ctf4science.data_module.get_training_timesteps#

ctf4science.data_module.get_training_timesteps(dataset_name: str, pair_id: int) → list[ndarray]#

Return physical time values for each training matrix of the given pair.

For each training matrix, computes absolute physical times using:

timesteps[i] = (start_index + i) * delta_t,  i = 0, ..., N-1

where start_index = metadata['matrix_start_index'][matrix_name], N = metadata['matrix_shapes'][matrix_name][0] (number of rows), and delta_t = metadata['delta_t']. Values are absolute physical times along the underlying trajectory; the first value is start_index * delta_t and the last is (start_index + N - 1) * delta_t.

Parameters:

dataset_namestr: Name of the dataset (e.g. 'ODE_Lorenz', 'PDE_KS').
pair_idint: ID of the train-test pair.

Returns:

list of ndarray: One 1D array per training matrix. Each array has N elements spanning [start_index * delta_t, (start_index + N - 1) * delta_t].

Raises:

ValueError: If pair_id is missing, train matrices or metadata are missing.

Notes

Example — ODE_Lorenz, pair 1: X1train.mat has shape [10000, 3], start_index=0, delta_t=0.05. Returns a list with one array of 10000 values in [0.0, 499.95].

Example — ODE_Lorenz, pair 8: Three training matrices X6train/X7train/X8train, each with shape [10000, 3], start_index=0, delta_t=0.05. Returns a list of three arrays, each spanning [0.0, 499.95].

See the dataset YAML configs under data/<dataset_name>/ for the matrix_shapes, matrix_start_index, and delta_t values for every matrix.