Identifiability beyond Kruskal's bound for symmetric tensors of degree 4

Abstract

We show how methods of algebraic geometry can produce criteria for the identifiability of specific tensors that reach beyond the range of applicability of the celebrated Kruskal criterion. More specifically, we deal with the symmetric identifiability of symmetric tensors in Sym${}^4({\mathbb{C}}^{n+1})$, i.e., quartic hypersurfaces in a projective space ${\mathbb{P}}^n$, that have a decomposition in 2n+1 summands of rank 1. This is the first case where the reshaped Kruskal criterion no longer applies. We present an effective algorithm, based on efficient linear algebra computations, that checks if the given decomposition is minimal and unique. The criterion is based on the application of advanced geometric tools, like Castelnuovo's lemma for the existence of rational normal curves passing through a finite set of points, and the Cayley-Bacharach condition on the postulation of finite sets. In order to apply these tools to our situation, we prove a reformulation of these results, hereby extending classical results such as Castelnuovo's lemma and the analysis of Geramita, Kreuzer, and Robbiano, "Cayley-Bacharach schemes and their canonical modules", Trans. Amer. Math. Soc. 339:443-452, 1993.

16 pages

References in corpus (3)

• Identifiability of parameters in latent structure models with many observed variables
• Learning Overcomplete HMMs
• Effective identifiability criteria for tensors and polynomials

Cited by in corpus (8)

• Tensor rank and entanglement of pure quantum states
• On the identifiability of ternary forms
• Identifiability for a class of symmetric tensors
• Decompositions and Terracini loci of cubic forms of low rank
• Minimality and uniqueness for decompositions of specific ternary forms
• Quaternary quartic forms and Gorenstein rings
• Toward a generalization of Kruskal's theorem on tensor decomposition
• Partially complex ranks for real projective varieties