boss_bench.data

Find and read the BoSS data files.

boss_bench.data.list_files(data_dir, kind='confinement', split='test')

List the data files of one kind and split.

Parameters:

• data_dir (str) – the folder with the downloaded files; it is searched recursively.

• kind (str) – "confinement" or "trajectory".

• split (str) – "train" or "test".

Returns:

the file paths, sorted.

Return type:

list of str

boss_bench.data.load(paths, fields)

Read chosen fields of every configuration in a list of files.

Every field can be read except the trajectory fields themselves: traj_data, traj_offsets and config_offsets. To load trajectories, use trajectory.read_trajectories. Fields are returned as stored, with one row per configuration and gridded fields flattened: modB, for example, has shape (n, 15 * 41 * 41).

Parameters:

• paths (list of str) – the files.

• fields (list of str) – the names of the fields to read.

Returns:

maps each field name to an array with one row per configuration, with the configurations of all files stacked in order.

Return type:

dict

boss_bench.data.load_record(f, config, fields=None)

Read the fields of one configuration.

Every field can be read except the trajectory fields themselves: traj_data, traj_offsets and config_offsets. To load trajectories, use trajectory.read_trajectories. Only the requested fields are read from disk, and gridded fields are returned flattened, as stored.

Parameters:

• f (h5py.File) – an open file, e.g. h5py.File(path, "r").

• config (int or str) – the row index in the file or the uuid of the configuration.

• fields (list of str, optional) – the names of the fields to read. By default, every field of the configuration.

Returns:

maps each field name to its value for the configuration.

Return type:

dict

Raises:

KeyError – if no configuration in the file has this uuid.