Struct MatchOrderResult 

pub struct MatchOrderResult {
    pub ordering: Perm<usize>,
    pub scaling: Vec<f64>,
    pub matched: usize,
    pub condensed_dim: usize,
    pub singletons: usize,
    pub two_cycles: usize,
}

Result of the combined matching-ordering pipeline.

Contains both the fill-reducing ordering (for symbolic analysis) and the MC64 scaling factors (for numeric factorization), plus diagnostics about the condensation process.

Usage

use rivrs_sparse::ordering::match_order_metis;
use faer::sparse::{SparseColMat, Triplet};

let triplets = vec![
    Triplet::new(0, 0, 4.0),
    Triplet::new(0, 1, 1.0),
    Triplet::new(1, 0, 1.0),
    Triplet::new(1, 1, 3.0),
];
let matrix = SparseColMat::try_new_from_triplets(2, 2, &triplets).unwrap();

let result = match_order_metis(&matrix).unwrap();
assert_eq!(result.scaling.len(), 2);
// Use ordering for symbolic analysis:
// SymmetricOrdering::Custom(result.ordering.as_ref())

Fields

ordering: Perm<usize>

Fill-reducing permutation with matched pair adjacency guarantee. Use with SymmetricOrdering::Custom(result.ordering.as_ref()).

scaling: Vec<f64>

MC64 symmetric scaling factors (linear domain). Apply as A_scaled[i,j] = scaling[i] * A[i,j] * scaling[j].

matched: usize

Number of matched entries from MC64. Equals n for structurally nonsingular matrices.

condensed_dim: usize

Dimension of the condensed graph passed to METIS. Strictly less than n when 2-cycles exist.

singletons: usize

Number of singleton nodes (self-matched).

two_cycles: usize

Number of 2-cycle pairs in the decomposed matching.