Dense vector with semi-sparse vector for updatesIn many algorithms vectors are updated in every iteration, by adding a multiple of another vector to it, i.e., given a vector x, a scalar $\alpha$ and another vector $\delta$ , the update to x constists of substituting it by $x \leftarrow x + \alpha\cdot\delta$ . More...

#include <updatevector.h>

Inheritance diagram for UpdateVector:

[legend]

Public Member Functions
Constructors / destructors
	UpdateVector (int p_dim, Real p_eps)
	default constructor.

	~UpdateVector ()

	UpdateVector (const UpdateVector &)
	copy constructor

UpdateVector &	operator= (const DVector &rhs)
	assignment from DVector

UpdateVector &	operator= (const Vector &rhs)
	assignment from Vector

UpdateVector &	operator= (const UpdateVector &rhs)
	assignment

Access
Real &	value ()
	update multiplicator $\alpha$ , writeable

Real	value () const
	update multiplicator $\alpha$

SSVector &	delta ()
	update vector $\delta$ , writeable

const SSVector &	delta () const
	update vector $\delta$

const IdxSet &	idx () const
	nonzero indices of update vector $\delta$

Modification
void	update ()
	Perform the update.

void	clear ()
	clear vector and update vector

void	clearUpdate ()
	clear $\delta$ , $\alpha$

void	reDim (int newdim)
	reset dimension

Consistency check
bool	isConsistent () const

Public Member Functions inherited from DVector
void	reDim (int newdim)
	resets DVector's dimension to `newdim`.

void	reSize (int newsize)
	resets DVector's memory size to `newsize`.

void	reSize (int newsize, int newdim)
	resets DVector's memory size to `newsize` and dimension to `newdim`.

int	memSize () const
	returns DVector's memory size.

bool	isConsistent () const
	consistency check.

	DVector (int dim=0)
	default constructor. `dim` is the initial dimension.

	DVector (const Vector &old)
	copy constructor.

	DVector (const Vector_exact &old)
	cast from exact vector.

	DVector (const DVector &old)
	copy constructor.

DVector &	operator= (const Vector &vec)
	assignment operator.

DVector &	operator= (const DVector &vec)
	assignment operator.

DVector &	operator= (const SVector &vec)
	assingment operator.

	~DVector ()
	destructor.

Public Member Functions inherited from Vector
	Vector (int p_dimen, Real *p_val)
	construction

Vector &	operator= (const Vector &vec)
	Assignment operator.

Vector &	operator= (const SVector &vec)
	Assignment operator.

Vector &	operator= (const SSVector &vec)
	Assignment operator.

Vector &	operator= (const Vector_exact &vec)
	Assignment operator.

Vector &	assign (const SVector &sv)
	Assign values of `sv`.

Vector &	assign (const SSVector &sv)
	Assign values of `sv`.

int	dim () const
	dimension of vector

Real &	operator[] (int n)
	return `n` 'th value by reference

Real	operator[] (int n) const
	return `n` 'th value

Vector &	operator+= (const Vector &vec)
	vector addition

Vector &	operator+= (const SVector &vec)
	vector addition

Vector &	operator+= (const SSVector &vec)
	vector addition

Vector &	operator-= (const Vector &vec)
	vector difference

Vector &	operator-= (const SVector &vec)
	vector difference

Vector &	operator-= (const SSVector &vec)
	vector difference

Vector &	operator*= (Real x)
	scaling

Real	operator* (const SSVector &v) const
	inner product.

Real	operator* (const SVector &v) const
	inner product.

Real	operator* (const Vector &v) const
	inner product.

Real	maxAbs () const
	absolute biggest element (infinity norm).

Real	minAbs () const
	absolute smallest element.

Real	length () const
	euclidian norm.

Real	length2 () const
	squared norm.

Vector &	multAdd (Real x, const SVector &vec)
	addition of scaled vector

Vector &	multAdd (Real x, const SSVector &svec)
	addition of scaled vector

Vector &	multAdd (Real x, const Vector &vec)
	addition of scaled vector

Real *	get_ptr ()
	Conversion to C-style pointer.

const Real *	get_const_ptr () const
	Conversion to C-style pointer.

bool	isConsistent () const
	consistency check.

void	clear ()
	set vector to 0.

Private Attributes
Data
Real	theval
	update multiplicator

SSVector	thedelta
	update vector

Additional Inherited Members
Protected Attributes inherited from Vector
int	dimen
	dimension of vector

Real *	val
	values of a vector

Arithmetic operations (friends)
I/O

Detailed Description

Dense vector with semi-sparse vector for updates

In many algorithms vectors are updated in every iteration, by adding a multiple of another vector to it, i.e., given a vector x, a scalar $\alpha$ and another vector $\delta$ , the update to x constists of substituting it by $x \leftarrow x + \alpha\cdot\delta$ .

While the update itself can easily be expressed with methods of the class Vector, it is often desirable to save the last update vector $\delta$ and value $\alpha$ . This is provided by class UpdateVector.

UpdateVectors are derived from DVector and provide additional methods for saving and setting the multiplicator $\alpha$ and the update vector $\delta$ . Further, it allows for efficient sparse updates, by providing an IdxSet idx() containing the nonzero indices of $\delta$ .

Definition at line 53 of file updatevector.h.