Generalized Kruskal conjecture for product-vector sums

At least 7 years old · documented by

Let n≥2n\geq 2, m≥1m\geq 1, and d1,…,dm≥1d_1,\dots,d_m\geq 1 be integers. Let X1,…,Xm\mathcal{X}_1,\dots,\mathcal{X}_m be vector spaces over a field F\mathbb{F}, and let

{xa=xa,1⊗⋯⊗xa,m:a∈{1,…,n}}⊂X1⊗⋯⊗Xm\{x_a=x_{a,1}\otimes\dots\otimes x_{a,m}:a\in\{1,\dots,n\}\}\subset \mathcal{X}_1\otimes\dots\otimes\mathcal{X}_m

be non-zero product vectors such that, for each j∈{1,…,m}j\in\{1,\dots,m\},

dim⁡span⁡{xa,j:a∈{1,…,n}}≥dj.\dim\operatorname{span}\{x_{a,j}:a\in\{1,\dots,n\}\}\geq d_j.

If n−1≤∑j=1m(dj−1)n-1\leq\sum_{j=1}^m(d_j-1) and

∑a=1nxa=0,\sum_{a=1}^n x_a=0,

then there exists a non-empty strict subset Γ⊂{1,…,n}\Gamma\subset\{1,\dots,n\} such that

∑a∈Γxa=0.\sum_{a\in\Gamma}x_a=0.

Generalized Kruskal conjecture. Under these hypotheses, a nontrivial zero-sum subcollection must exist. This relaxes general position to lower bounds on the dimensions of the subsystem spans and replaces pairwise cancellation by cancellation of an arbitrary non-empty strict subcollection. The stated consequence was previously proved, so this conjectural formulation is solved.

References

Primary source

Benjamin Lovitz, “Toward a generalization of Kruskal's theorem on tensor decomposition”, arXiv:1812.00264 (2020).

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.