Struct Graph 

pub struct Graph {
    pub n_vertices: usize,
    pub edges: Vec<(usize, usize, f64)>,
}

Graph for MaxCut warm-start.

Fields

n_vertices: usize

Number of vertices.

edges: Vec<(usize, usize, f64)>

Weighted edge list: (u, v, weight).

Implementations

impl Graph

pub fn laplacian(&self) -> Array2<f64>

Compute the graph Laplacian L = D - A.

D is the diagonal degree matrix (weighted degrees) and A is the weighted adjacency matrix.

pub fn fiedler_vector(&self) -> Result<Vec<f64>>

Compute the Fiedler vector (second-smallest eigenvector of the Laplacian).

Uses inverse power iteration with a small shift λ = 0.1 to avoid the trivial zero eigenvalue, deflating the constant component at each step so the iteration converges to the Fiedler vector rather than the all-ones eigenvector.

Returns a unit vector whose entries are the vertex assignments in [-1, 1].

Errors

Returns MLError::NumericalError if the shifted Laplacian is singular.

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

Compute the MaxCut value for the given vertex assignments.

For each edge (u, v, w), adds w * (1 − sign(a_u) * sign(a_v)) / 2 to the total.

Trait Implementations

impl Clone for Graph

fn clone(&self) -> Graph

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Graph

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.