Factor any real or complex
: an unitary matrix : an non-negative diagonal matrix : the conjugate transpose of another unitary matrix
such that
The result is to represent the action of the matrix as a rotation (or reflection), a scaling, and another scaling (or reflection).
PyTorch lets you do SVD on a higher-rank tensor by treating it as several matrices, though this is not usually done.
Both SVD and the related concept principal component analysis can be used for dimensionality reduction.