Struct RouterConfig 

pub struct RouterConfig {
    pub neumann_spectral_radius_threshold: f64,
    pub cg_condition_threshold: f64,
    pub sparsity_sublinear_threshold: f64,
    pub true_batch_threshold: usize,
    pub push_graph_size_threshold: usize,
}

Configuration thresholds that govern the routing decision tree.

All thresholds have sensible defaults; override them when benchmarks on your workload indicate a different crossover point.

Example

use ruvector_solver::router::RouterConfig;

let config = RouterConfig {
    cg_condition_threshold: 50.0,
    ..Default::default()
};

Fields

neumann_spectral_radius_threshold: f64

Maximum spectral radius for which the Neumann series is attempted.

If the estimated spectral radius exceeds this value the router will not select Neumann even for diagonally dominant matrices.

Default: 0.95.

cg_condition_threshold: f64

Maximum condition number for which CG is preferred over BMSSP.

CG converges in O(sqrt(kappa)) iterations; when kappa is too large a preconditioned method (BMSSP) is cheaper.

Default: 100.0.

sparsity_sublinear_threshold: f64

Maximum density (fraction of non-zeros) for the Neumann sublinear fast-path.

Neumann is only worthwhile when the matrix is truly sparse.

Default: 0.05 (5%).

true_batch_threshold: usize

Minimum batch size for which the TRUE solver is preferred over CG in BatchLinearSystem queries.

Default: 100.

push_graph_size_threshold: usize

Graph size threshold (number of rows) above which PageRankPairwise switches from ForwardPush to HybridRandomWalk.

Default: 1_000.

Trait Implementations

impl Clone for RouterConfig

fn clone(&self) -> RouterConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RouterConfig

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

Formats the value using the given formatter.

impl Default for RouterConfig

fn default() -> Self

Returns the “default value” for a type.