技能创建器
skill-creator
创建、改进和评估自定义技能,把专业流程转化为可重复使用的 AI 能力。
pyquil【数据与研究】Guides PyQuil and Quil quantum-programming workflows: author programs, simulate locally, compile and execute through QVM or QPU services, model noise and experiments, and inspect processor topologies and ISA metadata.
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Use this repo skill when a task names PyQuil, pyquil, Quil, Rigetti Forest,
Program, get_qc, QVM, QPU, quilc, QuantumComputer, PyQVM, Pauli
experiments, noise models, or compiler ISA/topology metadata. It is a
versioned, self-contained operating guide for PyQuil 4.18.0; it is not a
replacement for QCS credentials, QVM/quilc binaries, a reservation, or a
physical processor.
program-authoring. It owns
Program, gates, declarations, memory, placeholders, control flow, Quil-T,
and serialization.simulation. Prefer PyQVM or the
reference/NumPy simulators. Do not confuse them with the service-backed
WavefunctionSimulator.compile-execute. It owns
QuantumComputer, compiler/QVM/QPU selection, result maps, batching,
endpoints, services, and QCS configuration.noise-experiments. It owns
construction and validation; execution remains a separate backend step.processor-isa. It owns NetworkX
topologies, CompilerISA, QCS ISA conversion, and processor compatibility.For a task crossing routes, start with the first route that creates the artifact, then follow its explicit handoff. Keep these distinctions visible:
| Local artifact or operation | What it proves | What it does not prove |
|---|---|---|
Program(...).out() | Quil construction and serialization | Compiler acceptance or execution |
PyQVM / reference simulator | In-process behavior on the selected simulator | QVM/quilc availability or QPU behavior |
qc.compile() / qc.run() | Backend interaction only when the configured services respond | Physical correctness beyond the returned evidence |
NoiseModel, Experiment, or ISA object | Model/metadata construction and validation | A run, calibration, reservation, or hardware result |
Use a supported Python version from the package metadata (Python 3.11 or 3.12 for this release) in an isolated environment:
python -m pip install pyquil
python -c "import pyquil; print(pyquil.__version__)"
The optional latex extra supplies IPython for interactive display helpers:
python -m pip install 'pyquil[latex]'
The package depends on compiled Rust-backed distributions on common platforms;
less common platforms may need a Rust toolchain. QVM/quilc are separate Forest
SDK services and are not installed by the Python package. QCS/QPU workflows
also require authorized configuration and access. Read
references/concepts-and-runtime-boundaries.md
for the dependency and service boundary before installing broad extras.
Run the bundled, service-free installation check from any working directory:
python scripts/check_pyquil_install.py --help
python scripts/check_pyquil_install.py
It reports package metadata/imports and small local capability checks without
starting services, reading secrets, contacting QCS, or writing files. Read
references/troubleshooting.md when the
import, optional dependency, configuration, compiler, service, or credential
boundary is unclear. Read
references/repo-provenance.md before using
this skill against a changed checkout; a changed commit or public API surface
calls for refresh-repo-skill.
Program before handing it to a simulator or compiler.quilc, or QVM is missing.All references and helpers linked by the routes above are bundled inside this skill. Review/test records, native candidate maps, environment reports, and verification notes live outside the runtime tree under the construction artifact directory; do not make later Researcher runs depend on them.