Structure

SparseNumericFactorOptions

A structure that contains options that affect the numerical stage of a sparse factorization.

Declaration

typedef struct {
    ...
} SparseNumericFactorOptions;

Overview

SparseNumericFactorOptions allows for the following types of scaling to be performed:

SparseScalingDefault

Default scaling—at present, SparseScalingEquilibriationInf if LDLᵀ, or no scaling if Cholesky.

SparseScalingUser

User scaling if scaling is nonnull, otherwise no scaling.

SparseScalingEquilibriationInf

Norm equilibration scaling using infinity norm.

Note that pivotTolerance is clamped to range [0,0.5].

Topics

Instance Properties

control

Flags controlling the computation.

SparseControl_t

Options that control the computation.

scalingMethod

Scaling method to use.

scaling

User-supplied array for scaling.

SparseScaling_t

Options that define which scaling algorithm to use.

pivotTolerance

Pivot tolerance used by threshold partial pivoting.

zeroTolerance

Zero tolerance used by some pivoting modes.

See Also

Supporting Types

SparseFactorization_t

A structure that defines the factorization type.

SparseSymbolicFactorOptions

A structure that contains options that affect the symbolic stage of a sparse factorization.

SparseOpaqueFactorization_Double

A structure representing the factorization of a matrix of double-precision, floating-point values.

SparseOpaqueFactorization_Float

A structure representing the factorization of a matrix of single-precision, floating-point values.