MathIdeasResearch in progressRepository ↗
← Research catalogOriginal Markdown ↓
On this page

Quantum optimization regime and resource audit

Initial decision: Prototype evidence workflow. First dossier, 9 October 2026 Australia/Brisbane. No buyer validation or profitability evidence has been established.

Research finding

The QAOA manuscript claims SK limiting expected energy optimality with system size tending to infinity before depth. It explicitly provides neither a quantitative required depth nor an efficient angle-selection procedure. Its selected formal scope supplies SK variational value identities, not QAOA circuit-energy convergence. Related families distinguish polynomial oracle calls from an unrestricted target-dependent truth table and query counts from unrestricted intervening computation.

Problem and buyer

A quantum optimization R&D group needs a reviewable explanation of what a theory claim predicts for a finite graph, circuit depth, parameter-search budget and noisy hardware run. A procurement or research decision can be skewed by an asymptotic guarantee whose unmeasured costs are decisive.

What the finding could enable

A benchmark and claim-review app can track limit order, angle training, circuit synthesis, oracle cost and observed solution quality. It could introduce a consistent comparison contract for QAOA, classical optimization and oracle-based constructions. The new results supply useful scope records; they do not provide a finite-instance performance advantage.

Technical and commercial limits

The SK model is Gaussian and zero field, with expected energy per spin in an ordered limit. The associated random-regular MaxCut conclusion also has ordered size/degree limits. Arbitrary constrained business instances, finite-device error and optimal training angles are not covered. Family 283 gives a fixed diamond-norm tolerance of one half and does not construct the target oracle efficiently.

Minimal architecture

Instance distribution and objective -> theory-contract resolver -> finite size/depth grid -> angle-search and compile logs -> classical baseline -> noise and sampling metadata -> time/cost/quality report. A declared oracle must include construction and implementation cost or remain an explicitly unpriced dependency.

Existing alternatives and differentiation

Qiskit-community’s QAOA implementation already exposes depth, classical optimizer, starting parameters, callback and transpiler hooks. An adapter should collect those existing measurements rather than replace the algorithm. Its repository states that IBM no longer officially supports Qiskit Algorithms; an integration must pin and maintain its own dependencies. QAOA API, project repository.

Monetization hypothesis

Hypothesis: AUD 8,000–25,000 for an independent finite-workload assessment, followed by AUD 1,000–3,000 monthly only if repeated comparisons are valuable. A hypothetical AUD 15,000 assessment with 55 expert hours at AUD 180/hour leaves AUD 5,100 before hardware access, maintenance, sales and overhead. Paid quantum compute is not used in this research.

Validation experiment

Use small seeded SK instances and one graph family with exact classical optima, train several low depths using existing software, and record all parameter-search evaluations. Compare total measured effort and quality, while verifying that the report does not label the formal SK-value result a formal QAOA performance guarantee.

Conditions to reject or defer

Reject if buyers want guaranteed business-instance speedups that the source does not supply, if existing benchmarking already records these costs, or if custom adapter maintenance consumes the fee. Defer claims about required depth, ideal optimal angles or hardware advantage.

Next concrete action

Specify the benchmark export schema and extract all theorem quantifiers before running a small simulator-only comparison.

Pinned research sources