mlcolvar.data.graph.create_dataset_from_configurations

class mlcolvar.data.graph.create_dataset_from_configurations(config: List[Configuration], atomic_numbers: AtomicNumberTable, cutoff: float, buffer: float = 0.0, long_range_cutoff: float = -1.0, atom_names: List = None, remove_isolated_nodes: bool = False, show_progress: bool = True)[source]

Bases:

Build DictDataset object containing torch_geometric graph data objects from configurations.

Parameters:

  • config (mlcolvar.graph.utils.atomic.Configurations) – The configurations from whihc to generate the dataset

  • atomic_numbers (mlcolvar.graph.utils.atomic.AtomicNumberTable) – The atomic number table used to build the node attributes

  • cutoff (float) – The graph cutoff radius

  • buffer (float) – Buffer size used in finding active environment atoms if restricting the neighborhood to a subsystem (i.e., system + environment), see also mlcolvar.data.grap.neighborhood.get_neighborhood

  • long_range_cutoff (float) – Cutoff radius for the long-range edges defined on subsystem atoms. If negative, no long-range interactions are considered, by default -1.0

  • atom_names, List[str] – Names of the system atoms. If using truncated graphs, use the system atoms only

  • remove_isolated_nodes (bool) – If to remove isolated nodes from the dataset

  • show_progress (bool) – If to show the progress bar

__init__(**kwargs)