Computes $Ma(x)b(x)\cdots\int\vec{s}(x)\cdot\nabla\phi(x)\,dx$ . More...

#include <Panzer_Integrator_GradBasisDotTensorTimesVector_decl.hpp>

Inheritance diagram for panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >:

Classes
struct	FieldMultTag
	This empty struct allows us to optimize `operator()()` depending on the number of field multipliers. This is the version that does not use shared memory. More...

struct	SharedFieldMultTag
	This empty struct allows us to optimize `operator()()` depending on the number of field multipliers. This is the shared memory version. More...

Public Member Functions
	Integrator_GradBasisDotTensorTimesVector (const panzer::EvaluatorStyle &evalStyle, const std::string &resName, const std::string &fluxName, const panzer::BasisIRLayout &basis, const panzer::IntegrationRule &ir, const std::string &tensorName, const Teuchos::RCP< PHX::DataLayout > &vecDL=Teuchos::null)
	Main Constructor. More...

	Integrator_GradBasisDotTensorTimesVector (const Teuchos::ParameterList &p)
	`ParameterList` Constructor. More...

void	postRegistrationSetup (typename Traits::SetupData d, PHX::FieldManager< Traits > &fm)
	Post-Registration Setup. More...

void	evaluateFields (typename Traits::EvalData d)
	Evaluate Fields. More...

KOKKOS_INLINE_FUNCTION void	operator() (const FieldMultTag &tag, const Kokkos::TeamPolicy< PHX::exec_space >::member_type &team) const
	Perform the integration. More...

KOKKOS_INLINE_FUNCTION void	operator() (const SharedFieldMultTag &tag, const Kokkos::TeamPolicy< PHX::exec_space >::member_type &team) const
	Perform the integration. More...

Public Member Functions inherited from panzer::EvaluatorWithBaseImpl< Traits >
void	setDetailsIndex (const int di)
	An evaluator builder sets the details index. More...

Public Member Functions inherited from PHX::EvaluatorWithBaseImpl< Traits >
virtual void	evaluateFields (typename Traits::EvalData d) override=0

Public Member Functions inherited from PHX::Evaluator< Traits >
	Evaluator ()

virtual	~Evaluator ()

virtual void	postRegistrationSetup (typename Traits::SetupData d, PHX::FieldManager< Traits > &vm)=0

virtual const std::vector < Teuchos::RCP< FieldTag > > &	evaluatedFields () const =0

virtual const std::vector < Teuchos::RCP< FieldTag > > &	contributedFields () const =0

virtual const std::vector < Teuchos::RCP< FieldTag > > &	dependentFields () const =0

virtual const std::vector < Teuchos::RCP< FieldTag > > &	unsharedFields () const =0

virtual void	preEvaluate (typename Traits::PreEvalData d)=0

virtual void	postEvaluate (typename Traits::PostEvalData d)=0

virtual const std::string &	getName () const =0

virtual void	bindField (const PHX::FieldTag &ft, const PHX::any &f)=0

virtual PHX::DeviceEvaluator < Traits > *	createDeviceEvaluator () const =0

virtual void	rebuildDeviceEvaluator (PHX::DeviceEvaluator< Traits > *e) const =0

virtual void	deleteDeviceEvaluator (PHX::DeviceEvaluator< Traits > *e) const =0

virtual void	printFieldValues (std::ostream &os) const =0

Public Member Functions inherited from panzer::DomainEvaluator
	DomainEvaluator (DomainType domain=ALL)
	Constructor. More...

virtual	~DomainEvaluator ()=default
	Default destructor. More...

void	setDomain (const DomainType domain)
	Set the domain for the evaluator. More...

DomainType	getDomain ()
	Get the domain for the evaluator. More...

virtual int	cellStartIndex (const panzer::Workset &workset) const
	Returns the starting cell for the specified domain for a given workset. More...

virtual int	cellEndIndex (const panzer::Workset &workset) const
	Returns the non-inclusive end cell for the specified domain for a given workset. More...

Private Types
using	ScalarT = typename EvalT::ScalarT
	The scalar type. More...

using	scratch_view = Kokkos::View< ScalarT *,typename PHX::DevLayout< ScalarT >::type, typename PHX::exec_space::scratch_memory_space, Kokkos::MemoryUnmanaged >
	Type for shared memory. More...

Private Member Functions
Teuchos::RCP < Teuchos::ParameterList >	getValidParameters () const
	Get Valid Parameters. More...

Private Attributes
const panzer::EvaluatorStyle	evalStyle_
	An `enum` determining the behavior of this `Evaluator`. More...

PHX::MDField< ScalarT, panzer::Cell, panzer::BASIS >	field_
	A field to which we'll contribute, or in which we'll store, the result of computing this integral. More...

PHX::MDField< const ScalarT, panzer::Cell, panzer::IP, panzer::Dim >	vector_
	A field representing the vector-valued function we're integrating ( $\vec{s}$ ). More...

PHX::MDField< const ScalarT, panzer::Cell, panzer::IP, panzer::Dim, panzer::Dim >	tensor_
	The tensor field( ). More...

PHX::View< const ScalarT **** >	kokkosTensor_
	The `PHX::View` representation of the tensor fields that are multiplied out in front of the integral ( ). More...

std::string	basisName_
	The name of the basis we're using. More...

std::size_t	basisIndex_
	The index in the `Workset` bases for our particular `BasisIRLayout` name. More...

PHX::MDField< double, panzer::Cell, panzer::BASIS, panzer::IP, panzer::Dim >	basis_
	The gradient vector basis information necessary for integration. More...

PHX::View< ScalarT * >	tmp_
	Temporary used when shared memory is disabled. More...

Additional Inherited Members
Public Types inherited from panzer::DomainEvaluator
enum	DomainType : int { OWNED =0, GHOST =1, REAL =2, VIRTUAL =3, EXTERNAL =4, ALL =5 }
	Domain types supported by worksets. More...

Protected Attributes inherited from panzer::EvaluatorWithBaseImpl< Traits >
WorksetDetailsAccessor	wda

Detailed Description

template<typename EvalT, typename Traits>
class panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >

Computes $Ma(x)b(x)\cdots\int\vec{s}(x)\cdot\nabla\phi(x)\,dx$ .

Evaluates the integral

$\int (a(x)\cdots\vec{s}(x)) \cdot \nabla\phi(x)\,dx,$

where $ a(x) $ is a tensor valued field that depends on position, $\vec{s}$ is some vector- valued function, and $\phi$ is some scalar basis.

Definition at line 78 of file Panzer_Integrator_GradBasisDotTensorTimesVector_decl.hpp.

Member Typedef Documentation

template<typename EvalT , typename Traits >

using panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::ScalarT = typename EvalT::ScalarT

private

The scalar type.

Definition at line 278 of file Panzer_Integrator_GradBasisDotTensorTimesVector_decl.hpp.

template<typename EvalT , typename Traits >

using panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::scratch_view = Kokkos::View<ScalarT* ,typename PHX::DevLayout<ScalarT>::type,typename PHX::exec_space::scratch_memory_space,Kokkos::MemoryUnmanaged>

private

Type for shared memory.

Definition at line 281 of file Panzer_Integrator_GradBasisDotTensorTimesVector_decl.hpp.

Constructor & Destructor Documentation

template<typename EvalT , typename Traits >

panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::Integrator_GradBasisDotTensorTimesVector	(	const panzer::EvaluatorStyle &	evalStyle,
		const std::string &	resName,
		const std::string &	fluxName,
		const panzer::BasisIRLayout &	basis,
		const panzer::IntegrationRule &	ir,
		const std::string &	tensorName,
		const Teuchos::RCP< PHX::DataLayout > &	vecDL = `Teuchos::null`
	)

Main Constructor.

Creates an Evaluator to evaluate the integral

$\int (a(x)\cdots\vec{s}(x)) \cdot \nabla\phi(x)\,dx,$

where $ a(x) $ is a tensor valued field that depends on position, $\vec{s}$ is some vector- valued function, and $\phi$ is some scalar basis.

Parameters

[in]	evalStyle	An `enum` declaring the behavior of this `Evaluator`, which is to either: compute and contribute (`CONTRIBUTES`), or compute and store (`EVALUATES`).
[in]	resName	The name of either the contributed or evaluated field, depending on `evalStyle`.
[in]	fluxName	The name of the vector-valued function being integrated ( $\vec{s}$ ).
[in]	basis	The basis that you'd like to use ( $\phi$ ).
[in]	ir	The integration rule that you'd like to use.
[in]	tensorName	The tensors name ( ).
[in]	vecDL	The vector data layout that you'd like to use. If not specified, this defaults to `Teuchos::null` and the vector data layout from the given `ir` is used.

Exceptions

std::invalid_argument	If any of the inputs are invalid.
std::logic_error	If the `basis` supplied does not support the gradient operator, or if the dimension of the space exceeds the dimension of the vector data layout.

Definition at line 67 of file Panzer_Integrator_GradBasisDotTensorTimesVector_impl.hpp.

template<typename EvalT , typename Traits >

panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::Integrator_GradBasisDotTensorTimesVector ( const Teuchos::ParameterList & p )

ParameterList Constructor.

Creates an Evaluator to evaluate the integral

$Ma(x)b(x)\cdots\int\vec{s}(x)\cdot\nabla\phi(x)\,dx,$

where $ M $ is some constant, $ a(x) $ , $ b(x) $ , etc., are some fields that depend on position, $\vec{s}$ is some vector-valued function, and $\phi$ is some basis.

Note: This constructor exists to preserve the older way of creating an Evaluator with a ParameterList; however, it is strongly advised that you not use this ParameterList Constructor, but rather that you favor the main constructor with its compile-time argument checking instead.

Parameters

[in]

p

A ParameterList of the form

<ParameterList>
  <Parameter name = "Residual Name"      type = "std::string"                         value = (required)    />
  <Parameter name = "Flux Name"          type = "std::string"                         value = (required)    />
  <Parameter name = "Basis"              type = "RCP<panzer::BasisIRLayout>"          value = (required)    />
  <Parameter name = "IR"                 type = "RCP<panzer::IntegrationRule>"        value = (required)    />
  <Parameter name = "Tensor Name"       type = "std::string"                         value = (required)    />
  <Parameter name = "Vector Data Layout" type = "RCP<PHX::DataLayout>"                value = null (default)/>
</ParameterList>

where

"Residual Name" is the name for the term this Evaluator is evaluating,
"Flux Name" is the name of the vector-valued function being integrated ( $\vec{s}$ ),
"Basis" is the basis that you'd like to use ( $\phi$ ),
"IR" is the integration rule that you'd like to use,
"Tensor Name" is the name of the tensor field ( ).
"Vector Data Layout" is the vector data layout that you'd like to use.

Definition at line 143 of file Panzer_Integrator_GradBasisDotTensorTimesVector_impl.hpp.

Member Function Documentation

template<typename EvalT , typename Traits >

void panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::postRegistrationSetup	(	typename Traits::SetupData	d,
		PHX::FieldManager< Traits > &	fm
	)

Post-Registration Setup.

Sets the basis index, and gets the PHX::View versions of the field multiplier, if there are any.

Parameters

[in]	sd	Essentially a list of `Workset`s, which are collections of cells (elements) that all live on a single process.
[in]	fm	This is an unused part of the `Evaluator` interface.

Definition at line 173 of file Panzer_Integrator_GradBasisDotTensorTimesVector_impl.hpp.

template<typename EvalT , typename Traits >

void panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::evaluateFields ( typename Traits::EvalData d )

Evaluate Fields.

This actually performs the integration by calling operator()() in a Kokkos::parallel_for over the cells in the Workset.

Parameters

[in] workset The Workset on which you're going to do the integration.

Definition at line 317 of file Panzer_Integrator_GradBasisDotTensorTimesVector_impl.hpp.

template<typename EvalT , typename Traits >

KOKKOS_INLINE_FUNCTION void panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::operator()	(	const FieldMultTag &	tag,
		const Kokkos::TeamPolicy< PHX::exec_space >::member_type &	team
	)		const

Perform the integration.

Generally speaking, for a given cell in the Workset, this routine loops over quadrature points, vector dimensions, and bases to perform the integration, scaling the vector field to be integrated by the multiplier ( $ M $ ) and any field multipliers ( $ a(x) $ , $ b(x) $ , etc.).

Note: Optimizations are made for the cases in which we have no field multipliers or only a single one.

Parameters

[in]	tag	An indication of the number of field multipliers we have; either 0, 1, or something else.
[in]	cell	The cell in the `Workset` over which to integrate.

Definition at line 207 of file Panzer_Integrator_GradBasisDotTensorTimesVector_impl.hpp.

template<typename EvalT , typename Traits >

KOKKOS_INLINE_FUNCTION void panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, Traits >::operator()	(	const SharedFieldMultTag &	tag,
		const Kokkos::TeamPolicy< PHX::exec_space >::member_type &	team
	)		const

Perform the integration.

Generally speaking, for a given cell in the Workset, this routine loops over quadrature points, vector dimensions, and bases to perform the integration, scaling the vector field to be integrated by the multiplier ( $ M $ ) and any field multipliers ( $ a(x) $ , $ b(x) $ , etc.). Uses Shared memory.

Note: Optimizations are made for the cases in which we have no field multipliers or only a single one.

Parameters

[in]	tag	An indication of the number of field multipliers we have; either 0, 1, or something else.
[in]	cell	The cell in the `Workset` over which to integrate.

Definition at line 250 of file Panzer_Integrator_GradBasisDotTensorTimesVector_impl.hpp.

template<typename EvalT , typename TRAITS >

Teuchos::RCP< Teuchos::ParameterList > panzer::Integrator_GradBasisDotTensorTimesVector< EvalT, TRAITS >::getValidParameters ( ) const

private

Get Valid Parameters.

Get all the parameters that we support such that the ParameterList Constructor can do some validation of the input ParameterList.