Generate ML data from quantum programs. Compare software stacks. Search for better representations.
WestQuant Open is an open research infrastructure for generating machine-learning data from quantum programs, comparing quantum software stacks, and searching for better circuit representations and compilation strategies across frameworks and hardware targets.
The unique layer:
Quantum program -> representation -> transformation sequence -> hardware mapping -> result
- Generate ML training data from quantum algorithms
- Search representation/transformation schedules across frameworks
- Compare Qiskit / PennyLane / Cirq compilation pipelines
- Optimize circuits for different hardware targets
- Record complete transformation provenance for reproducibility
pip install westquant[qiskit]from westquant import optimize
result = optimize(circuit, framework="qiskit", backend=backend, budget=64)
print(result.pareto_front)Estimate required QPU budget from problem structure — before any quantum execution:
pip install westquant-qcscfrom qcsc import QPUBudgetPredictor, ProblemProfile
profile = ProblemProfile(problem="MaxCut", n_qubits=16, p=3, graph_family="GEO")
predictor = QPUBudgetPredictor()
budget = predictor.estimate(profile)
# 1024 shots, 2070x reduction vs naive, quality=0.9961| Package | Description |
|---|---|
| westquant | Umbrella package — representation search |
| westquant-core | WQIR, RepGraph, plugin contracts |
| westquant-qcsc | Semantic QPU Minimization + QPU Budget Predictor |
| Package | Description |
|---|---|
| westquant-qiskit | Qiskit transpiler plugins |
| westquant-pytket | pytket compiler pass search |
| westquant-pennylane | PennyLane transform search |
| westquant-pulser | Pulser neutral-atom search |
| Package | Description |
|---|---|
| westquant-orchestrator | Cross-framework normalization |
| westquant-bridges | Cirq/Braket/QIR/OpenQASM bridges |
| Repo | Description |
|---|---|
| qpu-mini | Paper A — QPU-efficient quantum graph optimization (510-2070x reduction) |
| representation-stack | Artifact #001 — Representation search for quantum optimization |
┌─────────────────────────────────────────┐
│ WestQuant Open Layer │
│ │
│ Quantum program │
│ │ │
│ ▼ │
│ ┌─────────┐ ┌──────────────┐ │
│ │ WQIR │───▶│ RepGraph │ │
│ │ (IR) │ │ (search │ │
│ └────┬────┘ │ space) │ │
│ │ └──────┬───────┘ │
│ ▼ │ │
│ ┌─────────────┐ │ │
│ │ Transform │◀───────┘ │
│ │ Registry │ │
│ └─────┬───────┘ │
│ ▼ │
│ ┌─────────────┐ ┌──────────────┐ │
│ │ Verifier │───▶│ Planner │ │
│ └─────────────┘ └──────────────┘ │
│ │ │
│ ▼ │
│ Optimized circuit + ML dataset + │
│ provenance trace + benchmark results │
└─────────────────────────────────────────┘
▲ ▲ ▲
┌────┴────┐ ┌────┴────┐ ┌────┴────┐
│ Qiskit │ │PennyLane │ │ Cirq │
└─────────┘ └─────────┘ └─────────┘
See full architecture.
WestQuant asks three questions in order:
- Does this operation need to happen? (Eliminate)
- Does it need to be quantum? (Replace with classical)
- Only then: which resource should execute it? (Schedule)
Optimize the Optimization: QPU-Efficient Quantum Graph Optimization through HPC-First Experimentation Submitted to IEEE Transactions on Quantum Engineering.
Key finding: QPU requirement is a structured, predictable function of problem structure — not a fixed constant. 510-2070x shot reduction with ≤0.4% quality loss.
Systematic mapping review of QPU minimization and quantum workflow systems. Submitted to ACM Computing Surveys.
WestQuantOpen: Cross-Framework Quantum Program Optimization, Representation Scheduling, and ML Dataset Generation
Four contributions:
- Cross-framework abstraction (Qiskit, PennyLane, Cirq)
- Transformation provenance (every step recorded)
- ML dataset generation (before/after pairs, actions, metrics)
- Representation-policy search (systematic transformation sequence search)
10 algorithm families, ~2,000 circuits:
| Family | Sizes |
|---|---|
| GHZ | 4-100 qubits |
| QFT | 3-30 |
| Grover | 3-20 |
| Bernstein-Vazirani | 4-50 |
| QPE | 3-20 |
| QAOA MaxCut | multiple graph families |
| VQE ansatze | multiple molecules |
| Trotter simulation | varying steps |
| Quantum arithmetic | adders/multipliers |
| Random Clifford | varying density |
See benchmarks.
- Discussions: Join the conversation
- Contributing: See CONTRIBUTING.md
- Code of Conduct: See CODE_OF_CONDUCT.md
- Releases: Monthly on the last Friday — see release schedule
- Project plan: See WQO_PROJECT_PLAN.md
Apache-2.0 (core packages) / MIT (qcsc)
David Vesterlund — Vesterlund Ventures, Industrial Research / WestQuant Open Source Project
- Email: david@vesterlundventures.se
- ORCID: 0009-0000-6455-1141
- Website: davidvesterlund.com