MA27 Interface
- class InteriorPointMA27Interface(cntl_options=None, icntl_options=None, iw_factor=1.2, a_factor=2)
Bases:
LinearSolverInterfaceAn interface to HSL’s MA27 routines for use with Parapint’s interior point algorithm. See http://www.hsl.rl.ac.uk/archive/specs/ma27.pdf for details on the use of MA27.
Note
The pivot tolerance, cntl(1), should be selected carefully. Larger values result in better precision but smaller values result in better performance.
- Parameters
- cntl_options: dict
- icntl_options: dict
- iw_factor: float
The factor for memory allocation of the integer working arrays used by MA27. This value is increased by the increase_memory_allocation method.
- a_factor: float
The factor for memory allocation of the A array used by MA28. This value is increased by the increase_memory_allocation_method.
- do_symbolic_factorization(matrix, raise_on_error=True, timer=None)
Perform symbolic factorization. This calls the MA27A/AD routines.
- Parameters
- matrix: scipy.sparse.spmatrix or pyomo.contrib.pynumero.sparse.block_matrix.BlockMatrix
The matrix to factorize
- raise_on_error: bool
If False, an error will not be raised if an error occurs during symbolic factorization. Instead the status attribute of the results object will indicate an error ocurred.
- timer: HierarchicalTimer
- Returns
- res: LinearSolverResults
A LinearSolverResults object with a status attribute for the LinearSolverStatus
- do_numeric_factorization(matrix, raise_on_error=True, timer=None)
Perform numeric factorization. This calls the MA27B/BD routines.
- Parameters
- matrix: scipy.sparse.spmatrix or pyomo.contrib.pynumero.sparse.block_matrix.BlockMatrix
The matrix to factorize
- raise_on_error: bool
If False, an error will not be raised if an error occurs during numeric factorization. Instead the status attribute of the results object will indicate an error ocurred.
- timer: HierarchicalTimer
- Returns
- res: LinearSolverResults
A LinearSolverResults object with a status attribute for the LinearSolverStatus
- increase_memory_allocation(factor)
Increas the memory allocation for factorization. This method should only be called if the results status from do_symbolic_factorization or do_numeric_factorization is LinearSolverStatus.not_enough_memory.
- Parameters
- factor: float
The factor by which to increase memory allocation. Should be greater than 1.
- do_back_solve(rhs)
Performs a back solve with the factorized matrix. Should only be called after do_numeric_factorization.
- Parameters
- rhs: numpy.ndarray or BlockVector
- Returns
- result: numpy.ndarray or BlockVector
- get_inertia()
Get the inertia. Should only be called after do_numeric_factorization.
- Returns
- num_pos: int
The number of positive eigenvalues of A
- num_neg: int
The number of negative eigenvalues of A
- num_zero: int
The number of zero eigenvalues of A
- set_icntl(key, value)
Set the value for an icntl option.
- Parameters
- key: int
- value: int
- set_cntl(key, value)
Set the value for a cntl option.
- Parameters
- key: int
- value: float
- get_icntl(key)
Get the value for an icntl option.
- Parameters
- key: int
- Returns
- val: int
- get_cntl(key)
Get the value for a cntl option.
- Parameters
- key: int
- Returns
- val: float