Struct SimdTraversal 

pub struct SimdTraversal;

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

Provides optimized initialization, update, and reduction operations on the distance/level arrays used in graph traversals.

Implementations

impl SimdTraversal

pub fn init_distances(n: usize, source: usize) -> Array1<f64>

Initialize a distance vector with infinity (or a sentinel value).

Sets all entries to f64::INFINITY except the source node.

Arguments

• n - Number of nodes
• source - Source node index

Returns

Distance vector with distances[source] = 0.0 and all others infinity.

pub fn init_distances_multi_source(n: usize, sources: &[usize]) -> Array1<f64>

Initialize a distance vector for multi-source BFS.

Arguments

• n - Number of nodes
• sources - Slice of source node indices

Returns

Distance vector with distances[s] = 0.0 for each source s.

pub fn relax_neighbors( distances: &mut Array1<f64>, adjacency_row: &ArrayView1<'_, f64>, current_distance: f64, ) -> Vec<usize>

SIMD-accelerated BFS frontier relaxation.

For each node in the current frontier, attempts to relax the distance of its neighbors. This is the inner loop of BFS.

Arguments

• distances - Current distance vector (modified in place)
• adjacency_row - Adjacency row for the current node
• current_distance - Distance of the current node

Returns

Vector of neighbor indices that were relaxed (for the next frontier).

pub fn eccentricity(distances: &ArrayView1<'_, f64>) -> f64

Compute the eccentricity of a node from its distance vector.

The eccentricity is the maximum finite distance to any reachable node.

Arguments

• distances - Distance vector from a single source

Returns

The eccentricity value. Returns 0.0 if no nodes are reachable.

pub fn count_reachable(distances: &ArrayView1<'_, f64>) -> usize

Count the number of reachable nodes from a distance vector using SIMD.

A node is reachable if its distance is finite.

Arguments

• distances - Distance vector from a single source

Returns

Number of reachable nodes (including the source itself).

pub fn sum_finite_distances(distances: &ArrayView1<'_, f64>) -> f64

Compute the sum of finite distances using SIMD.

Used for computing closeness centrality: C(v) = (n-1) / sum(d(v, w)).

Arguments

• distances - Distance vector from a single source

Returns

Sum of all finite distances.

pub fn element_wise_min( a: &ArrayView1<'_, f64>, b: &ArrayView1<'_, f64>, ) -> Array1<f64>

SIMD-accelerated element-wise minimum of two distance vectors.

Useful for merging BFS trees or computing all-pairs shortest paths.

Arguments

• a - First distance vector
• b - Second distance vector

Returns

Element-wise minimum vector.

pub fn diameter_from_distance_matrix( distance_matrix: &ArrayView2<'_, f64>, ) -> f64

Compute the diameter of a graph from an all-pairs distance matrix.

The diameter is the maximum eccentricity over all nodes.

Arguments

• distance_matrix - All-pairs distance matrix (n x n)

Returns

The graph diameter. Returns 0.0 for empty or disconnected graphs.