regpy.functionals.ngsolve¶

Special NGSolve functionals defined on the regpy.vecsps.ngsolve.NgsVectorSpace.

Classes¶

`SignumFilter`
`NgsL1`	Implementation of the \(L^1\)-norm on a given NgsVectorSpace. It is registered under the
`NgsTV`	Implementation of the total variation functional \(TV\) on a given NgsVectorSpace. It is

Module Contents¶

class regpy.functionals.ngsolve.SignumFilter(space, vec)[source]¶

Bases: ngsolve.la.BaseMatrix

gf¶

gf_out¶

gf_help¶

Update(new_vec)[source]¶

Mult(x, y)[source]¶

Height()[source]¶

Width()[source]¶

class regpy.functionals.ngsolve.NgsL1(domain)[source]¶

Bases: regpy.functionals.base.Functional

Implementation of the \(L^1\)-norm on a given NgsVectorSpace. It is registered under the Abstract functional L1 and should not be called directly but rather used by defining the abstract L1 functional as the penalty or data_fid when initializing the regularization setting by calling regpy.solvers.Setting.

Parameters:: domain (NgsVectorSpace) – The underlying ngsolve space.

sign¶

class regpy.functionals.ngsolve.NgsTV(domain)[source]¶

Bases: regpy.functionals.base.Functional

Implementation of the total variation functional \(TV\) on a given NgsVectorSpace. It is registered under the Abstract functional TV and should not be called directly but rather used by defining the abstract TV functional as the penalty or data_fid when initializing the regularization setting by calling regpy.solvers.Setting.

Parameters:

domain (NgsVectorSpace) – The underlying ngsolve space.
h_domain (HilbertSpace) – The Hilbert space wrt which the proximal gets computed.