- class SimPEG.regularization.BaseSparse(mesh, norm=2.0, irls_scaled=True, irls_threshold=1e-08, **kwargs)#
Base class for sparse-norm regularization.
BaseSparseclass defines properties and methods inherited by sparse-norm regularization classes. Sparse-norm regularization in SimPEG is implemented using an iteratively re-weighted least squares (IRLS) approach. The
BaseSparseclass however, is not directly used to define the regularization for the inverse problem.
Mesh on which the regularization is discretized. This is not necessarily the same as the mesh on which the simulation is defined.
Boolean array defining the set of
RegularizationMeshcells that are active in the inversion. If
None, all cells are active.
The mapping from the model parameters to the active cells in the inversion. If
None, the mapping is the identity map.
Reference model values used to constrain the inversion. If
None, the starting model is set as the reference model.
Units for the model parameters. Some regularization classes behave differently depending on the units; e.g. ‘radian’.
Weight multipliers to customize the least-squares function. Each key points to a (n_cells, ) numpy.ndarray that is defined on the
The norm used in the regularization function. Must be between within the interval [0, 2].
True, scale the IRLS weights to preserve magnitude of the regularization function. If
False, do not scale.
Constant added to IRLS weights to ensures stability in the algorithm.
Scale IRLS weights.
Stability constant for computing IRLS weights.
Norm for the sparse regularization.
Compute and return iteratively re-weighted least-squares (IRLS) weights.