Quantum State Tomography

An interactive guide to reconstructing quantum states from Pauli measurements

1 The Pauli Operators

The Pauli matrices are a set of three 2×2 complex matrices (plus the identity) that form the most natural language for describing single-qubit quantum states and measurements. Together with the identity I, they form a basis for all 2×2 Hermitian matrices—which is exactly the space in which density matrices live.

σ₀ = I

Identity

σ_x = X

Bit-flip / NOT gate

σ_y = Y

Bit & phase flip

0-i

σ_z = Z

Phase-flip

0-1

Key Properties

Hermitian and Unitary: Each Pauli matrix equals its own conjugate-transpose (P = P†) and its own inverse (P² = I). This means they are simultaneously valid as observables and as quantum gates.
Eigenvalues ±1: Every Pauli matrix has exactly two eigenvalues, +1 and −1. A measurement always returns one of these two outcomes.
Traceless (except I): Tr(σ_k) = 0 for k ∈ {x, y, z}. This makes them ideal for parameterizing the traceless part of a density matrix.
Orthogonal basis: The set {I, X, Y, Z} satisfies Tr(σ_jσ_k) = 2δ_jk, forming an orthogonal basis for the real vector space of 2×2 Hermitian matrices.

Eigenstates

Each Pauli operator defines a measurement axis on the Bloch sphere:

Operator	+1 Eigenstate	−1 Eigenstate	Bloch Axis
Z	\|0⟩	\|1⟩	Poles (north / south)
X	\|+⟩ = (\|0⟩+\|1⟩)/√2	\|−⟩ = (\|0⟩−\|1⟩)/√2	Equator (right / left)
Y	\|+i⟩ = (\|0⟩+i\|1⟩)/√2	\|−i⟩ = (\|0⟩−i\|1⟩)/√2	Equator (front / back)

2 From Pauli Operators to the Bloch Sphere

Because {I, X, Y, Z} form a complete basis, any single-qubit density matrix can be written as:

ρ = ½(I + r_xX + r_yY + r_zZ)
where r = (r_x, r_y, r_z) is the Bloch vector with ||r|| ≤ 1

The three Bloch vector components are directly the expectation values of the Pauli operators:

r_x = ⟨X⟩ = Tr(Xρ), r_y = ⟨Y⟩ = Tr(Yρ), r_z = ⟨Z⟩ = Tr(Zρ)

    This is the central insight of quantum state tomography:
    if you can estimate the three expectation values ⟨X⟩, ⟨Y⟩, ⟨Z⟩
    from measurements, you can reconstruct the full density matrix.
    The Pauli operators provide exactly the right "coordinate axes" for this reconstruction.
  

3 Interactive Explorer

Adjust the qubit state using the θ and φ sliders (or pick a preset). The Bloch sphere, expectation values, and density matrix all update in real time.

Bloch Sphere

True state Estimate

Drag to rotate the view

State Controls

|ψ⟩ = |0⟩

θ (polar) 0.00

φ (azimuthal) 0.00

Purity ||r|| 1.00

Expectation Values

⟨X⟩

0.00

⟨Y⟩

0.00

⟨Z⟩

1.00

Density Matrix ρ

4 Tomography Simulation

In a real experiment, you cannot read expectation values directly—you collect individual measurement outcomes (±1) from many copies of the state and estimate the expectation values from the statistics. This is quantum state tomography.

Prepare many identical copies of the unknown state ρ.
Measure a fraction in the X, Y, and Z bases.
Compute the sample average of outcomes in each basis to estimate ⟨X⟩, ⟨Y⟩, ⟨Z⟩.
Reconstruct: ρ̂ = ½(I + r̂_xX + r̂_yY + r̂_zZ).

    Try it yourself: the state from the explorer above is the "true" state.
    The dashed ghost bars show the true expectation values for comparison.
  

Speed:

Noise: 0% depolarization

Shots per basis: 0 | Total measurements: 0

X̂

—

Ŷ

—

Ẑ

—

Reconstructed ρ̂: no data yet

Trace distance vs. shots per basis — dashed line shows 1/√N scaling

    Why 1/√N? Each Pauli expectation value is estimated by averaging
    N independent ±1 outcomes. By the central limit theorem, the standard error
    is σ/√N where σ ≤ 1. So doubling the accuracy requires
    four times as many measurements—a fundamental statistical cost.
  

5 Mystery State Challenge

Think you understand tomography? In this challenge, a random unknown state is prepared and the Bloch sphere is hidden. Use only your measurement data to reconstruct the state, then reveal the answer to see how close you got!

6 Scaling Up: Multi-Qubit Tomography

For an n-qubit system, the density matrix lives in a 2ⁿ×2ⁿ space. The Pauli operators generalize via tensor products:

Basis: {σ_j₁ ⊗ σ_j₂ ⊗ … ⊗ σ_{j_n}} where j_k ∈ {I, X, Y, Z}

This gives 4ⁿ basis elements, but the identity component is fixed by Tr(ρ) = 1, so you need to measure 4ⁿ − 1 independent expectation values.

The Exponential Cost

Qubits	Parameters	Measurement Settings
1	3	3 (X, Y, Z)
2	15	9 (XX, XY, … ZZ)
3	63	27
n	4ⁿ − 1	3ⁿ

This exponential scaling is why full state tomography is practical only for small systems. For larger systems, techniques like compressed sensing, matrix product state tomography, or shadow tomography dramatically reduce the required measurements.

Interactive: 2-Qubit Pauli Correlations

Select a 2-qubit state to see all 15 Pauli expectation values. Notice how entangled states have non-zero correlation terms that product states lack.

Product state: correlations equal products of single-qubit values.

7 Self-Check

Test your understanding of the key concepts covered above.

1. How many independent real parameters are needed to fully describe a single-qubit density matrix?

2. Why are the Pauli matrices the natural choice for tomography measurements?

3. If you double the number of measurement shots per basis, the statistical error approximately:

4. For a 5-qubit system, how many independent Pauli expectation values must be measured?

5. In a 2-qubit tomography experiment, you measure ⟨ZZ⟩ = −1. Can you conclude the qubits are entangled?