Template Numerical Library: TNL Namespace Reference

The main TNL namespace. More...

Namespaces
	Algorithms
	Namespace for fundamental TNL algorithms.

	Allocators
	Namespace for TNL allocators.

	Assert
	Internal namespace for helper classes used in the TNL_ASSERT_* macros.

	Communicators
	Namespace for TNL communicators.

	Containers
	Namespace for TNL containers.

	Devices
	Namespace for TNL execution models.

	Exceptions
	Namespace for TNL exceptions.

	Images
	Namespace for image processing.

	Matrices
	Namespace for matrix formats.

	Meshes
	Namespace for numerical meshes and related objects.

	Pointers
	Namespace for TNL pointers.

Classes
class	Atomic

class	Atomic< T, Devices::Cuda >

class	Atomic< T, Devices::Host >

struct	CacheSizes

class	File
	This class serves for binary IO. It allows to do IO even for data allocated on GPU together with on-the-fly data type conversion. More...

class	FileName
	Helper class for the construction of file names based on name, index and extension. More...

class	HasAddAssignmentOperator
	Type trait for checking if T has operator+= taking one argument of type T. More...

struct	HasConstexprGetSizeMethod
	Type trait for checking if T has a constexpr getSize method. More...

class	HasGetArrayDataMethod
	Type trait for checking if T has getArrayData method. More...

class	HasGetSizeMethod
	Type trait for checking if T has getSize method. More...

class	HasSetSizeMethod
	Type trait for checking if T has setSize method. More...

class	HasStaticGetSerializationType
	Type trait for checking if T has a static getSerializationType method. More...

class	HasSubscriptOperator
	Type trait for checking if T has operator[] taking one index argument. More...

struct	IsArrayType
	Type trait for checking if T is an array type, e.g. Containers::Array or Containers::Vector. More...

struct	IsStaticArrayType
	Type trait for checking if T is a static array type. More...

struct	IsVectorType
	Type trait for checking if T is a vector type, e.g. Containers::Vector or Containers::VectorView. More...

struct	IsViewType
	Type trait for checking if T is a view type. More...

class	Logger
	Creates calculations log in the form of a table. More...

class	Object
	Basic class for majority of TNL objects like matrices, meshes, grids, solvers, etc.. More...

class	String
	Class for managing strings. More...

class	SystemInfo

class	Timer
	Class for real time, CPU time and CPU cycles measuring. More...

Functions
template<class T , std::enable_if_t< ! std::is_unsigned< T >::value &&! std::is_class< T >::value, bool > = true>
__cuda_callable__ T	abs (const T &n)
	This function returns absolute value of given number n. More...

template<typename T >
__cuda_callable__ auto	acos (const T &value) -> decltype(std::acos(value))
	This function returns the arc cosine of the given value.

template<typename T >
__cuda_callable__ auto	acosh (const T &value) -> decltype(std::acosh(value))
	This function returns the inverse hyperbolic cosine of the given value.

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
Result	addAndReduce (Vector &lhs, const Containers::Expressions::BinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
Result	addAndReduce (Vector &lhs, const Containers::Expressions::DistributedBinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
Result	addAndReduce (Vector &lhs, const Containers::Expressions::DistributedUnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
__cuda_callable__ Result	addAndReduce (Vector &lhs, const Containers::Expressions::StaticBinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
__cuda_callable__ Result	addAndReduce (Vector &lhs, const Containers::Expressions::StaticUnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
Result	addAndReduce (Vector &lhs, const Containers::Expressions::UnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
Result	addAndReduceAbs (Vector &lhs, const Containers::Expressions::BinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
Result	addAndReduceAbs (Vector &lhs, const Containers::Expressions::DistributedBinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
Result	addAndReduceAbs (Vector &lhs, const Containers::Expressions::DistributedUnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
__cuda_callable__ Result	addAndReduceAbs (Vector &lhs, const Containers::Expressions::StaticBinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
__cuda_callable__ Result	addAndReduceAbs (Vector &lhs, const Containers::Expressions::StaticUnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
Result	addAndReduceAbs (Vector &lhs, const Containers::Expressions::UnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type>
__cuda_callable__ ResultType	argAbsMax (const T1 &a, const T2 &b)

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type>
__cuda_callable__ ResultType	argAbsMin (const T1 &a, const T2 &b)

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type>
__cuda_callable__ ResultType	argMax (const T1 &a, const T2 &b)

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type>
__cuda_callable__ ResultType	argMin (const T1 &a, const T2 &b)

template<typename T >
__cuda_callable__ auto	asin (const T &value) -> decltype(std::asin(value))
	This function returns the arc sine of the given value.

template<typename T >
__cuda_callable__ auto	asinh (const T &value) -> decltype(std::asinh(value))
	This function returns the inverse hyperbolic sine of the given value.

template<typename T >
__cuda_callable__ auto	atan (const T &value) -> decltype(std::atan(value))
	This function returns the arc tangent of the given value.

template<typename T >
__cuda_callable__ auto	atanh (const T &value) -> decltype(std::atanh(value))
	This function returns the inverse hyperbolic tangent of the given value.

template<typename T >
__cuda_callable__ auto	cbrt (const T &value) -> decltype(std::cbrt(value))
	This function returns cubic root of the given value.

template<typename T >
__cuda_callable__ auto	ceil (const T &value) -> decltype(std::ceil(value))
	This function returns the smallest integer value not less than the given value.

template<>
String	convertToString (const bool &b)
	Specialization of function convertToString for boolean. More...

template<typename T >
String	convertToString (const T &value)
	Converts value of type T to a String. More...

template<typename T >
__cuda_callable__ auto	cos (const T &value) -> decltype(std::cos(value))
	This function returns cosine of the given value.

template<typename T >
__cuda_callable__ auto	cosh (const T &value) -> decltype(std::cosh(value))
	This function returns the hyperbolic cosine of the given value.

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
Result	evaluateAndReduce (Vector &lhs, const Containers::Expressions::BinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
Result	evaluateAndReduce (Vector &lhs, const Containers::Expressions::DistributedBinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
Result	evaluateAndReduce (Vector &lhs, const Containers::Expressions::DistributedUnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename T2 , typename Operation , typename Reduction , typename Result >
__cuda_callable__ Result	evaluateAndReduce (Vector &lhs, const Containers::Expressions::StaticBinaryExpressionTemplate< T1, T2, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
__cuda_callable__ Result	evaluateAndReduce (Vector &lhs, const Containers::Expressions::StaticUnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename Vector , typename T1 , typename Operation , typename Reduction , typename Result >
Result	evaluateAndReduce (Vector &lhs, const Containers::Expressions::UnaryExpressionTemplate< T1, Operation > &expression, const Reduction &reduction, const Result &zero)

template<typename T >
__cuda_callable__ auto	exp (const T &value) -> decltype(std::exp(value))
	This function returns the base-e exponential of the given value.

bool	fileExists (const String &fileName)
	Returns true if the file exists and false otherwise. More...

template<typename T >
__cuda_callable__ auto	floor (const T &value) -> decltype(std::floor(value))
	This function returns largest integer value not greater than the given value.

String	getFileExtension (const String fileName)
	Returns extension of given file name, i.e. part after the last dot. More...

String	getObjectType (const String &fileName)
	Does the same as getObjectType but with a fileName parameter instead of file. More...

String	getObjectType (File &file)
	Extracts object type from a binary file. More...

template<typename T , std::enable_if_t< ! HasStaticGetSerializationType< T >::value, bool > = true>
String	getSerializationType ()
	Returns a string identifying a type for the purpose of serialization. More...

String	getSerializationType (Algorithms::Segments::ElementsOrganization Organization)

template<typename T >
String	getType ()
	Returns a human-readable string representation of given type. More...

template<typename T >
String	getType (T &&obj)
	Returns a human-readable string representation of given object's type. More...

__cuda_callable__ bool	isPow2 (int x)
	This function checks if x is an integral power of two. More...

__cuda_callable__ bool	isPow2 (long int x)
	This function checks if x is an integral power of two. More...

template<typename Real >
__cuda_callable__ bool	isSmall (const Real &v, const Real &tolerance=1.0e-5)
	This function tests whether the given real number is small. More...

template<typename T >
__cuda_callable__ auto	log (const T &value) -> decltype(std::log(value))
	This function returns the natural logarithm of the given value.

template<typename T >
__cuda_callable__ auto	log10 (const T &value) -> decltype(std::log10(value))
	This function returns the common logarithm of the given value.

template<typename T >
__cuda_callable__ auto	log2 (const T &value) -> decltype(std::log2(value))
	This function returns the binary logarithm of the given value.

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type, std::enable_if_t< ! HasSubscriptOperator< T1 >::value &&! HasSubscriptOperator< T2 >::value, bool > = true>
__cuda_callable__ ResultType	max (const T1 &a, const T2 &b)
	This function returns maximum of two numbers. More...

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type, std::enable_if_t< ! HasSubscriptOperator< T1 >::value &&! HasSubscriptOperator< T2 >::value, bool > = true>
__cuda_callable__ ResultType	min (const T1 &a, const T2 &b)
	This function returns minimum of two numbers. More...

void	mpiReceive (String &str, int source, int tag=0, MPI_Comm mpi_comm=MPI_COMM_WORLD)
	Receives a string from the target MPI process. More...

void	mpiSend (const String &str, int target, int tag=0, MPI_Comm mpi_comm=MPI_COMM_WORLD)
	Sends the string to the target MPI process. More...

String	operator+ (char string1, const String &string2)
	Returns concatenation of string1 and string2.

String	operator+ (const char *string1, const String &string2)
	Returns concatenation of string1 and string2.

String	operator+ (const std::string &string1, const String &string2)
	Returns concatenation of string1 and string2.

File &	operator<< (File &file, const std::string &str)
	Serialization of strings.

File &	operator>> (File &file, std::string &str)
	Deserialization of strings.

std::vector< String >	parseObjectType (const String &objectType)
	Parses the object type. More...

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type, std::enable_if_t< ! std::is_class< T1 >::value &&! std::is_class< T2 >::value, bool > = true>
__cuda_callable__ ResultType	pow (const T1 &base, const T2 &exp)
	This function returns the result of base to the power of exp.

String	removeFileNameExtension (String fileName)
	Cuts off the file extension. More...

__cuda_callable__ int	roundToMultiple (int number, int multiple)
	This function rounds up number to the nearest multiple of number multiple. More...

__cuda_callable__ int	roundUpDivision (const int num, const int div)
	This function divides num by div and rounds up the result. More...

void	saveObjectType (File &file, const String &type)
	Saves object type into a binary file. More...

template<class T , std::enable_if_t< ! HasSubscriptOperator< T >::value, bool > = true>
__cuda_callable__ T	sign (const T &a)
	This function represents the signum function. More...

template<typename T >
__cuda_callable__ auto	sin (const T &value) -> decltype(std::sin(value))
	This function returns sine of the given value.

template<typename T >
__cuda_callable__ auto	sinh (const T &value) -> decltype(std::sinh(value))
	This function returns the hyperbolic sine of the given value.

template<typename T >
__cuda_callable__ auto	sqrt (const T &value) -> decltype(std::sqrt(value))
	This function returns square root of the given value.

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type>
__cuda_callable__ ResultType	sum (const T1 &a, const T2 &b)

template<typename Type >
__cuda_callable__ void	swap (Type &a, Type &b)
	This function swaps values of two parameters. More...

template<typename T >
__cuda_callable__ auto	tan (const T &value) -> decltype(std::tan(value))
	This function returns tangent of the given value.

template<typename T >
__cuda_callable__ auto	tanh (const T &value) -> decltype(std::tanh(value))
	This function returns the hyperbolic tangent of the given value.

Detailed Description

The main TNL namespace.

Function Documentation

◆ abs()

template<class T , std::enable_if_t< ! std::is_unsigned< T >::value &&! std::is_class< T >::value, bool > = true>

__cuda_callable__ T TNL::abs ( const T & n )

inline

This function returns absolute value of given number n.

This function returns the absolute value of given unsigned number n, i.e. n.

◆ convertToString() [1/2]

template<>

String TNL::convertToString ( const bool & b )

inline

Specialization of function convertToString for boolean.

The boolean type is converted to 'true' or 'false'.

◆ convertToString() [2/2]

template<typename T >

String TNL::convertToString ( const T & value )

Converts value of type T to a String.

Template Parameters

T	can be any type fir which operator << is defined.

◆ fileExists()

bool TNL::fileExists ( const String & fileName )

inline

Returns true if the file exists and false otherwise.

Finds out if the file fileName exists.

Parameters

fileName Name of the file to check.

Returns: returns true if the file exists and false othervise

◆ getFileExtension()

String TNL::getFileExtension ( const String fileName )

inline

Returns extension of given file name, i.e. part after the last dot.

Parameters

fileName Input file name.

Returns: Extension of the given file name.

◆ getObjectType() [1/2]

String TNL::getObjectType ( const String & fileName )

inline

Does the same as getObjectType but with a fileName parameter instead of file.

Throws Exceptions::FileDeserializationError if the object type cannot be detected.

Parameters

fileName name of a file where the object is stored

Returns: string with the object type

◆ getObjectType() [2/2]

String TNL::getObjectType ( File & file )

inline

Extracts object type from a binary file.

Throws Exceptions::FileDeserializationError if the object type cannot be detected.

Parameters

file	is file where the object is stored

Returns: string with the object type

◆ getSerializationType()

template<typename T , std::enable_if_t< ! HasStaticGetSerializationType< T >::value, bool > = true>

String TNL::getSerializationType ( )

Returns a string identifying a type for the purpose of serialization.

Specialization of getSerializationType for types which provide a static getSerializationType method to override the default behaviour.

By default, this function returns the same string as getType. However, if a user-defined class has a static getSerializationType method, it is called instead. This is useful for overriding the default typeid name, which may be necessary e.g. for class templates which should have the same serialization type for multiple devices.

◆ getType() [1/2]

template<typename T >

String TNL::getType ( )

Returns a human-readable string representation of given type.

Note that since we use the typeid operator internally, the top-level cv-qualifiers are always ignored. See https://stackoverflow.com/a/8889143 for details.

◆ getType() [2/2]

template<typename T >

String TNL::getType ( T && obj )

Returns a human-readable string representation of given object's type.

Note that since we use the typeid operator internally, the top-level cv-qualifiers are always ignored. See https://stackoverflow.com/a/8889143 for details.

◆ isPow2() [1/2]

__cuda_callable__ bool TNL::isPow2 ( int x )

inline

This function checks if x is an integral power of two.

Returns true if x is a power of two. Otherwise returns false.

Parameters

x Integer.

◆ isPow2() [2/2]

__cuda_callable__ bool TNL::isPow2 ( long int x )

inline

This function checks if x is an integral power of two.

Returns true if x is a power of two. Otherwise returns false.

Parameters

x	Long integer.

◆ isSmall()

template<typename Real >

__cuda_callable__ bool TNL::isSmall	(	const Real &	v,
		const Real &	tolerance = `1.0e-5`
	)

This function tests whether the given real number is small.

It tests whether the number v is in tolerance, in other words, whether v in absolute value is less then or equal to tolerance.

Parameters

v	Real number.
tolerance	Critical value which is set to 0.00001 by defalt.

◆ max()

template<typename T1 , typename T2 , typename ResultType = typename std::common_type< T1, T2 >::type, std::enable_if_t< ! HasSubscriptOperator< T1 >::value &&! HasSubscriptOperator< T2 >::value, bool > = true>

__cuda_callable__ ResultType TNL::max	(	const T1 &	a,
		const T2 &	b
	)

This function returns maximum of two numbers.

GPU device code uses the functions defined in the CUDA's math_functions.h, host uses the STL functions.

◆ min()

__cuda_callable__ ResultType TNL::min	(	const T1 &	a,
		const T2 &	b
	)

inline

This function returns minimum of two numbers.

GPU device code uses the functions defined in the CUDA's math_functions.h, host uses the STL functions.

◆ mpiReceive()

void TNL::mpiReceive	(	String &	str,
		int	source,
		int	tag = `0`,
		MPI_Comm	mpi_comm = `MPI_COMM_WORLD`
	)

inline

Receives a string from the target MPI process.

Parameters

str	says where the received string is to be saved to
source	source MPI process ID
tag	MPI tag
mpi_comm	MPI communication group

◆ mpiSend()

void TNL::mpiSend	(	const String &	str,
		int	target,
		int	tag = `0`,
		MPI_Comm	mpi_comm = `MPI_COMM_WORLD`
	)

inline

Sends the string to the target MPI process.

Parameters

str	string to be sent
target	target MPI process ID
tag	MPI tag
mpi_comm	MPI communication group

◆ parseObjectType()

std::vector< String > TNL::parseObjectType ( const String & objectType )

inline

Parses the object type.

Parameters

objectType is a string with the object type to be parsed.

Returns: vector of strings where the first one is the object type and the next strings are the template parameters.

Example: #include <iostream>

#include <TNL/Object.h>

using namespace TNL;

int main()

{

auto parsedObjectType = parseObjectType( String( "MyObject< Value, Device, Index >" ) );

for( auto &token : parsedObjectType )

std::cout << token << std::endl;

}

Output: MyObject

Value

Device

Index

◆ removeFileNameExtension()

String TNL::removeFileNameExtension ( String fileName )

inline

Cuts off the file extension.

Parameters

file_name Input file name.

Returns: String with the file name without extension.

◆ roundToMultiple()

__cuda_callable__ int TNL::roundToMultiple	(	int	number,
		int	multiple
	)

inline

This function rounds up number to the nearest multiple of number multiple.

Parameters

number	Integer we want to round.
multiple	Integer.

◆ roundUpDivision()

__cuda_callable__ int TNL::roundUpDivision	(	const int	num,
		const int	div
	)

inline

This function divides num by div and rounds up the result.

Parameters

num	An integer considered as dividend.
div	An integer considered as divisor.

◆ saveObjectType()

void TNL::saveObjectType	(	File &	file,
		const String &	type
	)

inline

Saves object type into a binary file.

Throws Exceptions::FileDeserializationError if the object type cannot be detected.

Parameters

file	is the file where the object will be saved
type	is the object type to be saved

◆ sign()

template<class T , std::enable_if_t< ! HasSubscriptOperator< T >::value, bool > = true>

__cuda_callable__ T TNL::sign ( const T & a )

This function represents the signum function.

It extracts the sign of the number a. In other words, the signum function projects negative numbers to value -1, positive numbers to value 1 and zero to value 0.

◆ swap()

template<typename Type >

__cuda_callable__ void TNL::swap	(	Type &	a,
		Type &	b
	)

This function swaps values of two parameters.

It assigns the value of a to the parameter b and vice versa.

Namespaces

Classes

Functions

Detailed Description

Function Documentation

◆ abs()

◆ convertToString() [1/2]

◆ convertToString() [2/2]

◆ fileExists()

◆ getFileExtension()

◆ getObjectType() [1/2]

◆ getObjectType() [2/2]

◆ getSerializationType()

◆ getType() [1/2]

◆ getType() [2/2]

◆ isPow2() [1/2]

◆ isPow2() [2/2]

◆ isSmall()

◆ max()

◆ min()

◆ mpiReceive()

◆ mpiSend()

◆ parseObjectType()

◆ removeFileNameExtension()

◆ roundToMultiple()

◆ roundUpDivision()

◆ saveObjectType()

◆ sign()

◆ swap()