Reference documentation for deal.II version Git cb0bd54b52 2019-09-21 16:31:22 -0400
\(\newcommand{\dealcoloneq}{\mathrel{\vcenter{:}}=}\)
Public Member Functions | Private Attributes | List of all members
SparseMatrixIterators::Accessor< number, false >::Reference Class Reference

Public Member Functions

 Reference (const Accessor *accessor, const bool dummy)
 
 operator number () const
 
const Referenceoperator= (const number n) const
 
const Referenceoperator+= (const number n) const
 
const Referenceoperator-= (const number n) const
 
const Referenceoperator*= (const number n) const
 
const Referenceoperator/= (const number n) const
 

Private Attributes

const Accessoraccessor
 

Detailed Description

template<typename number>
class SparseMatrixIterators::Accessor< number, false >::Reference

Reference class. This is what the accessor class returns when you call the value() function. The reference acts just as if it were a reference to the actual value of a matrix entry, i.e. you can read and write it, you can add and multiply to it, etc, but since the matrix does not give away the address of this matrix entry, we have to go through functions to do all this.

The constructor takes a pointer to an accessor object that describes which element of the matrix it points to. This creates an ambiguity when one writes code like iterator->value()=0 (instead of iterator->value()=0.0), since the right hand side is an integer that can both be converted to a number (i.e., most commonly a double) or to another object of type Reference. The compiler then complains about not knowing which conversion to take.

For some reason, adding another overload operator=(int) doesn't seem to cure the problem. We avoid it, however, by adding a second, dummy argument to the Reference constructor, that is unused, but makes sure there is no second matching conversion sequence using a one-argument right hand side.

The testcase oliver_01 checks that this actually works as intended.

Definition at line 219 of file sparse_matrix.h.

Constructor & Destructor Documentation

◆ Reference()

template<typename number >
SparseMatrixIterators::Accessor< number, false >::Reference::Reference ( const Accessor< number, false > *  accessor,
const bool  dummy 
)

Constructor. For the second argument, see the general class documentation.

Member Function Documentation

◆ operator number()

template<typename number >
SparseMatrixIterators::Accessor< number, false >::Reference::operator number ( ) const

Conversion operator to the data type of the matrix.

◆ operator=()

template<typename number >
const Reference& SparseMatrixIterators::Accessor< number, false >::Reference::operator= ( const number  n) const

Set the element of the matrix we presently point to to n.

◆ operator+=()

template<typename number >
const Reference& SparseMatrixIterators::Accessor< number, false >::Reference::operator+= ( const number  n) const

Add n to the element of the matrix we presently point to.

◆ operator-=()

template<typename number >
const Reference& SparseMatrixIterators::Accessor< number, false >::Reference::operator-= ( const number  n) const

Subtract n from the element of the matrix we presently point to.

◆ operator*=()

template<typename number >
const Reference& SparseMatrixIterators::Accessor< number, false >::Reference::operator*= ( const number  n) const

Multiply the element of the matrix we presently point to by n.

◆ operator/=()

template<typename number >
const Reference& SparseMatrixIterators::Accessor< number, false >::Reference::operator/= ( const number  n) const

Divide the element of the matrix we presently point to by n.

Member Data Documentation

◆ accessor

template<typename number >
const Accessor* SparseMatrixIterators::Accessor< number, false >::Reference::accessor
private

Pointer to the accessor that denotes which element we presently point to.

Definition at line 268 of file sparse_matrix.h.


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