turi::optimization::non_negative_orthant Class Reference

#include <ml/optimization/constraints-inl.hpp>

Public Member Functions
	non_negative_orthant (const size_t &_variables)
	~non_negative_orthant ()
void	project (DenseVector &point) const
void	project_block (DenseVector &point, const size_t block_start, const size_t block_size) const
bool	is_satisfied (const DenseVector &point) const
double	first_order_optimality_conditions (const DenseVector &point, const DenseVector &gradient) const
virtual void	project (DenseVector &point) const =0
virtual void	project_block (DenseVector &point, const size_t block_start, const size_t block_size) const =0
virtual bool	is_satisfied (const DenseVector &point) const =0
virtual double	first_order_optimality_conditions (const DenseVector &point, const DenseVector &gradient) const =0

Protected Attributes
size_t	variables

Detailed Description

Interface for non-negative constriants. x >= 0

Definition at line 40 of file constraints-inl.hpp.

Member Function Documentation

first_order_optimality_conditions()

virtual double turi::optimization::constraint_interface::first_order_optimality_conditions ( const DenseVector &  point, const DenseVector &  gradient  ) const

pure virtualinherited

A measure of the first order optimality conditions.

Parameters

[in]	point	Point which we are querying.
[in]	gradient	Gradient at that point for a given function

If you don't know what to do, then use ||P_C(x - grad) - x|| where x is the point, grad is the gradient and P_C is the projection to the set in consideration.

Implemented in turi::optimization::box_constraints.

is_satisfied()

virtual bool turi::optimization::constraint_interface::is_satisfied ( const DenseVector &  point ) const

pure virtualinherited

Boolean function to determine if a dense point is present in a constraint space.

Parameters

[in]

point

Point which we are querying.

Implemented in turi::optimization::box_constraints.

project()

virtual void turi::optimization::constraint_interface::project ( DenseVector &  point ) const

pure virtualinherited

Project a dense point into the constraint space.

Parameters

[in,out]

point

Point which we are using to project the point.

Given a convex set X, the projection operator is given by

P(y) = \min_{x \in X} || x - y ||^2

Implemented in turi::optimization::box_constraints.

project_block()

virtual void turi::optimization::constraint_interface::project_block ( DenseVector &  point, const size_t  block_start, const size_t  block_size  ) const

pure virtualinherited

Project a dense a block of co-ordinates point into the constraint space.

Parameters

[in,out]	point	A block project the point.
[in]	block_start	Start index of the block
[in]	block_size	Size of the block

Given a convex set X, the projection operator is given by

P(y) = \min_{x \in X} || x - y ||^2

Implemented in turi::optimization::box_constraints.

---

The documentation for this class was generated from the following file:

• ml/optimization/constraints-inl.hpp