Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Allow creating subvectors only supplying local indices #3857

Merged
merged 2 commits into from
May 24, 2024
Merged

Allow creating subvectors only supplying local indices #3857

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
254e961
Allow creating subvectors only supplying local indices
lindsayad May 22, 2024
6bdf108
Add doxygen for boolean param
lindsayad May 23, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 9 additions & 5 deletions include/numerics/numeric_vector.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -717,13 +717,17 @@ class NumericVector : public ReferenceCountedObject<NumericVector<T>>,
}

/**
* Fills in \p subvector from this vector using the indices in \p
* rows. Similar to the \p create_submatrix() routine for the
* SparseMatrix class, it is currently only implemented for
* PetscVectors.
* Fills in \p subvector from this vector using the indices in \p rows.
* Similar to the \p create_submatrix() routine for the SparseMatrix class, it
* is currently only implemented for PetscVectors. The boolean parameter
* communicates whether the supplied vector of rows corresponds to all the
* rows that should be used in the subvector's index set, e.g. whether the
* rows correspond to the global collective. If the rows supplied are only the
* local indices, then the boolean parameter should be set to false
*/
virtual void create_subvector(NumericVector<T> & ,
const std::vector<numeric_index_type> &) const
const std::vector<numeric_index_type> &,
bool = true) const
Comment on lines -726 to +730
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you add some documentation for the new parameter? Since it doesn't even have a name here in the base class declaration, it's a bit mysterious as to what it might be for...

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

My very poor excuse for why I didn't do this initially is that I will have to comment out the parameter name to avoid unused warnings. Will a commented out parameter name still show up in the doxygen? Either way I can still articulate something in the doxygen wrt that parameter

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

relevant issue: doxygen/doxygen#6926

{
libmesh_not_implemented();
}
Expand Down
3 changes: 2 additions & 1 deletion include/numerics/petsc_vector.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -327,7 +327,8 @@ class PetscVector final : public NumericVector<T>
virtual void print_matlab(const std::string & name = "") const override;

virtual void create_subvector(NumericVector<T> & subvector,
const std::vector<numeric_index_type> & rows) const override;
const std::vector<numeric_index_type> & rows,
bool supplying_global_rows = true) const override;

virtual void swap (NumericVector<T> & v) override;

Expand Down
3 changes: 0 additions & 3 deletions include/numerics/trilinos_epetra_vector.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -253,9 +253,6 @@ class EpetraVector final : public NumericVector<T>
virtual void pointwise_mult (const NumericVector<T> & vec1,
const NumericVector<T> & vec2) override;

virtual void create_subvector (NumericVector<T> & subvector,
const std::vector<numeric_index_type> & rows) const override;

virtual void swap (NumericVector<T> & v) override;

virtual std::size_t max_allowed_id() const override;
Expand Down
51 changes: 38 additions & 13 deletions src/numerics/petsc_vector.C
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1187,10 +1187,15 @@ void PetscVector<T>::print_matlab (const std::string & name) const

template <typename T>
void PetscVector<T>::create_subvector(NumericVector<T> & subvector,
const std::vector<numeric_index_type> & rows) const
const std::vector<numeric_index_type> & rows,
const bool supplying_global_rows) const
{
parallel_object_only();

libmesh_error_msg_if(
subvector.type() == GHOSTED,
"We do not support scattering parallel information to ghosts for subvectors");

this->_restore_array();

// PETSc data structures
Expand All @@ -1204,21 +1209,32 @@ void PetscVector<T>::create_subvector(NumericVector<T> & subvector,
// If not, we use the appropriate PETSc routines to initialize it.
if (!petsc_subvector->initialized())
{
// Initialize the petsc_subvector to have enough space to hold
// the entries which will be scattered into it. Note: such an
// init() function (where we let PETSc decide the number of local
// entries) is not currently offered by the PetscVector
// class. Should we differentiate here between sequential and
// parallel vector creation based on this->n_processors() ?
ierr = VecCreateMPI(this->comm().get(),
PETSC_DECIDE, // n_local
cast_int<PetscInt>(rows.size()), // n_global
&(petsc_subvector->_vec));
libmesh_assert(petsc_subvector->_type == AUTOMATIC || petsc_subvector->_type == PARALLEL);

if (supplying_global_rows)
// Initialize the petsc_subvector to have enough space to hold
// the entries which will be scattered into it. Note: such an
// init() function (where we let PETSc decide the number of local
// entries) is not currently offered by the PetscVector
// class. Should we differentiate here between sequential and
// parallel vector creation based on this->n_processors() ?
ierr = VecCreateMPI(this->comm().get(),
PETSC_DECIDE, // n_local
cast_int<PetscInt>(rows.size()), // n_global
&(petsc_subvector->_vec));
else
ierr = VecCreateMPI(this->comm().get(),
cast_int<PetscInt>(rows.size()),
PETSC_DETERMINE,
&(petsc_subvector->_vec));
LIBMESH_CHKERR(ierr);

ierr = VecSetFromOptions (petsc_subvector->_vec);
LIBMESH_CHKERR(ierr);

// We created a parallel vector
petsc_subvector->_type = PARALLEL;

// Mark the subvector as initialized
petsc_subvector->_is_initialized = true;
}
Expand All @@ -1227,9 +1243,16 @@ void PetscVector<T>::create_subvector(NumericVector<T> & subvector,
petsc_subvector->_restore_array();
}

// Use iota to fill an array with entries [0,1,2,3,4,...rows.size()]
std::vector<PetscInt> idx(rows.size());
std::iota (idx.begin(), idx.end(), 0);
if (supplying_global_rows)
std::iota (idx.begin(), idx.end(), 0);
else
{
PetscInt start;
ierr = VecGetOwnershipRange(petsc_subvector->_vec, &start, nullptr);
LIBMESH_CHKERR(ierr);
std::iota (idx.begin(), idx.end(), start);
}

// Construct index sets
WrappedPetsc<IS> parent_is;
Expand Down Expand Up @@ -1258,6 +1281,8 @@ void PetscVector<T>::create_subvector(NumericVector<T> & subvector,

// Actually perform the scatter
VecScatterBeginEnd(this->comm(), scatter, this->_vec, petsc_subvector->_vec, INSERT_VALUES, SCATTER_FORWARD);

petsc_subvector->_is_closed = true;
}


Expand Down
9 changes: 0 additions & 9 deletions src/numerics/trilinos_epetra_vector.C
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -576,15 +576,6 @@ void EpetraVector<T>::localize_to_one (std::vector<T> & v_local,
}



template <typename T>
void EpetraVector<T>::create_subvector(NumericVector<T> & /* subvector */,
const std::vector<numeric_index_type> & /* rows */) const
{
libmesh_not_implemented();
}


/*********************************************************************
* The following were copied (and slightly modified) from
* Epetra_FEVector.h in order to allow us to use a standard
Expand Down