olb::SuperLatticeRefinementMetricKnudsen2D< T, DESCRIPTOR > Class Template Reference

SuperLatticeRefinementMetricKnudsen2D suggests a per-block grid refinement factor. More...

#include <superLatticeRefinementMetricF2D.h>

Inheritance diagram for olb::SuperLatticeRefinementMetricKnudsen2D< T, DESCRIPTOR >:

Collaboration diagram for olb::SuperLatticeRefinementMetricKnudsen2D< T, DESCRIPTOR >:

Public Member Functions
	SuperLatticeRefinementMetricKnudsen2D (SuperLattice< T, DESCRIPTOR > &lattice, const UnitConverter< T, DESCRIPTOR > &converter)
	Constructor.
bool	operator() (T output[], int glob)
void	print ()
	Convenience method for printing per-block refinement factors.
Public Member Functions inherited from olb::SuperLatticeF2D< T, DESCRIPTOR >
SuperLattice< T, DESCRIPTOR > &	getSuperLattice ()
bool	operator() (T output[], const int input[])
Public Member Functions inherited from olb::SuperF2D< T, T >
SuperF2D< T, T > &	operator- (SuperF2D< T, T > &rhs)
SuperF2D< T, T > &	operator+ (SuperF2D< T, T > &rhs)
SuperF2D< T, T > &	operator* (SuperF2D< T, T > &rhs)
SuperF2D< T, T > &	operator/ (SuperF2D< T, T > &rhs)
SuperStructure< T, 2 > &	getSuperStructure ()
int	getBlockFSize () const
BlockF2D< T > &	getBlockF (int iCloc)
bool	operator() (T output[], const int input[]) override
Public Member Functions inherited from olb::GenericF< T, S >
virtual	~GenericF ()=default
int	getSourceDim () const
	read only access to member variable _m
int	getTargetDim () const
	read only access to member variable _n
std::string &	getName ()
	read and write access to name
std::string const &	getName () const
	read only access to name
virtual bool	operator() (T output[], const S input[])=0
	has to be implemented for 'every' derived class
bool	operator() (T output[])
	wrapper that call the pure virtual operator() (T output[], const S input[]) from above
bool	operator() (T output[], S input0)
bool	operator() (T output[], S input0, S input1)
bool	operator() (T output[], S input0, S input1, S input2)
bool	operator() (T output[], S input0, S input1, S input2, S input3)

Additional Inherited Members
Public Types inherited from olb::SuperF2D< T, T >
using	identity_functor_type
Public Types inherited from olb::GenericF< T, S >
using	targetType = T
using	sourceType = S
Public Attributes inherited from olb::GenericF< T, S >
std::shared_ptr< GenericF< T, S > >	_ptrCalcC
	memory management, frees resouces (calcClass)
Protected Member Functions inherited from olb::SuperLatticeF2D< T, DESCRIPTOR >
	SuperLatticeF2D (SuperLattice< T, DESCRIPTOR > &superLattice, int targetDim)
Protected Member Functions inherited from olb::SuperF2D< T, T >
	SuperF2D (SuperStructure< T, 2 > &superStructure, int targetDim)
Protected Member Functions inherited from olb::GenericF< T, S >
	GenericF (int targetDim, int sourceDim)
Protected Attributes inherited from olb::SuperLatticeF2D< T, DESCRIPTOR >
SuperLattice< T, DESCRIPTOR > &	_sLattice
Protected Attributes inherited from olb::SuperF2D< T, T >
SuperStructure< T, 2 > &	_superStructure
std::vector< std::unique_ptr< BlockF2D< T > > >	_blockF
	Super functors may consist of several BlockF2D<W> derived functors.

Detailed Description

template<typename T, typename DESCRIPTOR>
class olb::SuperLatticeRefinementMetricKnudsen2D< T, DESCRIPTOR >

SuperLatticeRefinementMetricKnudsen2D suggests a per-block grid refinement factor.

This functor is a direct implementation of the refinement criterion described in "Automatic grid refinement criterion for lattice Boltzmann method" [arXiv:1507.06767] by D. Lagrava et al.

Definition at line 55 of file superLatticeRefinementMetricF2D.h.

Constructor & Destructor Documentation

SuperLatticeRefinementMetricKnudsen2D()

template<typename T , typename DESCRIPTOR >

olb::SuperLatticeRefinementMetricKnudsen2D< T, DESCRIPTOR >::SuperLatticeRefinementMetricKnudsen2D	(	SuperLattice< T, DESCRIPTOR > &	lattice,
		const UnitConverter< T, DESCRIPTOR > &	converter )

Constructor.

Parameters

converter	Required to get the theoretical Knudsen number as reference
rounding	Configures the quality of the grid by rounding the output

Definition at line 52 of file superLatticeRefinementMetricF2D.hh.

  : SuperLatticeF2D<T, DESCRIPTOR>(lattice, 1)
{
  this->getName() = "refinementMetricKnudsen";
 
  for (int iC = 0; iC < lattice.getLoadBalancer().size(); ++iC) {
    this->_blockF.emplace_back(
      new BlockLatticeRefinementMetricKnudsen2D<T, DESCRIPTOR>(
        lattice.getBlock(iC), converter)
    );
  }
}

References olb::SuperF2D< T, T >::_blockF, olb::SuperLattice< T, DESCRIPTOR >::getBlock(), olb::SuperStructure< T, D >::getLoadBalancer(), and olb::GenericF< T, S >::getName().

Here is the call graph for this function:

Member Function Documentation

operator()()

template<typename T , typename DESCRIPTOR >

bool olb::SuperLatticeRefinementMetricKnudsen2D< T, DESCRIPTOR >::operator()	(	T	output[],
		int	glob )

Parameters

output	output[0] == 0 suggests sufficient grid resolution, output[0] > 0 suggests the grid to be refined output[0] times
glob	cuboid to be considered

Definition at line 68 of file superLatticeRefinementMetricF2D.hh.

{
  auto& load = this->_sLattice.getLoadBalancer();
 
  if (load.isLocal(glob)) {
    const int loc = load.loc(glob);
 
    return this->getBlockF(loc)(output);
  }
  else {
    return false;
  }
}

print()

template<typename T , typename DESCRIPTOR >

void olb::SuperLatticeRefinementMetricKnudsen2D< T, DESCRIPTOR >::print

(

)

Convenience method for printing per-block refinement factors.

Parameters

output

output[0] == 0 suggests sufficient grid resolution, output[0] > 0 suggests the grid to be refined output[0] times \input input grid cell to be considered

Definition at line 85 of file superLatticeRefinementMetricF2D.hh.

{
  const int nC = this->_sLattice.getCuboidGeometry().getNc();
 
  std::vector<T> factors(nC, T{});
  T output[1] = { };
 
  auto& load = this->_sLattice.getLoadBalancer();
 
  for (int iC = 0; iC < load.size(); ++iC) {
    this->getBlockF(iC)(output, iC);
 
    factors[load.glob(iC)] = output[0];
  }
 
  OstreamManager clout(std::cout, "refinement");
 
  for (int i = 0; i < nC; ++i) {
#ifdef PARALLEL_MODE_MPI
    singleton::mpi().sendToMaster(&factors[i], 1, load.isLocal(i));
#endif
 
    clout << "factors[" << i << "]: " << factors[i] << std::endl;
  }
}

References olb::singleton::mpi(), and olb::singleton::MpiManager::sendToMaster().

Here is the call graph for this function:

---

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

• src/functors/lattice/superLatticeRefinementMetricF2D.h
• src/functors/lattice/superLatticeRefinementMetricF2D.hh