Quantum optimization learning route¶
Define the classical problem before choosing a circuit. These five lessons follow its encoding through optimization, final sampling and comparison with a classical reference. Each lesson includes a prediction, runnable source, output interpretation and exercises.
| Lesson | Run from the repository root | What to check |
|---|---|---|
| Beginner: graph problems | python3 examples/user/tracks/optimization_researcher/01_beginner_graph_problem_en.py |
Node order, feasible sets and an enumerated optimum. |
| QUBO and Hamiltonian | python3 examples/user/tracks/optimization_researcher/02_foundation_qubo_hamiltonian_en.py |
All four basis energies, including the constant and sign convention. |
| QAOA workflow | python3 examples/user/tracks/optimization_researcher/03_applied_qaoa_workflow_en.py |
Objective budget, stopping reason and final sampled candidates. |
| VQE workflow | python3 examples/user/tracks/optimization_researcher/04_advanced_vqe_workflow_en.py |
Expectation energy versus optimal-state probability. |
| Expert: baseline analysis | python3 examples/user/tracks/optimization_researcher/05_expert_baseline_analysis_en.py |
Candidate and expectation gaps, repeated sampling and execution cost. |
These scripts use local ideal-state objectives, explicit initial parameters and small problems. Final sampling has finite shots; the optimization objective is exact. Graph positions here describe the problem and do not establish a hardware placement.
For finite-shot VQE, noise, gradient methods, measurement grouping or independent candidate confirmation, consult variational algorithms. Those features require their own configurations; adding final shots does not activate them. Sampled QAOA, chemistry mapping and live hardware execution are not provided.
After the route, compare three ways to solve the same MIS. Keep each method's model assumptions and budget visible when interpreting its results.