SearchSpaceFactor¶

Type:: float
Range:: [1, inf]
Default:: 1.0
Appearance:: simple

In a nutshell, an eigensolver approximate eigenvectors by a weighted sum over basis vectors of a lower-dimensional subspace (search space). At least the search space has a dimension of the number of sought eigenvectors (SearchSpaceFactor=1), but the larger the search space the faster is the convergence. This parameter increases the search space dimension ( $\mathrm{dim}_{\mathrm{s}}$ ) as follows:

$\begin{eqnarray*} \mathrm{dim}_{\mathrm{s}} = (f_0+f) \mathrm{N}_{\mathrm{ev}}, \end{eqnarray*}$

where $f_0$ is an implementation dependent offset factor, $f$ is the user defined SearchSpaceFactor, $\mathrm{N}_{\mathrm{ev}}$ is the number of desired eigenvalues.