Generic subtypical Segre identifiability conjecture

Let d3d \ge 3 and n1nd2n_1 \ge \cdots \ge n_d \ge 2. Let SF=Seg(PFn1××PFnd)\mathcal{S}_{\mathbb{F}}=\operatorname{Seg}(\mathbb{P}\mathbb{F}^{n_1}\times\cdots\times\mathbb{P}\mathbb{F}^{n_d}) be the Segre variety in P(Fn1Fnd)\mathbb{P}(\mathbb{F}^{n_1}\otimes\cdots\otimes\mathbb{F}^{n_d}). A decomposition is generically rr-identifiable when a generic tensor of rank rr has a unique decomposition. Generic subtypical Segre identifiability conjecture. The variety SF\mathcal{S}_{\mathbb{F}} is generically rr-identifiable for every strictly subtypical rank r<rSFr<\overline r_{\mathcal{S}_{\mathbb{F}}}, except in the six listed cases: (1) n1>k=2dnkk=2d(nk1)n_1>\prod_{k=2}^d n_k-\sum_{k=2}^d(n_k-1) and rk=2dnkk=2d(nk1)r\ge\prod_{k=2}^d n_k-\sum_{k=2}^d(n_k-1); (2) Seg(PF4×PF4×PF3)\operatorname{Seg}(\mathbb{P}\mathbb{F}^{4}\times\mathbb{P}\mathbb{F}^{4}\times\mathbb{P}\mathbb{F}^{3}) with r=5r=5; (3) Seg(PFn×PFn×PF2×PF2)\operatorname{Seg}(\mathbb{P}\mathbb{F}^{n}\times\mathbb{P}\mathbb{F}^{n}\times\mathbb{P}\mathbb{F}^{2}\times\mathbb{P}\mathbb{F}^{2}) with r=2n1r=2n-1; (4) Seg(PF4×PF4×PF4)\operatorname{Seg}(\mathbb{P}\mathbb{F}^{4}\times\mathbb{P}\mathbb{F}^{4}\times\mathbb{P}\mathbb{F}^{4}) with r=6r=6; (5) Seg(PF6×PF6×PF3)\operatorname{Seg}(\mathbb{P}\mathbb{F}^{6}\times\mathbb{P}\mathbb{F}^{6}\times\mathbb{P}\mathbb{F}^{3}) with r=8r=8; or (6) Seg(PF2)×5\operatorname{Seg}(\mathbb{P}\mathbb{F}^{2})^{\times 5} with r=5r=5. The first three cases generically admit infinitely many decompositions; case (4) generically admits two complex decompositions, case (5) is expected to generically admit six complex decompositions, and case (6) generically admits two complex decompositions. The conjecture was initially stated over C\mathbb{C}; computational results establish it there when n1nd15000n_1\cdots n_d\le 15000, with an extension to R\mathbb{R} described in the source.

Sources & referencesView supporting material

Primary source

Luca Chiantini, Giorgio Ottaviani and Nick Vannieuwenhoven, “Effective criteria for specific identifiability of tensors and forms”, arXiv:1609.00123 (2016).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.