qimpy.dft.electrons.CheFSI

class CheFSI(*, electrons, checkpoint_in=(None, ''), n_iterations=100, eig_threshold=1e-08, filter_order=10, init_threshold=0.1)

Chebyshev Filter Subspace Iteration (CheFSI) diagonalization.

Parameters:

__init__(*, electrons, checkpoint_in=(None, ''), n_iterations=100, eig_threshold=1e-08, filter_order=10, init_threshold=0.1)

Initialize with stopping criteria and filter order.

Parameters:

n_iterations (int) – [Input file] Number of diagonalization iterations. This only affects fixed-Hamiltonian calculations because the self-consistent field method overrides this when diagonalizing in an inner loop.
eig_threshold (float) – [Input file] Convergence threshold on eigenvalues in Hartrees. Stop when the maximum change in any eigenvalue between iterations falls below this threshold. This only affects fixed-Hamiltonian calculations because the self-consistent field method overrides this when diagonalizing in an inner loop.
filter_order (int) – [Input file] Order of the Chebyshev filter. This amounts to the number of Hamiltonian evaluations per band per eigenvalue iteration.
init_threshold (float) – [Input file] Eigenvalue threshold for initial Davidson steps. This controls the accuracy of a coarse Davidson diagonalization that creates a reasonable starting point for CheFSI.
electrons (Electrons) –
checkpoint_in (CheckpointPath) –

Return type:

None

Methods

`__init__`	Initialize with stopping criteria and filter order.
`add_child`	Construct child object self.`attr_name` of type cls.
`add_child_one_of`	Invoke add_child on one of several child options in args.
`diagonalize`	Diagonalize Kohn-Sham Hamiltonian in electrons.
`get_Eband`	Compute the sum over band eigenvalues, averaged over k
`precondition`	Inverse-kinetic preconditioner on the Cerr in eigenpairs, using the per-band kinetic energy KEref
`save_checkpoint`	Save self and all children in hierarchy to cp_path.

Attributes

`filter_order`	Order of Chebyshev filter
`init_threshold`	Threshold for Davidson initialization
`electrons`	Electronic system to diagonalize
`n_iterations`	Number of diagonalization iterations
`eig_threshold`	Eigenvalue convergence threshold (in \(E_h\))
`child_names`	Names of attributes with child objects.

diagonalize(n_iterations=None, eig_threshold=None)

Diagonalize Kohn-Sham Hamiltonian in electrons. Also available as __call__() to make CheFSI callable.

Parameters:

Return type:

None