Public Member Functions | Private Member Functions | List of all members
oomph::HelmholtzFGMRESMG< MATRIX > Class Template Reference

The FGMRES method, i.e. the flexible variant of the GMRES method which allows for nonconstant preconditioners [see Saad Y, "Iterative methods for sparse linear systems", p.287]. Note, FGMRES can only cater to right preconditioning; if the user tries to switch to left preconditioning they will be notified of this. More...

#include <complex_smoother.h>

+ Inheritance diagram for oomph::HelmholtzFGMRESMG< MATRIX >:

Public Member Functions

 HelmholtzFGMRESMG ()
 Constructor (empty) More...
 
virtual ~HelmholtzFGMRESMG ()
 Destructor (cleanup storage) More...
 
 HelmholtzFGMRESMG (const HelmholtzFGMRESMG &)
 Broken copy constructor. More...
 
void operator= (const HelmholtzFGMRESMG &)
 Broken assignment operator. More...
 
void set_preconditioner_LHS ()
 Overloaded function to let the user know that left preconditioning is not possible with FGMRES, only right preconditioning. More...
 
void solve (Problem *const &problem_pt, DoubleVector &result)
 Solver: Takes pointer to problem and returns the results vector which contains the solution of the linear system defined by the problem's fully assembled Jacobian and residual vector. More...
 
- Public Member Functions inherited from oomph::HelmholtzGMRESMG< MATRIX >
 HelmholtzGMRESMG ()
 Constructor. More...
 
virtual ~HelmholtzGMRESMG ()
 Destructor (cleanup storage) More...
 
 HelmholtzGMRESMG (const HelmholtzGMRESMG &)
 Broken copy constructor. More...
 
void operator= (const HelmholtzGMRESMG &)
 Broken assignment operator. More...
 
void disable_resolve ()
 Overload disable resolve so that it cleans up memory too. More...
 
void preconditioner_solve (const DoubleVector &r, DoubleVector &z)
 Implementation of the pure virtual base class function. The function has been broken because this is meant to be used as a linear solver. More...
 
void setup ()
 Implementation of the pure virtual base class function. This accompanies the preconditioner_solve function and so is also broken. More...
 
void solve (DoubleMatrixBase *const &matrix_pt, const DoubleVector &rhs, DoubleVector &solution)
 Linear-algebra-type solver: Takes pointer to a matrix and rhs vector and returns the solution of the linear system. More...
 
void solve (DoubleMatrixBase *const &matrix_pt, const Vector< double > &rhs, Vector< double > &result)
 Linear-algebra-type solver: Takes pointer to a matrix and rhs vector and returns the solution of the linear system Call the broken base-class version. If you want this, please implement it. More...
 
void resolve (const DoubleVector &rhs, DoubleVector &result)
 Re-solve the system defined by the last assembled Jacobian and the rhs vector specified here. Solution is returned in the vector result. More...
 
unsigned iterations () const
 Number of iterations taken. More...
 
void set_preconditioner_LHS ()
 Set left preconditioning (the default) More...
 
void set_preconditioner_RHS ()
 Enable right preconditioning. More...
 
- Public Member Functions inherited from oomph::IterativeLinearSolver
 IterativeLinearSolver ()
 Constructor: Set (default) trivial preconditioner and set defaults for tolerance and max. number of iterations. More...
 
 IterativeLinearSolver (const IterativeLinearSolver &)
 Broken copy constructor. More...
 
void operator= (const IterativeLinearSolver &)
 Broken assignment operator. More...
 
virtual ~IterativeLinearSolver ()
 Destructor (empty) More...
 
Preconditioner *& preconditioner_pt ()
 Access function to preconditioner. More...
 
Preconditioner *const & preconditioner_pt () const
 Access function to preconditioner (const version) More...
 
double & tolerance ()
 Access to convergence tolerance. More...
 
unsigned & max_iter ()
 Access to max. number of iterations. More...
 
void enable_doc_convergence_history ()
 Enable documentation of the convergence history. More...
 
void disable_doc_convergence_history ()
 Disable documentation of the convergence history. More...
 
void open_convergence_history_file_stream (const std::string &file_name, const std::string &zone_title="")
 Write convergence history into file with specified filename (automatically switches on doc). Optional second argument is a string that can be used (as a zone title) to identify what case we're running (e.g. what combination of linear solver and preconditioner or parameter values are used). More...
 
void close_convergence_history_file_stream ()
 Close convergence history output stream. More...
 
double jacobian_setup_time () const
 returns the time taken to assemble the jacobian matrix and residual vector More...
 
double linear_solver_solution_time () const
 return the time taken to solve the linear system More...
 
virtual double preconditioner_setup_time () const
 returns the the time taken to setup the preconditioner More...
 
void enable_setup_preconditioner_before_solve ()
 Setup the preconditioner before the solve. More...
 
void disable_setup_preconditioner_before_solve ()
 Don't set up the preconditioner before the solve. More...
 
void enable_error_after_max_iter ()
 Throw an error if we don't converge within max_iter. More...
 
void disable_error_after_max_iter ()
 Don't throw an error if we don't converge within max_iter (default). More...
 
void enable_iterative_solver_as_preconditioner ()
 
void disable_iterative_solver_as_preconditioner ()
 
- Public Member Functions inherited from oomph::LinearSolver
 LinearSolver ()
 Empty constructor, initialise the member data. More...
 
 LinearSolver (const LinearSolver &dummy)
 Broken copy constructor. More...
 
void operator= (const LinearSolver &)
 Broken assignment operator. More...
 
virtual ~LinearSolver ()
 Empty virtual destructor. More...
 
void enable_doc_time ()
 Enable documentation of solve times. More...
 
void disable_doc_time ()
 Disable documentation of solve times. More...
 
bool is_doc_time_enabled () const
 Is documentation of solve times enabled? More...
 
bool is_resolve_enabled () const
 Boolean flag indicating if resolves are enabled. More...
 
virtual void enable_resolve ()
 Enable resolve (i.e. store matrix and/or LU decomposition, say) Virtual so it can be overloaded to perform additional tasks. More...
 
virtual void enable_computation_of_gradient ()
 function to enable the computation of the gradient required for the globally convergent Newton method More...
 
void disable_computation_of_gradient ()
 function to disable the computation of the gradient required for the globally convergent Newton method More...
 
void reset_gradient ()
 function to reset the size of the gradient before each Newton solve More...
 
void get_gradient (DoubleVector &gradient)
 function to access the gradient, provided it has been computed More...
 
- Public Member Functions inherited from oomph::DistributableLinearAlgebraObject
 DistributableLinearAlgebraObject ()
 Default constructor - create a distribution. More...
 
 DistributableLinearAlgebraObject (const DistributableLinearAlgebraObject &matrix)
 Broken copy constructor. More...
 
void operator= (const DistributableLinearAlgebraObject &)
 Broken assignment operator. More...
 
virtual ~DistributableLinearAlgebraObject ()
 Destructor. More...
 
LinearAlgebraDistributiondistribution_pt () const
 access to the LinearAlgebraDistribution More...
 
unsigned nrow () const
 access function to the number of global rows. More...
 
unsigned nrow_local () const
 access function for the num of local rows on this processor. More...
 
unsigned nrow_local (const unsigned &p) const
 access function for the num of local rows on this processor. More...
 
unsigned first_row () const
 access function for the first row on this processor More...
 
unsigned first_row (const unsigned &p) const
 access function for the first row on this processor More...
 
bool distributed () const
 distribution is serial or distributed More...
 
bool distribution_built () const
 
void build_distribution (const LinearAlgebraDistribution *const dist_pt)
 setup the distribution of this distributable linear algebra object More...
 
void build_distribution (const LinearAlgebraDistribution &dist)
 setup the distribution of this distributable linear algebra object More...
 
- Public Member Functions inherited from oomph::BlockPreconditioner< MATRIX >
 BlockPreconditioner ()
 Constructor. More...
 
virtual ~BlockPreconditioner ()
 Destructor. More...
 
 BlockPreconditioner (const BlockPreconditioner &)
 Broken copy constructor. More...
 
void operator= (const BlockPreconditioner &)
 Broken assignment operator. More...
 
MATRIX * matrix_pt () const
 Access function to matrix_pt. If this is the master then cast the matrix pointer to MATRIX*, error check and return. Otherwise ask the master for its matrix pointer. More...
 
void turn_on_recursive_debug_flag ()
 Toggles on the recursive debug flag. The change goes up the block preconditioning hierarchy. More...
 
void turn_off_recursive_debug_flag ()
 Toggles off the recursive debug flag. The change goes up the block preconditioning hierarchy. More...
 
void turn_on_debug_flag ()
 Toggles on the debug flag. More...
 
void turn_off_debug_flag ()
 Toggles off the debug flag. More...
 
void turn_into_subsidiary_block_preconditioner (BlockPreconditioner< MATRIX > *master_block_prec_pt, const Vector< unsigned > &doftype_in_master_preconditioner_coarse)
 Function to turn this preconditioner into a subsidiary preconditioner that operates within a bigger "master block preconditioner (e.g. a Navier-Stokes 2x2 block preconditioner dealing with the fluid sub-blocks within a 3x3 FSI preconditioner. Once this is done the master block preconditioner deals with the block setup etc. The vector doftype_in_master_preconditioner_coarse must specify the dof number in the master preconditioner that corresponds to a dof number in this preconditioner. 1. The length of the vector is used to determine the number of blocks in this preconditioner therefore it must be correctly sized. 2. block_setup(...) should be called in the master preconditioner before this method is called. 3. block_setup(...) should be called in the corresponding subsidiary preconditioner after this method is called. More...
 
void turn_into_subsidiary_block_preconditioner (BlockPreconditioner< MATRIX > *master_block_prec_pt, const Vector< unsigned > &doftype_in_master_preconditioner_coarse, const Vector< Vector< unsigned > > &doftype_coarsen_map_coarse)
 Function to turn this preconditioner into a subsidiary preconditioner that operates within a bigger "master block preconditioner (e.g. a Navier-Stokes 2x2 block preconditioner dealing with the fluid sub-blocks within a 3x3 FSI preconditioner. Once this is done the master block preconditioner deals with the block setup etc. The vector doftype_in_master_preconditioner_coarse must specify the dof number in the master preconditioner that corresponds to a dof number in this preconditioner. 1. The length of the vector is used to determine the number of blocks in this preconditioner therefore it must be correctly sized. 2. block_setup(...) should be called in the master preconditioner before this method is called. 3. block_setup(...) should be called in the corresponding subsidiary preconditioner after this method is called. More...
 
virtual void block_setup ()
 Determine the size of the matrix blocks and setup the lookup schemes relating the global degrees of freedom with their "blocks" and their indices (row/column numbers) in those blocks. The distributions of the preconditioner and the internal blocks are automatically specified (and assumed to be uniform) at this stage. This method should be used if the identity dof-to-block mapping is okay, i.e. dof number 0 corresponds to block number 0 dof number 1 corresponds to block number 1 dof number 2 corresponds to block number 2 etc... More...
 
void block_setup (const Vector< unsigned > &dof_to_block_map)
 Determine the size of the matrix blocks and setup the lookup schemes relating the global degrees of freedom with their "blocks" and their indices (row/column numbers) in those blocks. The distributions of the preconditioner and the blocks are automatically specified (and assumed to be uniform) at this stage. This method should be used if anything other than the identity dof-to-block mapping is required. The argument vector dof_to_block_map should be of length ndof. The indices represents the dof types whilst the value represents the block types. In general we want: More...
 
void get_block (const unsigned &i, const unsigned &j, MATRIX &output_matrix, const bool &ignore_replacement_block=false) const
 Put block (i,j) into output_matrix. This block accounts for any coarsening of dof types and any replaced dof-level blocks above this preconditioner. More...
 
MATRIX get_block (const unsigned &i, const unsigned &j, const bool &ignore_replacement_block=false) const
 Return block (i,j). If the optional argument ignore_replacement_block is true, then any blocks in Replacement_dof_block_pt will be ignored throughout the preconditioning hierarchy. More...
 
void set_master_matrix_pt (MATRIX *in_matrix_pt)
 Set the matrix_pt in the upper-most master preconditioner. More...
 
void get_block_other_matrix (const unsigned &i, const unsigned &j, MATRIX *in_matrix_pt, MATRIX &output_matrix)
 Get a block from a different matrix using the blocking scheme that has already been set up. More...
 
void get_blocks (DenseMatrix< bool > &required_blocks, DenseMatrix< MATRIX * > &block_matrix_pt) const
 Get all the block matrices required by the block preconditioner. Takes a pointer to a matrix of bools that indicate if a specified sub-block is required for the preconditioning operation. Computes the required block matrices, and stores pointers to them in the matrix block_matrix_pt. If an entry in block_matrix_pt is equal to NULL on return, that sub-block has not been requested and is therefore not available. More...
 
void get_dof_level_block (const unsigned &i, const unsigned &j, MATRIX &output_block, const bool &ignore_replacement_block=false) const
 Gets dof-level block (i,j). If Replacement_dof_block_pt(i,j) is not null, then the replacement block is returned via a deep copy. More...
 
MATRIX get_concatenated_block (const VectorMatrix< BlockSelector > &selected_block)
 Returns a concatenation of the block matrices specified by the argument selected_block. The VectorMatrix selected_block must be correctly sized as it is used to determine the number of sub block matrices to concatenate. More...
 
void get_concatenated_block_vector (const Vector< unsigned > &block_vec_number, const DoubleVector &v, DoubleVector &b)
 Takes the naturally ordered vector and extracts the blocks indicated by the block number (the values) in the Vector block_vec_number all at once, then concatenates them without communication. Here, the values in block_vec_number is the block number in the current preconditioner. This is a non-const function because distributions may be created and stored in Auxiliary_block_distribution_pt for future use. More...
 
void return_concatenated_block_vector (const Vector< unsigned > &block_vec_number, const DoubleVector &b, DoubleVector &v) const
 Takes concatenated block ordered vector, b, and copies its entries to the appropriate entries in the naturally ordered vector, v. Here the values in block_vec_number indicates which blocks the vector b is a concatenation of. The block number are those in the current preconditioner. If the preconditioner is a subsidiary block preconditioner the other entries in v that are not associated with it are left alone. More...
 
void get_block_vectors (const Vector< unsigned > &block_vec_number, const DoubleVector &v, Vector< DoubleVector > &s) const
 Takes the naturally ordered vector and rearranges it into a vector of sub vectors corresponding to the blocks, so s[b][i] contains the i-th entry in the vector associated with block b. Note: If the preconditioner is a subsidiary preconditioner then only the sub-vectors associated with the blocks of the subsidiary preconditioner will be included. Hence the length of v is master_nrow() whereas the total length of the s vectors is the sum of the lengths of the individual block vectors defined in block_vec_number. More...
 
void get_block_vectors (const DoubleVector &v, Vector< DoubleVector > &s) const
 Takes the naturally ordered vector and rearranges it into a vector of sub vectors corresponding to the blocks, so s[b][i] contains the i-th entry in the vector associated with block b. Note: If the preconditioner is a subsidiary preconditioner then only the sub-vectors associated with the blocks of the subsidiary preconditioner will be included. Hence the length of v is master_nrow() whereas the total length of the s vectors is Nrow. This is simply a wrapper around the other get_block_vectors(...) function where the block_vec_number Vector is the identity, i.e. block_vec_number is [0, 1, ..., nblock_types - 1]. More...
 
void return_block_vectors (const Vector< unsigned > &block_vec_number, const Vector< DoubleVector > &s, DoubleVector &v) const
 Takes the vector of block vectors, s, and copies its entries into the naturally ordered vector, v. If this is a subsidiary block preconditioner only those entries in v that are associated with its blocks are affected. The block_vec_number indicates which block the vectors in s came from. The block number corresponds to the block numbers in this preconditioner. More...
 
void return_block_vectors (const Vector< DoubleVector > &s, DoubleVector &v) const
 Takes the vector of block vectors, s, and copies its entries into the naturally ordered vector, v. If this is a subsidiary block preconditioner only those entries in v that are associated with its blocks are affected. The block_vec_number indicates which block the vectors in s came from. The block number corresponds to the block numbers in this preconditioner. This is simply a wrapper around the other return_block_vectors(...) function where the block_vec_number Vector is the identity, i.e. block_vec_number is [0, 1, ..., nblock_types - 1]. More...
 
void get_block_vector (const unsigned &n, const DoubleVector &v, DoubleVector &b) const
 Takes the naturally ordered vector, v and returns the n-th block vector, b. Here n is the block number in the current preconditioner. More...
 
void return_block_vector (const unsigned &n, const DoubleVector &b, DoubleVector &v) const
 Takes the n-th block ordered vector, b, and copies its entries to the appropriate entries in the naturally ordered vector, v. Here n is the block number in the current block preconditioner. If the preconditioner is a subsidiary block preconditioner the other entries in v that are not associated with it are left alone. More...
 
void get_block_ordered_preconditioner_vector (const DoubleVector &v, DoubleVector &w)
 Given the naturally ordered vector, v, return the vector rearranged in block order in w. This function calls get_concatenated_block_vector(...) with the identity block mapping. More...
 
void return_block_ordered_preconditioner_vector (const DoubleVector &w, DoubleVector &v) const
 Takes the block ordered vector, w, and reorders it in natural order. Reordered vector is returned in v. Note: If the preconditioner is a subsidiary preconditioner then only the components of the vector associated with the blocks of the subsidiary preconditioner will be included. Hence the length of v is master_nrow() whereas that of the vector w is of length this->nrow(). More...
 
unsigned nblock_types () const
 Return the number of block types. More...
 
unsigned ndof_types () const
 Return the total number of DOF types. More...
 
const Meshmesh_pt (const unsigned &i) const
 Access to i-th mesh (of the various meshes that contain block preconditionable elements of the same number of dof type). More...
 
unsigned nmesh () const
 Return the number of meshes in Mesh_pt. More...
 
int block_number (const unsigned &i_dof) const
 Return the block number corresponding to a global index i_dof. More...
 
int index_in_block (const unsigned &i_dof) const
 Given a global dof number, returns the index in the block it belongs to. This is the overall index, not local block (in parallel). More...
 
const LinearAlgebraDistributionblock_distribution_pt (const unsigned &b) const
 Access function to the block distributions (const version). More...
 
LinearAlgebraDistributionblock_distribution_pt (const unsigned b)
 Access function to the block distributions (non-const version). More...
 
LinearAlgebraDistributiondof_block_distribution_pt (const unsigned &b)
 Access function to the dof-level block distributions. More...
 
const LinearAlgebraDistributionmaster_distribution_pt () const
 Access function to the distribution of the master preconditioner. If this preconditioner does not have a master preconditioner then the distribution of this preconditioner is returned. More...
 
unsigned ndof_types_in_mesh (const unsigned &i) const
 Return the number of DOF types in mesh i. WARNING: This should only be used by the upper-most master block preconditioner. An error is thrown is this function is called from a subsidiary preconditioner. They (and since every block preconditioner can in principle be used as s subsidiary preconditioner: all block preconditioners) should store local copies of "their meshes" (if they're needed for anything) More...
 
bool is_subsidiary_block_preconditioner () const
 Return true if this preconditioner is a subsidiary preconditioner. More...
 
bool is_master_block_preconditioner () const
 Return true if this preconditioner is the master block preconditioner. More...
 
void set_block_output_to_files (const std::string &basefilename)
 Set the base part of the filename to output blocks to. If it is set then all blocks will be output at the end of block_setup. If it is left empty nothing will be output. More...
 
void disable_block_output_to_files ()
 Turn off output of blocks (by clearing the basefilename string). More...
 
bool block_output_on () const
 Test if output of blocks is on or not. More...
 
void output_blocks_to_files (const std::string &basefilename, const unsigned &precision=8) const
 
void post_block_matrix_assembly_partial_clear ()
 A helper method to reduce the memory requirements of block preconditioners. Once the methods get_block(...), get_blocks(...) and build_preconditioner_matrix(...) have been called in this and all subsidiary block preconditioners this method can be called to clean up. More...
 
BlockPreconditioner< MATRIX > * master_block_preconditioner_pt () const
 Access function to the master block preconditioner pt. More...
 
void clear_block_preconditioner_base ()
 Clears all BlockPreconditioner data. Called by the destructor and the block_setup(...) methods. More...
 
void document ()
 debugging method to document the setup. Should only be called after block_setup(...). More...
 
Vector< Vector< unsigned > > doftype_coarsen_map_fine () const
 Access function for the Doftype_coarsen_map_fine variable. More...
 
Vector< unsigned > get_fine_grain_dof_types_in (const unsigned &i) const
 Returns the most fine grain dof types in a (possibly coarsened) dof type. More...
 
unsigned nfine_grain_dof_types_in (const unsigned &i) const
 Access function for the number of most fine grain dof types in a (possibly coarsened) dof type. More...
 
MapMatrix< unsigned,
CRDoubleMatrix * > 
replacement_dof_block_pt () const
 Access function to the replaced dof-level blocks. More...
 
void setup_matrix_vector_product (MatrixVectorProduct *matvec_prod_pt, CRDoubleMatrix *block_pt, const Vector< unsigned > &block_col_indices)
 Setup a matrix vector product. matvec_prod_pt is a pointer to the MatrixVectorProduct, block_pt is a pointer to the block matrix, block_col_indices is a vector indicating which block indices does the RHS vector we want to multiply the matrix by. More...
 
void setup_matrix_vector_product (MatrixVectorProduct *matvec_prod_pt, CRDoubleMatrix *block_pt, const unsigned &block_col_index)
 Setup matrix vector product. This is simply a wrapper around the other setup_matrix_vector_product function. More...
 
void internal_get_block_ordered_preconditioner_vector (const DoubleVector &v, DoubleVector &w) const
 Given the naturally ordered vector, v, return the vector rearranged in block order in w. This is a legacy function from the old block preconditioning framework. Kept alive in case it may be needed again. More...
 
void internal_return_block_ordered_preconditioner_vector (const DoubleVector &w, DoubleVector &v) const
 Takes the block ordered vector, w, and reorders it in the natural order. Reordered vector is returned in v. Note: If the preconditioner is a subsidiary preconditioner then only the components of the vector associated with the blocks of the subsidiary preconditioner will be included. Hence the length of v is master_nrow() whereas that of the vector w is of length this->nrow(). More...
 
unsigned internal_nblock_types () const
 Return the number internal blocks. This should be the same as the number of internal dof types. Internally, the block preconditioning framework always work with the most fine grain blocks. I.e. it always deal with the most fine grain dof-level blocks. This allows for coarsening of dof types. When we extract a block, we look at the Block_to_dof_map_fine vector to find out which most fine grain dof types belongs to this block. More...
 
unsigned internal_ndof_types () const
 Return the number of internal dof types. This is the number of most fine grain dof types. The preconditioner writer should not have to concern him/her-self with the internal dof/block types. Thus this fuction is moved to private. We have kept this function alive since it it still used deep within the inner workings of the block preconditioning framework. More...
 
void internal_return_block_vector (const unsigned &n, const DoubleVector &b, DoubleVector &v) const
 Takes the n-th block ordered vector, b, and copies its entries to the appropriate entries in the naturally ordered vector, v. Here n is the block number in the current block preconditioner. If the preconditioner is a subsidiary block preconditioner the other entries in v that are not associated with it are left alone. More...
 
void internal_get_block_vector (const unsigned &n, const DoubleVector &v, DoubleVector &b) const
 A helper function, takes the naturally ordered vector, v, and extracts the n-th block vector, b. Here n is the block number in the current preconditioner. NOTE: The ordering of the vector b is the same as the ordering of the block matrix from internal_get_block(...). More...
 
void internal_get_block_vectors (const Vector< unsigned > &block_vec_number, const DoubleVector &v, Vector< DoubleVector > &s) const
 Takes the naturally ordered vector and rearranges it into a vector of sub vectors corresponding to the blocks, so s[b][i] contains the i-th entry in the vector associated with block b. The block_vec_number indicates which blocks we want. These blocks and vectors are those corresponding to the internal blocks. Note: If the preconditioner is a subsidiary preconditioner then only the sub-vectors associated with the blocks of the subsidiary preconditioner will be included. Hence the length of v is master_nrow() whereas the total length of the s vectors is the sum of the Nrow of the sub vectors. More...
 
void internal_get_block_vectors (const DoubleVector &v, Vector< DoubleVector > &s) const
 A helper function, takes the naturally ordered vector and rearranges it into a vector of sub vectors corresponding to the blocks, so s[b][i] contains the i-th entry in the vector associated with block b. The block_vec_number indicates which blocks we want. These blocks and vectors are those corresponding to the internal blocks. Note: If the preconditioner is a subsidiary preconditioner then only the sub-vectors associated with the blocks of the subsidiary preconditioner will be included. Hence the length of v is master_nrow() whereas the total length of the s vectors is the sum of the Nrow of the sub vectors. This is simply a wrapper around the other internal_get_block_vectors(...) function with the identity block_vec_number vector. More...
 
void internal_return_block_vectors (const Vector< unsigned > &block_vec_number, const Vector< DoubleVector > &s, DoubleVector &v) const
 A helper function, takes the vector of block vectors, s, and copies its entries into the naturally ordered vector, v. If this is a subsidiary block preconditioner only those entries in v that are associated with its blocks are affected. More...
 
void internal_return_block_vectors (const Vector< DoubleVector > &s, DoubleVector &v) const
 A helper function, takes the vector of block vectors, s, and copies its entries into the naturally ordered vector, v. If this is a subsidiary block preconditioner only those entries in v that are associated with its blocks are affected. This is simple a wrapper around the other internal_return_block_vectors(...) function with the identity block_vec_number vector. More...
 
void internal_get_block (const unsigned &i, const unsigned &j, MATRIX &output_block) const
 Gets block (i,j) from the matrix pointed to by Matrix_pt and returns it in output_block. This is associated with the internal blocks. Please use the other get_block(...) function. More...
 
int internal_block_number (const unsigned &i_dof) const
 Return the block number corresponding to a global index i_dof. This returns the block number corresponding to the internal blocks. What this means is that this returns the most fine grain dof-block number which this global index i_dof corresponds to. Since the writer of the preconditioner does not need to care about the internal block types, this function should not be used and thus moved to private. This function should not be removed since it is still used deep within the inner workings of the block preconditioning framework. More...
 
int internal_index_in_block (const unsigned &i_dof) const
 Return the index in the block corresponding to a global block number i_dof. The index returned corresponds to the internal blocks, which is the most fine grain dof blocks. More...
 
const LinearAlgebraDistributioninternal_block_distribution_pt (const unsigned &b) const
 Access function to the internal block distributions. More...
 
void insert_auxiliary_block_distribution (const Vector< unsigned > &block_vec_number, LinearAlgebraDistribution *dist_pt)
 insert a Vector<unsigned> and LinearAlgebraDistribution* pair into Auxiliary_block_distribution_pt. The Auxiliary_block_distribution_pt should only contain pointers to distributions concatenated at this block level. We try to ensure this by checking if the block_vec_number vector is within the range nblock_types(). Of course, this does not guarantee correctness, but this is the least we can do. More...
 
void block_matrix_test (const unsigned &i, const unsigned &j, const MATRIX *block_matrix_pt) const
 Private helper function to check that every element in the block matrix (i,j) matches the corresponding element in the original matrix. More...
 
template<typename myType >
int get_index_of_value (const Vector< myType > &vec, const myType val, const bool sorted=false) const
 Get the index of first occurrence of value in a vector. If the element does not exist, -1 is returned. The optional parameter indicates of the Vector is sorted or not. Complexity: if the Vector is sorted, then on average, logarithmic in the distance between first and last: Performs approximately log2(N)+2 element comparisons. Otherwise, up to linear in the distance between first and last: Compares elements until a match is found. More...
 
template<>
void internal_get_block (const unsigned &block_i, const unsigned &block_j, CRDoubleMatrix &output_block) const
 
template<>
void get_dof_level_block (const unsigned &block_i, const unsigned &block_j, CRDoubleMatrix &output_block, const bool &ignore_replacement_block) const
 
- Public Member Functions inherited from oomph::Preconditioner
 Preconditioner ()
 Constructor. More...
 
 Preconditioner (const Preconditioner &)
 Broken copy constructor. More...
 
void operator= (const Preconditioner &)
 Broken assignment operator. More...
 
virtual ~Preconditioner ()
 Destructor (empty) More...
 
void setup (DoubleMatrixBase *matrix_pt)
 Setup the preconditioner: store the matrix pointer and the communicator pointer then call preconditioner specific setup() function. More...
 
void setup (const Problem *problem_pt, DoubleMatrixBase *matrix_pt)
 Compatability layer for old preconditioners where problem pointers were needed. The problem pointer is only used to get a communicator pointer. More...
 
virtual void set_matrix_pt (DoubleMatrixBase *matrix_pt)
 Set the matrix pointer. More...
 
virtual const OomphCommunicatorcomm_pt () const
 Get function for comm pointer. More...
 
virtual void set_comm_pt (const OomphCommunicator *const comm_pt)
 Set the communicator pointer. More...
 
double setup_time () const
 Returns the time to setup the preconditioner. More...
 
virtual void turn_into_subsidiary_block_preconditioner (BlockPreconditioner< CRDoubleMatrix > *master_block_prec_pt, const Vector< unsigned > &doftype_in_master_preconditioner_coarse)
 
virtual void turn_into_subsidiary_block_preconditioner (BlockPreconditioner< CRDoubleMatrix > *master_block_prec_pt, const Vector< unsigned > &doftype_in_master_preconditioner_coarse, const Vector< Vector< unsigned > > &doftype_coarsen_map_coarse)
 

Private Member Functions

void solve_helper (DoubleMatrixBase *const &matrix_pt, const DoubleVector &rhs, DoubleVector &solution)
 General interface to solve function. More...
 
void update (const unsigned &k, const Vector< Vector< std::complex< double > > > &hessenberg, const Vector< std::complex< double > > &s, const Vector< Vector< DoubleVector > > &z_m, Vector< DoubleVector > &x)
 Helper function to update the result vector. More...
 

Additional Inherited Members

- Protected Member Functions inherited from oomph::HelmholtzGMRESMG< MATRIX >
void solve_helper (DoubleMatrixBase *const &matrix_pt, const DoubleVector &rhs, DoubleVector &solution)
 General interface to solve function. More...
 
void clean_up_memory ()
 Cleanup data that's stored for resolve (if any has been stored) More...
 
void complex_matrix_multiplication (Vector< CRDoubleMatrix * > const matrices_pt, const Vector< DoubleVector > &x, Vector< DoubleVector > &soln)
 Helper function to calculate a complex matrix-vector product. Assumes the matrix has been provided as a Vector of length two; the first entry containing the real part of the system matrix and the second entry containing the imaginary part. More...
 
void update (const unsigned &k, const Vector< Vector< std::complex< double > > > &hessenberg, const Vector< std::complex< double > > &s, const Vector< Vector< DoubleVector > > &v, Vector< DoubleVector > &x)
 Helper function to update the result vector. More...
 
void generate_plane_rotation (std::complex< double > &dx, std::complex< double > &dy, std::complex< double > &cs, std::complex< double > &sn)
 Helper function: Generate a plane rotation. This is done by finding the value of $ \cos(\theta) $ (i.e. cs) and the value of $ \sin(\theta) $ (i.e. sn) such that:

\[ \begin{bmatrix} \overline{\cos\theta} & \overline{\sin\theta} \\ -\sin\theta & \cos\theta \end{bmatrix} \begin{bmatrix} dx \\ dy \end{bmatrix} = \begin{bmatrix} r \\ 0 \end{bmatrix}, \]

where $ r=\sqrt{pow(|dx|,2)+pow(|dy|,2)} $. The values of a and b are given by: The values of dx and dy are given by:

\[ \cos\theta&=\dfrac{dx}{\sqrt{|dx|^2+|dy|^2}}, \]

and

\[ \sin\theta&=\dfrac{dy}{\sqrt{|dx|^2+|dy|^2}}. \]

Taken from: Saad Y."Iterative methods for sparse linear systems", p.193. We also check to see that sn is always a real (nonnegative) number. See pp.193-194 for an explanation. More...

 
void apply_plane_rotation (std::complex< double > &dx, std::complex< double > &dy, std::complex< double > &cs, std::complex< double > &sn)
 Helper function: Apply plane rotation. This is done using the update:

\[ \begin{bmatrix} dx \\ dy \end{bmatrix} \leftarrow \begin{bmatrix} \overline{\cos\theta} & \overline{\sin\theta} \\ -\sin\theta & \cos\theta \end{bmatrix} \begin{bmatrix} dx \\ dy \end{bmatrix}. \]

Taken from: Saad Y."Iterative methods for sparse linear systems", p.193. More...

 
- Protected Member Functions inherited from oomph::DistributableLinearAlgebraObject
void clear_distribution ()
 clear the distribution of this distributable linear algebra object More...
 
- Protected Member Functions inherited from oomph::BlockPreconditioner< MATRIX >
void set_nmesh (const unsigned &n)
 Specify the number of meshes required by this block preconditioner. Note: elements in different meshes correspond to different types of DOF. More...
 
void set_mesh (const unsigned &i, const Mesh *const mesh_pt, const bool &allow_multiple_element_type_in_mesh=false)
 Set the i-th mesh for this block preconditioner. Note: The method set_nmesh(...) must be called before this method to specify the number of meshes. By default, it is assumed that each mesh only contains elements of the same type. This condition may be relaxed by setting the boolean allow_multiple_element_type_in_mesh to true, however, each mesh must only contain elements with the same number of dof types. More...
 
void set_replacement_dof_block (const unsigned &block_i, const unsigned &block_j, CRDoubleMatrix *replacement_dof_block_pt)
 Set replacement dof-level blocks. Only dof-level blocks can be set. This is important due to how the dof type coarsening feature operates. More...
 
bool any_mesh_distributed () const
 Check if any of the meshes are distributed. This is equivalent to problem.distributed() and is used as a replacement. More...
 
int internal_dof_number (const unsigned &i_dof) const
 Return the number of the block associated with global unknown i_dof. If this preconditioner is a subsidiary block preconditioner then the block number in the subsidiary block preconditioner is returned. If a particular global DOF is not associated with this preconditioner then -1 is returned. More...
 
unsigned internal_index_in_dof (const unsigned &i_dof) const
 Return the row/column number of global unknown i_dof within it's block. More...
 
unsigned internal_block_dimension (const unsigned &b) const
 Return the number of degrees of freedom in block b. Note that if this preconditioner acts as a subsidiary preconditioner then b refers to the block number in the subsidiary preconditioner not the master block preconditioner. More...
 
unsigned internal_dof_block_dimension (const unsigned &i) const
 Return the size of the dof "block" i, i.e. how many degrees of freedom are associated with it. Note that if this preconditioner acts as a subsidiary preconditioner, then i refers to the block number in the subsidiary preconditioner not the master block preconditioner. More...
 
unsigned master_nrow () const
 Return the number of dofs (number of rows or columns) in the overall problem. The prefix "master_" is sort of redundant when used as a stand-alone block preconditioner but is required to avoid ambiguities. The latter is stored (and maintained) separately for each specific block preconditioner regardless of its role. More...
 
unsigned internal_master_dof_number (const unsigned &b) const
 Takes the block number within this preconditioner and returns the corresponding block number in the master preconditioner. If this preconditioner does not have a master block preconditioner then the block number passed is returned. More...
 
const LinearAlgebraDistributioninternal_preconditioner_matrix_distribution_pt () const
 access function to the internal preconditioner matrix distribution pt. preconditioner_matrix_distribution_pt always returns the concatenation of the internal block distributions. Since the writer of the preconditioner does not need to concern themselves with the internal dof/block, please use preconditioner_matrix_distribution_pt(). More...
 
const LinearAlgebraDistributionpreconditioner_matrix_distribution_pt () const
 Access function to the preconditioner matrix distribution pointer. This is the concatenation of the block distributions with the identity ordering. I.e. if this preconditioner has three block types, with the three associated block distributions dist_b0, dist_b1 and dist_b2, then this distribution is: LinearAlgebraDistributionHelpers::concatenate(dist_b0, dist_b1, dist_b2). More...
 
- Protected Attributes inherited from oomph::HelmholtzGMRESMG< MATRIX >
unsigned Iterations
 Number of iterations taken. More...
 
Vector< CRDoubleMatrix * > Matrices_storage_pt
 Vector of pointers to the real and imaginary part of the system matrix. More...
 
bool Resolving
 Boolean flag to indicate if the solve is done in re-solve mode, bypassing setup of matrix and preconditioner. More...
 
bool Matrix_can_be_deleted
 Boolean flag to indicate if the matrix pointed to be Matrix_pt can be deleted. More...
 
bool Preconditioner_LHS
 boolean indicating use of left hand preconditioning (if true) or right hand preconditioning (if false) More...
 
- Protected Attributes inherited from oomph::IterativeLinearSolver
bool Doc_convergence_history
 Flag indicating if the convergence history is to be documented. More...
 
std::ofstream Output_file_stream
 Output file stream for convergence history. More...
 
double Tolerance
 Convergence tolerance. More...
 
unsigned Max_iter
 Maximum number of iterations. More...
 
PreconditionerPreconditioner_pt
 Pointer to the preconditioner. More...
 
double Jacobian_setup_time
 Jacobian setup time. More...
 
double Solution_time
 linear solver solution time More...
 
double Preconditioner_setup_time
 Preconditioner setup time. More...
 
bool Setup_preconditioner_before_solve
 indicates whether the preconditioner should be setup before solve. Default = true; More...
 
bool Throw_error_after_max_iter
 Should we throw an error instead of just returning when we hit the max iterations? More...
 
bool Use_iterative_solver_as_preconditioner
 Use the iterative solver as preconditioner. More...
 
bool First_time_solve_when_used_as_preconditioner
 
- Protected Attributes inherited from oomph::LinearSolver
bool Enable_resolve
 Boolean that indicates whether the matrix (or its factors, in the case of direct solver) should be stored so that the resolve function can be used. More...
 
bool Doc_time
 Boolean flag that indicates whether the time taken. More...
 
bool Compute_gradient
 flag that indicates whether the gradient required for the globally convergent Newton method should be computed or not More...
 
bool Gradient_has_been_computed
 flag that indicates whether the gradient was computed or not More...
 
DoubleVector Gradient_for_glob_conv_newton_solve
 DoubleVector storing the gradient for the globally convergent Newton method. More...
 
- Protected Attributes inherited from oomph::BlockPreconditioner< MATRIX >
MapMatrix< unsigned,
CRDoubleMatrix * > 
Replacement_dof_block_pt
 The replacement dof-level blocks. More...
 
Vector
< LinearAlgebraDistribution * > 
Block_distribution_pt
 The distribution for the blocks. More...
 
Vector< Vector< unsigned > > Block_to_dof_map_coarse
 Mapping for block types to dof types. These are the dof types the writer of the preconditioner expects. For the upper-most master block preconditioner, this would be the sum of the dof types in the meshes. For subsidiary block preconditioners, this is determined by the parent preconditioner when passing in the doftype_coarsen_map_coarse vector in turn_into_subsidiary_block_preconditioner(...). More...
 
Vector< Vector< unsigned > > Block_to_dof_map_fine
 Mapping for the block types to the most fine grain dof types. More...
 
Vector< Vector< unsigned > > Doftype_coarsen_map_coarse
 Mapping for dof types within THIS precondition. This is usually passed down from the parent preconditioner. This list is used to tell which does types should be considered as a single dof type within this preconditioner. I.e. we "coarsen" the dof types. The values are local to this preconditioner, for example, even if the Doftype_in_master_preconditioner_coarse = [2,3,4], the vector Doftype_coarsen_map_coarse = [[0],[1,2]], saying your local dof types 0 should be considered as dof type 0 and dof types 1 and 2 are considered as dof type 1. More...
 
Vector< Vector< unsigned > > Doftype_coarsen_map_fine
 Mapping the dof types within this preconditioner. The values in here refers to the most grain dof types. This list is automatically generated either in block_setup(...) (for the top-most preconditioner) or the turn_into_subsidiary_block_preconditioner(...) function. Please refer to the comment above Doftype_coarsen_map_coarse for more details. More...
 
Vector
< LinearAlgebraDistribution * > 
Internal_block_distribution_pt
 Storage for the default distribution for each internal block. More...
 
Vector
< LinearAlgebraDistribution * > 
Dof_block_distribution_pt
 Storage for the default distribution for each dof block at this level. More...
 
Vector< unsigned > Allow_multiple_element_type_in_mesh
 Vector of unsigned to indicate which meshes contain multiple element types. More...
 
Vector< const Mesh * > Mesh_pt
 Vector of pointers to the meshes containing the elements used in the block preconditioner. Const pointers to prevent modification of the mesh by the preconditioner (this could be relaxed if needed). If this is a subsidiary preconditioner, then the information is looked up in the master preconditioner. More...
 
Vector< unsigned > Ndof_types_in_mesh
 Storage for number of types of degree of freedom of the elements in each mesh. More...
 
unsigned Internal_nblock_types
 Number of different block types in this preconditioner. Note that this information is maintained if used as a subsidiary or stand-alone block preconditioner, in the latter case it stores the number of blocks within the subsidiary preconditioner. More...
 
unsigned Internal_ndof_types
 Number of different DOF types in this preconditioner. Note that this information is maintained if used as a subsidiary or stand-alone block preconditioner, in the latter case it stores the number of dofs within the subsidiary preconditioner. More...
 
- Static Protected Attributes inherited from oomph::IterativeLinearSolver
static IdentityPreconditioner Default_preconditioner
 Default preconditioner: The base class for preconditioners is a fully functional (if trivial!) preconditioner. More...
 

Detailed Description

template<typename MATRIX>
class oomph::HelmholtzFGMRESMG< MATRIX >

The FGMRES method, i.e. the flexible variant of the GMRES method which allows for nonconstant preconditioners [see Saad Y, "Iterative methods for sparse linear systems", p.287]. Note, FGMRES can only cater to right preconditioning; if the user tries to switch to left preconditioning they will be notified of this.

Definition at line 3272 of file complex_smoother.h.

Constructor & Destructor Documentation

template<typename MATRIX >
oomph::HelmholtzFGMRESMG< MATRIX >::HelmholtzFGMRESMG ( )
inline

Constructor (empty)

Definition at line 3278 of file complex_smoother.h.

References oomph::HelmholtzGMRESMG< MATRIX >::Preconditioner_LHS.

template<typename MATRIX >
virtual oomph::HelmholtzFGMRESMG< MATRIX >::~HelmholtzFGMRESMG ( )
inlinevirtual

Destructor (cleanup storage)

Definition at line 3285 of file complex_smoother.h.

References oomph::HelmholtzGMRESMG< MATRIX >::clean_up_memory().

template<typename MATRIX >
oomph::HelmholtzFGMRESMG< MATRIX >::HelmholtzFGMRESMG ( const HelmholtzFGMRESMG< MATRIX > &  )
inline

Broken copy constructor.

Definition at line 3292 of file complex_smoother.h.

References oomph::BrokenCopy::broken_copy().

Member Function Documentation

template<typename MATRIX >
void oomph::HelmholtzFGMRESMG< MATRIX >::operator= ( const HelmholtzFGMRESMG< MATRIX > &  )
inline

Broken assignment operator.

Definition at line 3298 of file complex_smoother.h.

References oomph::BrokenCopy::broken_assign().

template<typename MATRIX >
void oomph::HelmholtzFGMRESMG< MATRIX >::set_preconditioner_LHS ( )
inline

Overloaded function to let the user know that left preconditioning is not possible with FGMRES, only right preconditioning.

Definition at line 3305 of file complex_smoother.h.

template<typename MATRIX >
void oomph::HelmholtzFGMRESMG< MATRIX >::solve ( Problem *const &  problem_pt,
DoubleVector result 
)
inlinevirtual
template<typename MATRIX >
void oomph::HelmholtzFGMRESMG< MATRIX >::solve_helper ( DoubleMatrixBase *const &  matrix_pt,
const DoubleVector rhs,
DoubleVector solution 
)
private

General interface to solve function.

Linear-algebra-type solver: Takes pointer to a matrix and rhs vector and returns the solution of the linear system. based on the algorithm presented in Templates for the Solution of Linear Systems: Building Blocks for Iterative Methods, Barrett, Berry et al, SIAM, 2006 and the implementation in the IML++ library : http://math.nist.gov/iml++/

Definition at line 3575 of file complex_smoother.h.

References oomph::DoubleVector::build(), oomph::DoubleVector::built(), oomph::LinearAlgebraDistribution::distributed(), oomph::DistributableLinearAlgebraObject::distribution_pt(), i, oomph::DoubleVector::initialise(), oomph::BlackBoxFDNewtonSolver::Max_iter, oomph::DistributableLinearAlgebraObject::nrow(), oomph::oomph_info, s, oomph::Global_string_for_annotation::string(), and oomph::TimingHelpers::timer().

Referenced by oomph::HelmholtzFGMRESMG< MATRIX >::solve().

template<typename MATRIX >
void oomph::HelmholtzFGMRESMG< MATRIX >::update ( const unsigned &  k,
const Vector< Vector< std::complex< double > > > &  hessenberg,
const Vector< std::complex< double > > &  s,
const Vector< Vector< DoubleVector > > &  z_m,
Vector< DoubleVector > &  x 
)
inlineprivate

Helper function to update the result vector.

Definition at line 3471 of file complex_smoother.h.

References oomph::DistributableLinearAlgebraObject::distribution_pt(), i, and s.


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