All Classes Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
Ewoms::FvBaseDiscretization< TypeTag > Class Template Reference

The base class for the finite volume discretization schemes. More...

#include <fvbasediscretization.hh>

Inheritance diagram for Ewoms::FvBaseDiscretization< TypeTag >:
Ewoms::EcfvDiscretization< TypeTag > Ewoms::VcfvDiscretization< TypeTag >

Public Member Functions

 FvBaseDiscretization (Simulator &simulator)
 
void finishInit ()
 Apply the initial conditions to the model.
 
bool enableGridAdaptation () const
 Returns whether the grid ought to be adapted to the solution during the simulation.
 
void applyInitialSolution ()
 Applies the initial solution for all degrees of freedom to which the model applies.
 
void prefetch (const Element &elem OPM_UNUSED) const
 Allows to improve the performance by prefetching all data which is associated with a given element.
 
NewtonMethodnewtonMethod ()
 Returns the newton method object.
 
const NewtonMethodnewtonMethod () const
 Returns the newton method object. More...
 
const IntensiveQuantities * thermodynamicHint (unsigned globalIdx, unsigned timeIdx) const
 Return the thermodynamic hint for a entity on the grid at given time. More...
 
const IntensiveQuantities * cachedIntensiveQuantities (unsigned globalIdx, unsigned timeIdx) const
 Return the cached intensive quantities for a entity on the grid at given time. More...
 
void updateCachedIntensiveQuantities (const IntensiveQuantities &intQuants, unsigned globalIdx, unsigned timeIdx) const
 Update the intensive quantity cache for a entity on the grid at given time. More...
 
void setIntensiveQuantitiesCacheEntryValidity (unsigned globalIdx, unsigned timeIdx, bool newValue) const
 Invalidate the cache for a given intensive quantities object. More...
 
void invalidateIntensiveQuantitiesCache (unsigned timeIdx) const
 Invalidate the whole intensive quantity cache for time index. More...
 
void shiftIntensiveQuantityCache (unsigned numSlots=1)
 Move the intensive quantities for a given time index to the back. More...
 
bool enableStorageCache () const
 Returns true iff the storage term is cached. More...
 
const EqVector & cachedStorage (unsigned globalIdx, unsigned timeIdx) const
 Retrieve an entry of the cache for the storage term. More...
 
void updateCachedStorage (unsigned globalIdx, unsigned timeIdx, const EqVector &value) const
 Set an entry of the cache for the storage term. More...
 
Scalar globalResidual (GlobalEqVector &dest, const SolutionVector &u) const
 Compute the global residual for an arbitrary solution vector. More...
 
Scalar globalResidual (GlobalEqVector &dest) const
 Compute the global residual for the current solution vector. More...
 
void globalStorage (EqVector &storage, unsigned timeIdx=0) const
 Compute the integral over the domain of the storage terms of all conservation quantities. More...
 
void checkConservativeness (Scalar OPM_OPTIM_UNUSED tolerance=-1, bool OPM_OPTIM_UNUSED verbose=false) const
 Ensure that the difference between the storage terms of the last and of the current time step is consistent with the source and boundary terms. More...
 
Scalar dofTotalVolume (unsigned globalIdx) const
 Returns the volume $\mathrm{[m^3]}$ of a given control volume. More...
 
bool isLocalDof (unsigned globalIdx) const
 Returns if the overlap of the volume ofa degree of freedom is non-zero. More...
 
Scalar gridTotalVolume () const
 Returns the volume $\mathrm{[m^3]}$ of the whole grid which represents the spatial domain.
 
const SolutionVector & solution (unsigned timeIdx) const
 Reference to the solution at a given history index as a block vector. More...
 
SolutionVector & solution (unsigned timeIdx)
 const More...
 
const Linearizer & linearizer () const
 Returns the operator linearizer for the global jacobian of the problem.
 
Linearizer & linearizer ()
 Returns the object which linearizes the global system of equations at the current solution.
 
const LocalLinearizer & localLinearizer (unsigned openMpThreadId) const
 Returns the local jacobian which calculates the local stiffness matrix for an arbitrary element. More...
 
LocalLinearizer & localLinearizer (unsigned openMpThreadId)
 Returns the local jacobian which calculates the local stiffness matrix for an arbitrary element. const More...
 
const LocalResidual & localResidual (unsigned openMpThreadId) const
 Returns the object to calculate the local residual function.
 
LocalResidual & localResidual (unsigned openMpThreadId)
 Returns the object to calculate the local residual function. const More...
 
Scalar primaryVarWeight (unsigned globalDofIdx, unsigned pvIdx) const
 Returns the relative weight of a primary variable for calculating relative errors. More...
 
Scalar eqWeight (unsigned globalVertexIdx OPM_UNUSED, unsigned eqIdx OPM_UNUSED) const
 Returns the relative weight of an equation. More...
 
Scalar relativeDofError (unsigned vertexIdx, const PrimaryVariables &pv1, const PrimaryVariables &pv2) const
 Returns the relative error between two vectors of primary variables. More...
 
bool update ()
 Try to progress the model to the next timestep. More...
 
void syncOverlap ()
 Syncronize the values of the primary variables on the degrees of freedom that overlap with the neighboring processes. More...
 
void updateBegin ()
 Called by the update() method before it tries to apply the newton method. More...
 
void updateSuccessful ()
 Called by the update() method if it was successful.
 
void adaptGrid ()
 Called by the update() method when the grid should be refined.
 
void updateFailed ()
 Called by the update() method if it was unsuccessful. More...
 
void advanceTimeLevel ()
 Called by the problem if a time integration was successful, post processing of the solution is done and the result has been written to disk. More...
 
template<class Restarter >
void serialize (Restarter &res OPM_UNUSED)
 Serializes the current state of the model. More...
 
template<class Restarter >
void deserialize (Restarter &res OPM_UNUSED)
 Deserializes the state of the model. More...
 
template<class DofEntity >
void serializeEntity (std::ostream &outstream, const DofEntity &dof)
 Write the current solution for a degree of freedom to a restart file. More...
 
template<class DofEntity >
void deserializeEntity (std::istream &instream, const DofEntity &dof)
 Reads the current solution variables for a degree of freedom from a restart file. More...
 
size_t numGridDof () const
 Returns the number of degrees of freedom (DOFs) for the computational grid.
 
size_t numAuxiliaryDof () const
 Returns the number of degrees of freedom (DOFs) of the auxiliary equations.
 
size_t numTotalDof () const
 Returns the total number of degrees of freedom (i.e., grid plux auxiliary DOFs)
 
const DofMapper & dofMapper () const
 Mapper to convert the Dune entities of the discretization's degrees of freedoms are to indices.
 
const VertexMapper & vertexMapper () const
 Returns the mapper for vertices to indices.
 
const ElementMapper & elementMapper () const
 Returns the mapper for elements to indices.
 
void resetLinearizer ()
 Resets the Jacobian matrix linearizer, so that the boundary types can be altered.
 
std::string primaryVarName (unsigned pvIdx) const
 Given an primary variable index, return a human readable name. More...
 
std::string eqName (unsigned eqIdx) const
 Given an equation index, return a human readable name. More...
 
void updatePVWeights (const ElementContext &elemCtx OPM_UNUSED) const
 Update the weights of all primary variables within an element given the complete set of intensive quantities. More...
 
void addOutputModule (BaseOutputModule< TypeTag > *newModule)
 Add an module for writing visualization output after a timestep.
 
template<class VtkMultiWriter >
void addConvergenceVtkFields (VtkMultiWriter &writer, const SolutionVector &u, const GlobalEqVector &deltaU) const
 Add the vector fields for analysing the convergence of the newton method to the a VTK writer. More...
 
void prepareOutputFields () const
 Prepare the quantities relevant for the current solution to be appended to the output writers.
 
void appendOutputFields (BaseOutputWriter &writer) const
 Append the quantities relevant for the current solution to an output writer.
 
const GridView & gridView () const
 Reference to the grid view of the spatial domain.
 
void addAuxiliaryModule (std::shared_ptr< BaseAuxiliaryModule< TypeTag > > auxMod)
 Add a module for an auxiliary equation. More...
 
void clearAuxiliaryModules ()
 Causes the list of auxiliary equations to be cleared. More...
 
size_t numAuxiliaryModules () const
 Returns the number of modules for auxiliary equations.
 
std::shared_ptr
< BaseAuxiliaryModule< TypeTag > > 
auxiliaryModule (unsigned auxEqModIdx)
 Returns a given module for auxiliary equations.
 
std::shared_ptr< const
BaseAuxiliaryModule< TypeTag > > 
auxiliaryModule (unsigned auxEqModIdx) const
 Returns a given module for auxiliary equations.
 
const Ewoms::TimerprePostProcessTimer () const
 
const Ewoms::TimerlinearizeTimer () const
 
const Ewoms::TimersolveTimer () const
 
const Ewoms::TimerupdateTimer () const
 

Static Public Member Functions

static void registerParameters ()
 Register all run-time parameters for the model.
 
static std::string discretizationName ()
 Returns a string of discretization's human-readable name.
 
static bool storeIntensiveQuantities ()
 Returns true if the cache for intensive quantities is enabled.
 

Protected Member Functions

SolutionVector & mutableSolution (unsigned timeIdx) const
 const More...
 
void resizeAndResetIntensiveQuantitiesCache_ ()
 
template<class Context >
void supplementInitialSolution_ (PrimaryVariables &priVars OPM_UNUSED, const Context &context OPM_UNUSED, unsigned dofIdx OPM_UNUSED, unsigned timeIdx OPM_UNUSED)
 
void registerOutputModules_ ()
 Register all output modules which make sense for the model. More...
 
LocalResidual & localResidual_ ()
 Reference to the local residal object.
 
bool verbose_ () const
 Returns whether messages should be printed.
 
Implementation & asImp_ ()
 
const Implementation & asImp_ () const
 

Static Protected Member Functions

static bool enableIntensiveQuantitiesCache_ ()
 
static bool enableThermodynamicHints_ ()
 

Protected Attributes

Simulatorsimulator_
 
GridView gridView_
 
ElementMapper elementMapper_
 
VertexMapper vertexMapper_
 
std::vector< std::shared_ptr
< BaseAuxiliaryModule< TypeTag > > > 
auxEqModules_
 
NewtonMethod newtonMethod_
 
Ewoms::Timer prePostProcessTimer_
 
Ewoms::Timer linearizeTimer_
 
Ewoms::Timer solveTimer_
 
Ewoms::Timer updateTimer_
 
std::vector< LocalLinearizer > localLinearizer_
 
Linearizer * linearizer_
 
IntensiveQuantitiesVector intensiveQuantityCache_ [historySize]
 
std::vector< bool > intensiveQuantityCacheUpToDate_ [historySize]
 
DiscreteFunctionSpace space_
 
std::array< std::unique_ptr
< DiscreteFunction >
, historySize > 
solution_
 
std::list< BaseOutputModule
< TypeTag > * > 
outputModules_
 
Scalar gridTotalVolume_
 
std::vector< Scalar > dofTotalVolume_
 
std::vector< bool > isLocalDof_
 
bool enableGridAdaptation_
 
GlobalEqVector storageCache_ [historySize]
 
bool enableStorageCache_
 

Detailed Description

template<class TypeTag>
class Ewoms::FvBaseDiscretization< TypeTag >

The base class for the finite volume discretization schemes.

Member Function Documentation

template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::addAuxiliaryModule ( std::shared_ptr< BaseAuxiliaryModule< TypeTag > >  auxMod)
inline

Add a module for an auxiliary equation.

This module can add additional degrees of freedom and additional off-diagonal elements, but the number of equations per DOF needs to be the same as for the "main" model.

For example, auxiliary modules can be used to specify non-neighboring connections, well equations or model couplings via mortar DOFs. Auxiliary equations are completely optional, though.

template<class TypeTag >
template<class VtkMultiWriter >
void Ewoms::FvBaseDiscretization< TypeTag >::addConvergenceVtkFields ( VtkMultiWriter writer,
const SolutionVector &  u,
const GlobalEqVector &  deltaU 
) const
inline

Add the vector fields for analysing the convergence of the newton method to the a VTK writer.

Parameters
writerThe writer object to which the fields should be added.
uThe solution function
deltaUThe delta of the solution function before and after the Newton update
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::advanceTimeLevel ( )
inline

Called by the problem if a time integration was successful, post processing of the solution is done and the result has been written to disk.

This should prepare the model for the next time integration.

template<class TypeTag >
const IntensiveQuantities* Ewoms::FvBaseDiscretization< TypeTag >::cachedIntensiveQuantities ( unsigned  globalIdx,
unsigned  timeIdx 
) const
inline

Return the cached intensive quantities for a entity on the grid at given time.

Attention
If no up-to date intensive quantities are available, this method will return 0.
Parameters
globalIdxThe global space index for the entity where a hint is requested.
timeIdxThe index used by the time discretization.
template<class TypeTag >
const EqVector& Ewoms::FvBaseDiscretization< TypeTag >::cachedStorage ( unsigned  globalIdx,
unsigned  timeIdx 
) const
inline

Retrieve an entry of the cache for the storage term.

This is supposed to represent a DOF's total amount of conservation quantities per volume unit at a given time. The user is responsible for making sure that the value of this is correct and that it can be used before this method is called.

Parameters
globalDofIdxThe index of the relevant degree of freedom in a grid-global vector
timeIdxThe relevant index for the time discretization
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::checkConservativeness ( Scalar OPM_OPTIM_UNUSED  tolerance = -1,
bool OPM_OPTIM_UNUSED  verbose = false 
) const
inline

Ensure that the difference between the storage terms of the last and of the current time step is consistent with the source and boundary terms.

This method is purely intented for debugging purposes. If the program is compiled with optimizations enabled, it becomes a no-op.

template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::clearAuxiliaryModules ( )
inline

Causes the list of auxiliary equations to be cleared.

Note that this method implies recreateMatrix()

template<class TypeTag >
template<class Restarter >
void Ewoms::FvBaseDiscretization< TypeTag >::deserialize ( Restarter &res  OPM_UNUSED)
inline

Deserializes the state of the model.

Template Parameters
RestarterThe type of the serializer class
Parameters
resThe serializer object
template<class TypeTag >
template<class DofEntity >
void Ewoms::FvBaseDiscretization< TypeTag >::deserializeEntity ( std::istream &  instream,
const DofEntity &  dof 
)
inline

Reads the current solution variables for a degree of freedom from a restart file.

Parameters
instreamThe stream from which the vertex data should be deserialized from
dofThe Dune entity which's data should be deserialized
template<class TypeTag >
Scalar Ewoms::FvBaseDiscretization< TypeTag >::dofTotalVolume ( unsigned  globalIdx) const
inline

Returns the volume $\mathrm{[m^3]}$ of a given control volume.

Parameters
globalIdxThe global index of the degree of freedom
template<class TypeTag >
bool Ewoms::FvBaseDiscretization< TypeTag >::enableStorageCache ( ) const
inline

Returns true iff the storage term is cached.

Be aware that calling the *CachedStorage() methods if the storage cache is disabled will crash the program.

template<class TypeTag >
std::string Ewoms::FvBaseDiscretization< TypeTag >::eqName ( unsigned  eqIdx) const
inline

Given an equation index, return a human readable name.

Parameters
eqIdxThe index of the conservation equation of interest.
template<class TypeTag >
Scalar Ewoms::FvBaseDiscretization< TypeTag >::eqWeight ( unsigned globalVertexIdx  OPM_UNUSED,
unsigned eqIdx  OPM_UNUSED 
) const
inline

Returns the relative weight of an equation.

Parameters
globalVertexIdxThe global index of the vertex
eqIdxThe index of the equation
template<class TypeTag >
Scalar Ewoms::FvBaseDiscretization< TypeTag >::globalResidual ( GlobalEqVector &  dest,
const SolutionVector &  u 
) const
inline

Compute the global residual for an arbitrary solution vector.

Parameters
destStores the result
uThe solution for which the residual ought to be calculated
template<class TypeTag >
Scalar Ewoms::FvBaseDiscretization< TypeTag >::globalResidual ( GlobalEqVector &  dest) const
inline

Compute the global residual for the current solution vector.

Parameters
destStores the result
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::globalStorage ( EqVector &  storage,
unsigned  timeIdx = 0 
) const
inline

Compute the integral over the domain of the storage terms of all conservation quantities.

Parameters
storageStores the total amount of each conserved quantity inside the domain.
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::invalidateIntensiveQuantitiesCache ( unsigned  timeIdx) const
inline

Invalidate the whole intensive quantity cache for time index.

Parameters
timeIdxThe index used by the time discretization.
template<class TypeTag >
bool Ewoms::FvBaseDiscretization< TypeTag >::isLocalDof ( unsigned  globalIdx) const
inline

Returns if the overlap of the volume ofa degree of freedom is non-zero.

Parameters
globalIdxThe global index of the degree of freedom
template<class TypeTag >
const LocalLinearizer& Ewoms::FvBaseDiscretization< TypeTag >::localLinearizer ( unsigned  openMpThreadId) const
inline

Returns the local jacobian which calculates the local stiffness matrix for an arbitrary element.

The local stiffness matrices of the element are used by the jacobian linearizer to produce a global linerization of the problem.

template<class TypeTag >
LocalLinearizer& Ewoms::FvBaseDiscretization< TypeTag >::localLinearizer ( unsigned  openMpThreadId)
inline

Returns the local jacobian which calculates the local stiffness matrix for an arbitrary element. const

The local stiffness matrices of the element are used by the jacobian linearizer to produce a global linerization of the problem. const

template<class TypeTag >
LocalResidual& Ewoms::FvBaseDiscretization< TypeTag >::localResidual ( unsigned  openMpThreadId)
inline

Returns the object to calculate the local residual function. const

const

template<class TypeTag >
SolutionVector& Ewoms::FvBaseDiscretization< TypeTag >::mutableSolution ( unsigned  timeIdx) const
inlineprotected

const

const

template<class TypeTag >
const NewtonMethod& Ewoms::FvBaseDiscretization< TypeTag >::newtonMethod ( ) const
inline

Returns the newton method object.

template<class TypeTag >
std::string Ewoms::FvBaseDiscretization< TypeTag >::primaryVarName ( unsigned  pvIdx) const
inline

Given an primary variable index, return a human readable name.

Parameters
pvIdxThe index of the primary variable of interest.
template<class TypeTag >
Scalar Ewoms::FvBaseDiscretization< TypeTag >::primaryVarWeight ( unsigned  globalDofIdx,
unsigned  pvIdx 
) const
inline

Returns the relative weight of a primary variable for calculating relative errors.

Parameters
globalDofIdxThe global index of the degree of freedom
pvIdxThe index of the primary variable
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::registerOutputModules_ ( )
inlineprotected

Register all output modules which make sense for the model.

This method is supposed to be overloaded by the actual models, or else only the primary variables can be written to the result files.

template<class TypeTag >
Scalar Ewoms::FvBaseDiscretization< TypeTag >::relativeDofError ( unsigned  vertexIdx,
const PrimaryVariables &  pv1,
const PrimaryVariables &  pv2 
) const
inline

Returns the relative error between two vectors of primary variables.

Parameters
vertexIdxThe global index of the control volume's associated vertex
pv1The first vector of primary variables
pv2The second vector of primary variables
template<class TypeTag >
template<class Restarter >
void Ewoms::FvBaseDiscretization< TypeTag >::serialize ( Restarter &res  OPM_UNUSED)
inline

Serializes the current state of the model.

Template Parameters
RestarterThe type of the serializer class
Parameters
resThe serializer object
template<class TypeTag >
template<class DofEntity >
void Ewoms::FvBaseDiscretization< TypeTag >::serializeEntity ( std::ostream &  outstream,
const DofEntity &  dof 
)
inline

Write the current solution for a degree of freedom to a restart file.

Parameters
outstreamThe stream into which the vertex data should be serialized to
dofThe Dune entity which's data should be serialized
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::setIntensiveQuantitiesCacheEntryValidity ( unsigned  globalIdx,
unsigned  timeIdx,
bool  newValue 
) const
inline

Invalidate the cache for a given intensive quantities object.

Parameters
globalIdxThe global space index for the entity where a hint is to be set.
timeIdxThe index used by the time discretization.
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::shiftIntensiveQuantityCache ( unsigned  numSlots = 1)
inline

Move the intensive quantities for a given time index to the back.

This method should only be called by the time discretization.

Parameters
numSlotsThe number of time step slots for which the hints should be shifted.
template<class TypeTag >
const SolutionVector& Ewoms::FvBaseDiscretization< TypeTag >::solution ( unsigned  timeIdx) const
inline

Reference to the solution at a given history index as a block vector.

Parameters
timeIdxThe index of the solution used by the time discretization.
template<class TypeTag >
SolutionVector& Ewoms::FvBaseDiscretization< TypeTag >::solution ( unsigned  timeIdx)
inline

const

const

template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::syncOverlap ( )
inline

Syncronize the values of the primary variables on the degrees of freedom that overlap with the neighboring processes.

By default, this method does nothing...

template<class TypeTag >
const IntensiveQuantities* Ewoms::FvBaseDiscretization< TypeTag >::thermodynamicHint ( unsigned  globalIdx,
unsigned  timeIdx 
) const
inline

Return the thermodynamic hint for a entity on the grid at given time.

The hint is defined as a IntensiveQuantities object which is supposed to be "close" to the IntensiveQuantities of the current solution. It can be used as a good starting point for non-linear solvers when having to solve non-linear relations while updating the intensive quantities. (This may yield a major performance boost depending on how the physical models require.)

Attention
If no up-to date intensive quantities are available, or if hints have been disabled, this method will return 0.
Parameters
globalIdxThe global space index for the entity where a hint is requested.
timeIdxThe index used by the time discretization.
template<class TypeTag >
bool Ewoms::FvBaseDiscretization< TypeTag >::update ( )
inline

Try to progress the model to the next timestep.

Parameters
solverThe non-linear solver
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::updateBegin ( )
inline

Called by the update() method before it tries to apply the newton method.

This is primary a hook which the actual model can overload.

template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::updateCachedIntensiveQuantities ( const IntensiveQuantities &  intQuants,
unsigned  globalIdx,
unsigned  timeIdx 
) const
inline

Update the intensive quantity cache for a entity on the grid at given time.

Parameters
intQuantsThe IntensiveQuantities object hint for a given degree of freedom.
globalIdxThe global space index for the entity where a hint is to be set.
timeIdxThe index used by the time discretization.
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::updateCachedStorage ( unsigned  globalIdx,
unsigned  timeIdx,
const EqVector &  value 
) const
inline

Set an entry of the cache for the storage term.

This is supposed to represent a DOF's total amount of conservation quantities per volume unit at a given time. The user is responsible for making sure that the storage cache is enabled before this method is called.

Parameters
globalDofIdxThe index of the relevant degree of freedom in a grid-global vector
timeIdxThe relevant index for the time discretization
valueThe new value of the cache for the storage term
template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::updateFailed ( )
inline

Called by the update() method if it was unsuccessful.

This is primary a hook which the actual model can overload.

template<class TypeTag >
void Ewoms::FvBaseDiscretization< TypeTag >::updatePVWeights ( const ElementContext &elemCtx  OPM_UNUSED) const
inline

Update the weights of all primary variables within an element given the complete set of intensive quantities.


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