boltzkit.evaluation.sample_based.tica

Functions

`get_tica_hist`(tica_proj, **kwargs)	Compute a 2D histogram of TICA projections.
`get_tica_wasserstein_1_2`(...[, include_w1, ...])	Compute the Wasserstein distances (W1 and W2) between two sets of TICA projections.
`visualize_tica`(tica_hist, vis_mode, ax, show)
`visualize_tica_true_and_pred`(tica_hist_true, ...)

boltzkit.evaluation.sample_based.tica.get_tica_hist(tica_proj: ndarray, **kwargs)[source]

Compute a 2D histogram of TICA projections.

Parameters:

tica_proj (np.ndarray) – Array of shape (batch, 2), containing the first two TICA components per frame.
**kwargs – Additional keyword arguments passed to get_histogram_2d.

Returns:

A Histogram2D object containing the relative histogram counts and bin edges.

Return type:

Histogram2D

boltzkit.evaluation.sample_based.tica.visualize_tica(tica_hist: Histogram2D, vis_mode: VisualizationMode = <function plot_as_log_density>, ax: Axes | None = None, show: bool = False)[source]

boltzkit.evaluation.sample_based.tica.visualize_tica_true_and_pred(tica_hist_true: Histogram2D, tica_hist_pred: Histogram2D, vis_mode: VisualizationMode = <function plot_as_log_density>, show: bool = False, clip_pred_to_true_range: bool = True)[source]

boltzkit.evaluation.sample_based.tica.get_tica_wasserstein_1_2(tica_projections_true: ndarray, tica_projections_pred: ndarray, include_w1=False, include_w2=True)[source]

Compute the Wasserstein distances (W1 and W2) between two sets of TICA projections.

Parameters:

tica_projections_true (np.ndarray) – Reference TICA projections of shape (batch, n_tica_components).
tica_projections_pred (np.ndarray) – Predicted TICA projections of shape (batch, n_tica_components).
include_w1 (bool, optional) – Whether to compute the W1 distance (default is False).
include_w2 (bool, optional) – Whether to compute the W2 distance (default is True).

Returns:

Tuple (W1, W2) of Wasserstein distances. Each is either a float or None depending on include_w1 and include_w2.

Return type:

tuple