Skip to content

Guide: Qiskit Sampler (QAOA)

This guide walks through demo/02_qiskit_sampler.py — a parameter sweep over 10 (γ, β) configurations for a 2-qubit QAOA circuit targeting MaxCut on the edge (0, 1).

Run it directly:

python demo/02_qiskit_sampler.py

The Sampler proxy

from hilbertbench.integrations.qiskit import HilbertSamplerProxy
from hilbertbench.recorder.tape import HilbertTape

with HilbertTape("runs/sampler_qaoa", tags={"algorithm": "qaoa"}) as tape:
    sampler = HilbertSamplerProxy(tape)
    # sampler.run(...) is identical to StatevectorSampler.run(...)

HilbertSamplerProxy records each sampler.run(...) call as a span. The span outcome stores the bitstring count distribution.

Reading bitstring counts

Bitstring counts are stored as a JSON dict in the span outcome:

trace = HilbertTrace(tape.dir_path)
for span in trace.completed():
    counts = span.outcome       # e.g. {'00': 512, '11': 512}
    print(counts)

Extending this example

  • Run a real MaxCut optimization by looping over configurations and selecting the best (γ, β)
  • Add more QAOA layers and compare the bitstring distributions
  • Use circuit_structure to verify the circuit depth matches your QAOA layer count