Python API

Python implementations of PHY algorithms and AI models for rapid prototyping, algorithm validation, and integration testing. These components can then be lowered into TensorRT engines for real-time execution in the Runtime API.

Note

The Python API may be subject to change as more PHY channels are added in future releases.

Components

• Constants (Constants) - 5G NR physical layer constants and parameters

• Datasets (Datasets) - Channel simulation dataset utilities using Sionna

• MLIR-TensorRT Wrapper (MLIR-TensorRT Wrapper) - Python wrapper for MLIR-TensorRT compiler and execution

• PHY JAX (JAX PHY) - JAX-based PHY implementations including AI integration and TensorRT lowering

• PHY NumPy (NumPy PHY) - Pure NumPy reference PHY implementations

• TensorRT Plugins (TensorRT Plugins) - Custom TensorRT plugins for specialized operations (DMRS, FFT, Cholesky)

• Types (Types) - Type definitions and data structure aliases

• Utils (RAN Utils) - Helper utilities for configuration, I/O, and timing

Python Components

• Constants
  • Overview
  • API Reference
• Datasets
  • Overview
  • API Reference
• MLIR-TensorRT Wrapper
  • Overview
  • API Reference
• JAX PHY
  • Overview
  • Components
• NumPy PHY
  • Overview
  • Components
• TensorRT Plugins
  • Overview
  • Usage
  • Python API
  • C++ API Reference
• Types
  • Overview
  • API Reference
• RAN Utils
  • Overview
  • API Reference