Skip to content

Quantum Measurement

Time: 35 minutes | Difficulty: 🟡 Intermediate

Overview

Measurement is where quantum mechanics meets reality. It's the bridge between the probabilistic quantum world and definite classical outcomes. Understanding measurement is crucial for quantum computing because it determines how we extract information from quantum algorithms.

What you'll learn: - What happens during quantum measurement - Different types of measurements (projective, POVM) - How measurement basis affects outcomes - Quantum state tomography - Why measurement is both powerful and limiting


The Measurement Postulate

What Is Quantum Measurement?

Classical measurement: - Read a value without disturbing the system - Example: Check thermometer temperature - Can repeat measurement, get same answer

Quantum measurement: - Fundamentally disturbs the system - Collapses superposition to definite state - Cannot repeat and get same answer (state changed!) - Probabilistic outcomes

The Born Rule

Fundamental law of quantum measurement:

If a qubit is in state |ψ⟩ = α|0⟩ + β|1⟩, then:

Probability of measuring |0⟩ = |α|² = α*α
Probability of measuring |1⟩ = |β|² = β*β

where * denotes complex conjugate

Normalization: |α|² + |β|² = 1 (probabilities must sum to 100%)

Example:

|ψ⟩ = (1/2)|0⟩ + (√3/2)|1⟩

P(0) = |1/2|² = 1/4 = 25%
P(1) = |√3/2|² = 3/4 = 75%

Wave Function Collapse

Before measurement:

|ψ⟩ = α|0⟩ + β|1⟩

Superposition of both states
Qubit has both 0-ness and 1-ness

During measurement:

System interacts with measurement apparatus
Entanglement with environment
Irreversible process begins

After measurement:

If measured 0: State becomes |0⟩
If measured 1: State becomes |1⟩

Superposition destroyed!
All information about α and β lost (except their magnitudes)

This is irreversible: You cannot "unmeasure" and restore the original superposition.


Computational Basis Measurement

Standard Z-Basis Measurement

The most common measurement in quantum computing measures in the computational basis {|0⟩, |1⟩}.

Measurement operators:

M₀ = |0⟩⟨0| = [1  0]  Projects onto |0⟩
              [0  0]

M₁ = |1⟩⟨1| = [0  0]  Projects onto |1⟩
              [0  1]

Example:

State: |+⟩ = (|0⟩ + |1⟩)/√2

Z-basis measurement:
P(0) = |⟨0|+⟩|² = |(|0⟩/√2)|² = 1/2
P(1) = |⟨1|+⟩|² = |(|1⟩/√2)|² = 1/2

Result: 50% chance each, state collapses to |0⟩ or |1⟩

Multi-Qubit Measurement

For n qubits:

Computational basis has 2ⁿ possible outcomes.

Example (2 qubits):

|ψ⟩ = (1/2)|00⟩ + (1/2)|01⟩ + (1/√2)|10⟩

Measurement probabilities:
P(00) = |1/2|² = 1/4 = 25%
P(01) = |1/2|² = 1/4 = 25%
P(10) = |1/√2|² = 1/2 = 50%
P(11) = 0

Partial measurement:

You can measure just some qubits!

State: (|00⟩ + |11⟩)/√2

Measure only first qubit:
- If result is 0: State becomes |00⟩ (second qubit must be 0)
- If result is 1: State becomes |11⟩ (second qubit must be 1)

Entanglement ensures correlation!

Different Measurement Bases

Why Different Bases Matter

Key insight: The choice of measurement basis determines what information you extract!

Analogy: Measuring position vs momentum - Both are valid measurements - Give different information - Cannot measure both precisely (uncertainty principle)

X-Basis (Hadamard Basis)

Measurement basis: {|+⟩, |−⟩}

|+⟩ = (|0⟩ + |1⟩)/√2
|−⟩ = (|0⟩ − |1⟩)/√2

How to measure in X-basis: 1. Apply Hadamard gate 2. Measure in Z-basis (computational basis) 3. Interpret: 0 → |+⟩, 1 → |−⟩

Example:

State: |0⟩

X-basis measurement:
|0⟩ = (|+⟩ + |−⟩)/√2

P(+) = |⟨+|0⟩|² = 1/2
P(−) = |⟨−|0⟩|² = 1/2

Result: 50% chance each

But if state was |+⟩:

P(+) = |⟨+|+⟩|² = 1 = 100%
P(−) = 0

Lesson: |+⟩ is indefinite in Z-basis but definite in X-basis!

Y-Basis

Measurement basis: {|↻⟩, |↺⟩}

|↻⟩ = (|0⟩ + i|1⟩)/√2  (right circular)
|↺⟩ = (|0⟩ − i|1⟩)/√2  (left circular)

How to measure in Y-basis: 1. Apply S† gate, then Hadamard 2. Measure in Z-basis 3. Interpret results

Used in: Quantum state tomography, error correction

Arbitrary Basis

General single-qubit measurement:

Any orthonormal basis {|a⟩, |b⟩} where:

⟨a|b⟩ = 0 (orthogonal)
⟨a|a⟩ = ⟨b|b⟩ = 1 (normalized)

On Bloch sphere: - Measurement basis = Two opposite points - Measurement projects state onto one pole - Uncertainty maximized at equator


Projective Measurements

Mathematical Formalism

Projective measurement is defined by:

  1. Measurement operators: Set of projectors {Πₘ}
  2. Completeness: ΣΠₘ = I (identity)
  3. Orthogonality: ΠₘΠₙ = δₘₙΠₘ

For computational basis:

Π₀ = |0⟩⟨0|
Π₁ = |1⟩⟨1|

Π₀ + Π₁ = I ✓
Π₀Π₁ = 0 ✓

Measurement Process

Given state |ψ⟩ and measurement {Πₘ}:

  1. Probability of outcome m:

    P(m) = ⟨ψ|Πₘ|ψ⟩ = ||Πₘ|ψ⟩||²
    

  2. Post-measurement state:

    |ψₘ⟩ = Πₘ|ψ⟩ / √P(m)
    
    (Normalized projection)
    

Example:

State: |ψ⟩ = (3/5)|0⟩ + (4/5)|1⟩
Measure in computational basis

Outcome 0:
P(0) = |3/5|² = 9/25
Post-measurement: |0⟩

Outcome 1:
P(1) = |4/5|² = 16/25
Post-measurement: |1⟩

Observable Measurements

Alternative formalism: Measurements as observables (Hermitian operators)

For Z-basis measurement:

Z = |0⟩⟨0| − |1⟩⟨1| = [1   0]
                       [0  -1]

Eigenvalues: +1 (for |0⟩), -1 (for |1⟩)
Eigenvectors: |0⟩, |1⟩

Expectation value:

⟨Z⟩ = ⟨ψ|Z|ψ⟩ = |α|² − |β|²

For |ψ⟩ = α|0⟩ + β|1⟩

Example:

|+⟩ = (|0⟩ + |1⟩)/√2

⟨Z⟩ = 1/2 − 1/2 = 0

Makes sense: Equal probability for ±1, so average is 0


Generalized Measurements (POVM)

Beyond Projective Measurements

POVM (Positive Operator-Valued Measure) allows more general measurements.

Why needed: - Measure in overcomplete basis - Discriminate between non-orthogonal states (optimally) - Model realistic measurement processes

POVM Elements

Set of positive operators {Eₘ}:

Eₘ ≥ 0 (positive semi-definite)
ΣEₘ = I (completeness)

But: Eₘ need not be projectors!

Measurement probability:

P(m) = ⟨ψ|Eₘ|ψ⟩ = Tr(Eₘ|ψ⟩⟨ψ|)

Key difference: POVM doesn't specify post-measurement state!

Example: Trine POVM

Three measurement outcomes for a qubit:

E₀ = (2/3)|0⟩⟨0|
E₁ = (2/3)|+120°⟩⟨+120°|
E₂ = (2/3)|−120°⟩⟨−120°|

where states are separated by 120° on Bloch sphere

Properties: - Three outcomes (not just two!) - Optimal for certain state discrimination tasks - Cannot be implemented as projective measurement

Applications: - Quantum cryptography (BB84 protocol) - Optimal state discrimination - Modeling noisy measurements

Naimark's Theorem

Important result: Every POVM can be realized as a projective measurement in a larger Hilbert space.

Practical: Introduce ancilla qubit, do controlled operations, measure ancilla.


Sequential Measurements

Measurement Changes the State

First measurement:

|ψ⟩ = α|0⟩ + β|1⟩

Z-measurement → Outcome 0 → State becomes |0⟩

Second measurement (same basis):

State: |0⟩

Z-measurement → Outcome 0 → State remains |0⟩

Always get same answer!

Second measurement (different basis):

State: |0⟩ (after first Z-measurement)

X-measurement: |0⟩ = (|+⟩ + |−⟩)/√2
P(+) = P(−) = 1/2

Different answer possible!

Non-Commuting Observables

Key concept: Order matters if observables don't commute!

Commuting observables:

[A, B] = AB − BA = 0

Can measure simultaneously
Example: Z₁ and Z₂ (measurements on different qubits)

Non-commuting observables:

[X, Z] = XZ − ZX ≠ 0

Cannot measure simultaneously
Uncertainty principle applies!

Example:

X = [0  1],  Z = [1   0]
    [1  0]       [0  -1]

XZ = [0  -1],  ZX = [0   1]
     [1   0]        [-1  0]

XZ ≠ ZX → Don't commute!

Complementary Measurements

Definition: Two measurements are complementary if: - Eigenbases are mutually unbiased - Knowing outcome of one gives no information about the other

Example: X and Z

Z eigenstates: |0⟩, |1⟩
X eigenstates: |+⟩, |−⟩

|0⟩ = (|+⟩ + |−⟩)/√2 → Equal probability in X-basis
|+⟩ = (|0⟩ + |1⟩)/√2 → Equal probability in Z-basis

Perfectly complementary!

Applications: - Quantum cryptography (BB84) - Measuring trade-offs - Complete state characterization


Quantum State Tomography

The Challenge

Problem: A single measurement only gives partial information.

State: |ψ⟩ = α|0⟩ + β|1⟩

Z-measurement gives:
- Probability |α|² or |β|²
- But not α and β themselves (which include phase!)

Need multiple measurements to reconstruct full state!

Single-Qubit Tomography

Recipe to reconstruct any qubit state:

  1. Prepare many copies of |ψ⟩
  2. Measure in three bases:
  3. Z-basis: Get ⟨Z⟩
  4. X-basis: Get ⟨X⟩
  5. Y-basis: Get ⟨Y⟩

  6. Reconstruct state:

    |ψ⟩ uniquely determined by (⟨X⟩, ⟨Y⟩, ⟨Z⟩)
    
    These are the Bloch sphere coordinates!
    

Example:

Measurements give:
⟨X⟩ = 0.6
⟨Y⟩ = 0.0
⟨Z⟩ = 0.8

Bloch vector: (0.6, 0.0, 0.8)

This corresponds to:
|ψ⟩ ≈ 0.95|0⟩ + 0.32|1⟩

Multi-Qubit Tomography

For n qubits:

Need to measure in 3ⁿ different bases!

1 qubit: 3 measurements (X, Y, Z)
2 qubits: 9 measurements (XX, XY, XZ, YX, YY, YZ, ZX, ZY, ZZ)
3 qubits: 27 measurements
10 qubits: 59,049 measurements!

Challenge: Exponentially expensive!

Workarounds: - Compressed sensing (fewer measurements) - Assume specific structure (MPS, low rank) - Partial tomography (subset of observables)

Process Tomography

Goal: Characterize a quantum gate, not just a state.

Method: 1. Prepare complete set of input states 2. Perform operation 3. Measure output states (full tomography) 4. Reconstruct process matrix

Cost: Even more expensive than state tomography!

Used for: Quantum gate calibration, error characterization


Measurement in Quantum Algorithms

Measurement Strategies

Different algorithms use measurement differently:

1. Final measurement only (most common)

Example: Grover's algorithm
- Prepare superposition
- Apply quantum operations
- Measure at the end
- Get answer!

2. Mid-circuit measurement

Example: Quantum error correction
- Measure syndrome qubits during computation
- Use results to conditionally apply corrections
- Continue computation

3. Measurement-based quantum computing

Example: One-way quantum computing
- Prepare large entangled cluster state
- Computation is just measurements!
- Measurement order and basis determines computation

Extracting Information

The measurement challenge:

State before measurement: Exponential information (2ⁿ amplitudes)
After measurement: n classical bits

Where did the information go?

Answer: Most quantum information is destroyed!

Algorithm design: - Use interference to amplify correct answer - Make wrong answers destructively interfere - Final measurement extracts the answer with high probability

Repeated Measurements

Statistical accumulation:

Single measurement: Probabilistic outcome
Many measurements: Statistical distribution

Example: |ψ⟩ = 0.6|0⟩ + 0.8|1⟩

Theory: P(0) = 0.36, P(1) = 0.64

Experiment (1000 shots):
Measured 0: 357 times (≈36%)
Measured 1: 643 times (≈64%)

Current quantum computers: - Shots: 1000-10,000 typical - More shots = Better statistics - Trade-off: Time vs accuracy


Weak Measurements

What Are Weak Measurements?

Standard (strong) measurement: - Completely collapses state - Get definite answer - Destroys superposition

Weak measurement: - Gentle interaction - Partial information - Minimally disturbs state

How it works:

1. Weakly couple system to meter
2. Measure meter (not system directly)
3. System only slightly disturbed
4. Can repeat many times

Weak Values

Definition: Expectation value conditioned on both pre- and post-selection.

Aweak = ⟨ψf|A|ψi⟩ / ⟨ψf|ψi⟩

where:
|ψi⟩ = initial state (pre-selected)
|ψf⟩ = final state (post-selected)
A = observable being measured

Weird property: Weak values can be outside eigenvalue range!

Example:

Spin measurement: Eigenvalues are ±1
But weak value can be 100!

"How" is subtle (involves post-selection)

Applications: - Amplification of tiny effects - Precise measurements - Quantum foundations research

Note: Controversial topic, still being researched!


Measurement-Induced Entanglement

Measurement Can Create Entanglement!

Surprising fact: Measuring can entangle previously independent qubits.

Example: Bell state measurement

Start: Two separate qubits in |+⟩

Apply CNOT and Hadamard (measurement in Bell basis)

Result: Entangled state based on measurement outcome!

Measurement-based quantum computing: - Start with product state - Selective measurements create entanglement - Pattern of measurements implements algorithm

Measurement-Induced Phase Transitions

Recent discovery: In quantum many-body systems:

Few measurements: System remains entangled
Many measurements: Entanglement destroyed

Critical measurement rate → Phase transition!

Implications: - Quantum error correction threshold - Dynamics of quantum information - Active research area (2020s)


Practical Measurement

Real-World Implementation

Superconducting qubits:

Method: Dispersive readout
- Send microwave pulse
- Qubit state affects reflected signal
- Measure phase/amplitude of reflection
- Distinguish |0⟩ vs |1⟩

Trapped ions:

Method: Fluorescence detection
- Shine laser on ion
- |0⟩ state: Fluoresces (emits photons)
- |1⟩ state: Dark (no photons)
- Use CCD camera to detect photons

Photonic qubits:

Method: Single-photon detectors
- Photon hits detector → Click!
- No photon → No click
- Detect which path photon took

Measurement Errors

Real measurements aren't perfect:

Types of errors: 1. Preparation error: State isn't what you think 2. Thermal population: Excited states from environment 3. Relaxation during measurement: T₁ decay 4. Readout error: Misclassify the outcome

Example: Readout confusion matrix

Prepare |0⟩, measure:
- Get 0: 97% (correct)
- Get 1: 3% (error)

Prepare |1⟩, measure:
- Get 0: 5% (error)
- Get 1: 95% (correct)

Mitigation: - Calibration and characterization - Error mitigation techniques - Multiple measurements and majority vote


Measurement Backaction

The Observer Effect

Key principle: Measurement inherently disturbs quantum systems.

Not just technical limitation - Fundamental to quantum mechanics!

Example: Position measurement

Precisely measure position → Large uncertainty in momentum
(Heisenberg uncertainty principle)

Quantum Zeno Effect

Paradox: Watched pot never boils!

Quantum version:

System evolves: |ψ⟩ → |φ⟩ (slow transition)

Continuous measurement: State kept in |ψ⟩!

Measurement prevents evolution!

Why: Each measurement projects back to initial state.

Applications: - Error suppression - Quantum control - Slowing decoherence

Quantum Anti-Zeno Effect

Opposite: Frequent measurements can accelerate decay!

Depends on: - Measurement frequency - System Hamiltonian - Timing relative to natural oscillations


Summary: Quantum Measurement

Key concepts:

  1. Born rule: P(m) = |⟨m|ψ⟩|²
  2. Wave function collapse: Superposition → Definite state
  3. Basis dependence: Different bases extract different information
  4. Projective measurements: Orthogonal projectors
  5. POVM: Generalized measurements
  6. Tomography: Reconstruct state from measurements
  7. Backaction: Measurement disturbs the system

Measurement in quantum computing: - Extracts classical answer from quantum state - Only gives partial information (n bits from n qubits) - Algorithm design must account for measurement - Repeated measurements give statistics

Practical considerations: - Imperfect readout (errors) - Time required (shots × measurement time) - Mid-circuit measurement (more capability, more complexity)


Check Your Understanding

Question 1: If |ψ⟩ = (1/√5)|0⟩ + (2/√5)|1⟩, what's the probability of measuring 1?

Answer P(1) = |2/√5|² = 4/5 = 80%

Question 2: After measuring |+⟩ in the Z-basis and getting 0, what is the new state?

Answer The state collapses to |0⟩. The original superposition is destroyed, and all information about the state being |+⟩ is lost.

Question 3: How many different basis measurements do you need to completely characterize a single qubit?

Answer Three: X, Y, and Z basis measurements. These give you the three Bloch sphere coordinates needed to uniquely determine the state.

Question 4: Can you measure X and Z simultaneously with perfect precision?

Answer No! X and Z don't commute ([X,Z] ≠ 0), so they're complementary observables. Uncertainty principle applies - knowing one exactly means complete uncertainty in the other.

Question 5: If you measure a 10-qubit system, how many classical bits do you get?

Answer 10 bits (one per qubit). Even though the quantum state before measurement contained 2¹⁰ = 1024 complex amplitudes of information, measurement only extracts 10 classical bits.

What's Next?

Congratulations! You've completed the foundations of quantum computing. You now understand: - Classical vs quantum computing - Quantum mechanics basics - Qubits and the Bloch sphere - Entanglement - Measurement

Next steps:

Continue learning: - Part 2: Quantum Gates & Circuits - Build quantum programs - Part 3: Quantum Stack - Hardware to software - Part 4: Quantum Algorithms - Powerful algorithms

Get hands-on: - Setup Guide - Install quantum frameworks - First Quantum Circuit - Write code!


Additional Resources

Books: - Quantum Measurement by Paul Busch et al. - The Quantum Handshake by Yakir Aharonov (weak measurements)

Papers: - Born: "Zur Quantenmechanik der Stoßvorgänge" (1926) - Original Born rule - Busch: "Measurement in quantum mechanics" (2016 handbook)

Videos: - Quantum Measurement Problem - PBS Space Time - Measurement in Quantum Computing - Qiskit

Interactive: - Measurement Simulator - Quirk circuit simulator - Try state tomography experiments in Qiskit!


💡 Final insight: Measurement is where quantum probability becomes classical reality. Understanding measurement helps you understand both the power and limitations of quantum computing - you can compute with exponentially large superpositions, but extraction of the answer is the bottleneck!


🎉 Congratulations on completing Part 1: Quantum Foundations! You now have the conceptual foundation to understand quantum computing. Time to build circuits and write algorithms! 🚀