cirq
Use Cirq to build, simulate, transform, serialize, validate, and provider-package quantum circuits.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- cirq
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Use Cirq to build, simulate, transform, serialize, validate, and provider-package quantum circuits.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/package-overview.md3.7 KB
- 参考文档references/repo-provenance.md3.2 KB
- 参考文档references/repo-routing-metadata.json317 B
- 参考文档references/troubleshooting.md4.7 KB
- 脚本scripts/check_cirq_environment.py4.0 KB
03 SKILL.md 正文
Cirq Repo Skill
Use this skill when a task is about Cirq, the Python framework for quantum
circuits and near-term quantum computing workflows. It covers the Cirq package
family: cirq/cirq-core, cirq-google, cirq-ionq, cirq-aqt,
cirq-pasqal, and cirq-web.
First checks
- For package layout, Python support, optional dependencies, and route selection, read references/package-overview.md.
- For install/import, provider credential, serialization, or environment issues, read references/troubleshooting.md.
- For source freshness, read references/repo-provenance.md before relying on this skill for a different checkout or package version.
- To verify an active Python environment without credentials or network calls,
run scripts/check_cirq_environment.py
with
--helpor--text.
Install stance
For ordinary package use, prefer public packages:
python -m pip install cirq # full Cirq family
python -m pip install cirq-core # core circuits/simulation only
python -m pip install cirq-google # add Google Quantum AI package
Cirq requires Python 3.11+. Local circuit construction and CPU simulation do not require CUDA, ROCm, MPS, TPU, cloud credentials, or live quantum hardware. Provider packages can be imported offline, but live service execution requires provider credentials and account/project/target access.
Route by task
| User task | Read next |
|---|---|
| Build circuits, choose qubit classes, work with gates/operations/moments, debug measurement keys, parameters, diagrams, custom gates, JSON/QASM basics | core-circuits-and-ops |
| Run local simulations, sample measurements, sweep parameters, inspect results/histograms/state vectors/density matrices, add noise/channels | simulation-study-and-noise |
| Optimize, decompose, transform, route, or compile circuits to a gateset, topology, or provider constraint | transformers-and-compilation |
| Implement or validate algorithm examples, QFT/phase estimation/Grover/QAOA-like circuits, Pauli observables, expectation values | algorithms-and-observables |
Use cirq_google, IonQ, AQT, Pasqal, cirq_web, provider serializers, credentials, offline provider validation, or JSON custom resolvers |
hardware-providers-and-serialization |
Operating guidance
- Start with the narrowest sub-skill that matches the user intent. Avoid answering provider, simulator, transformer, or algorithm questions only from the root router.
- Prefer verified public APIs and bundled examples over memory. The sub-skill references include inspected signatures, workflow recipes, and failure modes.
- Use bundled scripts only for safe local checks. They do not contact cloud services, download data, or require the original repository checkout.
- For live provider execution, stop and verify credentials, project/account, target/processor, queue/service availability, and provider-specific supported gates before calling remote APIs.
- If a circuit fails provider serialization or device validation, route first
through
transformers-and-compilationto decompose, route, or target-gateset optimize the circuit. - If a sampled or simulated result looks wrong, check measurement key order, qubit order, unresolved parameters, random seeds, simulator choice, and noise model assumptions before changing algorithm logic.
Bundled root assets
- references/package-overview.md — package family, dependency stance, and top-level route map.
- references/troubleshooting.md — cross-cutting install/import, optional dependency, provider credential, and serialization recovery steps.
- references/repo-provenance.md — source commit, version, and evidence baseline for staleness checks.
- references/repo-routing-metadata.json — structured scenario metadata for managed repo-skill import.
- scripts/check_cirq_environment.py — local import/version/tiny simulation/JSON roundtrip diagnostic.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).