SparseLapack
The sparse factorizations, the seam a host sparse solver plugs into.
Named for symmetry with com.eignex.koblas.dense.Lapack rather than for accuracy — LAPACK has no sparse half. The real world here is UMFPACK, KLU, SuperLU and CHOLMOD, and the symmetry is worth more than the pedantry: a caller looking for the sparse counterpart of factor finds it where they expect.
Returns SparseFactorization rather than a concrete SparseLu, which is the design decision that makes the seam usable at all — see that interface for why no sparse factor format is interchangeable.
Inheritors
Properties
Relative preference among simultaneously available backends: automatic selection through registerBackend — JVM classpath discovery, native startup registration — picks the highest per half. The portable reference is 0; native-accelerated backends rank above it (koblas-openblas 100, koblas-cblas 90).
Functions
factor
Factorize the square a into something solvable, never null.
A singular matrix comes back as a factorization reporting singular, matching the dense contract where com.eignex.koblas.dense.LuDecomposition carries the flag. There is no nullable result to unwrap and no exception on the ordinary path: a simplex whose basis went singular wants to inspect and repair it, not catch something.
With equilibrate, rows are first scaled by a power of two so pivoting is better conditioned; the scaling is undone transparently in the solves and the determinant.
dropTolerance trades accuracy for sparsity: an entry the elimination produces whose magnitude is below this fraction of the matrix's largest is discarded rather than stored, so the factors stay sparser than the true L·U. The result is then an incomplete factorization — a solve against it is an approximation, not a solve — which is why it defaults to zero and has to be asked for. It is worth asking for when fill is the binding constraint and a residual around the tolerance is acceptable; on a diagonally dominant matrix 1e-9 has been measured cutting fill threefold at a residual of 1e-11 against 1e-15. A backend that cannot honor it must fall back rather than ignore it.