Noise Modeling in Python

Noise Modeling in Python

Real quantum hardware is imperfect. To write quantum programs that work on real devices, you must model and account for noise. Python gives us the tools to build realistic noise models and test circuits against them before submitting to hardware.

Types of Quantum Noise

Coherent Errors Systematic rotation errors, e.g., a gate that should apply RX(π) actually applies RX(π + 0.01). Coherent errors add up across a circuit (they don't average out).

Modeling: Add a small random rotation after each gate.

Depolarizing Noise The most commonly used simplified model. After a gate, with probability p, apply a uniformly random Pauli error (X, Y, or Z).

def depolarizing_channel(density_matrix, p):
    rho = density_matrix
    X = [[0,1],[1,0]]; Y = [[0,-1j],[1j,0]]; Z = [[1,0],[0,-1]]
    rho_new = (1-p)*rho + (p/3)*(X@rho@X + Y@rho@Y + Z@rho@Z)
    return rho_new

In state vector simulation, depolarizing is equivalent to: with probability p, sample and apply X, Y, or Z.

T1 Relaxation (Energy Decay) |1⟩ spontaneously decays to |0⟩ with characteristic time T1. The amplitude damping channel:

A₀ = [[1, 0], [0, √(1-γ)]], A₁ = [[0, √γ], [0, 0]]

where $\gamma =1-e^{-t/T_1}$ is the decay probability for gate time t.

Typical values: T1 ≈ 100 μs to 1 ms (superconducting qubits in 2024).

T2 Dephasing (Phase Decay) Random phase kicks accumulate, destroying phase coherence. The dephasing channel adds Z noise:

Characteristic time T2 ≤ 2T1 (T2 cannot exceed 2T1 by quantum mechanics).

For a gate of duration t, dephasing probability: $p_Z=(1-e^{-t/T_2})/2$.