Effectiveness of sample postprocessing for error removal in LR-QAOA benchmarking

Determine whether postprocessing of LR-QAOA output samples can remove or mitigate errors in quantum processing unit benchmarking and, if it does, specify the adjustments required for the random-sampler baseline to maintain fair comparisons.

Background

The LR-QAOA benchmark uses the approximation ratio relative to an optimal solution and validates meaningful performance by comparison to a random-sampling baseline. The authors discuss potential improvement strategies, including dynamical decoupling, circuit encoding, and routing optimizations.

They explicitly note that they have not found evidence that postprocessing of samples removes errors, and suggest that—if postprocessing proves beneficial—the same approach should be applied to the random baseline to preserve fairness. Clarifying the effectiveness and proper use of postprocessing remains an unresolved question impacting benchmark integrity.