simpeg.potential_fields.base.BaseEquivalentSourceLayerSimulation.getJ#

BaseEquivalentSourceLayerSimulation.getJ(m, f=None)[source]#

Returns the full Jacobian.

The general definition of the linear forward simulation is:

$$\mathbf{d}=\mathbf{G}\mathbf{f}(\mathbf{m})$$

where $\mathbf{f}$ is a mapping operator (optional) from the model space to a user-defined parameter space, and $\mathbf{G}$ is an (n_data, n_param) linear operator. The getJ method forms and returns the full Jacobian:

$$\mathbf{J}(\mathbf{m})=\mathbf{G}\frac{\partial \mathbf{f}}{\partial \mathbf{m}}$$

for the model $\mathbf{m}$ provided. When $\mathbf{f}$ is the identity map (default), the Jacobian is no longer model-dependent and reduces to:

$$\mathbf{J}=\mathbf{G}$$

Parameters:

mnumpy.ndarray

The model vector.

fNone

Precomputed fields are not used to speed up the computation of the Jacobian for linear problems.

Returns:

J(n_data, n_param) numpy.ndarray

$J=G\frac{\partial f}{\partial \mathbf{m}}$. Where $f$ is model_map.