Training¶
Routine to train all GPs in parallel to emulate the 3D Power Spectrum
-
training.
main
(directory: str = 'semigps') → None[source]¶ Main function to train all the Gaussian Process models.
- Param
directory (str) - directory where the GPs are stored.
- Returns
None
-
training.
parallel_training
(arguments: list) → None[source]¶ Call the parallel processing routine here
- Param
arguments (list) - list of arguments (inputs) to train the GPs
- Returns
None
-
training.
train
(cosmologies: numpy.ndarray, target: numpy.ndarray, folder_name: str, fname: str, kwargs: dict) → None[source]¶ Function to train GPs
- Param
cosmologies (np.ndarray) : array of size N_train x N_dim for the inputs to the GP
- Param
target (np.ndarray) : an array for the targets (function)
- Param
folder_name (str) : name of the folder where the outputs are stored
- Param
fname (str) : name of the GP output
- Param
kwargs (dict) : a dictionary with the settings for the GPs, for example, lambda_cap = 1000