inversion_ideas.LinearRegressor

class inversion_ideas.LinearRegressor(matrix, *, build_jacobian=True, cache=True)

Linear regressor simulation.

Implements a linear regressor that generates data values as the product between its matrix and a given model vector.

Important

This class is included mainly as an example of a simple simulation class that could be created and used within the inversion framework.

Parameters:

matrix(n_data, n_params) array

Matrix used in the definition of the linear regressor.

build_jacobianbool, optional

Whether the Jacobian matrix will be created as a dense matrix (True) or as a LinearOperator (False). Default to True.

cachebool, optional

Whether to cache the results of the __call__() method for the last model vector or not. Default to True.

Attributes

n_data	Number of data values.
n_params	Number of model parameters.

Methods

__call__(model)	Evaluate simulation for a given model.
create_random(n_data, n_params, *[, seed])	Create a linear regressor with a random matrix.
jacobian(model)	Jacobian matrix for a given model.

Notes

Given the matrix \(\mathbf{X}\), the linear regressor simulation computes the data vector \(\mathbf{y}\) for a given model vector \(\mathbf{m}\) as follows:

\[\mathbf{y} = \mathbf{X} \cdot \mathbf{m}\]

Attributes

LinearRegressor.n_data

LinearRegressor.n_params

Methods

Methods documentation

LinearRegressor.__call__(model)

Evaluate simulation for a given model.

Parameters:

model(n_params) array

Array with model values.

Returns:

(n_data)

Array with predicted data values for the given model.

---

classmethod LinearRegressor.create_random(n_data, n_params, *, seed=None, **kwargs)

Create a linear regressor with a random matrix.

Parameters:

n_dataint

Number of data values that the simulation will generate.

n_paramsint

Number of elements in the model vector.

seedint or None, optional

Random seed or random state used to generate the matrix.

**kwargs

Keyword arguents passed to the constructor of LinearRegressor.

---

LinearRegressor.jacobian(model)

Jacobian matrix for a given model.

Parameters:

model(n_params) array

Array with model values.

Returns:

(n_data, n_params) array or LinearOperator

Jacobian matrix as a dense or sparse array, or as a LinearOperator.

---