WrathfulSpatula
commented
1 year ago We have some significant work on "externally applied" depolarizing noise channels, now, for both "weak" and "strong" simulation conditions. See d5fdf00, for example. We can close this, though improvements on that work continue.
Qrack has long concerned itself with being an HPC simulator for ideally noiseless simulations. However, there is likely a demand for Qrack as a noisy simulator, for various noise models including depolarizing noise. The basic idea is that a single noise or fidelity parameter causes a qubit to relax from a pure state into a combination of the originally pure state and a maximally mixed state. The maximally mixed component is entirely random between the two state poles along any rotation of the Bloch sphere for the single qubit, (effectively at the dead center of the 3-ball volume interior to the 2-sphere Bloch sphere of perfectly "pure" quantum states).
With new approximation methods in vm6502q.v6.0.0, we might effectively enable simulating error correction after depolarizing noise arises, by clamping states that locally look partially mixed in secant volumes to exact Pauli poles of the local single qubit Bloch sphere. The appearance of a locally mixed state, for a single qubit, can be created through ideal "pure state" entanglement with other qubits. (For example, either end of an ideal Bell pair looks locally maximally mixed.) This quickly becomes prohibitive for simulating noise in the ideal, but it's a jumping-off point for thinking about novel ways to use Qrack for noise and error correction simulation.