agedi.api._display¶

Rich UI / console display utilities.

Functions¶

`_fmt_target`(→ str)	Return the short class name from a dotted `_target_` string.
`_extract_data_info`(→ Dict)	Extract summary information from a list of ASE Atoms objects.
`_check_head_dimensions`(→ list)	Validate that head input dimensions match feature_size + conditioning output dims.
`_render_config_tree`(→ None)	Recursively populate a Rich `Tree` with config key/values.
`_extract_diffusion_display_info`(→ dict)	Extract human-readable display values from a nested Hydra diffusion config.
`_get_device_info`(→ str)	Return a human-readable string describing the available compute device.
`_print_training_config`(→ None)	Print a Rich-formatted training configuration panel.
`_print_log_path`(→ None)	Print the resolved log directory for this training run.
`_print_loaded_model_info`(→ None)	Print a Rich-formatted summary of a loaded diffusion model.
`_print_sampling_config`(→ None)	Print a Rich-formatted sampling configuration panel.

agedi.api._display._fmt_target(target: str) → str¶: Return the short class name from a dotted _target_ string.

agedi.api._display._extract_data_info(data: Sequence[Atoms]) → Dict¶

Extract summary information from a list of ASE Atoms objects.

Parameters:

data (Sequence[Atoms]) – List of ASE Atoms objects to inspect.

Returns:

A dictionary with the following keys:

"cell" – flattened 9-element list of the (shared) unit-cell matrix, or None if cells differ between structures.
"symbols" – list of unique chemical symbols present in the data.
"n_training_data" – total number of structures.

Return type:

dict

agedi.api._display._check_head_dimensions(diffusion_cfg: dict) → list¶

Validate that head input dimensions match feature_size + conditioning output dims.

Parameters:: diffusion_cfg (dict) – The diffusion sub-dict from hparams.yaml.
Returns:: Warning messages for any dimension mismatches, empty if all match.
Return type:: list of str

agedi.api._display._render_config_tree(data, tree: rich.tree.Tree) → None¶

Recursively populate a Rich Tree with config key/values.

Dicts: each key becomes a leaf (scalar) or sub-branch (dict/list). The _target_ key is skipped — callers show it as the branch label.
Lists: each item becomes a sub-branch labelled by its _target_ class name (or a numeric index for plain scalars).
Scalars and None: shown as key value; None is omitted.

agedi.api._display._extract_diffusion_display_info(diffusion_cfg: dict) → dict¶

Extract human-readable display values from a nested Hydra diffusion config.

Parameters:: diffusion_cfg (dict) – The diffusion sub-dict from hparams.yaml, as returned by get_hparams().
Returns:: Flat dict with display-friendly keys: model, feature_size, n_blocks, cutoff, noisers, conditionings, lr, lr_factor, lr_patience, weight_decay.
Return type:: dict

agedi.api._display._get_device_info() → str¶: Return a human-readable string describing the available compute device.

agedi.api._display._print_training_config(hparams: dict) → None¶

Print a Rich-formatted training configuration panel.

Parameters:: hparams (dict) – The hparams dict as saved to hparams.yaml. Must contain a "diffusion" key with the nested Hydra config. Metadata keys (n_train, n_val, batch_size, etc.) are read directly.

agedi.api._display._print_log_path(trainer: Trainer) → None¶: Print the resolved log directory for this training run.

agedi.api._display._print_loaded_model_info(params: dict, checkpoint_path: pathlib.Path, device) → None¶: Print a Rich-formatted summary of a loaded diffusion model.

agedi.api._display._print_sampling_config(n_samples: int, steps: int, eps: float, batch_size: int, formula=None, n_atoms=None, template=None, cell=None, confinement=None, property=None, force_field_guidance: float = 0.0) → None¶: Print a Rich-formatted sampling configuration panel.