slope::ScreeningRule Class Reference

Base class for screening rules in SLOPE. More...

#include <screening.h>

Inheritance diagram for slope::ScreeningRule:

Public Member Functions
virtual	~ScreeningRule ()=default
	Virtual destructor.
virtual std::vector< int >	initialize (const std::vector< int > &full_set, int alpha_max_ind)=0
	Initialize the screening rule at the start of the path algorithm.
virtual std::vector< int >	screen (Eigen::VectorXd &gradient, const Eigen::ArrayXd &lambda_curr, const Eigen::ArrayXd &lambda_prev, const Eigen::VectorXd &beta, const std::vector< int > &full_set)=0
	Screen for the next path step.
virtual bool	checkKktViolations (Eigen::VectorXd &gradient, const Eigen::VectorXd &beta, const Eigen::ArrayXd &lambda_curr, std::vector< int > &working_set, const Eigen::MatrixXd &x, const Eigen::MatrixXd &residual, const Eigen::VectorXd &x_centers, const Eigen::VectorXd &x_scales, JitNormalization jit_normalization, const std::vector< int > &full_set)=0
	Check for KKT violations and update working set if necessary.
virtual bool	checkKktViolations (Eigen::VectorXd &gradient, const Eigen::VectorXd &beta, const Eigen::ArrayXd &lambda_curr, std::vector< int > &working_set, const Eigen::SparseMatrix< double > &x, const Eigen::MatrixXd &residual, const Eigen::VectorXd &x_centers, const Eigen::VectorXd &x_scales, JitNormalization jit_normalization, const std::vector< int > &full_set)=0
	Check for KKT violations with sparse matrix input.
virtual std::string	toString () const =0
	Get string representation of the screening rule.

Protected Attributes
std::vector< int >	strong_set
	Strong set of variables.

Detailed Description

Base class for screening rules in SLOPE.

This abstract class defines the interface for screening rules to accelerate optimization by reducing the problem size.

Definition at line 47 of file screening.h.

Member Function Documentation

checkKktViolations() [1/2]

virtual bool slope::ScreeningRule::checkKktViolations ( Eigen::VectorXd &  gradient, const Eigen::VectorXd &  beta, const Eigen::ArrayXd &  lambda_curr, std::vector< int > &  working_set, const Eigen::MatrixXd &  x, const Eigen::MatrixXd &  residual, const Eigen::VectorXd &  x_centers, const Eigen::VectorXd &  x_scales, JitNormalization  jit_normalization, const std::vector< int > &  full_set  )

pure virtual

Check for KKT violations and update working set if necessary.

Parameters

gradient	The gradient vector
beta	Current beta coefficients
lambda_curr	Current lambda values
working_set	Current working set (will be updated if violations found)
x	Design matrix
residual	Current residuals
x_centers	Centers for normalization
x_scales	Scales for normalization
jit_normalization	Whether to use JIT normalization
full_set	Full set of features

Returns

True if no violations found, false otherwise

Implemented in slope::NoScreening, and slope::StrongScreening.

checkKktViolations() [2/2]

virtual bool slope::ScreeningRule::checkKktViolations ( Eigen::VectorXd &  gradient, const Eigen::VectorXd &  beta, const Eigen::ArrayXd &  lambda_curr, std::vector< int > &  working_set, const Eigen::SparseMatrix< double > &  x, const Eigen::MatrixXd &  residual, const Eigen::VectorXd &  x_centers, const Eigen::VectorXd &  x_scales, JitNormalization  jit_normalization, const std::vector< int > &  full_set  )

pure virtual

Check for KKT violations with sparse matrix input.

Parameters

gradient	The gradient vector
beta	Current beta coefficients
lambda_curr	Current lambda values
working_set	Current working set (will be updated if violations found)
x	Design matrix (sparse format)
residual	Current residuals
x_centers	Centers for normalization
x_scales	Scales for normalization
jit_normalization	Whether to use JIT normalization
full_set	Full set of features

Returns

True if no violations found, false otherwise

Implemented in slope::NoScreening, and slope::StrongScreening.

initialize()

virtual std::vector< int > slope::ScreeningRule::initialize ( const std::vector< int > &  full_set, int  alpha_max_ind  )

pure virtual

Initialize the screening rule at the start of the path algorithm.

Parameters

full_set	The full set of feature indices
alpha_max_ind	The index of the feature with maximum absolute gradient

Returns

The initial working set

Implemented in slope::NoScreening, and slope::StrongScreening.

screen()

virtual std::vector< int > slope::ScreeningRule::screen ( Eigen::VectorXd &  gradient, const Eigen::ArrayXd &  lambda_curr, const Eigen::ArrayXd &  lambda_prev, const Eigen::VectorXd &  beta, const std::vector< int > &  full_set  )

pure virtual

Screen for the next path step.

Parameters

gradient	The gradient vector
lambda_curr	Current lambda values
lambda_prev	Previous lambda values
beta	Current beta coefficients
full_set	Full set of features

Returns

Working set for the current path step

Implemented in slope::NoScreening, and slope::StrongScreening.

toString()

virtual std::string slope::ScreeningRule::toString ( ) const

pure virtual

Get string representation of the screening rule.

Returns

Name of the screening rule

Implemented in slope::NoScreening, and slope::StrongScreening.

Member Data Documentation

strong_set

std::vector<int> slope::ScreeningRule::strong_set

protected

Strong set of variables.

Definition at line 142 of file screening.h.

---

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

• src/slope/screening.h