Function prefer_direct_solver 

pub fn prefer_direct_solver(n: usize, density: f64) -> bool

Returns true if the direct sparse solver (dense LU) is preferred over iterative methods for the given system dimensions and density.

Heuristic: direct solver wins for small systems OR for dense/near-dense systems where iterative methods converge slowly.

Arguments

• n — system dimension.
• density — fill ratio in [0.0, 1.0] (nnz / (n * n)).

Examples

use oxicuda_solver::sparse::direct::prefer_direct_solver;

// Small system: always direct.
assert!(prefer_direct_solver(50, 0.01));
// Large sparse system: iterative (CG preferred for SPD).
assert!(!prefer_direct_solver(10_000, 0.001));
// Dense system: direct even if large-ish.
assert!(prefer_direct_solver(200, 0.8));