emukit.multi_fidelity package

Subpackages

• emukit.multi_fidelity.kernels package
  • Submodules
    • LinearMultiFidelityKernel
      • LinearMultiFidelityKernel.K()
      • LinearMultiFidelityKernel.Kdiag()
      • LinearMultiFidelityKernel.gradients_X()
      • LinearMultiFidelityKernel.gradients_X_diag()
      • LinearMultiFidelityKernel.update_gradients_diag()
      • LinearMultiFidelityKernel.update_gradients_full()
  • Module contents
• emukit.multi_fidelity.models package
  • Submodules
    • GPyLinearMultiFidelityModel
    • make_non_linear_kernels()
    • NonLinearMultiFidelityModel
      • NonLinearMultiFidelityModel.set_data()
      • NonLinearMultiFidelityModel.X
      • NonLinearMultiFidelityModel.Y
      • NonLinearMultiFidelityModel.n_samples
      • NonLinearMultiFidelityModel.predict()
      • NonLinearMultiFidelityModel.get_prediction_gradients()
      • NonLinearMultiFidelityModel.optimize()
      • NonLinearMultiFidelityModel.get_f_minimum()
  • Module contents

Submodules

This module provides functions to convert from a list representation of multi-fidelity data to an array representation.

The list based representation is a list of numpy arrays, with a numpy array for every fidelity. The list is ordered from the lowest fidelity to the highest fidelity.

The array representation is one array for all inputs where the last column of the X array is a zero-based index indicating the fidelity.

emukit.multi_fidelity.convert_lists_to_array.convert_x_list_to_array(x_list)

Converts list representation of features to array representation :type x_list: List :param x_list: A list of (n_points x n_dims) numpy arrays ordered from lowest to highest fidelity :rtype: ndarray :return: An array of all features with the zero-based fidelity index appended as the last column

emukit.multi_fidelity.convert_lists_to_array.convert_y_list_to_array(y_list)

Converts list representation of outputs to array representation :type y_list: List :param y_list: A list of (n_points x n_outputs) numpy arrays representing the outputs

ordered from lowest to highest fidelity

Return type:

ndarray

Returns:

An array of all outputs

emukit.multi_fidelity.convert_lists_to_array.convert_xy_lists_to_arrays(x_list, y_list)

Converts list representation of targets to array representation :type x_list: List :param x_list: A list of (n_points x n_dims) numpy arrays ordered from lowest to highest fidelity :type y_list: List :param y_list: A list of (n_points x n_outputs) numpy arrays representing the outputs

ordered from lowest to highest fidelity

Return type:

Tuple[ndarray, ndarray]

Returns:

Tuple of (x_array, y_array) where x_array contains all inputs across all fidelities with the fidelity index appended as the last column and y_array contains all outputs across all fidelities.

Module contents