Struct GraphColoring 

pub struct GraphColoring {
    pub num_colors: usize,
    pub colors: Vec<u32>,
    pub color_offsets: Vec<usize>,
    pub color_order: Vec<u32>,
}

Result of a distance-2 graph coloring of a sparse matrix.

Contains the color assignment for each row, plus auxiliary data structures for efficiently iterating over rows of the same color.

Fields

num_colors: usize

Number of distinct colors used.

colors: Vec<u32>

Color assigned to each row: colors[i] is the color of row i.

color_offsets: Vec<usize>

Start index of each color group in color_order. color_offsets[c]..color_offsets[c+1] gives the range of row indices with color c in color_order.

color_order: Vec<u32>

Row indices sorted by color.

Implementations

impl GraphColoring

pub fn from_csr<T: GpuFloat>( _handle: &SparseHandle, matrix: &CsrMatrix<T>, ) -> SparseResult<Self>

Compute a distance-2 coloring of the adjacency graph of a CSR matrix.

The matrix must be square. The coloring considers the structural (non-zero) pattern, not the numerical values.

Arguments

• _handle — sparse handle (reserved for future GPU-accelerated variants).
• matrix — a square CSR matrix.

Errors

Returns SparseError::DimensionMismatch if the matrix is not square.

pub fn from_csr_host( row_ptr: &[i32], col_idx: &[i32], n: usize, ) -> SparseResult<Self>

Compute distance-2 coloring from host-side CSR structure.

This is the core algorithm, operating on host data.

pub fn rows_for_color(&self, color: usize) -> &[u32]

Returns the row indices for a given color.

Panics

This method does not panic; returns an empty slice if color >= num_colors.

pub fn parallel_ilu0<T: GpuFloat>( &self, _handle: &SparseHandle, matrix: &CsrMatrix<T>, ) -> SparseResult<(CsrMatrix<T>, CsrMatrix<T>)>

Apply coloring to parallelize ILU(0) factorization.

Performs ILU(0) on a CSR matrix, processing rows color-by-color. Within each color, rows are independent and can be processed in parallel.

Returns (L, U) where L is unit lower triangular and U is upper triangular (including diagonal).

Arguments

• _handle — sparse handle.
• matrix — the CSR matrix to factor (square, will be consumed for values).

Errors

Returns SparseError::SingularMatrix if a zero pivot is encountered.

pub fn validate(&self, row_ptr: &[i32], col_idx: &[i32], n: usize) -> bool

Validates the coloring: checks that no two vertices within distance 2 share the same color.

Returns true if the coloring is valid.