Open-source quantum software initiative EigenOrbit has launched eigenkit 1.0, an Apache 2.0-licensed Python library designed to write quantum algorithms once and execute them across multiple simulators and hardware backends without code modification.
Installing via pip install eigenkit, the framework abstracts hardware compilation behind a single provider argument, connecting PyTorch and scikit-learn ML pipelines to eight classical simulators (including NumPy, Qulacs, TensorCircuit, and NVIDIA CUDA-Q) and four QPU hardware access points (IBM Qiskit Runtime, IQM Resonance, Amazon Braket, and Quandela Perceval).
| [ Supported Backends & Execution Frameworks ] | |
|---|---|
| Category | Integrated Providers & Infrastructure |
| • Simulators (8) | • NumPy (built-in, zero dependencies), PennyLane, Qiskit, Qiskit Aer (Dense & MPS), Qulacs, qsim, TensorCircuit, and NVIDIA CUDA-Q. |
| • QPU Hardware (4) | • IBM Quantum (Qiskit Runtime), IQM (Resonance), Amazon Braket, and Quandela Perceval (photonic quantum computing). |
| • ML & Cost Tools | • PyTorch nn.Module layers, scikit-learn estimators, Wilcoxon statistical significance benchmarking, and QPU pre-submission cost warnings. |
The library also includes built-in classical versus quantum benchmarking tools and automated pre-submission cost warnings for billable QPU hardware jobs.
Review the source repository on GitHub here and examine the developer guidelines in the eigenkit Documentation here.
October 14, 2026