Welcome to qalgora-Q
A conceptual overview of a hybrid-quantum programming and co-simulation platform for CPUs, local simulators, planned GPU backends, and vendor-QPU workflows under one programming model. The platform is still early-stage, so these docs lay out the design and direction more than finished, production-ready software.
The current open reference build is a CPU-only NumPy statevector simulator; GPU builds, Docker images, C++ packages, cloud backends, and native QPU submission are planned interfaces or require external vendor SDKs.
Current capability status
| Capability | Status |
|---|---|
| CPU-only NumPy statevector simulator | Available now |
| Core gate-model examples | Available in the reference build |
| Browser playground | Available |
| GPU backend | Planned — no public package or benchmark yet |
| PyPI wheel / Docker image | Planned — not yet published |
| C++ API / package | Documented or planned — not yet shipped as a public package |
| Cloud and hardware targets | Adaptation / export workflows only |
| Native physical-QPU submission | Not built in |
| Performance claims | Theoretical expectations until public benchmarks are released |
Browse by topic
Quick Start →
Request access to the reference build and run your first local kernel.
Basics →
What a kernel is and how the programming model works.
Examples →
Gates, measurement, execution, optimizers, noise.
Applications →
VQE, QAOA, Grover, Shor, chemistry, and more.
Backends →
Available simulators, planned backends, and vendor-QPU adaptation notes.
Dynamics →
Time evolution, operators, integrators, batching.
qalgora-QX →
Planned high-level solver and quantum-error-correction interfaces.
Reference →
Language spec, Python & C++ API, operations, glossary.
Featured algorithms
Grover's Search
Quadratic speedup for unstructured search.
Shor's Factoring
Polynomial-time integer factorization.
Quantum Chemistry
Molecular ground states with VQE.
QAOA Max-Cut
Combinatorial optimization on graphs.
Teleportation
Move a qubit with a Bell pair and 2 bits.
Ising Dynamics
Trotterized real-time spin-chain evolution.
Preparing for vendor hardware workflows
The kernel abstraction is designed to align with major quantum-cloud instruction models. In the current reference build, this means simulation-level adaptation and export-oriented workflows. Submitting jobs to physical QPUs requires external vendor SDKs, credentials, and hardware-specific compilation; qalgora-Q does not currently provide native QPU submission.
The specs below are vendor-reported / publicly announced figures as of the documentation review date; they are shown for workflow orientation only — not qalgora-Q native execution targets — and may change as vendors update their hardware. Likewise, any algorithmic “speedups” in these docs are theoretical / query-complexity statements unless benchmarked on a concrete backend.
IBM Quantum
Heron 156q & Nighthawk 120q superconducting.
Google Quantum AI
Willow 105q; below-threshold surface-code error correction.
Amazon Braket
Aggregates IonQ, IQM, Rigetti, QuEra, AQT; adaptation/export here, not native submission.
Quantinuum
Helios, 98 qubits, 48 logical (2025).
Hello, quantum world
A minimal Bell-state kernel in Python:
import qalgora
@qalgora.kernel
def bell():
q = qalgora.qvector(2)
h(q[0])
x.ctrl(q[0], q[1]) # controlled-X (CNOT): control q[0], target q[1]
mz(q)
result = qalgora.sample(bell, shots_count=1000)
print(result) # expected distribution: roughly 50% "00" and 50% "11"- qalgora-Q is a prototype demo framework for learning quantum algorithms — not production-grade industrial software. A runnable pure-NumPy reference build is provided so the examples execute; the packaged distributions (PyPI wheel, GPU build, Docker image, C++ packages) are planned and not yet published.
- Its multi-hardware compatibility covers simulation-level instruction-set adaptation only; it does not support native task submission directly to physical quantum processors.
- All performance statements are theoretical expectations; no public benchmark data is available to verify them.
- For industrial quantum development, please use mature frameworks such as Qiskit, Cirq, PennyLane, CUDA-Q, or pyQPanda / MindQuantum.
Attribution: qalgora-Q's programming model is inspired by NVIDIA CUDA-Q (Apache-2.0); the kernel / qvector / control-flow idioms intentionally align with it for interoperability.
现有开源参考实现是 CPU-only 的 NumPy 态矢量模拟器；GPU 版本、Docker 镜像、C++ 包、云端后端和原生 QPU 提交能力属于规划接口，或需外部厂商 SDK 支持。
当前能力状态
| 能力 | 状态 |
|---|---|
| CPU-only NumPy 态矢量模拟器 | 现已可用 |
| 核心门模型示例 | 参考实现已支持 |
| 浏览器在线模拟 | 已可用 |
| GPU 后端 | 规划中——尚无公开包或基准测试 |
| PyPI wheel / Docker 镜像 | 规划中——尚未发布 |
| C++ API / 包 | 文档化或规划中——尚未作为公开包发布 |
| 云端与硬件目标 | 仅适配/导出工作流 |
| 原生真实 QPU 提交 | 不支持 |
| 性能描述 | 公开基准发布前均为理论预期 |
按主题浏览
快速开始 →
申请参考实现访问权限，运行第一个本地内核。
基础 →
什么是内核，以及编程模型如何运作。
示例 →
门、测量、执行、优化器、噪声。
应用 →
VQE、QAOA、Grover、Shor、量子化学等。
后端 →
已支持模拟器、规划中后端，以及厂商 QPU 适配说明。
动力学 →
时间演化、算符、积分器、批量模拟。
qalgora-QX →
规划中的高层求解器与量子纠错接口。
参考 →
语言规范、Python 与 C++ API、操作、术语表。
精选算法
Grover 搜索
无结构搜索的平方级加速。
Shor 因数分解
多项式时间的整数因数分解。
量子化学
用 VQE 求分子基态。
QAOA 最大割
图上的组合优化。
隐形传态
用一对 Bell 态加两个比特搬运量子态。
伊辛动力学
自旋链实时演化的特罗特分解。
面向厂商硬件流程的适配准备
qalgora-Q 的内核抽象在设计上尽量对齐主流量子云的指令模型。当前参考实现主要覆盖仿真层面的适配与面向导出的工作流。提交到真实 QPU 仍需外部厂商 SDK、账号凭据和硬件特定编译；qalgora-Q 当前不提供原生 QPU 提交能力。
下列规格为截至本文档核对日期的供应商公开发布数据，仅用于工作流适配说明，不代表 qalgora-Q 可原生提交任务，且可能随厂商硬件更新而变化。同理，文档中任何算法“加速”在具体后端基准测试之前，均为理论/查询复杂度层面的陈述。
IBM Quantum
Heron 156q 与 Nighthawk 120q 超导。
Google Quantum AI
Willow 105q；低于阈值的表面码纠错演示。
Amazon Braket
聚合 IonQ、IQM、Rigetti、QuEra、AQT；此处为适配/导出流程，而非原生提交。
Quantinuum
Helios，98 量子比特，48 个逻辑比特（2025）。
你好 量子世界
一个最简的 Bell 态内核（Python）：
import qalgora
@qalgora.kernel
def bell():
q = qalgora.qvector(2)
h(q[0])
x.ctrl(q[0], q[1]) # 受控 X 门（CNOT）：控制位 q[0]，目标位 q[1]
mz(q)
result = qalgora.sample(bell, shots_count=1000)
print(result) # 期望分布：约 50% 为 "00"，约 50% 为 "11"- qalgora-Q 仅为面向量子算法学习的原型演示框架，并非工业级生产软件。我们提供一个可运行的纯 NumPy 参考实现，使文档示例都能跑通；打包发行版（PyPI wheel、GPU 版本、Docker 镜像、C++ 包）尚在规划中，暂未发布。
- 所谓多硬件兼容仅覆盖仿真层面的指令集适配，不支持直接向真实量子处理器原生提交任务。
- 所有性能描述均为理论预期，目前没有可供核验的公开基准测试数据。
- 工业级量子开发请选用 Qiskit、Cirq、PennyLane、CUDA-Q、pyQPanda / MindQuantum 等成熟框架。
来源致谢：qalgora-Q 的编程模型受 NVIDIA CUDA-Q（Apache-2.0）启发；kernel / qvector 与控制流写法有意与其保持一致以便互操作。