Docs.rs

Backend-independent Celox simulation runtime contracts

Rust-core quantum variational optimization framework with a Python (PyO3) interface: circuit simulation, VQC/QML training and population-based optimizers (DE, PSO, QNG, CMA-ES). Python users: `pip install polypus-quantum` (the module imports as `polypus`).

Keep shells, REPLs and long-running jobs alive between tool calls, locally or over SSH

Source-independent testbench bytecode contracts for Celox

Execution-backend layer for Polypus: quantum circuit backends (native statevector, local, CUNQA, CESGA QMIO QPU) and the Planner that decides how a circuit runs.

Evaluation layer for Polypus: oracles that turn a parameter vector into scalar fitness (VQC/QML, variance, Python-callback observables), plus the OracleFactory assembly for training flows.

Orchestration layer for Polypus: the pure, Python-free policy middle (dispatch_optimizer, Resources/Scheduler/Flow) that binds a backend to its planner and runs circuit and training flows.

A pyo3-free QuantumBackend that bridges to a provider's Python SDK running in its own subprocess, over a versioned length-prefixed JSON protocol. Lets a third-party Python backend be used from Polypus without embedding the interpreter (no PyO3).

Pure-Rust quantum machine learning for Polypus: datasets, layered variational models (encoders, ansatz, convolution, pooling), Pauli readout and losses. Produces polypus-circuit circuits; execution and optimizers stay decoupled. No PyO3, no Python.

Pure-Rust particle-physics simulation layer for Polypus: classical Monte Carlo transport and quantum Hamiltonian definitions as Pauli sums. No Python, no external physics framework.

Pure-Rust statevector simulator for Polypus circuits: consumes polypus-circuit's ConcreteCircuit directly (no OpenQASM round-trip), GIL-free, with optional rayon parallelism. No Python.

Pure-Rust variational optimizers for Polypus (Differential Evolution, Particle Swarm Optimization, Quantum Natural Gradient, CMA-ES, Adam with an optional momentum rule, SPSA, Rotosolve), plus a bridge that runs an external ask/tell optimizer. Decoupled from circuits and Python via the EvaluationOracle/VarianceOracle/GradientOracle/AskTell contracts. No PyO3, no Python.

A reusable conformance test battery for polypus-backend QuantumBackend implementations: run it against your backend and get a pass/fail/skip report over the documented execution semantics (error classification, shot conservation, batching/waves, concurrency, result validation) — pyo3-free.

Shared, pure-Rust logging sink for the Polypus workspace: a configurable `log::Log` implementation (text/JSON, stdout/stderr/file) installed once per process. Library crates only emit through the `log` facade; this crate is depended on by whoever installs the logger.

Pure-Rust diagonal cost observables (QUBO/Ising) mapping measurement counts to expectation values. rayon-parallel over candidates, GIL-free. No Python.