Strong classical scheduling first. Constrained quantum sampling only in classically ambiguous feasible cores. Classical certification always.
| Goal | Start here |
|---|---|
| Understand the claim and scorecard evidence | Final reviewer evidence card |
| Reproduce the frozen headline result | Locked evaluation report |
| Inspect the preregistered decision gates | Locked protocol |
| Watch the project overview | 50.5-second cinematic demo · reproduction package |
Evidence boundary: The frozen result is a selective statevector sampling-advantage candidate on four-variable ternary retained cores under a matched 128-sample budget. It is not evidence of real-QPU, wall-clock, asymptotic, production, or universal quantum advantage. Every accepted schedule is classically certified; the final system produced zero hard-constraint breaches.
 QAOA circuit + probability concentration Depth-two gates pulse while probability moves toward the best feasible bitstring. |
 Cross-domain network morph Portfolio and Human–AI staffing semantics share a coupled combinatorial geometry. |
 Economics × OR → QUBO Value, interaction structure, and feasibility penalties become one auditable objective. |
 Selective routing + classical certification Quantum proposes candidates; the trusted classical layer decides what can be accepted. |
▶ Watch or download the final 50.5-second cinematic demo (MP4)
The four GIFs are fast silent windows into the full panorama. The H.264/AAC film includes an original code-generated stereo soundtrack.
The final cinematic connects computational economics, operations research, quantum information, and Responsible AI through a continuous asset-network → QUBO → QAOA → Human–AI skill-network → classical-certification sequence. Portfolio construction is presented as a cross-domain transfer case; Human–AI staffing remains the active locked challenge track. Headline numbers are the frozen statevector results, and the on-screen evidence boundary explicitly excludes real-QPU, wall-clock, asymptotic, production, and universal quantum-advantage claims.
Reproduce or inspect: complete ZIP package · rendering source · requirements · storyboard and scientific boundary · poster
Q-OPS Global extends the challenge’s staffing problem to a future multinational service organization in which a task may be assigned to a human, an AI agent, or a Human–AI collaboration mode. Greedy and cluster-based adaptive large-neighborhood search solve ordinary cases and expose classical disagreement. A constrained statevector-QAOA pilot is invoked only when the retained interaction core is strictly feasible, strong classical methods disagree, and a prespecified value gate is satisfied. Every accepted schedule is evaluated by the same classical scenario, service, capacity, review, and hard-constraint checker.
| Category | Track | Evidence status | Current conclusion |
|---|---|---|---|
| Archived | Frontier QAOA and early Adoption Router | simulated / research memory | Strong classical controls exposed where quantum quality or Router calibration was weak. |
| Active project-wide | Liquidity Sentinel | simulated-executed / query model | Clearest rare-event-estimation result under an explicit coherent-oracle assumption. |
| Challenge locked evidence | Q-OPS Global C1.2 | exact synthetic + feasible-subspace statevector | All frozen held-out gates passed; supports a statevector-level selective hybrid advantage candidate on this synthetic benchmark. |
| Planned | Noise/QPU Sentinel, Human–Robot Pilot 2, Meta-Router | roadmap | Raise evidence maturity without overstating hardware readiness. |
The frozen quantum-adoption rate is 28.13%. Most held-out instances remain classical-only. The system treats quantum computation as a costly optional resource rather than a branding requirement.
Single-task greedy moves cannot reliably unlock three-task Human–AI collaboration thresholds. Cluster destroy–repair ALNS beats greedy in 78.13% of held-out instances, producing a measurable classical-ambiguity region.
The locked evaluation reports:
- 93.75% strict feasible-core rate;
- 0 final hard-constraint violations;
- service, capacity, review, and human-mandatory checks applied to every accepted schedule.
Development seeds 11 and 23 were excluded. The locked protocol used seeds 101, 211, 307, and 401; all model, solver, QAOA, sample-budget, outcome, and gate choices were frozen before execution.
- C1.0: the first hard-core emulator created no positive value over the strongest greedy baseline.
- C1.1: QAOA concentrated probability, but strict feasible-core coverage and classical ambiguity were both zero; decision
DO_NOT_FREEZE. - C1.2: explicit collaboration, chance constraints, correlated disruption, and cluster complementarity passed development gates and then all locked gates.
Tasks, regions, priorities, risk, and review rules
↓
one-task greedy + cluster ALNS
↓
feasibility + classical-ambiguity + value gate
↓
four-variable ternary retained interaction core
↓
feasible-subspace QAOA, depth p=1 and p=2
↓
classical evaluation, rejection / repair, certificate
↓
manager output: Human / AI / collaboration / fallback
Decision modes:
HUMAN_ONLY
AI_ONLY
HUMAN_AI_COLLABORATION
The long-term policy layer also supports:
CLASSICAL_ONLY
QUANTUM_PILOT
ABSTAIN
HUMAN_REVIEW
| Locked metric | Result |
|---|---|
| Instances | 32 |
| Strict feasible-core rate | 93.75% |
| ALNS win rate over one-task greedy | 78.13% |
| Mean ALNS gain | 17.019 |
| Classical ambiguity activation | 78.13% |
| Selective quantum adoption | 28.13% |
| Greedy Human–AI collaboration rate | 47.48% |
| ALNS Human–AI collaboration rate | 66.41% |
| Uniform optimum-state probability | 0.022 |
| QAOA p=2 optimum-state probability | 0.152 |
| QAOA / uniform optimum-probability ratio | 6.831× |
| Uniform mean first-hit samples | 36.50 |
| QAOA mean first-hit samples | 12.94 |
| QAOA / uniform first-hit ratio | 0.354× |
| Uniform near-optimal coverage | 0.919 |
| QAOA near-optimal coverage | 0.964 |
| Final hard violations | 0 |
| Locked decision | LOCKED_SUCCESS |
All seven prespecified locked gates passed. Paired 95% bootstrap intervals were reported for every prespecified outcome, and null or non-adopted cases were retained.
Inside the same frozen retained feasible cores, depth-two constrained QAOA puts more probability on exact optimal states than uniform feasible sampling and reaches the first exact optimum in fewer samples under the same 128-sample budget.
The supported claim is:
A statevector-level selective hybrid advantage candidate on a frozen synthetic Human–AI staffing benchmark.
The project does not claim:
- current real-QPU utility;
- wall-clock speedup;
- asymptotic scaling advantage;
- production-ready workforce management;
- real-workplace safety;
- superiority over every advanced classical method;
- universal quantum advantage.
Each panel below is embedded with a private-repository-safe relative path. Click an image to open its full-size SVG.
 Constrained-QAOA circuit schematic Warm-start preparation, cost and mixer phases, constraint ancillas, readout, and trusted adoption logic. |
Locked held-out dashboard All seven prespecified gates, including feasibility, classical separation, selective adoption, and sample-efficiency ratios. |
 Feasible-subspace state distribution The depth-two QAOA probability landscape for a representative held-out retained core. |
 Quantum-state blooms A presentation-oriented view of warm-start steering, cost-phase shaping, mixer exploration, and the retained-core energy landscape. |
 Classical ambiguity map Held-out instances positioned by greedy–ALNS disagreement and the prespecified ambiguity score. |
 Trusted scheduling certificate Feasibility, classical-search separation, quantum sampling diagnostics, and evidence maturity in one audit-style card. |
- Reviewer journey
- Research-evolution timeline
- Literature–experiment–blueprint network
- All Q-OPS locked figures
The official staffing brief emphasizes a quantum-compatible mathematical formulation, synthetic demand and staffing data, service and cost trade-offs, manager controls, classical validation, explainability, speed, optimality, scalability, and clean technical/business communication.
Q-OPS is strongest on:
- quantum-compatible ternary assignment formulation;
- synthetic and privacy-safe data;
- configurable service, risk, capacity, collaboration, and adoption controls;
- strong classical validation and exact retained-core references;
- zero final hard breaches;
- held-out locked evidence;
- reproducibility, visual explanation, and transparent post-processing.
Transparent scope gaps:
- no full interval-level queue simulator for average speed of answer and abandonment;
- no explicit break and overtime schedule as the main benchmark;
- no real-QPU or hardware-aware timing result;
- small four-variable ternary quantum cores.
See the complete challenge alignment scorecard.
The workforce copilot returns more than a quantum bitstring. It produces:
- a recommended Human, AI, or collaboration mode;
- service and capacity checks;
- review and fallback requirements;
- a classical verification status;
- an evidence level and claim boundary;
- an adoption decision explaining why quantum was or was not called.
python -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements-liquidity.txt
python -m pip install numpy pytest
python -m unittest discover -s tests -v
pytest -q tests/test_q_ops_c11.py tests/test_q_ops_c12.py tests/test_q_ops_c12_locked_protocol.py
python scripts/check_literature_refresh.pyCanonical locked assets:
- locked protocol
- locked report
- locked configuration
- locked summary
- locked records CSV
- environment manifest
- C1.2 literature refresh
- three-minute demo
- final submission narrative
- presentation deck outline
- challenge scorecard
- AI-use and contributor disclosure
- final submission checklist
- submission literature refresh
Every material experiment, solver, roadmap, report, status, or claim change begins with a fresh interdisciplinary review.
Q-GATE Atlas is organized by economic decision structure rather than industry labels:
Risk estimation
↓
Allocation
↓
Scheduling and assignment
↓
Routing and dynamic control
↓
Governance and solver adoption
↓
Cross-domain meta-learning
The broader roadmap includes the active Liquidity Sentinel, the locked-evaluated Q-OPS track, a future noise/QPU Sentinel, Human–Robot Pilot 2, and a Meta-Router that learns when to use classical, quantum, quantum-inspired, hybrid, abstention, or Human Review policies.