spd_learn.functional.log_map_pcm#
- spd_learn.functional.log_map_pcm(X: Tensor, P: Tensor | None = None, theta: float = 0.5) Tensor[source]#
Compute Riemannian logarithmic map under the Power-Cholesky Metric.
- Parameters:
X (torch.Tensor) – Target SPD matrix of shape (…, n, n).
P (torch.Tensor, optional) – Base SPD matrix of shape (…, n, n). If None, identity \(I\) is used.
theta (float, default=0.5) – Power parameter (\(\theta \neq 0\)).
- Returns:
Deformed Cholesky coordinates of shape (…, n, n).
- Return type: