com.eignex.koblas/sparse/host/umfpack

sparse.host.umfpack

Types

data class UmfpackConfig(val libraryPath: String? = null, val factorizeMin: Int? = null, val iterativeRefinementSteps: Int = 0, val pivotTolerance: Double = 0.1, val scaling: UmfpackScaling = UmfpackScaling.SUM)

Policy for one UMFPACK backend instance.

UmfpackFactorization

class UmfpackFactorization : F64SparseFactorization

UMFPACK's numeric factors behind koblas's F64SparseFactorization. Holds its matrix alive because umfpack_di_solve takes Ap, Ai and Ax alongside the factors.

class UmfpackFactorization : F64SparseFactorization

Holds matrix alive because umfpack_di_solve takes Ap, Ai and Ax alongside the factors, and releases the factors, which are UMFPACK's own allocation, through nativeCleaner once this object is unreachable.

UmfpackScaling

enum UmfpackScaling : Enum<UmfpackScaling>

The row-scaling strategy UMFPACK applies while building numeric factors.

UmfpackSparseLu

class UmfpackSparseLu(val config: UmfpackConfig = UmfpackConfig()) : F64SparseLuAdapter

The sparse half backed by SuiteSparse's UMFPACK.

class UmfpackSparseLu(val config: UmfpackConfig = UmfpackConfig()) : F64SparseLuAdapter

Sparse factorizations backed by a host UMFPACK.