The next gap in the subrank of 3-tensors
Fulvio Gesmundo, Jeroen Zuiddam

TL;DR
This paper identifies the next possible value gap in the asymptotic subrank and slice rank of 3-tensors, extending previous work on the discrete set of these tensor parameters.
Contribution
It precisely determines the next gap in the set of possible asymptotic subrank and slice rank values for 3-tensors, building on prior discrete value classifications.
Findings
The asymptotic subrank and slice rank can be exactly 1, 1.88, 2, or at least 2.68.
The set of possible values of these parameters is discrete.
The paper extends the understanding of tensor rank gaps.
Abstract
Recent works of Costa-Dalai, Christandl-Gesmundo-Zuiddam, Blatter-Draisma-Rupniewski, and Bri\"et-Christandl-Leigh-Shpilka-Zuiddam have investigated notions of discreteness and gaps in the possible values that asymptotic tensor ranks can take. In particular, it was shown that the asymptotic subrank and asymptotic slice rank of any nonzero 3-tensor is equal to 1, equal to 1.88, or at least 2 (over any field), and that the set of possible values of these parameters is discrete (in several regimes). We determine exactly the next gap, showing that the asymptotic subrank and asymptotic slice rank of any nonzero 3-tensor is equal to 1, equal to 1.88, equal to 2, or at least 2.68.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsTensor decomposition and applications · Parallel Computing and Optimization Techniques · Computational Physics and Python Applications
