Performance impact of device-dependent parameter shrinking

Determine how shrinking individual QAOA parameters on highly connected or high-error qubits, as in multi-angle QAOA or device-dependent parameter scaling, affects the performance of the resulting quantum optimization circuit.

Background

The paper’s error-resilience bound indicates that errors on highly connected qubits can have a greater impact because the relevant cost-term contribution scales with the qubit degree. The authors therefore suggest using multi-angle-QAOA-style parameter shrinking for highly connected qubits or reducing parameters associated with hardware qubits known to have high error rates.

The unresolved issue is whether these parameter modifications, while potentially improving resilience to noise, degrade or otherwise alter the optimization performance of QAOA. The paper does not establish this trade-off.