Introduces new tensors and demonstrates their effective low-rank approximations, suggesting improvements for tensor analysis.
In this paper, we introduce the class of generally unitarily decomposable partially symmetric tensors, which extend conventional orthogonal tensors while retaining many of their desirable properties. We show that the best low-rank approximation of such tensors is directly determined by the largest coefficients in their decompositions. To compute these decompositions, we propose a successive rank-one approximation algorithm and a matrix-based method. Moreover, we study the approximation of general partially symmetric tensors using the proposed tensors and establish Comon’s conjecture for partially symmetric tensors whose canonical polyadic rank is smaller than their order. Numerical experiments demonstrate the effectiveness of the proposed algorithms.
No takes yet. Share an insight, caveat, or question.
Tang et al. (2026) studied this question.
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: