NewtonConjugateGradientOptimizer.Hessian Property

Namespace: Numerics.NET.Optimization
Assembly: Numerics.NET (in Numerics.NET.dll) Version: 10.9.0

C# Copy

public Func<Vector<double>, LinearOperator<double>?, LinearOperator<double>>? Hessian { get; set; }

Visual Basic Copy

Public Property Hessian As Func(Of Vector(Of Double), LinearOperator(Of Double), LinearOperator(Of Double))
	Get
	Set

C++/CLI Copy

public:
property Func<Vector<double>^, LinearOperator<double>^, LinearOperator<double>^>^ Hessian {
	Func<Vector<double>^, LinearOperator<double>^, LinearOperator<double>^>^ get ();
	void set (Func<Vector<double>^, LinearOperator<double>^, LinearOperator<double>^>^ value);
}

F# Copy

member Hessian : Func<Vector<float>, LinearOperator<float>, LinearOperator<float>> with get, set

Property Value

Func<Vector<Double>, LinearOperator<Double>, LinearOperator<Double>>
A function receiving the current point and reusable operator, or null.

The first call of each solve receives null. Return a square symmetric operator with the original objective sign, including for maximization. Matrices provide both multiplication and explicit entries. A multiplication-only operator supports Newton-CG without storing the Hessian. Keep the operator valid until the next callback; copy input data retained by an operator. When reusing a matrix, overwrite entries that change, including entries becoming zero. IC/MIC require a sparse matrix. Stored sparse positions, including zeros, define a fixed pattern for the solve. New positions are rejected.