Valid keywords

Keyword	Description	Type	Default
use_ml	use a machine learning model to predict the screening parameters	bool	False
n_max	The maximum expansion coefficient n for radial basis functions. If a list is provided in the convergence_ml-task, a grid search will be performed	int/list	4
l_max	The maximum angular expansion coefficient. If a list is provided in the convergence_ml-task, a grid search will be performed	int/list	4
r_min	The width of the narrowest radial basis function. If a list is provided in the convergence_ml-task, a grid search will be performed	float/list	0.5
r_max	The width of the broadest radial basis function. If a list is provided in the convergence_ml-task, a grid search will be performed	float/list	4.0
criterium	The criterium which has to be satisfied in order to use the ML-predicted screening coefficients instead of computing them ab-initio	str	after_fixed_num_of_snapshots (must be after_fixed_num_of_snapshots)
number_of_training_snapshots	Number of snapshots needed for the "after_fixed_num_of_snapshots"-criterium. In case of the convergence_ml task, this number is taken to be the highest number of training samples for the convergence analysis	int	1
current_snapshot	Number of snapshots already trained on	int	0
alphas_from_file	If true, read the screening coefficients from file instead of calculating them ab-initio. The files have to be provided in the snapshot_ folders	bool	False
train_on_the_fly	If true, the ML-model gets trained after the calculation of each orbital. If false, the ML-model gets trained at the end of each snapshot	bool	False
occ_and_emp_together	If true, there will be one ML model for both occupied and empty states. If False, there will be one ML Model for occupied states and one for empty states	bool	True
type_of_ml_model	Which ML model to use for making the predictions	str	ridge_regression (must be ridge_regression/linear_regression/mean)
input_data_for_ml_model	Which data to use in case of the ridge_regression or the linear-regression Model	str	orbital_density (must be orbital_density/self_hartree)
quantities_of_interest	Which quantities are we interested in the convergence_ml-task. Note that the eigenvalues (evs) require performing the final calculation afresh for every snapshot.	str/list	['alphas']