Identifiability beyond Kruskal's bound for symmetric tensors of degree 4 (Q1650838)

The authors consider quartic homogeneous polynomial in \(n+1\), i.e., elements of \(S^4\mathbb{C}^{n+1}\) which have a minimal additive decomposition as a sum of \((2n+1)\) \(4\)-powers of linear forms and find a criterion for the uniqueness of the decomposition in geometric terms (associating to each linear form a point of \(\mathbb {P}^n(\mathbb {C})\)). The degree 3 cases would have been the symmetric form of the celebrated Kruskal's uniqueness criterion for \(3\)-tensors. This was extended to any \(n\)-tensors and the symmetric version of it was recently proved [\textit{L. Chiantini} et al., Trans. Amer. Math. 369, No. 6, 4021--4042 (2017; Zbl 1360.14021)]. The paper under review goes to the next step, the first step not covered by the generalized (symmetric) Kruskal's criterion. The proof is geometric, use a study of the Hilbert function of a finite set in \(\mathbb {P}^n(\mathbb {C})\), the Castelnuovo lemma and an extension to the relative case of the Cayley-Bacharach condition.