Reference visual: IBM Quantum System Two at Ikerbasque, 14 Oct 2025. Author: Irekia / Eusko Jaurlaritza; CC BY 4.0.
Group notebooks, datasets, circuits, Hamiltonians, tags, quotas, and collaborators under one durable research workspace.
Start from curated GPU images for CUDA-Q, Qiskit Aer-GPU, PennyLane-Lightning-GPU, qsim, quimb, PySCF, and OpenFermion.
Prototype in Jupyter, then dispatch the same code to Slurm GPU queues, arrays, or campaign sweeps.
Store engine versions, resources, input files, result schemas, logs, costs, and run history with every job.
Deep-circuit simulation, QAOA/VQE/QML campaigns, custom noise models, state vectors, expectation values, and cross-engine benchmarking.
Hamiltonian time evolution, MPS/MPO workflows, contraction planning, quimb/cotengra/JAX notebooks, and distributed tensor-network experiments.
OpenFermion and PySCF workflows, fermionic mappings, UCCSD templates, GPU-accelerated classical references, and reproducible chemistry notebooks.
project: molecule-vqe
env: cudaq-cuquantum-12
queue: gpu-a100
items: 128 circuits
status: RUNNING
result: expectation + logsqBraid-style environment catalog with subject facets, kernel registration, clone-from-curated workflows, and health checks.
Stable project-scoped job IDs, status history, cancellation, retries, quota depletion states, and partial-result recovery.
Parameter sweeps and circuit batches map to Slurm arrays while staying visible as a single campaign.
Normalized counts, statevector, expectation, and artifact outputs with schema versions, logs, metrics, and exportable handles.
Estimate GPU seconds, VRAM, storage, and wallet impact before a job queues.
Run on WECORE-managed OpenStack and Slurm resources; keep sensitive circuits, chemistry inputs, and results inside the operator boundary.
No. QuFabric is a quantum-simulation platform. It runs simulators and research engines on GPU/HPC infrastructure you control. Optional real-QPU pass-through may come later, but the first release focuses on simulation.
Those are hosted cloud services. QuFabric borrows the polished workspace pattern — projects, environments, jobs, and results — but runs on WECORE-managed OpenStack, Slurm, and GPU resources so data stays resident.
The first curated images target CUDA-Q, NVIDIA cuQuantum, Qiskit Aer-GPU, PennyLane-Lightning-GPU, qsim, quimb, PySCF/gpu4pyscf, and OpenFermion.
Yes. The intended path is notebook-first: prototype in JupyterLab or Open OnDemand, then submit the same environment to Slurm-backed jobs or campaigns.
Every run captures project metadata, input artifacts, engine versions, selected environment, resource request, result schema, logs, timing, and quota impact.
QuFabric — our self-hosted quantum simulation platform — isn't quite ready yet. We're putting the finishing touches on it and it will be available soon. Feel free to look around in the meantime.