petsc_direct_solver.py

LinearSolver that uses PETSc for LU factor/solve.

class openmdao.solvers.linear.petsc_direct_solver.PETScDirectSolver(**kwargs)

Bases: DirectSolver

LinearSolver that uses PETSc for LU factor/solve.

Parameters:

**kwargsdict

Options dictionary.

Attributes:

_PETSc<petsc4py.PETSc>

A lazily imported petsc4py.PETSc module.

Methods

add_recorder(recorder)	Add a recorder to the solver's RecordingManager.
can_solve_cycle()	Return True if this solver can solve groups with cycles.
check_config(logger)	Perform optional error checks.
cleanup()	Clean up resources prior to exit.
does_recursive_applies()	Return False.
get_outputs_dir(*subdirs[, mkdir])	Get the path under which all output files of this solver are to be placed.
get_reports_dir()	Get the path to the directory where the report files should go.
preferred_sparse_format()	Return the preferred sparse format for the dr/do matrix of a split jacobian.
raise_petsc_error(e, system, matrix)	Raise an error based on the issue that PETSc had with the linearize.
record_iteration(**kwargs)	Record an iteration of the current Solver.
report_failure(msg)	Report a failure that has occurred.
solve(mode[, rel_systems])	Run the solver.
use_relevance()	Return True if relevance should be active.

SOLVER = 'LN: PETScDirect'

__init__(**kwargs)

Declare the solver options.

check_config(logger)

Perform optional error checks.

Parameters:

loggerobject

The object that manages logging output.

preferred_sparse_format()

Return the preferred sparse format for the dr/do matrix of a split jacobian.

Returns:

str

The preferred sparse format for the dr/do matrix of a split jacobian.

raise_petsc_error(e, system, matrix)

Raise an error based on the issue that PETSc had with the linearize.

Parameters:

eError

Error returned by PETSc.

system<System>

System containing the Directsolver.

matrixndarray

Matrix of interest.

solve(mode, rel_systems=None)

Run the solver.

Parameters:

modestr

‘fwd’ or ‘rev’.

rel_systemsset of str

Names of systems relevant to the current solve. Deprecated.

class openmdao.solvers.linear.petsc_direct_solver.PETScLU(A: spmatrix, sparse_solver_name: str = None, comm=None)

Bases: object

Wrapper for PETSc LU decomposition, using petsc4py.

Parameters:

Andarray or <scipy.sparse.csc_matrix>

Matrix to use in solving x @ A == b.

sparse_solver_namestr

Name of the direct solver from PETSc to use.

comm<mpi4py.MPI.Intracomm>

The system MPI communicator.

Attributes:

orig_Andarray or <scipy.sparse.csc_matrix>

Originally provided matrix.

A<petsc4py.PETSc.Mat>

Assembled PETSc AIJ (compressed sparse row format) matrix.

ksp<petsc4py.PETSc.KSP>

PETSc Krylov Subspace Solver context.

running_mpibool

Is the script currently being run under MPI (True) or not (False).

comm<mpi4py.MPI.Intracomm>

The system MPI communicator.

_x<petsc4py.PETSc.Vec>

Sequential (non-distributed) PETSc vector to store the solve solution.

_PETSc<petsc4py.PETSc>

Lazily imported petsc4py.PETSc module.

Methods

solve(b[, transpose])

Solve the linear system using only the preconditioner direct solver.

__init__(A: spmatrix, sparse_solver_name: str = None, comm=None)

Initialize and setup the PETSc LU Direct Solver object.

solve(b: ndarray, transpose: bool = False) → ndarray

Solve the linear system using only the preconditioner direct solver.

Parameters:

bndarray

Input data for the right hand side.

transposebool

Is A.T @ x == b being solved (True) or A @ x == b (False).

Returns:

ndarray

The solution array.

openmdao.solvers.linear.petsc_direct_solver.check_err_on_singular(name, value)

Check the value of the “err_on_singular” option on the PETScDirectSolver.

Parameters:

namestr

Name of the option being checked.

valuebool

Value of the option being checked.