Almost all subgeneric third-order Chow decompositions are identifiable

From MaRDI portal



Abstract: For real and complex homogeneous cubic polyomials in n+1 variables, we prove that the Chow variety of products of linear forms is generically complex identifiable for all ranks up to the generic rank minus two. By integrating fundamental results of [Oeding, Hyperdeterminants of polynomials, Adv. Math., 2012], [Casarotti and Mella, From non defectivity to identifiability, J. Eur. Math. Soc., 2021], and [Torrance and Vannieuwenhoven, All secant varieties of the Chow variety are nondefective for cubics and quaternary forms, Trans. Amer. Math. Soc., 2021] the proof is reduced to only those cases in up to 103 variables. These remaining cases are proved using the Hessian criterion for tangential weak defectivity from [Chiantini, Ottaviani, and Vannieuwenhoven, An algorithm for generic and low-rank specific identifiability of complex tensors, SIAM J. Matrix Anal. Appl., 2014]. We also establish that the smooth loci of the real and complex Chow varieties are immersed minimal submanifolds in their usual ambient spaces.


Let \(\mathbb{K} \in \{\mathbb{R},\mathbb{C}\}\). A Chow decomposition over \(\mathbb{K}\) expresses a homogeneous polynomial \(p \in S^{d}(\mathbb{K}^{n+1})\) of degree \(d\) in \(n+1\) variables as minimum-length sum of completely decomposable forms, i.e., \[ p(x_{0},\dots,x_{n}) = \sum_{j=1}^{r}\prod_{i=0}^{d-1}L_{i,j} = \sum_{j=1}^{r}\prod_{i=0}^{d-1}(a_{0,i,j}x_{0} + \dots + a_{n,i,j}x_{n}). \] The minimal number \(r\) is called the Chow rank of \(p\) over \(\mathbb{K}\). We have the following two natural problems about complex Chow decomposition: i) What is the minimal value of \(r\) so that almost all polynomials \(p \in S^{d}\mathbb{C}^{n+1}\) admit a Chow decomposition of rank \(r\), and ii) how many distinct complex Chow decompositions does a generic rank \(r\) polynomial \(p\) have? Consider the Zariski closure of the completely decomposable forms in \(n+1\) variables and degree \(d\), namely \[ \mathcal{C}_{d,n} = \overline{\{[L_{0} \cdots L_{d-1}] : L_{i} \text{ is linear}\}} \subset S^{d}\mathbb{P}^{n}_{\mathbb{C}}. \] It is an irreducible non-degenerate singular projective variety, and it is called either the Chow variety of zero-cycles, or the split variety. The \(d\)-forms \([p]\in S^{d}\mathbb{P}^{n}_{\mathbb{C}}\) such that \(p\) admits a rank \(r\) Chow decomposition as above form a Zariski-open subset of the \(r\)-secant variety \(\sigma_{r}(\mathcal{C}_{d,n})\subset S^{d}\mathbb{P}^{n}_{d}\). The main result of the paper under review, revolved around the second question on the above list, can be formulated as follows. Main Theorem. Let \(\mathcal{C}_{3,n} \subset S^{3}\mathbb{P}^{n}_{\mathbb{C}}\) be the degree \(3\) Chow variety of zero-cycles. Then \(\mathcal{C}_{3,n}\) is generically complex \(r\)-identifiable up to the generic rank minus \(2\), i.e., \[ r\leq r_{\mathrm{gen}}-2 = \bigg\lceil \frac{1}{3n+1}\binom{n+3}{3}\bigg\rceil -2. \] In other words, the complex Chow decomposition of a generic real or complex Chow rank \(r\) cubic \(p\) is unique in the sense that the set of completely decomposable forms \(\{\prod_{i=0}^{d-1}L_{i,1}, \dots,\prod_{i=0}^{d-1}L_{i,r}\}\) is uniquely determined by \(p\).



Cites work



Describes a project that uses

Uses Software






This page was built for publication: Almost all subgeneric third-order Chow decompositions are identifiable

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2088044)