Searcharxiv⌕ Search

arXiv · 2611.13823

Rank Additivity and Comon's Conjecture for Real Hankel Tensors

Abstract

Let $\mathscr{H}_m(n)$ denote the set of real Hankel tensors of order $m$ and dimension $n$, and let $\textbf{H}\in\mathscr{H}_m(n)$, where $m\ge3$ and $n\ge1$. Set $M=(m-1)(n-1)+1$. First, we prove that the CP rank $r(\textbf{H})\le M$ and that CP rank is additive under direct sums of $\textbf{H}$ with any real tensor $\textbf{T}$ of the same order. Second, we prove that partially symmetric rank $p_j(\textbf{H})=r(\textbf{H})$ for $1\le j\le m-1$ and that, if $r(\textbf{H})\le M-1$, these ranks also equal its symmetric rank $s(\textbf{H})$. Third, we construct a fourth-order Hankel tensor $\textbf{H}_*\in\mathscr{H}_4(4)$ that attains this CP rank bound and satisfies $r(\textbf{H}_*)=10$ and $s(\textbf{H}_*)=11$. Moreover, for the direct sum $\textbf{H}_*^{(k)}=\bigoplus_{a=1}^k\textbf{H}_*$, we further prove that $r(\textbf{H}_*^{(k)})=10k$ and $s(\textbf{H}_*^{(k)})=11k$. This provides a low-dimensional counterexample to Comon's conjecture over the real numbers and further shows that the Comon rank gap is unbounded among real symmetric tensors of order four.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Jianze Li. 2026-11-10. Rank Additivity and Comon's Conjecture for Real Hankel Tensors. https://searcharxiv.com/abs/2611.13823

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Extremely proximal actions and primitivity of group rings

We present a purely dynamical criterion for the primitivity of group rings. To be precise, we show that if $G$ is a countable group that admits a topologically free, extremely proximal action, then the group ring $KG$ is primitive for any field $K$. Our criterion recovers several of the existing results on primitivity and applies to new examples, such as the commensurator groups of torsion-free non-elementary hyperbolic groups, some Kac-Moody groups, and other weakly hyperbolic groups.

math.RA↗

Self-injective dimensions and Iyama grade dualities over Noether algebras

Let $A$ be an algebra module-finite over a commutative Noetherian ring. We first establish a local criterion for the $n$-Gorenstein condition in terms of the grades and flat dimensions of injective envelopes of simple modules over central localizations. As a main application, we give an affirmative answer, in this module-finite setting, to a question of Huang on the symmetry of finite self-injective dimension under the Auslander condition. We also show that $A$ is Auslander--Gorenstein if and only if it is Iwanaga--Gorenstein and has the fiberwise Iyama grade duality. For Cohen--Macaulay algebras, central regular reductions to Artin algebras allow both the Auslander condition and the grade duality to be tested on simple modules over $A$ itself.

math.RA↗

Classification problems for diassociative and dendriform algebras: a survey

Diassociative algebras (associative dialgebras) and dendriform algebras are vector spaces with two binary operations, introduced by J.-L. Loday in the study of Leibniz algebras; their operads are Koszul dual to each other. We survey the classification problem for these two classes, including the Zinbiel algebras, the commutative dendriform algebras, which are Koszul dual to Leibniz algebras. We recall the definitions, the main examples and the Loday diagram with its extension by left-symmetric algebras and dialgebras, and describe the methods used to classify algebras of small dimension: direct computation from lists of associative algebras, annihilator (central) extensions and geometric classification by degenerations. We then collect, by dimension, the known algebraic classifications of diassociative algebras, dendriform algebras and their commutative case, the Zinbiel algebras, point out lists that later work corrected, and summarise the cohomological results that the methods rely on. We close with a list of open problems.

math.RA↗