑓 ⟩ = |⟨𝜆𝑖 |𝜓⟩|2 (Von-Neumann measurement)

𝑒𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡
Q. Measurement ⇔ {𝑃𝑖 }

Example: Application: state distinguishability

Send signal 0, 1 (classical)

|𝜓⟩ = 𝑐0 |0⟩ + 𝑐1 |1⟩ (quantum)

Many different choices |𝜓1 ⟩, |𝜓2 ⟩, ⋯

1) When |𝜓1 ⟩ ⊥ |𝜓2 ⟩, they can be distinguished by a measurement

Take projector 𝑃1 = |𝜓1 ⟩⟨𝜓1 |

𝑃2 = |𝜓2 ⟩⟨𝜓2 | 𝑃1 𝑃2 = 0

2) When |𝜓1 ⟩, |𝜓2 ⟩ are not orthogonal, not distinguishable (impossible to find

two projectors) ⇒ no transfer of information

§ 2.2 Finite versus continuous-variable quantum systems

1. Finite-dim quantum system

Hilbert space has a finite dim, basis vectors |𝑒0 ⟩, |𝑒1 ⟩, ⋯ |𝑒𝑑−1 ⟩

d-dim

The case of 𝑑 = 2 is particularly important. (2-dim quantum bit system)

Examples: 1) spin-1/2 particles ↑ ↓

2) two level atoms

3) photon polarization
 4) quantum dots (nano structures)

5) Josephson junction (superconducting device)

6) Neutrino oscillations

2. Description of 2-D (qubit, spin) systems

Basis states |0⟩ = (10), |1⟩ = (01)

𝑐0
|𝜓⟩ = 𝑐0 |0⟩ + 𝑐1 |1⟩ = ( ) |𝑐0 |2 + |𝑐1 |2 = 1
𝑐1
𝜃 𝜃
= 𝑒 𝑖𝑟 (cos 2 |0⟩ + 𝑒 𝑖𝜙 sin 2 |1⟩)

r: Globe phase – not important 𝜙: relative phase 𝜃: angle ⇒ 3 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟𝑠

Geometric representation: Ω = (𝜃, 𝜙)

Operators: 2 × 2 𝐻𝑒𝑟𝑚𝑖𝑡𝑖𝑎𝑛 𝑚𝑎𝑡𝑟𝑖𝑥

1 0 0 1 0 −𝑖 1 0
Basis: 𝐼 = ( ) , 𝜎𝑥 = ( ) , 𝜎𝑦 = ( ) , 𝜎𝑧 = ( )
0 1 1 0 𝑖 0 0 −1
Pauli matrix

Any operator can be expressed as 𝐴 = 𝑟1 𝐼 + 𝑟2 𝜎𝑥 + 𝑟3 𝜎𝑦 + 𝑟4 𝜎𝑧 (4-parameters)

Commutator for Pauli matrix 𝜎𝑖2 = 𝐼, 𝜎𝑖+ = 𝜎𝑖 , 𝜎𝑦 = 𝑖𝜎𝑥 𝜎𝑧

[𝜎𝑖 , 𝜎𝑗 ] = 2𝑖𝜖𝑖𝑗𝑘 𝜎𝑘 {𝜎𝑖 , 𝜎𝑗 } ≡ 𝜎𝑖 𝜎𝑗 + 𝜎𝑗 𝜎𝑖 = 0

ℏ
Spin operator 𝑆⃗ ≡ 2 𝜎⃗ (definition)

Basis vector of 𝜎𝑥 , 𝜎𝑦 , 𝜎𝑧 form 𝜎𝑖 |𝜓⟩ = 𝜆𝑖 |𝜓⟩

 one finds 𝜆𝑖 = ±1 (as 𝜎𝑖2 = 𝐼)

eigenvectors:

|0⟩ (|0⟩ + |1⟩)/√2 (|0⟩ + 𝑖|1⟩)/√2

𝜎𝑧 𝜎𝑥 𝜎𝑦
|1⟩ (|0⟩ + |1⟩)/√2 (|0⟩ − 𝑖|1⟩)/√2

3. Continuous-variable systems:

Def. Hilbert space has an infinite-dim and some operator has continuous

eigenspectrum (eigenvalues)

Example. 𝑥̂ (𝑐𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒) 𝑎𝑛𝑑 𝑝̂ (𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚)

Spectro-decomposition:

𝑥̂ = ∫ 𝑥|𝑥⟩⟨𝑥|𝑑𝑥 ⟨𝑥|𝑥 ′ ⟩ = 𝛿(𝑥 − 𝑥 ′ ) normalization

𝑝̂ = ∫ 𝑝|𝑝⟩⟨𝑝|𝑑𝑝 ⟨𝑝|𝑝′ ⟩ = 𝛿(𝑝 − 𝑝′ )

Momentum eigenstate Completeness:∫ |𝑥⟩⟨𝑥|𝑑𝑥 = ∫ |𝑝⟩⟨𝑝|𝑑𝑝 = 𝐼̂

|𝑝⟩ in terms of |𝑥⟩

1 𝑖𝑝𝑥 𝑝
|𝑝⟩ = 𝑒 ℏ |𝑥⟩ (𝑎 𝑝𝑙𝑎𝑛𝑒 𝑤𝑎𝑣𝑒 𝑤𝑖𝑡ℎ 𝑘 = )
√2𝜋ℏ ℏ
De-Broglie relation
𝑖𝑝𝑥
1
So ⟨𝑥|𝑝⟩ = ⟨𝑝|𝑥⟩∗ = 𝑒 ℏ
√2𝜋ℏ

4. Coordinate and momentum basis representations: wave functions

|𝜓⟩ ⇒ 𝜓(𝑥) ≡ ⟨𝑥|𝜓⟩

State Wave-function in the coordinate basis

⇒ 𝜙(𝑝) ≡ ⟨𝑝|𝜓⟩

Wave-function in momentum basis

𝑖𝑝𝑥
−
𝑒 ℏ
𝜙(𝑝) = ⟨𝑝|𝜓⟩ = ∫⟨𝑝|𝑥⟩⟨𝑥|𝜓⟩𝑑𝑥 = ∫ 𝜓(𝑥) 𝑑𝑥 (𝐹𝑜𝑢𝑟𝑖𝑒𝑟 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚)
√2𝜋ℏ

Completeness identity 𝜓(𝑥)

 𝑖𝑝𝑥
𝑒 ℏ
Similarly 𝜓(𝑥) = ∫ 𝜙(𝑝) 𝑑𝑝
√2𝜋ℏ

𝑝̂ in the coordinate basis

𝑝̂ = ∫ 𝑝|𝑝⟩⟨𝑝|𝑑𝑝 (def.)

= ∫ 𝑝|𝑥⟩⟨𝑥|𝑝⟩⟨𝑝|𝑥 ′ ⟩⟨𝑥 ′ |𝑑𝑝𝑑𝑥𝑑𝑥 ′ (completeness)

′
𝑒 𝑖𝑝(𝑥−𝑥 )/ℏ
= ∫ 𝑝|𝑥⟩⟨𝑥 ′ | 𝑑𝑥𝑑𝑥 ′ 𝑑𝑝
2𝜋ℏ
′
𝑑 𝑒 𝑖𝑝(𝑥−𝑥 )/ℏ
= ∫ |𝑥⟩⟨𝑥 ′ |(𝑖ℏ 𝑑𝑥 ′ ) 𝑑𝑥𝑑𝑥 ′ 𝑑𝑝
2𝜋ℏ
𝑑
= ∫ |𝑥⟩(−𝑖ℏ 𝑑𝑥 ′ )⟨𝑥 ′ |𝛿(𝑥 − 𝑥 ′ )𝑑𝑥𝑑𝑥 ′ (partial integration)

𝑑
= ∫|𝑥⟩ (−𝑖ℏ 𝑑𝑥) ⟨𝑥|𝑑𝑥
𝑑
𝑝̂ ⇔ −𝑖ℎ (𝑖𝑛 𝑡ℎ𝑒 𝑐𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒 − 𝑏𝑎𝑠𝑖𝑠 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛)
𝑑𝑥
𝑑
Similarly, 𝑥̂ ⇔ 𝑖ℏ 𝑑𝑝 (𝑖𝑛 𝑡ℎ𝑒 𝑚𝑜𝑚𝑒𝑛𝑡 − 𝑏𝑎𝑠𝑖𝑠 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛)

𝑑 𝑑
From it: commutator [𝑥̂, 𝑝̂ ] = [𝑥, −𝑖ℏ 𝑑𝑥] = [𝑖ℏ 𝑑𝑝 , 𝑝] = 𝑖ℏ

Coordinate basis Momentum basis

§ 2.3 Uncertainty relation, 𝑺𝒄𝒉𝒓𝒐̈ 𝒅𝒊𝒏𝒈𝒆𝒓 versus

Heisenberg picture
1. Uncertainty relations

Theory: For any two operators A, B, with [𝐴, 𝐵] ≠ 0

1
⟨(Δ𝐴)2 ⟩⟨(Δ𝐵)2 ⟩ ≥ |⟨[𝐴, 𝐵]⟩|2
4
Def. ⟨𝐴⟩ = ⟨𝜓|𝐴|𝜓⟩ (𝑓𝑜𝑟 𝑎𝑛𝑦 𝑠𝑡𝑎𝑡𝑒 |𝜓⟩) ⟨(Δ𝐴)2 ⟩ ≡ ⟨𝐴2 ⟩ − ⟨𝐴⟩2

Examples:

1) 𝐴 = 𝑥̂ [𝐴, 𝐵] = [𝑥̂, 𝑝̂ ] = 𝑖ℏ
 ℏ2
𝐵 = 𝑝̂ So ⟨(Δ𝑥̂)2 ⟩⟨(Δ𝑝̂ )2⟩ ≥ (When equality holds ⇒ minimum
4

uncertainty state: coherent state)

2) 𝐴 = 𝜎̂𝑥 [𝐴, 𝐵] = [𝜎𝑥 , 𝜎𝑦 ] = 2𝑖𝜎𝑧

2
𝐵 = 𝜎̂𝑦 So ⟨(Δ𝜎̂𝑥 )2 ⟩ ⟨(Δ𝜎̂𝑦 ) ⟩ ≥ ⟨𝜎̂𝑧 ⟩2

2. Evolution in the 𝑆𝑐ℎ𝑟𝑜̈ 𝑑𝑖𝑛𝑔𝑒𝑟 picture

𝑆𝑐ℎ𝑟𝑜̈ 𝑑𝑖𝑛𝑔𝑒𝑟 𝑃𝑖𝑐𝑡𝑢𝑟𝑒: State evolves, but operator does not

State evolution: 𝑆𝑐ℎ𝑟𝑜̈ 𝑑𝑖𝑛𝑔𝑒𝑟 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛

𝜕|𝜓⟩ 𝑝̂ 2
𝑖ℏ ̂ |𝜓⟩
=𝐻 ̂ (𝑥̂, 𝑝̂ ) =
𝐹𝑜𝑟 𝐶𝑉 𝑠𝑦𝑠𝑡𝑒𝑚, 𝐻 + 𝑉(𝑥̂)
𝜕𝑡 2𝑚
Case a: in the coordinate representation.
𝜕⟨𝑥|𝜓⟩
𝑖ℏ ̂ (𝑥̂, 𝑝̂ )|𝜓⟩
= ⟨𝑥|𝐻
𝜕𝑡
⟨𝑥|𝜓⟩ = 𝜓(𝑥)

𝜕 ℏ2 𝜕 2
̂ (𝑥̂, 𝑝̂ )|𝜓⟩ = 𝐻 (𝑥, −𝑖ℏ
⟨𝑥|𝐻 ) 𝜓(𝑥) = [− + 𝑉(𝑥)] 𝜓(𝑥)
𝜕𝑥 2𝑚 𝜕𝑥 2

𝜕𝜓(𝑥) ℏ2 𝜕 2
So ⇒ 𝑖ℏ = [− 2𝑚 𝜕𝑥 2 + 𝑉(𝑥)] 𝜓(𝑥) (wave equation)
𝜕𝑡

(𝑆𝑐ℎ𝑟𝑜̈ 𝑑𝑖𝑛𝑔𝑒𝑟 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 in the coordinate representation.)

3𝐷 𝜕𝜓(𝑟⃗) ℏ2
⇒ 𝑖ℏ = [− 2𝑚 ∇2𝑟⃗ + 𝑉(𝑟⃗)] 𝜓(𝑟⃗)
𝜕𝑡

Case b: in the momentum |𝑝⟩ representation

𝜕⟨𝑝|𝜓⟩
𝑖ℏ = ⟨𝑝|𝐻(𝑥̂, 𝑝̂ )|𝜓⟩
𝜕𝑡
⟨𝑝|𝜓⟩ = 𝜙(𝑝)
𝜕
⟨𝑝|𝐻(𝑥̂, 𝑝̂ )|𝜓⟩ = 𝐻 (𝑖ℏ , 𝑝) 𝜙(𝑝)
𝜕𝑝
𝜕𝜙(𝑝) 𝑝2 𝜕
So, 𝑖ℏ = [− 2𝑚 + 𝑉 (𝑖ℏ 𝜕𝑝)] 𝜙(𝑝) (momentum-basis 𝑆𝑐ℎ𝑟𝑜̈ 𝑑𝑖𝑛𝑔𝑒𝑟 Eq.)
𝜕𝑡

Example: free particle 𝑉(𝑥) = 0

 𝜕𝜙(𝑝) 𝑝2 −
𝑖𝑝2 𝑡
𝑖ℏ =− 𝜙(𝑝) ⇒ 𝜙(𝑝, 𝑡) = 𝑒 2𝑚ℏ 𝜙(𝑝, 0)
𝜕𝑡 2𝑚

1 1 𝑖𝑝 𝑝𝑡
𝜓(𝑥, 𝑡) = ∫ 𝜙(𝑝, 𝑡)𝑒 𝑖𝑝𝑥/ℏ 𝑑𝑥 = ∫ 𝜙(𝑝, 0)𝑒 ℏ (𝑥− 𝑚 ) 𝑑𝑝
√2𝜋ℏ √2𝜋ℏ
Wave packet spreading dispersion

3. Evolution in the Heisenberg picture

State Operator Phys really observed property

SP ̂ (𝑡)|𝜓(0)⟩
|𝜓(𝑡)⟩𝑠 = 𝑈 𝐴𝑆 = 𝐴 ⟨𝐴⟩ = 𝑠⟨𝜓(𝑡)|𝐴|𝜓(𝑡)⟩𝑠
(t-independent) Independent of
picture (frame)
HP |𝜓⟩𝐻 ≡ |𝜓(0)⟩ ̂ + (𝑡)𝐴𝑈
𝐴𝐻 (𝑡) = 𝑈 ̂ (𝑡) ⟨𝐴⟩ = 𝐻⟨𝜓|𝐴(𝑡)|𝜓⟩𝐻

𝜕|𝜓(𝑡)⟩ ̂(𝑡)
𝜕𝑈
From 𝑖ℏ = 𝑖ℏ ̂ (𝑡)|𝜓(0)⟩
|𝜓(0)⟩ = 𝐻|𝜓(𝑡)⟩ = 𝐻𝑈
𝜕𝑡 𝜕𝑡

𝜕𝑈 ̂
̂
Valid for any |𝜓(0)⟩ ⇒ So 𝑖ℏ 𝜕𝑡 = 𝐻𝑈

̂ + (𝑡)𝐴𝑈
As 𝐴𝐻 (𝑡) = 𝑈 ̂ (𝑡)

𝑑𝐴𝐻 𝑑𝑈 + 𝑑𝑈
𝑖ℏ = 𝑖ℏ [ 𝐴𝑈 + 𝑈 + 𝐴 ] = [−𝑈 + 𝐻𝐴𝑈 + 𝑈 + 𝐴𝐻𝑈]
𝑑𝑡 𝑑𝑡 𝑑𝑡

𝜕𝐴
(Assumed = 0) [𝑈, 𝐻] = 0
𝜕𝑡

= [−𝐻𝑈 + 𝐴𝑈 + 𝑈 + 𝐴𝑈𝐻]

= 𝐴𝐻 (𝑡)𝐻 − 𝐻𝐴𝐻 (𝑡) = [𝐴𝐻 , 𝑈]

𝑑𝐴𝐻
So 𝑖ℏ = [𝐴𝐻 , 𝐻] (Heisenberg equation)
𝑑𝑡

|𝜓⟩𝐻 = |𝜓(0)⟩

4. Example: application of the Heisenberg equation

Ex. 1) Free particles

 𝑝2
𝐻 = 2𝑚 observables: 𝑥̂, 𝑝̂

𝑑𝑥̂ 𝑝̂
In HP. Heisenberg Eq. 𝑖ℏ 𝑑𝑡 = [𝑥̂, 𝐻] = 𝑖ℏ 𝑚

𝑑𝑝̂
𝑖ℏ 𝑑𝑡 = 0

𝑝̂ (𝑡) = 𝑝̂ (0)
𝑝̂ (0)
𝑥̂(𝑡) = 𝑥̂(0) + 𝑡
𝑚
[𝑥̂(𝑡), 𝑝̂ (𝑡)] = [𝑥̂(0), 𝑝̂ (0)] = 𝑖ℏ

Commutator does not change with t, why?

𝑑𝐴
Reason: Def. 𝐴 ≡ [𝑥̂, 𝑝̂ ] 𝑖ℏ 𝑑𝑡 = [𝐴, 𝐻] = 0 ⇒ 𝐴(𝑡) = 𝐴(0) = 𝑖ℏ

Variance: ⟨(Δ𝑝̂ (𝑡))2 ⟩ = ⟨(Δ𝑝̂ (0))2 ⟩

𝑡2
⟨(Δ𝑥̂(𝑡))2 ⟩ = ⟨(Δ𝑥̂(0))2 ⟩ + 2 ⟨(Δ𝑝̂ (0))2 ⟩ (assumed ⟨𝑥̂(0)𝑝̂ (0)⟩ = 0)
𝑚

Increase with t, spread of wave packet

Coherent state → non-coherent state

quantum dispersion of a macroscopic object. (HW)

Ex. 2) Harmonic oscillator

𝑝̂ 2 1
̂=
𝐻 + 𝑚𝜔2 𝑥̂
2𝑚 2
𝑑𝑥̂
Heisenberg Eqs.
1
̂ ] = 𝑝̂
= 𝑖ℏ [𝑥̂, 𝐻
𝑑𝑡 𝑚

𝑑𝑝̂ 1 𝑑2 𝑥̂
̂ ] = −𝑚𝜔2 𝑥̂
= 𝑖ℏ [𝑝̂ , 𝐻 ⇒ = −𝜔2 𝑥̂
𝑑𝑡 𝑑𝑡 2

𝑝̂(0)
⇒ 𝑥̂(𝑡) = 𝑥̂(0)𝑐𝑜𝑠𝜔𝑡 + 𝑠𝑖𝑛𝜔𝑡
𝑚𝜔
 𝑝̂ (𝑡) = 𝑝̂ (0)𝑐𝑜𝑠𝜔𝑡 − 𝑚𝜔𝑥̂(0)𝑠𝑖𝑛𝜔𝑡
2 1 2 ℏ
Assume initially 𝑚𝜔 ⟨(Δ𝑥̂(0)) ⟩ = 𝑚𝜔 ⟨(Δ𝑝̂ (0)) ⟩ = 2

(coherent state: minimum uncertainty state)

2 2 2 sin2 𝜔𝑡 2
Then ⟨(Δ𝑥̂(𝑡)) ⟩ = ⟨(Δ𝑥̂(0)) ⟩ cos2 𝜔𝑡 + ⟨(Δ𝑝̂ (0)) ⟩ = ⟨(Δ𝑥̂(0)) ⟩
𝑚2 𝜔 2

2 2 2 2
⟨(Δ𝑝̂ (𝑡)) ⟩ = ⟨(Δ𝑝̂ (0)) ⟩ cos2 𝜔𝑡 + 𝑚2 𝜔2 ⟨(Δ𝑥̂(0)) ⟩ sin2 𝜔𝑡 = ⟨(Δ𝑝̂ (0)) ⟩

Keep minimum uncertainty state!

Quantum Field Theory ⇔ a set of harmonic oscillators

§ 2.4 Quantum dynamics of 2-dimensional systems and its

applications
1. Evolution of 2-D systems
𝜔
2-D Hamiltonian 𝐻 = ℏ(𝛼𝐼̂ + 2 𝑛⃗⃗ ∙ 𝜎⃗) (general expansion of a Hermitian

operator) convention

Where 𝑛⃗⃗ ∙ 𝜎⃗ ≡ 𝑛𝑥 𝜎𝑥 + 𝑛𝑦 𝜎𝑦 + 𝑛𝑧 𝜎𝑧 |𝑛⃗⃗| = 1 (assume 𝛼, 𝜔 t-independent)

𝑖𝐻𝑡 𝜔𝑡
𝜔𝑡 𝜔𝑡
Evolution. 𝑈(𝑡) = 𝑒 − ℏ = 𝑒 −𝑖𝛼𝑡 𝑒 −𝑖 2 𝑛⃗⃗∙𝜎⃗⃗ = 𝑒 −𝑖𝛼𝑡 (𝐼̂ cos 2 − 𝑖𝑛⃗⃗ ∙ 𝜎⃗ sin 2 )

(globe phase – drop 0) Rabi oscillation with frequency 𝜔/2 (NMR)

Examples:

1) If 𝑛⃗⃗ ∙ 𝜎⃗ = 𝜎𝑦 (along y-direction)

𝜔𝑡 𝜔𝑡
𝜔𝑡 1 0 𝜔𝑡 −sin cos
0 −𝑖 2 2)
𝑈(𝑡) = cos ( )−𝑖( ) sin =( 𝜔𝑡 𝜔𝑡
2 0 1 𝑖 0 2
sin cos
2 2
𝜔𝑡 𝜔𝑡
|𝜓(𝑡)⟩ = 𝑈(𝑡)|𝜓(0)⟩ = cos |0⟩ − sin |1⟩
2 2
𝜋 𝜋 1 1 −1
When 𝜔𝑡 = 2 , (2 − 𝑝𝑢𝑙𝑠𝑒) 𝑈(𝑡) = ( )
√2 1 1
 1
|0⟩ ⟶ (|0⟩ − |1⟩)
√2
1
|1⟩ → (|0⟩ + |1⟩)
√2
When 𝜔𝑡 = 𝜋 (𝜋 − 𝑝𝑢𝑙𝑠𝑒)

|0⟩ → −|1⟩

|1⟩ → |0⟩

2) If 𝑛⃗⃗ ∙ 𝜎⃗ = 𝜎𝑧 (along z-direction)

𝑖𝜔𝑡
𝜔𝑡 1 0 𝜔𝑡 1 0 −
) = (𝑒 0 )
2
𝑈(𝑡) = cos ( ) − 𝑖 sin ( 𝑖𝜔𝑡
2 0 1 2 0 −1
0 𝑒 2
𝑖𝜔𝑡
|0⟩ → |0⟩𝑒 − 2

𝑖𝜔𝑡
|1⟩ → |1⟩𝑒 2 (a relative phase shift)

2. Atomic clocks and Ramsey method

Clocks: periodic motion

Atomic clocks ⇒ small period ⇒ precision measurement

ℏ𝜔0
𝐻= 𝜎𝑧
2

𝑖𝜔0 𝑡
|0⟩ → |0⟩𝑒 − 2

𝑖𝜔0 𝑡
|1⟩ → |1⟩𝑒 + 2 (periodic phase change with fixed frequency 𝜔0 )

How to measure phase change ⇒ Ramsey method!

Ramsey method:
 𝜋
−𝑝𝑢𝑙𝑠𝑒 (𝜎𝑦 −𝑏𝑎𝑠𝑖𝑠) 1
2
|1⟩ → (|1⟩ + |0⟩)
√2
𝑓𝑟𝑒𝑒 𝑒𝑣𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑓𝑜𝑟 𝑎 𝑡𝑖𝑚𝑒 𝑇 (𝜎𝑧 −𝑏𝑎𝑠𝑖𝑠) 1 𝑖𝜔0 𝑡 𝑖𝜔0 𝑡
→ (|1⟩𝑒 2 + |0⟩𝑒 − 2 )
√2
𝜋
−𝑝𝑢𝑙𝑠𝑒 𝑎𝑔𝑎𝑖𝑛 (𝜎𝑦 −𝑏𝑎𝑠𝑖𝑠) 1 𝑖𝜔0 𝑡 𝑖𝜔0 𝑡
2
→ [(|1⟩ + |0⟩)𝑒 2 + (|0⟩ − |1⟩)𝑒 − 2 ]
2
𝜔0 𝑡 𝜔0 𝑡
= |0⟩ cos + 𝑖|1⟩ sin
2 2
𝑀𝑒𝑎𝑠𝑢𝑟𝑒 𝑃1 𝜔0 𝑡
→ 𝑃1 = sin2
2

𝑃1
𝑇 → 𝜔0 (𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑)
𝑃1
𝜔0 → 𝑇 (𝑐𝑙𝑜𝑐𝑘)

§ 2.5 Multi-partite quantum systems, density matrix, and

quantum entanglement
1. Multi-partite quantum systems

Particles: Bosons symmetrized

Indistinguishable, identical wave function

(no trajectory) Fermions anti-symmetrized

Distinguishable parties: distinguished by modes

Ex.

1) Spin ↑, ↓ particle

2) Lattice spin

3) Photon

Multi-partite systems: The Hilbert-space can be decomposed as

𝐻 = 𝐻1 ⊗ 𝐻2 ⊗ ⋯ ⊗ 𝐻𝑛

Tensor product

Basis vectors can be written as tensor products

Ex. Basis vectors

1) 1 qubit (spin) |0⟩, |1⟩ dim-2

2) 2 qubits |00⟩, |01⟩, |1,0⟩, |11⟩ dim-22

3) n qubits |00 ⋯ 1⟩ dim-2𝑛

n bits exponentially expanded Hilbert

space dimensions (intractable)

Remarks: 1) Many-particle systems, when particles are distinguished by modes, can

be considered as multi-partite systems

2) Particular multi-partite state

 Product state |𝜓⟩1,2,⋯𝑛 = |𝜓1 ⟩1 ⊗ |𝜓2 ⟩2 ⊗ |𝜓3 ⟩3 ⋯ ⊗ |𝜓𝑛 ⟩𝑛

𝑐0 |0⟩ + 𝑐1 |1⟩ 𝑑0 |0⟩ + 𝑑1 |1⟩

Degrees of freedom: 2 + 2 + 2 ⋯ + 2 = 2𝑛

𝑟𝑒𝑑𝑢𝑐𝑒𝑑 𝑡𝑜
A special state with dim 2𝑛 → 2𝑛 (tractable)

∗ Mean-field theory corresponds to a product-state ansatz

1
∗ Non-product state, Ex (|00⟩ + |11⟩) ⇒ 𝑒𝑛𝑡𝑎𝑛𝑔𝑙𝑒𝑑 𝑠𝑡𝑎𝑡𝑒
√2

2. Density matrix operator

1) Statistic interpretation

State Prob.

|𝜓1 ⟩ 𝑝1 ⇒ density operator (Von Neumann)

|𝜓2 ⟩ 𝑝2 𝜌 = 𝑝1 |𝜓1 ⟩⟨𝜓1 |+𝑝2 |𝜓2 ⟩⟨𝜓2 |

For any operator A, ⟨𝐴⟩ ≡ 𝑡𝑟(𝐴𝜌) = 𝑝1 ⟨𝜓1 |𝐴|𝜓1 ⟩ + 𝑝2 ⟨𝜓2 |𝐴|𝜓2 ⟩

In general, 𝜌 = ∑𝑖 𝑝𝑖 |𝜓𝑖 ⟩⟨𝜓𝑖 | , (∑𝑖 𝑝𝑖 = 1) {𝑝𝑖 , |𝜓𝑖 ⟩} called an ensemble-

decomposition of 𝜌

A note about “trace”

Def. of “trace”:

For matrix/operator 𝐴 = (𝐴𝑖𝑗 ), 𝑡𝑟𝐴 ≡ ∑𝑖 𝐴𝑖𝑖 (sum of diagonal term)

Properties: 𝑡𝑟(𝐴𝐵) = 𝑡𝑟(𝐵𝐴)

𝑡𝑟(𝐴 + 𝐵) = 𝑡𝑟𝐴 + 𝑡𝑟𝐵

𝑡𝑟(𝐴𝐵𝐶) = 𝑡𝑟(𝐶𝐴𝐵)

𝑡𝑟(|𝜓⟩⟨𝜓|) = ⟨𝜓|𝜓⟩

𝑡𝑟(𝐴|𝜓⟩⟨𝜓|) = ⟨𝜓|𝐴|𝜓⟩
 𝑡𝑟(𝐴) = ∑𝑖⟨𝜇𝑖 |𝐴|𝜇𝑖 ⟩ ({|𝜇𝑖 ⟩} forms a basis)

2) Reduced state interpretation

Particles other side of galaxy

|𝜓⟩12 = 𝑐0 |00⟩ + 𝑐1 |11⟩

What is the state of particle 1?

For any operator 𝐴1 of particle 1,

⟨𝐴1 ⟩ = 12⟨𝜓|𝐴1 |𝜓⟩12 = 𝑡𝑟(𝐴|𝜓⟩12 12⟨𝜓|)

𝑡𝑟 = 𝑡𝑟12 = 𝑡𝑟1 𝑡𝑟2 (𝑝𝑎𝑟𝑡𝑖𝑎𝑙 𝑡𝑟𝑎𝑐𝑒)

Def. of “partial trace”, For operator 𝐵12

𝑡𝑟2 (𝐵12 ) ≡ ∑⟨𝜇2𝑖 |𝐵12 |𝜇2𝑖 ⟩

𝜇2𝑖

an operator on particle 1 basis vector for particle 2

so ⟨𝐴1 ⟩ = 𝑡𝑟1 [𝐴1 𝑡𝑟2 (|𝜓⟩12 12⟨𝜓|)] = 𝑡𝑟(𝜌1 𝐴1 ), where 𝜌1 ≡ 𝑡𝑟2 (|𝜓⟩12 12⟨𝜓|)

Reduced state 𝜌1 : completely characterize the state of the particle 1.

In general, 𝜌1 is a density operator

Ex. |𝜓⟩12 = 𝑐0 |00⟩12 + 𝑐1 |11⟩12

𝜌1 = 𝑡𝑟2 (|𝜓⟩12 12⟨𝜓|)

= ∑ 2⟨𝑖|[|𝑐0 |2 |00⟩12 ⟨00|+|𝑐1 |2 |11⟩12 ⟨11|+𝑐0 𝑐1∗ |00⟩12 ⟨11|+𝑐0∗ 𝑐1 |11⟩12 ⟨00|]|𝑖⟩2
𝑖=0,1

|𝑐 |2 0
= |𝑐0 |2 |0⟩1 ⟨0|+|𝑐1 |2 |1⟩1 ⟨1| = ( 0 )
0 |𝑐1 |2
Ensemble decomposition mixed

In general, for a n-party state |𝜓⟩1,2⋯𝑛

Reduced state of particle 1 𝜌1 = 𝑡𝑟2,3⋯𝑛 (|𝜓⟩1,2⋯𝑛 ⟨𝜓|)

 Reduced state of particle 1, 2, 𝜌12 = 𝑡𝑟3,4⋯𝑛 (|𝜓⟩1,2⋯𝑛 ⟨𝜓|)

3. Properties of density matrix.

Prop. 1 Characterization Th. Any Hermitian operator 𝜌 is a density operator iff

(if and only if) 1) 𝑡𝑟𝜌 = 1 2) 𝜌 ≥ 0 (𝑛𝑜𝑛 − 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒𝑠)

Proof: If 𝜌 = ∑𝑖 𝑝𝑖 |𝜓𝑖 ⟩⟨𝜓𝑖 | 𝑡𝑟𝜌 = ∑𝑖 𝑝𝑖 ⟨𝜓𝑖 |𝜓𝑖 ⟩ = 1

∑𝑖 𝑝𝑖 = 1 𝜌 ≥ 0 (𝑎𝑠 𝑝𝑖 ≥ 0)
𝑡𝑟𝜌 = 1 𝜆𝑖 ≥ 0
If ⇒ 𝜌 has a spectral-decomposition 𝜌 = ∑𝑖 𝜆𝑖 |𝜆𝑖 ⟩⟨𝜆𝑖 |
𝜌≥0 ∑𝑖 𝜆𝑖 = 𝑡𝑟𝜌 = 1

⇒ 𝑒𝑛𝑠𝑒𝑚𝑏𝑙𝑒 𝑑𝑒𝑐𝑜𝑚𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛

Prop. 2 Pure state criterion:

𝑡𝑟(𝜌2 ) ≤ 1, and 𝑡𝑟(𝜌2 ) = 1 iff 𝜌 is a pure state, with 𝜌 = |𝜓⟩⟨𝜓| (𝑡𝑟(𝜌2 ) < 1

is called mixed state)

Proof: 𝜌 has spectral-decomposition 𝜌 = ∑𝑖 𝜆𝑖 |𝜆𝑖 ⟩⟨𝜆𝑖 |,

∑𝑖 𝜆 𝑖 = 1

𝜆𝑖 ≥ 0 ⇒ 0 ≤ 𝜆𝑖 ≤ 1

𝑡𝑟(𝜌2 ) = 𝑡𝑟(∑𝑖 𝜆2𝑖 |𝜆𝑖 ⟩⟨𝜆𝑖 |) = ∑𝑖 𝜆2𝑖 ≤ ∑𝑖 𝜆𝑖 = 1

𝜆𝑖 = 0,1 𝜆𝜈0 = 1
If 𝑡𝑟(𝜌2 ) = 1, equality holds ⇔ ⇒ ⇒ 𝜌 = |𝜓𝜈0 ⟩
∑𝑖 𝜆𝑖 = 1 𝜆𝑖 = 0 𝑓𝑜𝑟 𝑖 ≠ 𝜈0

4. Quantum entanglement and Von Neumann’s entropy

1) Def. of entanglement: For pure state |𝜓⟩𝐴𝐵 of particles A and B,

if |𝜓⟩𝐴𝐵 ≠ |𝜙𝐴 ⟩𝐴 ⊗ |𝜙𝐵 ⟩𝐵 , |𝜓⟩𝐴𝐵 𝑖𝑠 𝑒𝑛𝑡𝑎𝑛𝑔𝑙𝑒𝑑

Ex. Product state entangled state

|00⟩𝐴𝐵 (|00⟩𝐴𝐵 + |11⟩𝐴𝐵 )/√2

(|0⟩ + |1⟩)𝐴 ⊗ |1⟩𝐵 /√2 (√𝑝|00⟩𝐴𝐵 + √1 − 𝑝|11⟩𝐴𝐵 )/√2 (𝑝 ≠ 0, 1)

(|0⟩ + |1⟩)𝐴 ⊗ (|0⟩ + |1⟩)𝐵 /2

= (|00⟩ + |01⟩ + |10⟩ + |11⟩)/2

2) Quantifying entanglement – Von Neumann’s entropy

𝑟𝑒𝑑𝑢𝑐𝑒𝑑 𝑠𝑡𝑎𝑡𝑒
|𝜓𝐴𝐵 ⟩ → 𝜌𝐴

If |𝜓⟩𝐴𝐵 is more entangled, 𝜌𝐴 should be more mixed. (closer to 𝐼𝐴 =

(|0⟩⟨0| + |1⟩⟨1|))

Characterization of mixedness → entropy

Def. of Von-Neumann entropy 𝑆(𝜌) ≡ −𝑡𝑟(𝜌𝑙𝑜𝑔𝜌) = −⟨𝑙𝑜𝑔𝜌⟩ = − ∑𝑖 𝜆𝑖 𝑙𝑜𝑔𝜆𝑖

Where 𝜆𝑖 are eigenvalues of 𝜌 . ( 𝑙𝑜𝑔 for finite-dim systems, 𝑙𝑛 for

continuous-variable systems)

1 𝜆0 = 1/2 1
Ex. 𝜌 = 2 (|0⟩⟨0| + |1⟩⟨1|) ⇒ 𝑆(𝜌) = − ⟨log 2 2⟩ = 1
𝜆1 = 1/2

Properties of 𝑆(𝜌): 1) 𝑆(𝑈𝜌𝑈 + ) = 𝑆(𝜌) (for unitary operation)

2) 𝑆(|𝜓⟩⟨𝜓|) = 0

3) 𝑆(𝜆1 𝜌1 + 𝜆2 𝜌2 ) ≥ 𝜆1 𝑆(𝜌1 ) + 𝜆2 𝑆(𝜌2 )

 4) Subadditivity: 𝑆(𝜌𝐴𝐵 ) ≤ 𝑆(𝜌𝐴 ) + 𝑆(𝜌𝐵 )(equality holds iff 𝜌𝐴𝐵 = 𝜌𝐴 ⊗ 𝜌𝐵 )

5)Strong subadditivity 𝑆(𝜌𝐴𝐵𝐶 ) + 𝑆(𝜌𝐵 ) ≤ 𝑆(𝜌𝐴𝐵 ) + 𝑆(𝜌𝐵𝐶 )

Entanglement measure: entanglement of a bi-partite pure state |𝜓⟩𝐴𝐵 is

measured by 𝐸(|𝜓𝐴𝐵 ⟩) = 𝑆(𝜌𝐴 ) = 𝑆(𝜌𝐵 )

5. Schmidt decomposition

Composite system

|𝜓⟩𝐴𝐵 = ∑ 𝑎𝑖𝜇 |𝑖⟩𝐴 |𝜇⟩𝐵

𝑖𝜇

Th. There exist a basis |𝑖⟩𝐴 , so that |𝜓⟩𝐴𝐵 = ∑𝑖 √𝑝𝑖 |𝑖⟩𝐴 |𝜇𝑖 ⟩𝐵 , with ∑𝑖 𝑝𝑖 = 1

Orthonormal Schmidt number

This basis is given by eigenbasis of 𝜌𝐴 , i.e. 𝜌𝐴 = ∑𝑖 𝑝𝑖 |𝑖⟩𝐴 ⟨𝑖|

Proof: |𝜓⟩𝐴𝐵 = ∑𝑖𝜇 𝑎𝑖𝜇 |𝑖⟩𝐴 |𝜇⟩𝐵 = ∑𝑖|𝑖⟩|𝑖̃⟩𝐵 , with |𝑖̃⟩𝐵 ≡ ∑𝜇 𝑎𝑖𝜇 |𝜇⟩𝐵 (not

necessarily orthonormal)

𝜌𝐴 = 𝑡𝑟𝐵 ( |𝜓⟩𝐴𝐵 ⟨𝜓| ) = 𝑡𝑟𝐵 (∑ |𝑖⟩𝐴 |𝑖̃⟩𝐵 𝐴⟨𝑗| 𝐵⟨𝑗̃|)

𝑖𝑗

= ∑ |𝑖⟩𝐴 𝐴⟨𝑗| 𝐵⟨𝑗̃|𝑖̃⟩𝐵

𝑖𝑗

If |𝑖⟩𝐴 is a eigenbasis of 𝜌𝐴 , ⇒ 𝜌𝐴 = ∑𝑖 𝑝𝑖 |𝑖⟩𝐴 ⟨𝑖|

𝐵⟨𝑗̃|𝑖̃⟩𝐵 = 𝑝𝑖 𝛿𝑖𝑗
1
Let |𝜇𝑖 ⟩ ≡ |𝑖̃⟩ ⇒ |𝜓⟩𝐴𝐵 = ∑𝑖 √𝑝𝑖 |𝑖⟩𝐴 |𝜇𝑖 ⟩𝐵
√𝑝 𝑖

Application: 1) 𝜌𝐵 = ∑𝑖 𝑝𝑖 |𝜇𝑖 ⟩𝐵 ⟨𝜇𝑖 |

Same eigenspectrum to 𝜌𝐴 and 𝜌𝐵

2) purification: any mixed state 𝜌𝐴 can be written in the form

 𝜌𝐴 = 𝑡𝑟𝐵 ( |𝜓⟩𝐴𝐵 ⟨𝜓|), |𝜓⟩𝐴𝐵 𝑖𝑠 𝑐𝑎𝑙𝑙𝑒𝑑 𝑝𝑢𝑟𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝜌𝐴

and dim(𝐻𝐵 ) is at most dim(𝐻𝐴 ) (extend Hilbert space)

3) Proof: 𝜌𝐴 = ∑𝑖 𝜆𝑖 |𝑖⟩𝐴 ⟨𝑖| = 𝑡𝑟𝐵 ( |𝜓⟩𝐴𝐵 ⟨𝜓|), with |𝜓⟩𝐴𝐵 = ∑𝑖|𝑖⟩|𝑖̃⟩𝐵

§ 2.6 Generalized evolution: superoperators

2. Derivation from the extended unitary representation

Define 𝜌𝐴′ = 𝑡𝑟𝐵 (𝜌𝐴𝐵

′
), relation between 𝜌𝐴′ and 𝜌𝐴 ?

With loss of generalization, assume B in a pure state, denoted as |𝐸⟩𝐵 (if mixed, do

purification)

Initial state of AB, 𝜌𝐴𝐵 = 𝜌𝐴 ⨂|𝐸⟩𝐵 ⟨𝐸|, (No entanglement between A and B)

′ +
After evolution 𝜌𝐴𝐵 = 𝑈𝐴𝐵 (𝜌𝐴 ⨂|𝐸⟩𝐵 ⟨𝐸| 𝑈𝐴𝐵

+ |𝜇⟩
𝜌𝐴′ = 𝑡𝑟𝐵 (𝜌𝐴𝐵
′ )
=∑ 𝐵⟨𝜇|𝑈𝐴𝐵 (𝜌𝐴 ⊗ |𝐸⟩𝐵 ⟨𝐸|) 𝑈𝐴𝐵 𝐵
𝜇

A basis of B

Define a operator 𝑀𝜇 ≡ 𝐵⟨𝜇|𝑈𝐴𝐵 |𝜇⟩𝐵 , an operator acting on Hilbert space 𝐻𝐴

So 𝜌𝐴′ = ∑𝜇 𝑀𝜇 𝜌𝐴 𝑀𝜇+ (operator-sum representation)

+ |𝜇⟩
Properties of 𝑀𝜇 : ∑𝜇 𝑀𝜇+ 𝑀𝜇 = ∑𝜇 𝐵⟨𝐸|𝑈𝐴𝐵 𝐵 𝐵⟨𝜇|𝑈𝐴𝐵 |𝐸⟩𝐵 = 𝐼𝐴

𝐼̂𝐵 𝐼𝐴𝐵
1. Superoperators

Definition: operator: a map from state to state A

Superoperator: a map from operator to operator $̂

Evolution superoperator $̂: 𝜌 → 𝜌′ ⇔ 𝜌′ = $̂(𝜌)

Properties of $̂:

0) linear: $̂(𝜆1 𝜌1 + 𝜆2 𝜌2 ) = 𝜆1 $̂(𝜌1 ) + 𝜆2 $̂(𝜌2 ). In general, not linear in state

Ex. |0⟩ → |0⟩ |1⟩ → |1⟩ |0⟩ + |1⟩ → |0⟩⟨0| + |1⟩⟨1| rephasing

1) hermiticity-preserving

2) trace-preserving 𝑡𝑟𝜌′ = 𝑡𝑟𝜌 = 1

3) positive: 𝜌′ is nonnegative if 𝜌 is positive

3′ ) completely positive (strong): any extensive $̂ ⊗ 𝐼̂𝐵 is positive

Example of a positive but not-completely-positive superoperator: partial transpose

Transposition 𝑇̂: 𝜌 → 𝜌𝑇 positive as it does not change eigenvalues

Partial transpose 𝑇̂𝐴 ⨂𝑇̂𝐵

Acting on |Φ⟩𝐴𝐵 = ∑𝑁
𝑖=1|𝑖⟩𝐴 ⨂|𝑖̃⟩𝐵 (normalized as ⟨Φ|Φ⟩ = 𝑁, N-dim EPR state)

𝜌 = |Φ⟩𝐴𝐵 ⟨Φ| correspondence

𝑇̂𝐴 ⨂𝐼̂𝐵 (𝜌) = 𝑇̂𝐴 ⨂𝐼̂𝐵 [∑ |𝑖⟩𝐴 ⟨𝑗|⨂|𝑖̃⟩𝐵 ⟨𝑗̃|] = ∑ |𝑗⟩𝐴 ⟨𝑖|⨂|𝑖̃⟩𝐵 ⟨𝑗̃| ≡ 𝜌′
𝑖𝑗 𝑖𝑗

State exchange-swap

Any state |𝜙⟩𝐴 ⨂|𝜙⟩𝐵 = ∑𝑖 𝑎𝑖 |𝑖⟩𝐴 ⨂ ∑𝑗 𝑏𝑗 |𝑗̃⟩𝐵

𝜌′ = |𝜙1 ⟩𝐴 ⨂|𝜙⟩𝐵 = ∑ 𝑎𝑖 𝑏𝑗 |𝑗⟩𝐴 |𝑖̃⟩𝐵 = |𝜙2 ⟩𝐴 ⨂|𝜙1 ⟩𝐵

𝑖𝑗

⇒ 𝜌′ is swap operator. Eigenvalue of 𝜌′ : +1 for symmetric state

−1 for antisymmetric state

 𝜌′ : 𝑛𝑜𝑛 − 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 ⇒ 𝑇̂𝐴 ⨂𝐼̂𝐵 𝑛𝑜𝑡 𝑐𝑜𝑚𝑝. 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒

2. The Kraus theorem and the operator-sum representation

Th. Any superoperator satisfying properties (0), (1), (2), (3′ ) has an operator-sum

representation, i.e.,

$̂𝐴 (𝜌𝐴 ) = ∑ 𝑀𝜇 𝜌𝐴 𝑀𝜇+ , 𝑤𝑖𝑡ℎ ∑ 𝑀𝜇+ 𝑀𝜇 = 𝐼𝐴 𝑣𝑒𝑟𝑦 𝑔𝑒𝑛𝑒𝑟𝑎𝑙

𝜇 𝜇

Proof: (see Preskill’s)

Say derivation from a unitary representation $̂𝐴 × $̂𝐵 ( |𝜓⟩𝐴𝐵 ⟨𝜓| ) =

∑𝜇 𝑞𝜇 |𝜙𝜇 ⟩ ⟨𝜙𝜇 |
𝐴𝐵

Comments: 1) how many 𝑀𝜇 (Kraus operators)?

At most 𝑁 2 (𝑁 = dim(𝐻𝐴 )) 𝑁 2 = dim(𝐻𝐴 ⨂𝐻𝐴 )

2) how ambiguous?

In unitary derivation 𝑀𝜇 = 𝐵⟨𝜇|𝑈𝐴𝐵 |𝐸⟩𝐵

Choose a different basis |𝜈⟩

𝑁𝜈 = 𝐵⟨𝜈|𝑈𝐴𝐵 |𝐸⟩𝐵 = ∑ 𝐵⟨𝜈|𝜇⟩⟨𝜇|𝑈𝐴𝐵 |𝐸⟩𝐵 = ∑ 𝑈𝜈𝜇 𝑀𝜇 , 𝑈𝜈𝜇 ≡ ⟨𝜈|𝜇⟩

𝜇 𝜇

Equivalent $̂ iff 𝑀𝜇 , 𝑁𝜈 are related by a unitary transfer.

Comment: correspondence (move to sec. 3)

mixed state superoperator

ensemble Rep. 𝜌𝐴 = ∑𝑖 |𝑖⟩⟨𝑖| ⇔ Kraus Rep. $̂(𝜌𝐴 ) = ∑𝑖 𝑀𝜇 𝜌𝑀𝜇+

purification 𝜌𝐴 = 𝑡𝑟𝐵 ( |𝜓⟩𝐴𝐵 ⟨𝜓| ) ⇔ $̂(𝜌𝐴 ) = 𝑡𝑟𝐵 (𝑈𝐴𝐵 (𝜌𝐴 ⊗ |𝐸⟩𝐵 ⟨𝐸|)𝑈𝐴𝐵
+
)

3. Derivation of the Master Equation from Kraus rep.

𝜌(𝑡 + 𝛿𝑡) = $̂(𝜌(𝑡)) = ∑𝜇 𝑀𝜇 𝜌(𝑡)𝑀𝜇+ (No initial entanglement, no memory for B)

Assume 𝛿𝑡 is small, 𝜌(𝑡 + 𝛿𝑡) = 𝜌(𝑡) + 𝛿𝑡𝜌̇

 One 𝑀𝜇 must be close to I, let 𝑀0 = 𝐼 + 𝛿𝑡(−𝑖𝐻 + 𝐾)

∑ 𝑀𝜇+ 𝑀𝜇 = 𝑀0+ 𝑀0 + ∑ 𝑀𝜇+ 𝑀𝜇 = 𝐼

𝜇 𝜇≠0

𝐼 + 2𝛿𝑡 ∙ 𝐾

So for 𝜇 ≠ 0 𝑀𝜇 ∼ √𝛿𝑡 Let 𝑀𝜇 = 𝐿𝜇 √𝛿𝑡

1
(∑𝜇 𝐿+ +
𝜇 𝐿𝜇 + 2𝐾 )𝛿𝑡 = 0, 𝐾 = − 2 ∑𝜇 𝐿𝜇 𝐿𝜇

Let 𝛿𝑡 → 0, Hamiltonian

𝜌(𝑡 + 𝛿𝑡) − 𝜌(𝑡) 1 1 +

𝜌̇ = = −𝑖[𝐻, 𝑃] + ∑ [𝐿𝜇 𝜌𝐿+𝜇 − 𝐿+
𝜇 𝐿𝜇 𝜌 − 𝜌𝐿𝜇 𝐿𝜇 ]
𝛿𝑡 𝜇 2 2

Master Eq. jump operators Lindblad form

4. Inverse theorem: Relation between operator-sum and unitary representation

Th: Any evolution $̂ on density operator 𝜌𝐴 with $̂(ρA ) = ∑𝜇 𝑀𝜇 𝜌𝐴 𝑀𝜇+ can be

implemented with a unitary evolution 𝑈𝐴𝐵 in the extended space 𝐻𝐴 ⨂𝐻𝐵 with

+
𝑡𝑟𝐵 [𝑈𝐴𝐵 (𝜌𝐴 ⨂|𝐸⟩𝐵 ⟨𝐸|)𝑈𝐴𝐵 ] = ∑𝜇 𝑀𝜇 𝜌𝐴 𝑀𝜇+

any ancilla state

Proof: Assume 𝜌𝐴 = ∑𝑖 𝑝𝑖 |𝜙𝑖 ⟩⟨𝜙𝑖 |

Define 𝑈𝐴𝐵 as 𝑈𝐴𝐵 (|𝜙𝑖 ⟩𝐴 ⨂|𝐸⟩𝐵 ) ≡ ∑𝜇 𝑀𝜇 |𝜙𝑖 ⟩𝐴 ⨂|𝜇⟩𝐵

(orthonormal states in 𝐻𝐵 )

Such 𝑈𝐴𝐵 exists because the above mapping is inner-product preserving

Inner product

𝑈𝐴𝐵 ( |𝜓𝑗 ⟩𝐴 ⨂|𝐸⟩𝐵 ) = ∑ 𝑀𝜈 |𝜙𝑗 ⟩𝐴 ⨂|𝜈⟩𝐵

𝜈
+
𝑟𝑖𝑔ℎ𝑡 = ∑ 𝐴⟨𝜙𝑗 |𝑀𝜇 𝑀𝜇 |𝜙𝑖 ⟩𝐴 = 𝐴⟨𝜙𝑗 |𝜙𝑖 ⟩𝐴 = 𝑙𝑒𝑓𝑡
𝜇

With such a 𝑈𝐴𝐵

 +
𝑡𝑟𝐵 [𝑈𝐴𝐵 (𝜌𝐴 ⨂|𝐸⟩𝐵 ⟨𝐸|)𝑈𝐴𝐵 ] = ∑ 𝑝𝑖 {𝑡𝑟𝐵 [∑ 𝑀𝜇 |𝜙𝑖 ⟩𝐴 ⨂|𝜇⟩𝐵 ⟨𝜈|⟨𝜙𝑖 |𝑀𝜇+ ]}
𝑖 𝜇,𝜈

= ∑ 𝑀𝜇 (∑ 𝑝𝑖 |𝜙𝑖 ⟩𝐴 ⟨𝜙𝑖 |) 𝑀𝜇+ = ∑ 𝑀𝜇 𝜌𝐴 𝑀𝜇+

𝜇 𝑖 𝜇

Comment: correspondence:

5. Examples of superoperator evolution: quantum channels.

$̂
𝜌 → $̂(𝜌) 𝜌 → 𝜌′ = $̂(𝜌)

1) Depolarizing channel: 1 − 𝑝 no error 𝐼̂

𝑝
For a qubit state a bit flip 𝜎𝑥 = 𝑋
3

𝑝
𝑝 a phase flip 𝜎𝑧 = 𝑍 error
3

𝑝
Error a bit – phase flip 𝜎𝑦 = 𝑌 = 𝑖𝑋𝑍
3

So Krause Rep. 𝜌 → $̂(𝜌) = ∑𝑖=0,1,2,3 𝑀𝜇 𝜌𝑀𝜇+ ,

𝑝 𝑝 𝑝
𝑤𝑖𝑡ℎ 𝑀0 = √1 − 𝑝𝐼, 𝑀1 = √ 𝑋, 𝑀2 = √ 𝑌, 𝑀3 = √ 𝑍
3 3 3
𝑝
𝜌′ = (1 − 𝑝)𝜌 + (𝑋𝜌𝑋 + 𝑌𝜌𝑌 + 𝑍𝜌𝑍)
3
1 𝑋𝑋𝑋=𝑋
In Bloch – sphere Rep. 𝜌 = 2 (𝐼 + 𝑝⃗ ∙ 𝜎⃗) (𝑋𝑌𝑋=−𝑌 )
1 4
𝜌′ = (𝐼 + 𝑝⃗′ ∙ 𝜎⃗), 𝑤𝑖𝑡ℎ 𝑝⃗′ = (1 − 𝑝)𝑝⃗
2 3
3
𝑝 = 4 complete randomized shrinking ball

𝜌̇ = 𝛾(𝑋𝜌𝑋 + 𝑌𝜌𝑌 + 𝑍𝜌𝑍 − 3𝜌)

2) Phase-damping channel

1 − 𝑝, no error, I

𝑝, a phase flip error, 𝜎𝑧 = 𝑍

Krouse operator 𝑀0 = √1 − 𝑝𝐼, 𝑀1 = √𝑝𝑍

𝜌00 𝜌01
𝜌 = (𝜌 𝜌11 )
10
 𝜌00 (1 − 2𝑝)𝜌01
$̂(𝜌) = (1 − 𝑝)𝜌 + 𝑝𝑍𝜌𝑍 = ( )
(1 − 2𝑝)𝜌10 𝜌11
dephasing decoherence

3) Amplitude-damping channel. (spontaneous emission)

Def. |0⟩𝐴 |𝐸⟩ → |0⟩𝐴 |𝐸⟩

|1⟩𝐴 |𝐸⟩ → √1 − 𝑝|1⟩𝐴 |𝐸⟩ + √𝑝|0⟩𝐴 |𝐸 ′ ⟩

1 0
Kraus operators 𝑀0 = ( ) , 𝑀1 = (0 √𝑝) ∝ 𝜎
−
0 √1 − 𝑝 0 0
𝑀0+ 𝑀0 + 𝑀1+ 𝑀1 = 𝐼

𝜌00 + 𝑝𝜌11 √1 − 𝑝𝜌01 𝑝→1 𝜌00 0

𝜌 → $̂(𝜌) = ( )→ ( )
√1 − 𝑝𝜌10 (1 − 𝑝)𝜌11 0 0
𝛿𝑡 𝑝 𝛿𝑡
1−𝑇 ≈1−2 𝑝=𝑇 mixed state → pure state
2 1

𝑇2 = 2𝑇1 (𝑓𝑜𝑟 𝑎𝑚𝑝𝑙𝑖𝑡𝑢𝑑𝑒 − 𝑑𝑎𝑚𝑝𝑖𝑛𝑔 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 1)

1 1
𝜌̇ = 𝛾(𝜎− 𝜌𝜎+ − 𝜎+ 𝜎− 𝜌 − 𝜌𝜎+ 𝜎− )
2 2
𝐻𝑊2 3.1 3.2 3.6

§ 2.7 Generalized measurements: POVM

POVM (positive-operator-valued-measure)

1. Measurements in composite systems and POVM

Review Von Neumann measurement: 𝑃𝑟𝑜𝑗𝑒𝑐𝑡𝑜𝑟𝑠 {𝐸𝑖 }

(orthogonal) 𝐸𝑖 𝐸𝑗 = 𝛿𝑖𝑗 𝐸𝑖

∑𝑖 𝐸𝑖 = 𝐼

|𝜓⟩ → 𝐸0 |𝜓⟩ 𝑝𝑖 = ⟨𝜓|𝐸𝑖 |𝜓⟩ = 𝑡𝑟(𝜌)

𝜌 → 𝐸𝑖 𝜌𝐸𝑖

For a mixed state 𝜌 → 𝜌′ = 𝐸𝑖 𝜌𝐸𝑖 , 𝑝𝑖 = 𝑡𝑟(𝐸𝑖 𝜌)

Extension to composite system

What is its form in reduced system A?

𝜌𝐴𝐵 = 𝜌𝐴 ⨂𝜌𝐵 (extension of 𝜌𝐴 , arbitrary 𝜌𝐵 )

Probability 𝑝𝑖 = 𝑡𝑟𝐴𝐵 (𝐸𝑖 𝜌𝐴𝐵 ) = 𝑡𝑟𝐴 [𝜌𝐴 𝑡𝑟𝐵 (𝐸𝑖 𝜌𝐵 )] ≡ 𝑡𝑟𝐴 (𝜌𝐴 𝐹𝑖 )

𝐹𝑖 ≡ 𝑡𝑟𝐴𝐵 (𝐸𝑖 𝜌𝐵 ) (an operator in space 𝐻𝐴 )

Properties of 𝐹𝑖 :

1) Hermitian 𝐹𝑖+ = 𝑡𝑟𝐵 (𝜌𝐵+ 𝐸𝑖+ ) = 𝐹𝑖

2) Positive, for any state |𝜙⟩𝐴 , let 𝜌𝐴 ≡ |𝜙⟩𝐴 ⟨𝜙|

𝐴⟨𝜙|𝐹𝑖 |𝜙⟩𝐴 = 𝑡𝑟𝐴𝐵 (𝐸𝑖 𝜌𝐵 ⨂𝜌𝐴 ) ≥ 0

Positive

3) Complete ∑𝑖 𝐹𝑖 = 𝑡𝑟𝐵 (∑𝑖 𝐸𝑖 𝜌𝐵 ) = 𝑡𝑟𝐵 (𝐼𝐴𝐵 𝜌𝐵 ) = 𝐼𝐴

A set of {𝐹𝑖 } with properties (1)-(3) is a POVM. In general, 𝐹𝑖 𝐹𝑗 ≠ 𝛿𝑖𝑗 𝐹𝑖

Remark: 1) States after POVM measurements?

In general, no simple answer

?
𝜌 → 𝐹𝑖 𝜌𝐹𝑖 (×)

Example: |0⟩, 𝜌 = |0⟩⟨0|

Extension |0⟩1 , |0⟩2

1
Bell Measurements |𝜓 ± ⟩ = (|01⟩12 ± |10⟩12 ) ← Bell Basis
√2
1
|𝜙 ± ⟩ = (|00⟩12 ± |11⟩12 )
√2
𝐸1 = |𝜓 + ⟩⟨𝜓 + |, 𝐸2 = |𝜓 − ⟩⟨𝜓 − |, 𝐸3 = |𝜙 + ⟩⟨𝜙 + |, 𝐸4 = |𝜙 − ⟩⟨𝜙 − |
 {𝐸0 }：Von-Neumann Projectors, 𝐹𝐵 = 𝑡𝑟2 (𝐸𝐵 |0⟩2 ⟨0|) = |0⟩2 ⟨0|

With outcome “B”, |00⟩12 → |𝜙 + ⟩12

1 1
𝜌11 = 𝑡𝑟2 (|𝜙 + ⟩12 ⟨𝜙 + |) = 2 (|0⟩1 ⟨0| + |1⟩1 ⟨1|) = 2 𝐼1 ≠ 𝐹𝐵 |0⟩ ⟨0|𝐹𝐵

2. Inverse (Neumark) theorem: Relation between POVM and projective

measurements

Th. Any POVM {𝐹𝑖 } in 𝐻𝐴 can be implemented with a projective

measurement in extended space 𝐻𝐴 ⨂𝐻𝐵 with the same probability for

outcomes, i.e. 𝑡𝑟(𝐹𝑖 𝜌𝐴 ) = 𝑝𝑖 = 𝑡𝑟𝐵 (𝐸𝑖 𝜌𝐴 ⨂|𝐸⟩⟨𝐸|)

Proof: Let 𝐹𝑖 = ∑𝑘 𝑝𝑖𝑘 |𝜙𝑖𝑘 ⟩𝐴 ⟨𝜙𝑖𝑘 |

Define |𝜓𝑖𝑘 ⟩𝐴𝐵 = √𝑝𝑖𝑘 |𝜙𝑖𝐴 ⟩𝐴 ⨂|𝐸⟩𝐵 + |0⟩𝐴 ⨂|𝜇𝑖𝑘 ⟩𝐵

Orthogonal to |𝐸⟩𝐵

Choose |𝜇𝑖𝑘 ⟩𝐵 so that 𝐴𝐵⟨𝜓𝑖 ′ 𝑘 ′ |𝜓𝑖𝑘 ⟩𝐴𝐵 = 𝛿𝑖𝑖 ′ 𝛿𝑘𝑘 ′

∑𝑖𝑘|𝜓𝑖𝑘 ⟩𝐴𝐵 ⟨𝜓𝑖𝑘 | = 𝐼𝐴𝐵

Let 𝐸𝑖 = ∑𝑘|𝜓𝑖𝑘 ⟩𝐴𝐵 ⟨𝜓𝑖𝑘 |

Then ∑𝑖 𝐸𝑖 = 𝐼𝐴𝐵 , 𝑡𝑟𝐵 (𝐸𝑖 |𝐸⟩𝐵 ⟨𝐸|) = 𝐹𝑖

𝐸𝑖 𝐸𝑗 = 𝐸𝑖 𝛿𝑖𝑗

Ex. A POVM for a qubit

𝑛⃗⃗1 + 𝑛⃗⃗2 + 𝑛⃗⃗3 = 0

 1
Define 𝐹𝑖 = 3 (𝐼 + 𝑛⃗⃗𝑖 ∙ 𝜎⃗) (𝑖 = 1,2,3)

∑𝑖 𝐹𝑖 = 𝐼 ⇒ {𝐹𝑖 } is a POVM

1 {𝐹𝑖 } 1 1
A state 𝜌 = 2 (𝐼 + 𝑝⃗ ∙ 𝜎⃗) ⇒ 𝑝(𝐹𝑖 ) = 𝑡𝑟(𝐹𝑖 𝜌) = 3 + 6 𝑡𝑟(𝑛⃗⃗𝑖 ∙ 𝜎⃗ 𝑝⃗ ∙ 𝜎⃗)
1
= (1 + 𝑛⃗⃗𝑖 ∙ 𝑝⃗)
3