Module simd_ops 

SIMD-accelerated operations for graph algorithms

This module provides SIMD-optimized implementations of performance-critical graph algorithm kernels, leveraging the scirs2-core SIMD infrastructure via the SimdUnifiedOps trait.

Accelerated Operations

• PageRank power iteration: vector-scalar multiply, L1 norm, convergence check
• Spectral methods: graph Laplacian construction, eigenvalue iteration helpers
• Adjacency matrix-vector products: dense and CSR-style sparse matvec
• Betweenness centrality accumulation: batch dependency accumulation
• BFS/DFS distance vector operations: distance initialization, level updates

Usage

All functions are feature-gated under #[cfg(feature = "simd")] and automatically dispatch to the optimal SIMD implementation available on the current platform (AVX-512, AVX2, SSE2, NEON, or scalar fallback).

use scirs2_graph::simd_ops::{SimdPageRank, SimdSpectral, SimdAdjacency};

Structs

SimdAdjacency

SIMD-accelerated adjacency matrix-vector product operations.

SimdBetweenness

SIMD-accelerated operations for betweenness centrality computation.

SimdGraphUtils

General SIMD-accelerated vector utilities for graph computations.

SimdPageRank

SIMD-accelerated operations for PageRank power iteration.

SimdSpectral

SIMD-accelerated operations for spectral graph methods.

SimdTraversal

SIMD-accelerated operations for BFS/DFS distance and level vectors.