krypy.recycling.factories - deflation vector factories¶
-
class
krypy.recycling.factories.RitzFactory(subset_evaluator, subsets_generator=None, mode='ritz', print_results=None)¶ Bases:
krypy.recycling.factories._DeflationVectorFactoryFactory of Ritz vectors for automatic recycling.
Parameters: - subset_evaluator – an instance of
_RitzSubsetEvaluatorthat evaluates a proposed subset of Ritz vectors for deflation. - subsets_generator – (optional) an instance of
_RitzSubsetsGeneratorthat generates lists of subsets of Ritz vectors for deflation. - print_results –
(optional) may be one of the following:
- None: nothing is printed.
- ’number’: the number of selected deflation vectors is printed.
- ’values’: the Ritz values corresponding to the selected Ritz vectors are printed.
- ’timings’: the timings of all evaluated subsets of Ritz vectors are printed.
-
get(deflated_solver)¶ Get deflation vectors.
Returns: numpy.array of shape (N,k)
- subset_evaluator – an instance of
-
class
krypy.recycling.factories.RitzFactorySimple(mode='ritz', n_vectors=0, which='sm')¶ Bases:
krypy.recycling.factories._DeflationVectorFactorySelects a fixed number of Ritz or harmonic Ritz vectors with respect to a prescribed criterion.
Parameters: - mode – See
modeparameter ofRitz. - n_vectors – number of vectors that are chosen. Actual number of
deflation vectors may be lower if the number of Ritz pairs is less
than
n_vectors. - which –
the
n_vectorsRitz vectors are chosen such that the corresponding Ritz values are the ones withlm: largest magnitude.sm: smallest magnitude.lr: largest real part.sr: smallest real part.li: largest imaginary part.si: smallest imaginary part.smallest_res: smallest Ritz residual norms.
-
get(solver)¶ Get deflation vectors.
Returns: numpy.array of shape (N,k)
- mode – See
-
class
krypy.recycling.factories.UnionFactory(factories)¶ Bases:
krypy.recycling.factories._DeflationVectorFactoryCombine a list of factories.
Parameters: factories – a list of factories derived from _DeflationVectorFactory.-
get(solver)¶ Get deflation vectors.
Returns: numpy.array of shape (N,k)
-