Density Matrix

qc/noise qc/math

A statevector $|\psi ⟩$ can only describe a qubit that’s in a definite (if possibly superposed) pure state. Once noise enters the picture — a qubit weakly coupled to its environment, an imperfect gate, a noisy measurement — you often need to describe a mixture of possible states, each with some classical probability. That’s what a density matrix $\rho$ is for.

For a pure state, $\rho =|\psi ⟩⟨\psi |$ — no new information, just a different notation. The real value shows up for mixed states, where $\rho$ is a weighted sum over several possible pure states:

$$\rho =\sum _i{p}_i|{\psi }_i⟩⟨{\psi }_i|$$

This is the natural language for describing noise: every noise model in this lab — Depolarizing Noise, Pauli Error Model, Thermal Relaxation — is really just a rule for how $\rho$ transforms under an imperfect operation.

Open vs. closed systems

Key insight: noise is what happens when your qubit isn’t actually isolated — it’s weakly entangled with an environment (“bath”) you don’t have access to. The density matrix $\rho$ is what you get by mathematically averaging out (formally, taking the partial trace over) that inaccessible bath, leaving only the qubit’s own reduced description. The general law governing how an open system’s $\rho$ evolves over time is the Lindblad master equation — it replaces the closed-system Schrödinger equation once you can’t ignore the environment. This is the formal justification underneath NoiseLearnerV3’s Pauli-Lindblad noise model: that model’s $\mathcal{L}(\rho )={\sum }_k{\lambda }_k(P_k\rho P_k-\rho )$ is a Lindblad-form generator, specialized to Pauli jump operators.

Related

• Depolarizing Noise
• Pauli Error Model
• Thermal Relaxation
• NoiseLearnerV3 and Pauli-Lindblad Models — the Pauli-Lindblad model this open-system framing justifies

Self-Check

• Could you explain why a statevector alone isn’t enough once noise enters the picture?
• What does $\rho =|\psi ⟩⟨\psi |$ tell you for a pure state, and why is it “no new information”?
• Why is the density matrix described as “the natural language for describing noise”?
• What does the partial trace conceptually do, and why does it produce exactly the density matrix?