MatBase< T, ROWS, COLS > Class Template Reference

Base class for fixed-size matrices. Provides shared element-wise arithmetic, scalar arithmetic, equality, transpose, mat*mat, and mat*vec. Storage is a flat row-major array mData[ROWS * COLS]. Derived classes supply constructors plus any dimension-specific extras (e.g. inverse() on Mat2). More...

#include <nanovdb/math/Math.h>

Public Types
using	ValueType = T

Public Member Functions
	MatBase () noexcept=default
	Default-construct (entries are left uninitialized for trivially-default-constructible T).
template<typename S>
__hostdev__ constexpr	MatBase (S *array) noexcept
	Read ROWS*COLS elements from array in row-major order, converting each element from S to T via static_cast.
__hostdev__ constexpr T *	operator[] (int row) noexcept
	2D row access. Returns a pointer to the start of row; the caller indexes into it with the column. Asserts 0 <= row < ROWS in debug builds; column index is the caller's responsibility (use 0 <= col < COLS).
__hostdev__ constexpr const T *	operator[] (int row) const noexcept
	Const overload of operator[](int); see non-const variant.
__hostdev__ constexpr T *	data () noexcept
	return a raw pointer to the underlying 1D storage (row-major)
__hostdev__ constexpr const T *	data () const noexcept
	return a const raw pointer to the underlying 1D storage (row-major)
template<typename Derived>
__hostdev__ constexpr Derived	plus (const Derived &rhs) const noexcept
	return *this + rhs as a Derived
template<typename Derived>
__hostdev__ constexpr Derived	minus (const Derived &rhs) const noexcept
	return *this - rhs as a Derived
template<typename Derived>
__hostdev__ constexpr Derived	negate () const noexcept
	return -(*this) as a Derived
template<typename Derived>
__hostdev__ constexpr Derived	scale (T s) const noexcept
	return s * (*this) as a Derived
__hostdev__ constexpr MatBase &	addAssign (const MatBase &rhs) noexcept
	Element-wise add rhs into *this and return *this.
__hostdev__ constexpr MatBase &	subAssign (const MatBase &rhs) noexcept
	Element-wise subtract rhs from *this and return *this.
__hostdev__ constexpr MatBase &	scaleAssign (T s) noexcept
	Multiply every element by scalar s in place; return *this. s is taken by value so m *= m[0][0] scales every element by the original value.
template<typename Derived>
__hostdev__ constexpr Derived	divideBy (T s) const noexcept
	return (*this) / s element-wise as a Derived. Floating-point T multiplies by the reciprocal 1/s; integer T divides each element, since 1/s would truncate to zero.
__hostdev__ constexpr MatBase &	divideAssignScalar (T s) noexcept
	Divide every element by scalar s in place; return *this. Uses the same floating-point reciprocal / integer per-element split as divideBy. s is taken by value so m /= m[0][0] is safe.
__hostdev__ constexpr bool	equals (const MatBase &rhs) const noexcept
	Return true iff every element compares equal to rhs.
template<typename Result>
__hostdev__ constexpr Result	transposeAs () const noexcept
	return the transpose of *this as the Result type.
template<typename Result, typename Rhs>
__hostdev__ constexpr Result	multiply (const Rhs &rhs) const noexcept
	return (*this) * rhs as the Result type.
template<typename VecResult, typename VecRhs>
__hostdev__ constexpr VecResult	multiplyVec (const VecRhs &v) const noexcept
	return (*this) * v as the VecResult type.

Static Public Member Functions
static __hostdev__ constexpr int	rows () noexcept
	Compile-time row count.
static __hostdev__ constexpr int	cols () noexcept
	Compile-time column count.
static __hostdev__ constexpr int	size () noexcept
	Compile-time element count, i.e. ROWS * COLS.

Protected Member Functions
template<typename... Args>
__hostdev__ constexpr	MatBase (Args... args) noexcept
	Variadic component-init ctor — used only by derived ctors (Mat2(T a, T b, T c, T d) : Base(a, b, c, d) {}). The member-init list directly initializes mData, avoiding the default-construct- then-assign that a body-assignment ctor would impose on a non-fundamental T.

Static Protected Member Functions
template<typename Result, typename F, int... I>
static __hostdev__ constexpr Result	generate (F f, std::integer_sequence< int, I... >) noexcept
	Return a Result whose row-major element idx is f(idx), built through its Count-element constructor. This is constexpr-valid in C++17 without first value-initializing a result and overwriting it.
template<typename Result, int Count = ROWS * COLS, typename F>
static __hostdev__ constexpr Result	generate (F f) noexcept

Protected Attributes
T	mData [ROWS *COLS]

Detailed Description

template<typename T, int ROWS, int COLS>
class nanovdb::math::MatBase< T, ROWS, COLS >

Base class for fixed-size matrices. Provides shared element-wise arithmetic, scalar arithmetic, equality, transpose, mat*mat, and mat*vec. Storage is a flat row-major array mData[ROWS * COLS]. Derived classes supply constructors plus any dimension-specific extras (e.g. inverse() on Mat2).

Member Typedef Documentation

ValueType

template<typename T, int ROWS, int COLS>

using ValueType = T

Constructor & Destructor Documentation

MatBase() [1/3]

template<typename T, int ROWS, int COLS>

MatBase ( )

defaultnoexcept

Default-construct (entries are left uninitialized for trivially-default-constructible T).

MatBase() [2/3]

template<typename T, int ROWS, int COLS>

template<typename S>

__hostdev__ constexpr MatBase ( S * array )

inlineconstexprnoexcept

Read ROWS*COLS elements from array in row-major order, converting each element from S to T via static_cast.

MatBase() [3/3]

template<typename T, int ROWS, int COLS>

template<typename... Args>

__hostdev__ constexpr MatBase ( Args... args )

inlineexplicitconstexprprotectednoexcept

Variadic component-init ctor — used only by derived ctors (Mat2(T a, T b, T c, T d) : Base(a, b, c, d) {}). The member-init list directly initializes mData, avoiding the default-construct- then-assign that a body-assignment ctor would impose on a non-fundamental T.

Member Function Documentation

addAssign()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr MatBase & addAssign ( const MatBase< T, ROWS, COLS > & rhs )

inlineconstexprnoexcept

Element-wise add rhs into *this and return *this.

cols()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr int cols ( )

inlinestaticnodiscardconstexprnoexcept

Compile-time column count.

data() [1/2]

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr const T * data ( ) const

inlineconstexprnoexcept

return a const raw pointer to the underlying 1D storage (row-major)

data() [2/2]

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr T * data ( )

inlineconstexprnoexcept

return a raw pointer to the underlying 1D storage (row-major)

divideAssignScalar()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr MatBase & divideAssignScalar ( T s )

inlineconstexprnoexcept

Divide every element by scalar s in place; return *this. Uses the same floating-point reciprocal / integer per-element split as divideBy. s is taken by value so m /= m[0][0] is safe.

divideBy()

template<typename T, int ROWS, int COLS>

template<typename Derived>

__hostdev__ constexpr Derived divideBy ( T s ) const

inlinenodiscardconstexprnoexcept

return (*this) / s element-wise as a Derived. Floating-point T multiplies by the reciprocal 1/s; integer T divides each element, since 1/s would truncate to zero.

equals()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr bool equals ( const MatBase< T, ROWS, COLS > & rhs ) const

inlinenodiscardconstexprnoexcept

Return true iff every element compares equal to rhs.

generate() [1/2]

template<typename T, int ROWS, int COLS>

template<typename Result, int Count = ROWS * COLS, typename F>

__hostdev__ constexpr Result generate ( F f )

inlinestaticconstexprprotectednoexcept

generate() [2/2]

template<typename T, int ROWS, int COLS>

template<typename Result, typename F, int... I>

__hostdev__ constexpr Result generate ( F f, std::integer_sequence< int, I... >  )

inlinestaticconstexprprotectednoexcept

Return a Result whose row-major element idx is f(idx), built through its Count-element constructor. This is constexpr-valid in C++17 without first value-initializing a result and overwriting it.

minus()

template<typename T, int ROWS, int COLS>

template<typename Derived>

__hostdev__ constexpr Derived minus ( const Derived & rhs ) const

inlinenodiscardconstexprnoexcept

return *this - rhs as a Derived

multiply()

template<typename T, int ROWS, int COLS>

template<typename Result, typename Rhs>

__hostdev__ constexpr Result multiply ( const Rhs & rhs ) const

inlinenodiscardconstexprnoexcept

return (*this) * rhs as the Result type.

Template Parameters

Result	a matrix type whose dimensions are (ROWS, Rhs::cols())
Rhs	a matrix type whose row count equals COLS

multiplyVec()

template<typename T, int ROWS, int COLS>

template<typename VecResult, typename VecRhs>

__hostdev__ constexpr VecResult multiplyVec ( const VecRhs & v ) const

inlinenodiscardconstexprnoexcept

return (*this) * v as the VecResult type.

Template Parameters

VecResult	a vector type whose SIZE equals ROWS
VecRhs	a vector type whose SIZE equals COLS

negate()

template<typename T, int ROWS, int COLS>

template<typename Derived>

__hostdev__ constexpr Derived negate ( ) const

inlinenodiscardconstexprnoexcept

return -(*this) as a Derived

operator[]() [1/2]

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr const T * operator[] ( int row ) const

inlineconstexprnoexcept

Const overload of operator[](int); see non-const variant.

operator[]() [2/2]

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr T * operator[] ( int row )

inlineconstexprnoexcept

2D row access. Returns a pointer to the start of row; the caller indexes into it with the column. Asserts 0 <= row < ROWS in debug builds; column index is the caller's responsibility (use 0 <= col < COLS).

plus()

template<typename T, int ROWS, int COLS>

template<typename Derived>

__hostdev__ constexpr Derived plus ( const Derived & rhs ) const

inlinenodiscardconstexprnoexcept

return *this + rhs as a Derived

rows()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr int rows ( )

inlinestaticnodiscardconstexprnoexcept

Compile-time row count.

scale()

template<typename T, int ROWS, int COLS>

template<typename Derived>

__hostdev__ constexpr Derived scale ( T s ) const

inlinenodiscardconstexprnoexcept

return s * (*this) as a Derived

scaleAssign()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr MatBase & scaleAssign ( T s )

inlineconstexprnoexcept

Multiply every element by scalar s in place; return *this. s is taken by value so m *= m[0][0] scales every element by the original value.

size()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr int size ( )

inlinestaticnodiscardconstexprnoexcept

Compile-time element count, i.e. ROWS * COLS.

subAssign()

template<typename T, int ROWS, int COLS>

__hostdev__ constexpr MatBase & subAssign ( const MatBase< T, ROWS, COLS > & rhs )

inlineconstexprnoexcept

Element-wise subtract rhs from *this and return *this.

transposeAs()

template<typename T, int ROWS, int COLS>

template<typename Result>

__hostdev__ constexpr Result transposeAs ( ) const

inlinenodiscardconstexprnoexcept

return the transpose of *this as the Result type.

Template Parameters

Result

a matrix type whose dimensions are (COLS, ROWS)

Member Data Documentation

mData

template<typename T, int ROWS, int COLS>

T mData[ROWS *COLS]

protected