Insider Brief
- EigenOrbit has released eigenkit 1.0, a free, open-source Python library designed to run quantum algorithms across multiple simulators and quantum hardware platforms without rewriting the algorithm for each backend.
- The library supports tools including Qiskit, PennyLane, NVIDIA CUDA-Q and Amazon Braket, alongside hardware from IBM and IQM, and includes algorithms for quantum machine learning, optimization, chemistry and finance.
- Released under the Apache 2.0 license, eigenkit includes quantum-versus-classical benchmarking and cost warnings before paid hardware jobs, with packages available through PyPI and source code on GitHub.
PRESS RELEASE — EigenOrbit has released eigenkit, a free open-source Python library for quantum computing. With eigenkit, a quantum algorithm is written once and runs on a simulator on an ordinary laptop or on real quantum computers from IBM, IQM and the vendors available through Amazon Braket. Moving from one machine to another takes a single setting, and the rest of the code stays as it was.
Quantum teams today often pick a simulator or a hardware vendor first and then write code against that vendor’s SDK. Switching later means a rewrite. Yet no single backend suits every workload. Simulators differ in speed depending on qubit count, circuit structure, entanglement and batch size. Quantum processors differ in qubit count, connectivity, native gates, error rates, price and queue times, so the best device can change from one circuit to the next. eigenkit separates the definition of an algorithm from the machine that executes it, which makes trying several backends a routine step.
What eigenkit does
- Many machines, one codebase. eigenkit runs on its own built-in simulator and on PennyLane, Qiskit, Qiskit Aer, Qulacs, qsim, TensorCircuit, NVIDIA CUDA-Q and Amazon Braket simulators. On hardware it supports IBM quantum computers, IQM quantum computers and the processors available on Amazon Braket. Photonic circuits run through Quandela’s Perceval.
- Familiar tools. eigenkit works with PyTorch and scikit-learn, the libraries most data scientists already use. A quantum model can be trained and tested inside existing machine-learning code.
- Ready-made algorithms. Machine learning: hybrid quantum neural networks, quantum kernel methods and reservoir computing. Optimization: QAOA and Max-Cut. Chemistry: VQE with real molecules. Finance: portfolio risk estimation. Also HHL for linear systems and quantum amplitude estimation.
- Training on any backend. Models can be trained on every supported simulator and on real quantum hardware in the same way.
- Comparison with classical methods. A built-in benchmark runs a quantum model and a classical model on the same data several times, so a team can see whether the quantum approach actually helps on its problem.
- Protection from accidental spending. Before a paid job goes to real quantum hardware, eigenkit warns about the cost and shows a preview of the job.
- A small install. The basic version runs on a laptop with no extra software. Support for each additional backend is added only when needed.
The library comes from a team that has built quantum software for industrial clients in finance, energy, healthcare and manufacturing for more than five years. EigenOrbit uses eigenkit as the foundation of its own platform for comparing quantum and classical approaches on industrial problems.
Example of one circuit, three simulators
The code below builds a four-qubit circuit once and runs it on three different simulators, which return the same result. To run it on a real quantum computer, the simulator name is replaced with the device settings, and the rest of the program stays the same.
from eigenkit import Circuit, CircuitBuilder, Z
def iqp(n, edges, angles):
qc = CircuitBuilder(n_qubits=n)
for q in range(n):
qc.h(q)
for (i, j), angle in zip(edges, angles):
qc.pauli_exponential((("Z", i), ("Z", j)), angle)
for q in range(n):
qc.h(q)
return qc.build()
spec = iqp(4, [(0, 1), (1, 2), (2, 3), (3, 0)], [0.7, 1.1, 0.4, 1.9])
observables = [Z(0) * Z(1), Z(1) * Z(2), Z(0) * Z(1) * Z(2) * Z(3)]
for provider in ["numpy", "qulacs", "qiskit-aer"]:
circuit = Circuit(spec, observables=observables, provider=provider)
print(provider, circuit([]))
# numpy [0.46548538 0.11841725 1. ]
# qulacs [0.46548538 0.11841725 1. ]
# qiskit-aer [0.46548538 0.11841725 1. ]Availability
eigenkit 1.0 is free and open source under the Apache 2.0 license, which allows commercial use. It is available now from PyPI, the standard Python package index, and supports Python 3.12 to 3.14. The documentation includes a user guide, explanations of each algorithm and worked examples, each also available as a Jupyter notebook.
- Package: pypi.org/project/eigenkit
- Source code: github.com/eigenorbit/eigenkit
- Documentation: docs.kit.eigenorbit.com
