check_data_name_validity(fem_material,noise_level):
filter_raw_data(rng, raw_data, filter_value):
Randomly sub-sampling filter_value degrees of freedom from the data available at all quadrature points in raw_data
Input Arguments
-rng - Random number generator
-raw_data - Contains D (derivatives of feature library) and y (inertia terms and reaction forces) evaluated at all quadrature points
-filter_value - Number of degrees of freedom subsampled from the original data
Output Arguments
RawData(A1, b1, A2, b2, dof_x, dof_y)- Object of classRawDatawhich is constructed from subsampled data
get_data(rng, fem_dir, prefix, fem_material, noise_level, loadstep, feature_filter):
predict_energy_path(chain, theta_gt, fem_mat, feature_filter, deformation):
Predicts the energy deformation evolution for predicted and true material models along six deformation paths: i.) uniaxial tension, ii.) uniaxial compression, iii.) simple shear, iv.) biaxial tension, v.) biaxial compression, vi.) pure shear
Input Arguments
-chain - Object of class Chain (see core_spike_slab file)
-theta_gt - The true set of feature coefficients for the benchmark material
-fem_mat - The name of the benchmark material to be tested
-feature_filter - The list of features to retain for constructing the Markov chain. Suppressed features will be highlighted with a red patch in the plot
-deformation - Name of deformation path to be evaluated
Output Arguments
-gamma - Numpy array of different values for the deformation parameter
-energy_mean - Energy corresponding to mean value of theta (feature coefficients) across different members of the Markov chain
-energy_plus - 97.5 percentile energy branch across different members of the Markov chain
-energy_minus - 2.5 percentile energy branch across different members of the Markov chain
-energy_gt - Energy corresponding to the true feature coefficients
-energy - Numpy array containing energy of all chain members at all deformation parameters