quapp-pennylane

Quapp PennyLane library supporting the Quapp Platform for Quantum Computing with providers, devices, circuit export, invocation, and job fetching utilities.

Overview

quapp-pennylane is a Python library that integrates the Quapp Platform with PennyLane-based quantum backends and simulators. It provides common abstractions for providers and devices, helpers for circuit construction and export, robust job submission and result fetching flows, and consistent job-scoped logging across all components.

Features

• Provider & device factories — PennyLane-compatible factories for local simulators (e.g., default.qubit) and vendor-backed hardware (Rigetti QVM, AWS Braket, IBM Quantum).
• Circuit export — PennylaneCircuitExportTask with a _PennylaneDrawableCircuit adapter that satisfies the CircuitExportTask.draw() contract using qml.draw_mpl.
• Job execution — QAppPennylaneDevice drives the full execution pipeline: tape construction, device selection, QNode creation, probability histogram generation, and timing metrics.
• Rigetti QVM support — dynamic device-name derivation for N-qubit QVM topologies (Nq-qvm, Nq-pyqvm) alongside fixed 9-qubit variants.
• Structured logging — job-scoped, instance-bound loggers emit structured key=value fields at every major step for traceability and debugging.
• Compatibility shims — quapp_pennylane/__init__.py patches API breaks between pennylane-rigetti==0.36.0 and PennyLane 0.37–0.42+ at import time (see Compatibility below).

Installation

pip install quapp-pennylane

Requirements

Project Structure

quapp_pennylane/
├── __init__.py                          # Compatibility shims (loaded at import)
├── async_tasks/
│   └── pennylane_circuit_export_task.py # PennylaneCircuitExportTask + adapter
├── component/
│   └── backend/
│       └── pennylane_invocation.py
├── factory/
│   ├── pennylane_device_factory.py
│   ├── pennylane_handler_factory.py
│   └── pennylane_provider_factory.py
├── handler/
│   └── invocation_handler.py
└── model/
    ├── device/
    │   ├── qapp_pennylane_device.py     # Core execution device
    │   ├── aws_braket_device.py
    │   └── ibm_quantum_device.py
    └── provider/
        ├── qapp_pennylane_provider.py
        ├── aws_braket_provider.py
        ├── ibm_cloud_provider.py
        └── ibm_quantum_provider.py

Compatibility

quapp_pennylane/__init__.py applies the following shims at import time so that pennylane-rigetti==0.36.0 works against PennyLane 0.37–0.42+:

pydantic is pinned to >=1.7.2,<2.0.0 in pyproject.toml to make the v1 requirement of qcs-api-client<0.22.0 explicit and prevent pip from resolving v2.

Recent Changes

0.0.5.dev6

• feat(error): Thêm quapp_pennylane/error_rules.py — 9 rule phân loại lỗi đặc thù PennyLane, đăng ký vào classifier của quapp-common khi package được import. Nhờ đó lỗi PennyLane có category / code / hướng xử lý cụ thể trong Job Log thay vì rơi về fallback chung COMPILATION_FAILED:

  • PENNYLANE_DEVICE_NOT_AVAILABLE — device không có trong qml.plugin_devices (thường do thiếu plugin hoặc viết sai tên device);
  • PENNYLANE_INVALID_QUANTUM_FUNCTION, PENNYLANE_WIRE_ERROR, PENNYLANE_QUEUING_ERROR, PENNYLANE_MEASUREMENT_SHAPE_ERROR — lỗi trong circuit do user viết (USER_CODE);
  • PENNYLANE_OPERATION_UNSUPPORTED, PENNYLANE_TAPE_ERROR, PENNYLANE_DEVICE_ERROR — giới hạn của SDK/device (SDK_RUNTIME);
  • PENNYLANE_DEVICE_CONFIG_INVALID — DeviceError về shots/wire, phân loại là INVALID_INPUT vì user sửa được.

  Rule match theo tên class dạng string, nên quapp-common không phải import PennyLane — bắt buộc, vì lib này kéo theo pennylane-rigetti và pyquil với ràng buộc version rất chặt.

• fix(device): QAppPennylaneDevice._get_shots() dùng getattr(job_result, 'shots') trên một dict nên luôn trả None — shots không bao giờ tới được Job Log dù _create_job() đã tính đúng. Nay đọc bằng .get() cho dict.

• fix(device): _create_job() đọc parts[1] cho device Rigetti QVM mà không kiểm tra độ dài, nên spec thiếu phần sau / gây IndexError: list index out of range. Nay raise ValueError nói rõ spec phải có dạng rigetti.qvm/<device_type>.

• build: quapp-common từ ==0.0.11.dev7 (exact pin, không có register_rules) thành >=0.0.14.dev1.

• test: Thêm tests/test_error_rules.py — 14 case, dựng exception thật từ PennyLane 0.37 để test vỡ khi PennyLane đổi tên exception.

Ghi chú version: nhảy từ 0.0.4.dev6 (bản develop đang giữ sau revert/pennylane-to-0.0.4.dev6) lên 0.0.5.dev6 vì 0.0.5.dev1–dev5 đã publish, và template js-pennylane / qs-pennylane pin quapp-pennylane>=0.0.5.dev2 nên không thể lùi về dòng 0.0.4.

0.0.5.dev2

• fix(export): Fix _PennylaneDrawableCircuit.draw() — qml.draw_mpl requires a QNode, not a bare callable; now captures the circuit into a QuantumTape, derives wires, creates a default.qubit device, and wraps in a QNode before rendering with qml.draw_mpl.
• build: Bump version to 0.0.5.dev2.

0.0.5.dev1

• feat(device): Add structured key=value log fields to QAppPennylaneDevice at every execution stage — initialization, tape diagnostics, device resolution, QNode setup, execution timing, histogram generation, and finalization.
• feat(device): Default shots fallback — shots=None or shots=0 in CircuitRunningOption falls back to 1024 with a warning instead of raising.
• feat(export): Introduce _PennylaneDrawableCircuit adapter in PennylaneCircuitExportTask to wrap PennyLane circuits for the CircuitExportTask.draw() contract using qml.draw_mpl.
• feat(export): Emit Python runtime information and duration timing in _transpile_circuit.
• fix(compat): Patch pydantic.BaseModel.__config__ at import time for Pydantic v2 environments — qcs-api-client<0.22.0 accesses self.__config__.env_prefix (a Pydantic v1 inner Config class removed in v2).
• fix(deps): Pin pydantic>=1.7.2,<2.0.0 in pyproject.toml to prevent pip from resolving Pydantic v2.
• fix(compat): Extend compatibility shims for pennylane-rigetti==0.36.0 against PennyLane 0.37–0.42+: measurements.Expectation, operation.Tensor, and pennylane.QubitDevice.
• build: Bump version to 0.0.5.dev1.

For detailed API references, refer to the in-code documentation or contact the maintainers at corp@citynow.vn.