simpeg.maps.LogisticSigmoidMap#
- class simpeg.maps.LogisticSigmoidMap(mesh=None, nP=None, lower_bound=0, upper_bound=1, **kwargs)[source]#
Bases:
IdentityMap
Mapping that computes the logistic sigmoid of the model parameters.
Where \(\mathbf{m}\) is a set of model parameters,
LogisticSigmoidMap
creates a mapping \(\mathbf{u}(\mathbf{m})\) that computes the logistic sigmoid of every element in \(\mathbf{m}\); i.e.:\[\mathbf{u}(\mathbf{m}) = sigmoid(\mathbf{m}) = \frac{1}{1+\exp{-\mathbf{m}}}\]LogisticSigmoidMap
transforms values onto the interval (0,1), but can optionally be scaled and shifted to the interval (a,b). This can be useful for inversion of data that varies over a log scale and bounded on some interval:\[\mathbf{u}(\mathbf{m}) = a + (b - a) \cdot sigmoid(\mathbf{m})\]- Parameters:
- mesh
discretize.BaseMesh
The number of parameters accepted by the mapping is set to equal the number of mesh cells.
- nP
int
Set the number of parameters accepted by the mapping directly. Used if the number of parameters is known. Used generally when the number of parameters is not equal to the number of cells in a mesh.
- lower_bound: float or (nP) numpy.ndarray
lower bound (a) for the transform. Default 0. Defined in mathbf{u} space.
- upper_bound: float or (nP) numpy.ndarray
upper bound (b) for the transform. Default 1. Defined in mathbf{u} space.
- mesh
Attributes
Determine whether or not this mapping is a linear operation.
The lower bound
The mesh used for the mapping
Number of parameters the mapping acts on.
Dimensions of the mapping operator
The upper bound
Methods
deriv
(m[, v])Derivative of mapping with respect to the input parameters.
dot
(map1)Multiply two mappings to create a
simpeg.maps.ComboMap
.inverse
(m)Apply the inverse of the mapping to an array.
test
([m, num, random_seed])Derivative test for the mapping.