A quantum vendor showed your team pretty charts. The board wants a budget answer. Most teams have nothing defensible to point at. The Audit gives you the artifact: we run the proposed algorithm against your strongest classical baseline under generous assumptions. If even the most generous quantum runtime estimate still loses, the real implementation cannot beat it. Manifest. Board memo. Redirect path.
If the most generous quantum runtime loses to your strongest classical baseline, the candidate cannot win in production.
The Audit derives a best-case quantum runtime under stated assumptions. Noise-free hardware. One cycle per oracle call. Favorable condition estimates. Favorable precision. If even the best-case runtime exceeds the classical wall clock, the real implementation cannot beat it.
Numbers are indicative for a 5,000-variable, 3% sparsity portfolio rebalancing LP with a full allocation vector at 1e-6 precision. The Audit captures your workload's real condition-number distribution, the actual baseline solver runtime, and the verdict sensitivity to each named assumption.
Newton-system condition number around 1e4. Output extraction across a 5,000-element vector at 1e-6 precision. The cycle lower bound exceeds 1e10 before iterative refinement. Tomography or repeated amplitude readout compounds the cost by a factor proportional to dimension over precision squared.
Under any plausible cycle-time-to-runtime conversion above 1 microsecond per logical cycle, the lower bound exceeds the classical wall clock by orders of magnitude.
REDIRECT
Two named routes inside the chassis.
No route closes the gap.
No reformulation closes the gap, no named threshold opens a future window. The verdict is bounded by the workload, the baselines, and the assumption envelope.
A named threshold reopens it.
The candidate cannot survive today. A specific named threshold, logical cycle time below X or condition number below Y, would change the answer.
A different route may survive.
The current candidate cannot beat its lower bound. A different formulation, a different output requirement, or a different baseline configuration could.
A bounded next step is justified.
Output extraction is not obviously fatal. A hardware path is named. The memo recommends a bounded next step, not a multi-quarter program.
It is doing nothing, running an internal pilot, or accepting a vendor's own benchmark. The Audit is the only one of the four that produces a third-party-reproducible artifact a board can sign on.
No defensible answer when the board asks why the budget moved or did not.
Selection bias. Vendor picks the workload, the baseline, the assumption envelope.
Single-vendor framing. Weak baseline by default. No reformulation pass.
Manifest, memo, redirect path, reformulation log, citations, reproducibility hash.
The Manifest is the load-bearing record. Everything else is a translation of the Manifest for one buyer or one external reviewer.
Signed, hashable JSON. Every input, assumption, solver run, cycle-count step, reformulation pass, and verdict.
Two pages. The verdict per route, the assumption envelope, what would change the verdict, what to fund instead.
Where the current candidate fails, the memo names a different route. Reformulation review is mandatory before any NO-GO ships.
The audit trail of every reformulation considered. Tested, discarded, or promoted to a route.
Every primary methodology cited with a versioned reference. The Manifest names every source it relied on.
Canonicalized SHA-256 over the full Manifest. The board signs on the hash. A re-run reproduces it.
Risk pricing, derivatives valuation, portfolio optimization. Candidate runs against a strong Monte Carlo baseline plus an optional commercial alternative.
Head of quant, risk modeling, derivatives valuation, or portfolio optimization at a Tier-1 EU bank or asset manager.
Linear programs and convex optimization. Candidate runs against HiGHS plus an optional commercial solver. Lower-bound model uses published methodology with full attribution.
Available inside every Audit; led at launch where the buyer's primary workload is LP or convex.
Informed by published lower-bound analysis of hybrid quantum interior point methods. DeployQuantum did not author that research.
Each route in the Audit chassis is informed by published research with full citation in every Manifest. Best-case exclusion methodology covers hybrid quantum interior point methods, dequantization frameworks, and published quantum-optimization survey papers. Amplitude estimation cites option-pricing resource estimates, threshold-advantage analysis, and quantum signal processing. Classical-baseline reviews cover quantum simulation, chemistry tensor methods, and the QOBLIB Decathlon benchmark.
The Audit operationalizes published decision logic into a customer-facing audit workflow, with citation per route in every Manifest.
The Manifest schema is open-sourced under CC BY 4.0. The public-benchmark runner is open-sourced under MIT. The Audit's pipeline is reimplemented independently in DQ-owned code; no third-party code is copied.
See the methodology page for the full citation list and the field-by-field schema reference.