#include <AbstractNonlinearAssemblerSolverHybrid.hpp>
Inheritance diagram for AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >:
Collaboration diagram for AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >:
Public Member Functions | |
| void | ComputeResidual (const Vec currentGuess, Vec residualVector) |
| void | ComputeJacobian (const Vec currentGuess, Mat *pJacobian) |
| AbstractNonlinearAssemblerSolverHybrid (AbstractTetrahedralMesh< ELEMENT_DIM, SPACE_DIM > *pMesh, BoundaryConditionsContainer< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM > *pBoundaryConditions, unsigned numQuadPoints=2) | |
| virtual | ~AbstractNonlinearAssemblerSolverHybrid () |
| virtual Vec | Solve (Vec initialGuess, bool useAnalyticalJacobian=false) |
| void | SetNonlinearSolver (AbstractNonlinearSolver *pNonlinearSolver) |
| bool | VerifyJacobian (double tol) |
Protected Member Functions | |
| void | ApplyDirichletConditions (Vec currentGuess, Vec residual, Mat *pMat) |
| void | ComputeJacobianNumerically (const Vec currentGuess, Mat *pJacobian) |
Protected Attributes | |
| AbstractTetrahedralMesh < ELEMENT_DIM, SPACE_DIM > * | mpMesh |
| BoundaryConditionsContainer < ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM > * | mpBoundaryConditions |
| AbstractNonlinearSolver * | mpSolver |
| bool | mWeAllocatedSolverMemory |
| bool | mUseAnalyticalJacobian |
| NaturalNeumannSurfaceTermAssembler < ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM > * | mpNeumannSurfaceTermsAssembler |
Detailed Description
template<unsigned ELEMENT_DIM, unsigned SPACE_DIM, unsigned PROBLEM_DIM>
class AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >
The ASSEMBLER-SOLVER class.
An abstract solver for solving nonlinear elliptic PDEs.
Definition at line 105 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
Constructor & Destructor Documentation
| AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::AbstractNonlinearAssemblerSolverHybrid | ( | AbstractTetrahedralMesh< ELEMENT_DIM, SPACE_DIM > * | pMesh, |
| BoundaryConditionsContainer< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM > * | pBoundaryConditions, | ||
| unsigned | numQuadPoints = 2 |
||
| ) |
Constructor.
- Parameters:
-
pMesh The mesh pBoundaryConditions The boundary conditions to apply numQuadPoints number of quadrature points (defaults to 2)
Definition at line 226 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
References AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mpMesh, AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mpNeumannSurfaceTermsAssembler, AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mpSolver, and AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mWeAllocatedSolverMemory.
| AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::~AbstractNonlinearAssemblerSolverHybrid | ( | ) | [virtual] |
Destructor.
Definition at line 252 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
Member Function Documentation
| void AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::ApplyDirichletConditions | ( | Vec | currentGuess, |
| Vec | residual, | ||
| Mat * | pMat | ||
| ) | [protected] |
Apply Dirichlet boundary conditions to either the residual or Jacobian.
- Parameters:
-
currentGuess the solution guess for the current nonlinear solver iteration residual the residual to apply boundary conditions to (can be NULL if bcs to be applied to Jacobian only) pMat the Jacobian to apply boundary conditions to (can be NULL if bcs to be applied to residual only)
Definition at line 262 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
| void AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::ComputeJacobian | ( | const Vec | currentGuess, |
| Mat * | pJacobian | ||
| ) |
Compute the Jacobian matrix given a current guess at the solution. Choose whether to use a numerical or analytical method based on a flag provided by the user (in Solve()).
- Parameters:
-
currentGuess The solution guess for the current iteration. pJacobian Pointer to object to fill with the Jacobian matrix.
NOTE: this method is called indirectly by the PETSc iterative solvers, so must be public.
Definition at line 298 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
References PetscMatTools::Finalise(), and PetscMatTools::SwitchWriteMode().
| void AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::ComputeJacobianNumerically | ( | const Vec | currentGuess, |
| Mat * | pJacobian | ||
| ) | [protected] |
Computes the Jacobian numerically i.e. an approximation, using numerical partial derivatives.
- Parameters:
-
currentGuess Independent variable, u in f(u), for example pJacobian A pointer to the Jacobian matrix
Definition at line 319 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
References PetscVecTools::AddToElement(), PetscTools::CreateVec(), PetscTools::Destroy(), PetscMatTools::Finalise(), PetscVecTools::Scale(), PetscMatTools::SetElement(), and PetscVecTools::WAXPY().
| void AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::ComputeResidual | ( | const Vec | currentGuess, |
| Vec | residualVector | ||
| ) |
Compute the residual vector given the current solution guess.
- Parameters:
-
currentGuess The solution guess for the current nonlinear solve iteration. residualVector We fill this with the residual vector.
NOTE: this method is called indirectly by the PETSc iterative solvers, so must be public.
Definition at line 277 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
References PetscVecTools::Finalise().
| void AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::SetNonlinearSolver | ( | AbstractNonlinearSolver * | pNonlinearSolver | ) |
SetNonlinearSolver - by default a SimplePetscNonlinearSolver is created and used in this class, this method can be called to use a different AbstractNonlinearSolver.
- Parameters:
-
pNonlinearSolver a nonlinear solver
Definition at line 403 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
| Vec AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::Solve | ( | Vec | initialGuess, |
| bool | useAnalyticalJacobian = false |
||
| ) | [virtual] |
Assemble and solve the system for a nonlinear elliptic PDE.
- Parameters:
-
initialGuess An initial guess for the iterative solver useAnalyticalJacobian Set to true to use an analytically calculated Jacobian matrix rather than a numerically approximated one.
- Returns:
- A PETSc vector giving the solution at each mesh node.
Definition at line 379 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
References EXCEPTION.
| bool AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::VerifyJacobian | ( | double | tol | ) |
A helper method for use when writing concrete assemblers. Once the user has calculated (on paper) the weak form and the form of the ComputeMatrixTerm method, they can check whether the analytic Jacobian matches the numerical Jacobian to verify the correctness of the code.
- Parameters:
-
tol A tolerance which defaults to 1e-5
- Returns:
- true if the componentwise difference between the matrices is less than the tolerance, false otherwise.
This method should NOT be run in simulations - it is only to verify the correctness of the concrete assembler code. Allocates dense matrices.
Definition at line 414 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
References PetscMatTools::CheckEquality(), PetscTools::CreateAndSetVec(), PetscTools::Destroy(), and PetscTools::SetupMat().
Member Data Documentation
BoundaryConditionsContainer<ELEMENT_DIM,SPACE_DIM,PROBLEM_DIM>* AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mpBoundaryConditions [protected] |
Boundary conditions container.
Definition at line 113 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
AbstractTetrahedralMesh<ELEMENT_DIM,SPACE_DIM>* AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mpMesh [protected] |
The mesh.
Reimplemented from AbstractFeVolumeIntegralAssembler< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM, true, true, NONLINEAR >.
Definition at line 110 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
Referenced by AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::AbstractNonlinearAssemblerSolverHybrid().
NaturalNeumannSurfaceTermAssembler<ELEMENT_DIM,SPACE_DIM,PROBLEM_DIM>* AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mpNeumannSurfaceTermsAssembler [protected] |
An assembler for the surface integral terms in the residual
Definition at line 125 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
Referenced by AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::AbstractNonlinearAssemblerSolverHybrid().
AbstractNonlinearSolver* AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mpSolver [protected] |
The nonlinear solver.
Definition at line 116 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
Referenced by AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::AbstractNonlinearAssemblerSolverHybrid().
bool AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mUseAnalyticalJacobian [protected] |
Whether to use an analytical expression for the Jacobian.
Definition at line 122 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
bool AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::mWeAllocatedSolverMemory [protected] |
Whether memory has been allocated for the solver.
Definition at line 119 of file AbstractNonlinearAssemblerSolverHybrid.hpp.
Referenced by AbstractNonlinearAssemblerSolverHybrid< ELEMENT_DIM, SPACE_DIM, PROBLEM_DIM >::AbstractNonlinearAssemblerSolverHybrid().
The documentation for this class was generated from the following file:
- pde/src/solver/AbstractNonlinearAssemblerSolverHybrid.hpp
