spd_learn.functional.modeig_forward#

spd_learn.functional.modeig_forward(X, applied_fct, *args)[source]#

Forward pass for the modified eigenvalue of a symmetric matrix.

This function computes the forward pass for a function that modifies the eigenvalues of a symmetric matrix.

Parameters:

  • X (torch.Tensor) – Symmetric matrix of shape (…, n, n).

  • applied_fct (callable) – Function to apply to the eigenvalues.

  • *args (tuple) – Additional arguments for the applied function.

Returns:

  • output (torch.Tensor) – Modified matrix after applying the function to the eigenvalues.

  • s (torch.Tensor) – Eigenvalues of X.

  • U (torch.Tensor) – Eigenvectors of X.

  • s_modified (torch.Tensor) – Modified eigenvalues after applying the function.