Workload intake
Accept a circuit, algorithm, optimization problem, or research question.
PlannedPlatform overview
A neutral control plane for hybrid quantum workflows.
Validate a workload once, then compare how it behaves across simulators, GPU paths, classical fallbacks, and — where authorized — quantum hardware. Everything on this page is proposed unless marked otherwise.
Pipeline
Five stages
STAGE 01
STAGE 02
Validation & normalization
Parse the workload, identify requirements, and produce a portable intermediate representation where possible.
PlannedSTAGE 03
Backend selection
Compare available simulators, GPU paths, classical fallbacks, and authorized quantum hardware.
PlannedSTAGE 04
Execution & orchestration
Schedule jobs, track status, preserve parameters, and collect reproducible outputs.
PlannedSTAGE 05
Benchmark & insight
Compare runtime, memory, fidelity proxies, cost, noise sensitivity, and result quality.
PlannedPRINCIPLE
Classical baseline required
Every comparison includes a credible classical baseline. QAISS measures value instead of assuming quantum advantage.
Backend availability, performance, and compatibility depend on licences, credentials, infrastructure, circuit size, noise, and vendor-specific limitations.
Core asset · proposed
The Workload Passport
An open experiment manifest that travels with every workload. It makes QAISS useful as a reproducibility and comparison layer — even before any hardware connector exists.
Specification draft| Field | Records |
|---|---|
| workload | circuit or algorithm reference + hash |
| framework | SDK and version |
| compiler | transpiler / compiler version |
| backend | simulator, GPU path, or provider device |
| config | hardware or simulator configuration |
| precision | fp32 / fp64 / mixed |
| noise_model | model id + parameters |
| seed | RNG seed(s) |
| runtime | wall-clock, queue time |
| cost | credits or currency, if any |
| output | result artifact hash |
| timestamp | ISO-8601, UTC |
Maturity tracks
Learn · Experiment · Deploy
Learn
Education, tutorials, and explainers. Nothing here is a compute result.
PlannedExperiment
Simulation and benchmarking with full reproducibility metadata.
PlannedDeploy
Authorized research or production backends, with consent for any fee-bearing job.
PlannedTechnology foundations
External technologies, treated as external.
Each requires licensing review, access, and verification before any integration is claimed.
Path 01 · NVIDIA cuQuantum (public developer tooling)
GPU-accelerated simulation
- Accelerated state-vector simulation (cuStateVec).
- Tensor-network computation and contraction-path optimization (cuTensorNet).
- Density-matrix and quantum-dynamics workflows (cuDensityMat).
- Stabilizer and Pauli-propagation methods (cuStabilizer, cuPauliProp).
- C and Python access paths.
- Integration with selected quantum frameworks.
- Single-GPU, multi-GPU, and multi-node — where supported by implementation and infrastructure.
Path 02 · Origin Quantum (public platform information)
Hardware, cloud & hybrid
- Quantum cloud access and simulators.
- Hardware–software integration through QPanda3 (pyqpanda3 0.5.0 on PyPI).
- OriginIR and OpenQASM 2.0 as documented circuit interchange formats.
- Origin Pilot’s documented orchestration across quantum and classical resources.
- Quantum–classical job management, scheduling, monitoring, and error-mitigation concepts.
- Application areas — optimization, finance, chemistry, AI, smart grids, education — as research domains, not guaranteed outcomes.
Licensing. The cuQuantum documentation states that its SDK and Python bindings carry different licence terms and that NVIDIA marks may not be used to endorse derived products without permission. See the licence page. Origin Quantum access requires credentials and permission from the provider.
Proposed modules
Product modules
Quantum Workspace
A project space for circuits, notebooks, algorithms, experiment settings, datasets, and result artifacts.
PlannedBackend Router
A policy-driven service that recommends a backend based on circuit requirements, simulator capability, cost, queue, noise, and user permissions.
PlannedBenchmark Lab
Reproducible comparisons of simulation speed, memory, accuracy proxies, depth, gate counts, noise sensitivity, and cost.
No benchmark is ever shown without hardware, software version, workload, precision, and date.
PrototypeQuantum AI Lab
Research on hybrid algorithms — VQE, QAOA, quantum kernels, quantum neural networks, parameter optimization.
All examples are research/prototype unless independently validated.
PlannedDeveloper API & SDK
Submit workloads, read job state, retrieve results, and export experiment manifests.
Mock data only until a production API exists.
PlannedLearning Hub
Plain-language explainers, notebooks, glossary, tutorials, and guided experiments.
PrototypeCommunity Research Board
Proposals, bounties, grants, benchmark challenges, and research contributions.
Community signals are not legal governance until a model is defined and reviewed.
PlannedEcosystem · proposed
Integration certification
Instead of listing vendors as partners, backends earn compatibility labels by passing published tests.
- Schema-compatibleAccepts and returns the QAISS workload format.
- Simulator-verifiedMatches reference simulator results within tolerance.
- Cloud-verifiedEnd-to-end job flow passes on the provider’s cloud.
- Hardware-verifiedResults on real hardware recorded with full Passport metadata.
Compatibility matrix
What has actually been tested
Certification levels are earned by passing published tests. Today only the in-browser reference simulator has run the draft workload format.
| Backend | Schema-compatible | Simulator-verified | Cloud-verified | Hardware-verified | Evidence |
|---|---|---|---|---|---|
| QAISS in-browser reference simulator (fp64 state vector) | Passed | Reference | N/A | N/A | Lab + unit tests |
| Open-source CPU simulators | Not tested | Not tested | N/A | N/A | Adapter not built |
| QAISS connector — CPU reference (pure Python) | Passed | Passed | N/A | N/A | Parity tests vs browser |
| QAISS connector — NVIDIA cuStateVec (GPU) | Passed | Not tested | N/A | N/A | Code ready, GPU test pending |
| QAISS connector — tensor network (CPU, NumPy) | Passed | Passed | N/A | N/A | Parity tests vs browser |
| QAISS connector — NVIDIA cuTensorNet (GPU) | Passed | Not tested | N/A | N/A | Code ready, GPU test pending |
| QAISS connector — Origin Quantum pyqpanda3 CPUQVM | Passed | Passed | N/A | N/A | Parity tests vs browser |
| QAISS connector — Origin Quantum Cloud (simulators / QPU) | Passed | Not tested | Not tested | Not tested | Adapter written; needs API key |
| Authorized quantum cloud providers | Not tested | Not tested | Not tested | Not tested | Access and permission required |
“Simulator-verified” for the browser simulator means it is the reference implementation checked by unit tests (Bell, Grover, rotation, seeded sampling) — not an independent certification.