API Reference

Comprehensive API documentation for the Deepcomet AI ecosystem components.

Aurelia Standard Library

The Aurelia standard library provides core types and functions for AI-native computation.

Tensor Module

Core tensor operations for neural computation:

    tensor_api.aul 
  // Create tensors
Tensor::zeros(shape: [usize]) -> Tensor<T, N>
Tensor::ones(shape: [usize]) -> Tensor<T, N>
Tensor::random(shape: [usize]) -> Tensor<T, N>

// Operations
tensor.relu() -> Tensor<T, N>
tensor.sigmoid() -> Tensor<T, N>
tensor.softmax(dim: usize) -> Tensor<T, N>
tensor.reshape(shape: [usize]) -> Tensor<T, M>

// Matrix operations
tensor @ tensor -> Tensor<T, N>  // matmul
tensor + tensor -> Tensor<T, N>  // element-wise add
tensor * tensor -> Tensor<T, N>  // element-wise mul  

Target Attributes

Functions can be annotated with target attributes for hardware-specific compilation:

    targets.aul 
  @target(cpu="x86-avx512")
fn cpu_compute() { ... }

@target(gpu="cuda")
fn gpu_compute() { ... }

@target(npu="qualcomm-hexagon")
fn npu_compute() { ... }  

Zenith Kernel API

System-level APIs for interacting with the Zenith Kernel:

Scheduling

zen::schedule::priority(task, level) — Set task scheduling priority
zen::schedule::predict(task) — Query predicted resource needs
zen::schedule::affinity(task, core_set) — Set CPU/NPU affinity

Security

zen::watchdog::register(handler) — Register anomaly detection handler
zen::watchdog::status() — Query current system security status