Struct SimdBetweenness 

pub struct SimdBetweenness;

SIMD-accelerated operations for betweenness centrality computation.

Accelerates the dependency accumulation phase of Brandes’ algorithm, which is the most compute-intensive part of betweenness centrality.

Implementations

impl SimdBetweenness

pub fn accumulate_dependencies( delta: &mut [f64], sigma: &[f64], predecessors: &[Vec<usize>], order: &[usize], )

SIMD-accelerated dependency accumulation for Brandes’ algorithm.

Updates the dependency values: delta[v] += (sigma[v] / sigma[w]) * (1 + delta[w]) for all predecessors v of w in the shortest-path DAG.

Arguments

• delta - Dependency accumulator vector (modified in place)
• sigma - Number of shortest paths through each node
• predecessors - For each node w, list of predecessor node indices
• order - Nodes in reverse BFS order (for back-propagation)

pub fn batch_update_centrality( centrality: &mut [f64], delta: &[f64], source: usize, )

Batch update betweenness scores from a single-source dependency vector.

Adds the dependency values to the centrality scores using SIMD.

Arguments

• centrality - Running betweenness centrality scores (modified in place)
• delta - Dependency values from a single-source computation
• source - Index of the source node (excluded from update)

pub fn normalize(centrality: &mut [f64], n: usize, directed: bool)

Normalize betweenness centrality scores.

For undirected graphs, divide by (n-1)(n-2). For directed graphs, divide by (n-1)(n-2) as well (Brandes’ convention).

Arguments

• centrality - Betweenness scores to normalize (modified in place)
• n - Number of nodes in the graph
• directed - Whether the graph is directed