Capability Coverage
What runs on the open reference build today, versus what is documented as a specification. Every example on a “specification API” page also carries an in-line note — this table is the bird's-eye summary.
How to read this
Runnable — executes on the reference build (pure NumPy/SciPy, CPU, no account).Specification — documented intended interface, not bundled in the open build yet.
Planned — a distribution or hardware path that does not exist yet.
The per-page in-line notes are the always-current source of truth; this page is rebuilt from the same runnability checks.
Runnable today
| Area | What works on the reference build |
|---|---|
| Gate-model kernels | H, X, Y, Z, S, T, Rx/Ry/Rz, R1, SWAP; controls, adjoint, compute/action/uncompute |
| Sampling & expectation | sample, observe (spin operators), get_state run today; the *_async forms are specification APIs (see below) |
| Noise modelling | NoiseModel, add_channel/add_all_qubit_channel; Depolarization, AmplitudeDamping, BitFlip/PhaseFlip, X/Y/ZError, KrausChannel; inline apply_noise; quantum-trajectory sampling |
| Open-system dynamics | evolve, Schrödinger & Lindblad master equation, collapse_operators, SuperOperator (left/right/sandwich), Schedule + time-dependent coefficients, integrators |
| Operator algebra | spin; boson incl. position, momentum, displace, squeeze, Kerr products; Jordan-Wigner fermions; custom define/instantiate |
| Simulator targets | cpu (an alias of qpp-cpu), density-matrix-cpu, and the other CPU simulator targets |
Specification APIs (documented, not in the open build)
| Area | Status |
|---|---|
| Generative Quantum Eigensolver (GQE) | Transformer-based eigensolver — interface documented |
| QEC decoders | qLDPC-GPU, tensor-network, sliding-window, LUT; real-time decoding |
| Torch-differentiable integrators | Dopri5/8, Bosh3, Magnus (the CPU integrators above run today) |
| Photonic GBS | squeeze/displace gates, interferometers, beam splitters |
| Hardware introspection | QPU topology, native gates, connectivity |
| Async execution | sample_async, observe_async, run_async (futures) |
run & dynamic circuits | run, mid-circuit measurement / conditioned (dynamic) circuits |
| Detector error models (DEM) | DEM construction and export |
| Built-in optimizers & gradients | bundled optimizers, parameter-shift / analytic gradients |
| Visualization & resources | draw, LaTeX export, plot, estimate_resources |
| Operation extension | register_operation, exp_pauli, PTSBE |
| Graph helpers | operators.from_google_matrix (PageRank) and similar |
Planned (not published yet)
| Area | Status |
|---|---|
| Packaged distribution | PyPI wheels, GPU build, Docker image, C++ API |
| Real hardware / cloud submission | IBM, IonQ, Braket, Quantinuum, Origin… — the reference build accepts the targets but runs locally instead |
能力覆盖
开放参考实现现在能跑哪些功能，哪些还只是接口规范。每个「规范接口」页面的示例都配有行内提示——本页则是一份全局速览。
如何阅读
可运行——在参考实现上直接就能跑（纯 NumPy/SciPy，只用 CPU，不需要账号）。规范——已写明的预期接口，开放参考实现暂时还没内置。
规划中——尚未提供的发行或硬件路径。
各页的行内提示始终最新，以它为准；本页由同一套可运行性检查自动生成。
现在就能跑
| 领域 | 参考实现上可用的能力 |
|---|---|
| 门模型内核 | H、X、Y、Z、S、T、Rx/Ry/Rz、R1、SWAP；受控、伴随、compute/action/uncompute |
| 采样与期望 | sample、observe（自旋算符）、get_state 当前可用；异步版本（*_async）属规范接口（见下） |
| 噪声建模 | NoiseModel、add_channel/add_all_qubit_channel；去极化、振幅阻尼、比特/相位翻转、X/Y/ZError、KrausChannel；内联 apply_noise；量子轨迹采样 |
| 开放系统动力学 | evolve、薛定谔与 Lindblad 主方程、collapse_operators、SuperOperator（左乘/右乘/夹乘）、Schedule + 时间依赖系数、积分器 |
| 算符代数 | 自旋；玻色（含 position、momentum、displace、squeeze、Kerr 乘积）；Jordan-Wigner 费米子；自定义 define/instantiate |
| 模拟器后端 | cpu（qpp-cpu 的别名）、density-matrix-cpu 及其它 CPU 模拟后端 |
规范接口（已写明，未内置）
| 领域 | 状态 |
|---|---|
| 生成式量子本征求解器 GQE | 基于 transformer 的本征求解器——接口已写明 |
| QEC 解码器 | qLDPC-GPU、张量网络、滑动窗口、LUT；实时解码 |
| Torch 可微积分器 | Dopri5/8、Bosh3、Magnus（上面那几个 CPU 积分器现在就能跑） |
| 光子 GBS | squeeze/displace 门、干涉仪、分束器 |
| 硬件自省 | QPU 拓扑、原生门、连通性 |
| 异步执行 | sample_async、observe_async、run_async（future 句柄） |
run 与动态线路 | run、中途测量 / 条件（动态）线路 |
| 探测器错误模型（DEM） | DEM 构造与导出 |
| 内置优化器与梯度 | 内置优化器、参数移位 / 解析梯度 |
| 可视化与资源估计 | draw、LaTeX 导出、plot、estimate_resources |
| 算符扩展 | register_operation、exp_pauli、PTSBE |
| 图算法辅助 | operators.from_google_matrix（PageRank）等 |
规划中（尚未发布）
| 领域 | 状态 |
|---|---|
| 打包发行 | PyPI wheel、GPU 版本、Docker 镜像、C++ API |
| 真实硬件 / 云提交 | IBM、IonQ、Braket、Quantinuum、Origin……参考实现接受这些目标，但实际改在本地运行 |