You’re staring at your job output, a perfectly formed Bell state, or so you think. The fidelity numbers look *almost* there, but something’s off. It’s not the readout errors or the visible crosstalk; it’s **unitary contamination**.
Unitary Contamination: A Deep NISQ Circuit’s Hidden Threat
The performance ceiling in deep NISQ circuits may not be dictated by gate or readout fidelity, but by a subtle form of decoherence: unitary contamination. This degrades your computation *before* readout even becomes a factor.
Deep NISQ Circuits and Unitary Contamination
Think about it. You’ve got “good” qubits and “poison qubits.” When poison qubits interact with unitary operations, they *contaminate* the coherent evolution of their neighbors. It’s coherent corruption, not just standard depolarizing noise.
Deep NISQ Circuits and Unitary Contamination’s Measurement Taint
Standard error correction is largely oblivious to this. The measurement outcome is a distorted echo, tainted by rogue elements. Your algorithms should prioritize high-fidelity connected subgraphs, and use measurement itself as a diagnostic tool.
Unitary Contamination in Deep NISQ Circuits
Re-evaluate your ‘good’ qubits. Watch the ‘Poison Qubit’ Threshold. Prioritize Islands of Coherence. Noise IS Signal (partially): deviations from expected noise models might be directly attributable to unitary contamination. Go run a benchmark with this in mind.
For More Check Out
