Class BackSolver<T extends MatrixMixin<T,?,U,?>,U extends VectorMixin<U,T,?,?>,V>

java.lang.Object

org.flag4j.linalg.solvers.exact.triangular.BackSolver<T,U,V>

Type Parameters:

T - Type of matrix to decompose.

U - Vector type equivalent of matrix.

V - Type of internal storage for the matrix and vector.

All Implemented Interfaces:

LinearMatrixSolver<T,U>, LinearSolver<T>

Direct Known Subclasses:

ComplexBackSolver, RealBackSolver

public abstract class BackSolver<T extends MatrixMixin<T,?,U,?>,U extends VectorMixin<U,T,?,?>,V>
extends Object
implements LinearMatrixSolver<T,U>

Base class for solvers which solve a linear system of equations $Ux=b$ or $UX=B$ where $U$ is an upper triangular matrix. This system is solved in an exact sense.

Field Summary

Fields

Modifier and Type	Field	Description
protected boolean	checkSingular	Flag indicating if an explicit check should be made that the matrix is singular (or near singular).
protected final boolean	enforceTriU	Flag indicating if determinant should be computed.
protected U	x	For storing vector results.
protected T	X	For storing matrix results.
protected V	xCol	For temporary storage of matrix columns to help improve cache performance.

Constructor Summary

Constructors

Constructor	Description
BackSolver(boolean enforceTriU)	Creates a solver for solving linear systems for upper triangular coefficient matrices.

Method Summary

Modifier and Type	Method	Description
protected void	checkParams(T coeff, Shape constantShape)	Ensures passed parameters are valid for the back solver.
protected void	checkSingular(double detAbs, int numRows, int numCols)	Checks if the coefficient matrix is singular based on the computed determinant.
protected BackSolver<T,U,V>	setCheckSingular(boolean checkSingular)	Sets a flag indicating if an explicit check should be made that the coefficient matrix is singular.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface org.flag4j.linalg.solvers.LinearMatrixSolver

solve, solve

Field Details

X

protected T extends MatrixMixin<T,?,U,?> X

For storing matrix results.

x

protected U extends VectorMixin<U,T,?,?> x

For storing vector results.

xCol

protected V xCol

For temporary storage of matrix columns to help improve cache performance.

enforceTriU

protected final boolean enforceTriU

Flag indicating if determinant should be computed.

checkSingular

protected boolean checkSingular

Flag indicating if an explicit check should be made that the matrix is singular (or near singular).

• If true, an explicit singularity check will be made.
• If false, no check will be made.

Constructor Details

BackSolver

public BackSolver(boolean enforceTriU)

Creates a solver for solving linear systems for upper triangular coefficient matrices.

Parameters:

enforceTriU - Flag indicating if an explicit check should be made that the coefficient matrix is upper triangular.

Method Details

setCheckSingular

protected BackSolver<T,U,V> setCheckSingular(boolean checkSingular)

Sets a flag indicating if an explicit check should be made that the coefficient matrix is singular.

Parameters:

checkSingular - Flag indicating if an explicit check should be made that the matrix is singular (or near singular).

• If true, an explicit singularity check will be made.
• If false, no check will be made.

Returns:

A reference to this back solver instance.

checkParams

protected void checkParams(T coeff,
 Shape constantShape)

Ensures passed parameters are valid for the back solver.

Parameters:

coeff - Coefficient matrix in the linear system.

constantRows - Shape of the constant vector or matrix.

Throws:

IllegalArgumentException - If coeff is not square, coeff.numRows()!=constantRows, or if enforceTriU is true and coeff is not upper triangular.

checkSingular

protected void checkSingular(double detAbs,
 int numRows,
 int numCols)

Checks if the coefficient matrix is singular based on the computed determinant.

Parameters:

detAbs - Absolute value of computed determinant.

numRows - Number of rows in the coefficient matrix.

numCols - Number of columns in the coefficient matrix.