ThresholdIncompleteLUPreconditioner<T> Class

public sealed class ThresholdIncompleteLUPreconditioner<T> : Preconditioner<T>

Public NotInheritable Class ThresholdIncompleteLUPreconditioner(Of T)
	Inherits Preconditioner(Of T)

generic<typename T>
public ref class ThresholdIncompleteLUPreconditioner sealed : public Preconditioner<T>

[<SealedAttribute>]
type ThresholdIncompleteLUPreconditioner<'T> = 
    class
        inherit Preconditioner<'T>
    end

Inheritance

Object  →  LinearOperator<T>  →  Preconditioner<T>  →  ThresholdIncompleteLUPreconditioner<T>

Type Parameters

T

The real or complex floating-point scalar type.

Lower multipliers of magnitude at most DropTolerance are dropped during elimination. Strict-upper coefficients of magnitude at most that tolerance times the mean magnitude of nonzero entries in the original shifted/scaled row are dropped afterward. Each strict triangular row part retains at most MaximumFillPerRow largest entries; ties prefer the smaller column. The diagonal is always retained. Fill-cap dropping of lower multipliers occurs after their row updates.

No numerical pivoting or automatic ordering is performed. This preconditioner is intended for nonsymmetric solvers such as GeneralizedMinimalResidualSolver<T> and does not ensure the positive definiteness required by CG. Call Refresh() after changing coefficients. Distinct-destination applications may run concurrently; refresh requires exclusive access.

Reference