linalg

Dense linear-algebra primitives shared by the orthogonalization and SVD sweeps.

truncated_svd, the directional left/right/up_svd and *_svd_pair factorizations (directions match the core-unfolding conventions), pad_safe_svd (the mask-aware SVD of a zero-padded matrix – the uniform layer’s replacement for a black-box SVD), and pad_or_truncate. Pure array-in, array-out helpers with no T3 semantics.

Functions

pad_or_truncate(array, pad_width[, mode])	Pad and/or truncate an array per axis, using signed (before, after) widths.
truncated_svd(A[, min_rank, max_rank, rtol, atol])	Compute (truncated) singular value decomposition of matrix A.
left_svd(G0_i_a_j[, min_rank, max_rank, rtol, atol])	Compute (truncated) singular value decomposition of 3-tensor left unfolding.
right_svd(G0_i_a_j[, min_rank, max_rank, rtol, atol])	Compute (truncated) singular value decomposition of 3-tensor right unfolding.
up_svd(G0_i_a_j[, min_rank, max_rank, rtol, atol])	Compute (truncated) singular value decomposition of 3-tensor up unfolding.
left_svd_pair(G0_i_a_j, G1_j_b_k[, min_rank, ...])	Compute (truncated) singular value decomposition of G0, pushing non-orthogonal remainder onto G1.
right_svd_pair(G0_i_a_j, G1_j_b_k[, min_rank, ...])	Compute (truncated) singular value decomposition of G1, pushing non-orthogonal remainder onto G0.
up_svd_pair(G_i_a_j, B_a_o[, min_rank, max_rank, ...])	Compute (truncated) singular value decomposition of G, pushing non-orthogonal remainder onto B.
down_svd_pair(G_i_a_j, B_a_o[, min_rank, max_rank, ...])	Compute (truncated) singular value decomposition of B, pushing non-orthogonal remainder onto G.
pad_safe_svd(A, row_mask, col_mask)	SVD of a zero-padded matrix whose null-space vectors are forced OFF the padding.