Intrepid2
Public Member Functions | Private Types | Private Attributes | List of all members
Intrepid2::BasisValues< Scalar, ExecSpaceType > Class Template Reference

The data containers in Intrepid2 that support sum factorization and other reduced-data optimizations distinguish between scalar-valued data that is a simple product of elements in tensor components, and vector-valued data that is made up of a series of such products. More...

#include <Intrepid2_BasisValues.hpp>

Public Member Functions

 BasisValues (TensorDataType tensorData)
 Constructor for scalar-valued BasisValues with a single family of values.
 
 numTensorDataFamilies_ (1)
 
 BasisValues (std::vector< TensorDataType > tensorDataFamilies)
 Constructor for scalar-valued BasisValues, with potentially multiple families of values. (Used, e.g., for op = DIV and functionSpace = HDIV.)
 
 BasisValues (VectorDataType vectorData)
 Constructor for vector-valued BasisValues.
 
 BasisValues ()
 Default constructor.
 
template<typename OtherExecSpaceType , class = typename std::enable_if<!std::is_same<ExecSpaceType, OtherExecSpaceType>::value>::type>
 BasisValues (const BasisValues< Scalar, OtherExecSpaceType > &basisValues)
 copy-like constructor for differing execution spaces. This does a deep copy of underlying views.
 
BasisValues< Scalar,
ExecSpaceType > 
basisValuesForFields (const int &fieldStartOrdinal, const int &numFields)
 field start and length must align with families in vectorData_ or tensorDataFamilies_ (whichever is valid).
 
KOKKOS_INLINE_FUNCTION int familyFieldOrdinalOffset (const int &familyOrdinal) const
 Returns the field ordinal offset for the specified family.
 
TensorDataTypetensorData ()
 TensorData accessor for single-family scalar data.
 
const TensorDataTypetensorData (const int &familyOrdinal) const
 TensorData accessor for multi-family scalar data.
 
KOKKOS_INLINE_FUNCTION int numFamilies () const
 For valid vectorData, returns the number of families in vectorData; otherwise, returns number of TensorData families.
 
KOKKOS_INLINE_FUNCTION int numTensorDataFamilies () const
 
KOKKOS_INLINE_FUNCTION int numFieldsInFamily (int familyOrdinal) const
 
const Kokkos::Array
< TensorDataType,
Parameters::MaxTensorComponents > & 
tensorDataFamilies () const
 TensorDataFamilies accessor.
 
const VectorDataTypevectorData () const
 VectorData accessor.
 
KOKKOS_INLINE_FUNCTION Scalar operator() (const int &fieldOrdinal, const int &pointOrdinal) const
 operator() for (F,P) scalar data; throws an exception if this is not a scalar-valued container
 
KOKKOS_INLINE_FUNCTION Scalar operator() (const int &fieldOrdinal, const int &pointOrdinal, const int &dim) const
 operator() for (F,P,D) vector data; throws an exception if this is not a vector-valued container
 
KOKKOS_INLINE_FUNCTION Scalar operator() (const int &cellOrdinal, const int &fieldOrdinal, const int &pointOrdinal, const int &dim) const
 operator() for (C,F,P,D) data, which arises in CVFEM; at present unimplemented, and only declared here to allow a generic setJacobian() method in CellTools to compile.
 
KOKKOS_INLINE_FUNCTION int extent_int (const int &i) const
 
KOKKOS_INLINE_FUNCTION size_t extent (const int &i) const
 
KOKKOS_INLINE_FUNCTION size_t rank () const
 
void setOrdinalFilter (Kokkos::View< ordinal_type *, ExecSpaceType > ordinalFilter)
 
Kokkos::View< ordinal_type
*, ExecSpaceType > 
ordinalFilter () const
 

Private Types

using TensorDataType = TensorData< Scalar, ExecSpaceType >
 
using VectorDataType = VectorData< Scalar, ExecSpaceType >
 

Private Attributes

Kokkos::Array< TensorDataType,
Parameters::MaxTensorComponents
tensorDataFamilies_
 
VectorDataType vectorData_
 
int numTensorDataFamilies_ = -1
 
Kokkos::View< ordinal_type
*, ExecSpaceType > 
ordinalFilter_
 

Detailed Description

template<class Scalar, typename ExecSpaceType>
class Intrepid2::BasisValues< Scalar, ExecSpaceType >

The data containers in Intrepid2 that support sum factorization and other reduced-data optimizations distinguish between scalar-valued data that is a simple product of elements in tensor components, and vector-valued data that is made up of a series of such products.

BasisValues contains both a VectorData and a TensorData object; only one of them will be valid in any BasisValues object. This allows us to maintain a common getValues() interface in the Basis class, rather than distinguishing at the interface level between operations that result in vector values and those that result in scalars.

Definition at line 69 of file Intrepid2_BasisValues.hpp.


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