Amazon Braket
Amazon Braket is a managed service that brokers access to QPUs from several vendors under one AWS account. qalgora-Q documents an intended export workflow to Braket devices through the Braket SDK, addressed by ARN-style name; the open reference build does not submit directly.
Devices on Braket
| Vendor | Device | Qubits | Modality | Region |
|---|---|---|---|---|
| IonQ | Forte-1 / Forte-Enterprise-1 | 36 #AQ | Trapped ion | us-east-1 |
| IQM | Garnet / Emerald | 20 / 54 | Superconducting | eu-north-1 |
| Rigetti | Ankaa-3 / Cepheus-1-108Q | 84 / 108 | Superconducting | us-west-1 |
| QuEra | Aquila | 256 | Neutral atom (analog) | us-east-1 |
| AQT | IBEX-Q1 | — | Trapped ion | eu-north-1 |
The device lineup shifts over time — OQC, Xanadu, and D-Wave have left; AQT IBEX-Q1 (a new ion-trap vendor) and Rigetti Cepheus-1-108Q are recent additions. For the current roster see the Amazon Braket supported devices page.
Credentials & submitting
Authenticate with standard AWS credentials (aws configure or an IAM role). Each
device has an ARN; results land in an S3 bucket. Jobs queue — use the async API.
import qalgora
# Planned Braket submission — requires the Braket SDK + AWS credentials (aws configure / IAM role).
# The open reference build does not submit directly; lines below are illustrative.
# bell is a predefined example kernel; the device ARN includes its region.
# qalgora.set_target("braket",
# machine="arn:aws:braket:us-east-1::device/qpu/ionq/Forte-1",
# region="us-east-1")
# job = qalgora.sample_async(bell, shots_count=1000)
# print(job.status())
# print(job.get())Regions, cost & simulators
| Item | Detail |
|---|---|
| Regions | Devices are pinned (e.g. IQM/AQT in eu-north-1, IonQ/QuEra in us-east-1, Rigetti in us-west-1) |
| Pricing | A per-task fee plus a per-shot fee that vary by device — see Amazon Braket Pricing for current rates |
| On-demand simulators | SV1 (statevector), DM1 (density matrix) |
| Program sets | Bundle up to 100 circuits (or a parametric sweep) in one task — 3×–24× faster on Rigetti / IQM |
| Braket Direct | Reserve a device by the hour, plus expert advice and experimental capabilities |
Validate locally on qalgora-Q's runnable qpp-cpu reference target (or a managed Braket
simulator) before paying for hardware shots.
Hybrid Jobs
Braket Hybrid Jobs run a classical-quantum loop next to the QPU. The same structure runs locally in qalgora-Q — iterate on the reference simulator, then (planned) submit the converged run to a Braket device.
qalgora.set_target("qpp-cpu") # develop locally (runnable reference target)
energy, params = qalgora.vqe(ansatz, hamiltonian, optimizer, parameter_count=n)
# Planned hardware submission (requires the Braket SDK + AWS credentials):
# qalgora.set_target("braket", machine="arn:aws:braket:eu-north-1::device/qpu/iqm/Garnet")
# final = qalgora.observe(ansatz, hamiltonian, params) # validate on hardwareReferences
- Amazon Braket, supported devices documentation. docs.aws.amazon.com
Amazon Braket
Amazon Braket 是一项托管服务，在统一的 AWS 账户下代理访问多家厂商的 QPU。qalgora-Q 记录的是经 Braket SDK 的预期导出工作流，按 ARN 风格的名称寻址；开放参考实现不会直接提交。
Braket 上的设备
| 厂商 | 设备 | 量子比特数 | 技术路线 | 区域 |
|---|---|---|---|---|
| IonQ | Forte-1 / Forte-Enterprise-1 | 36 #AQ | 离子阱 | us-east-1 |
| IQM | Garnet / Emerald | 20 / 54 | 超导 | eu-north-1 |
| Rigetti | Ankaa-3 / Cepheus-1-108Q | 84 / 108 | 超导 | us-west-1 |
| QuEra | Aquila | 256 | 中性原子（模拟式） | us-east-1 |
| AQT | IBEX-Q1 | — | 离子阱 | eu-north-1 |
设备阵容会随时间变化——OQC、Xanadu、D-Wave 已退出；AQT IBEX-Q1（新晋离子阱厂商）与 Rigetti Cepheus-1-108Q 是较新加入的设备。当前清单以 Amazon Braket supported devices 页面为准。
凭据与任务提交
使用标准 AWS 凭据（aws configure 或 IAM 角色）进行身份验证。每台设备都有对应的 ARN；结果存储在 S3 存储桶中。任务需要排队——请使用异步 API。
import qalgora
# 规划中的 Braket 提交——需要 Braket SDK + AWS 凭据（aws configure / IAM 角色）。
# 开放参考实现不会直接提交；以下为示意代码。
# bell 是预定义的示例内核；设备 ARN 含所在区域。
# qalgora.set_target("braket",
# machine="arn:aws:braket:us-east-1::device/qpu/ionq/Forte-1",
# region="us-east-1")
# job = qalgora.sample_async(bell, shots_count=1000)
# print(job.status())
# print(job.get())区域、费用与模拟器
| 项目 | 详情 |
|---|---|
| 区域 | 设备绑定特定区域（如 IQM/AQT 位于 eu-north-1，IonQ/QuEra 位于 us-east-1，Rigetti 位于 us-west-1） |
| 定价 | 每任务费 + 按次采样费，因设备而异——当前价格以 Amazon Braket Pricing 为准 |
| 按需模拟器 | SV1（态矢量）、DM1（密度矩阵） |
| Program sets | 将最多 100 条电路（或一组参数扫描）打包为单个任务——在 Rigetti / IQM 上提速 3×–24× |
| Braket Direct | 可按小时预约设备，并获得专家咨询与实验性能力 |
在付费使用真实硬件之前，请先在 qalgora-Q 可运行的 qpp-cpu 参考目标后端（或 Braket 托管模拟器）上完成验证。
Hybrid Jobs
Braket Hybrid Jobs 在 QPU 旁运行经典-量子混合循环。qalgora-Q 本地也支持相同的结构——先在参考模拟器上迭代，再（规划中）将收敛后的任务提交到 Braket 设备。
qalgora.set_target("qpp-cpu") # 本地开发（可运行的参考目标后端）
energy, params = qalgora.vqe(ansatz, hamiltonian, optimizer, parameter_count=n)
# 规划中的硬件提交（需要 Braket SDK + AWS 凭据）：
# qalgora.set_target("braket", machine="arn:aws:braket:eu-north-1::device/qpu/iqm/Garnet")
# final = qalgora.observe(ansatz, hamiltonian, params) # validate on hardware参考文献
- Amazon Braket, supported devices documentation. docs.aws.amazon.com