Struct Graph

Source

pub struct Graph { /* private fields */ }

Expand description

A weighted graph (directed or undirected).

Implementations§

Source §

impl Graph

Source

pub fn new(n_nodes: usize, directed: bool) -> Self

Create an empty graph with n_nodes nodes.

Source

pub fn add_edge(&mut self, from: usize, to: usize, weight: f64) -> Result<()>

Add a weighted edge.

§Errors

Returns an error if either node index is out of range.

Source

pub fn from_correlation_matrix(matrix: &[Vec<f64>], threshold: f64) -> Self

Build an undirected graph from a correlation matrix with a threshold.

Adds an edge between nodes i and j if |correlation[i][j]| ≥ threshold. Edge weight is the absolute correlation.

Source

pub fn from_sparse_matrix(matrix: &SparseMatrix) -> Self

Build an undirected graph from a sparse matrix.

For a symmetric matrix (e.g. adjacency/connectivity), each stored entry (i, j, w) with i < j becomes an edge. Entries with i >= j are skipped to avoid double-counting in undirected graphs.

Source

pub fn louvain_with_resolution(&self, resolution: f64) -> Community

Louvain community detection with a resolution parameter.

The resolution parameter γ controls community granularity:

γ < 1.0 → fewer, larger communities
γ = 1.0 → standard modularity
γ > 1.0 → more, smaller communities

Implements both Phase 1 (local moves) and Phase 2 (hierarchical aggregation).

Source

pub fn modularity_with_resolution( &self, assignments: &[usize], resolution: f64, ) -> f64

Compute modularity Q with a resolution parameter.

Q = Σ_c [L_c / m - γ * (d_c / 2m)²]

Source

pub fn n_nodes(&self) -> usize

Number of nodes.

Source

pub fn n_edges(&self) -> usize

Number of edges.

Source

pub fn neighbors(&self, node: usize) -> &[(usize, f64)]

Neighbors of a node with edge weights.

Source

pub fn degree_centrality(&self) -> Vec<f64>

Degree centrality: degree(v) / (n - 1).

Source

pub fn betweenness_centrality(&self) -> Vec<f64>

Betweenness centrality using Brandes’ algorithm.

Source

pub fn closeness_centrality(&self) -> Vec<f64>

Closeness centrality: (n-1) / Σ d(v, u).

Source

pub fn centrality(&self) -> CentralityScores

Compute all three centrality metrics.

Source

pub fn louvain(&self) -> Community

Louvain community detection.

Iteratively moves nodes between communities to maximize modularity.

Source

pub fn modularity(&self, assignments: &[usize]) -> f64

Compute modularity Q for a given community assignment.

Uses the community-level formula: Q = Σ_c [L_c / m - (d_c / 2m)²] where L_c = sum of edge weights within community c, m = total edge weight, and d_c = sum of node degrees in community c.