Tensor Rank: Matching Polynomials and Schur Rings

Dima Grigoriev, Mikhail Muzychuk, Ilya Ponomarenko

Research output: Contribution to journalArticlepeer-review

Abstract

We study the polynomial equations vanishing on tensors of a given rank. By means of polarization we reduce them to elements A of the group algebra ℚ[Sn × Sn] and describe explicit linear equations on the coefficients of A to vanish on tensors of a given rank. Further, we reduce the study to the Schur ring over the group Sn × Sn that arises from the diagonal conjugacy action of Sn. More closely, we consider elements of ℚ[Sn × Sn] vanishing on tensors of rank n-1 and describe them in terms of triples of Young diagrams, their irreducible characters, and nonvanishing of their Kronecker coefficients. Also, we construct a family of elements in ℚ[Sn × Sn] vanishing on tensors of rank n-1 and illustrate our approach by a sharp lower bound on the border rank of an explicitly produced tensor. Finally, we apply this construction to prove a lower bound 5n2/4 on the border rank of the matrix multiplication tensor (being, of course, weaker than the best known one (2-ε)n2, due to Landsberg, Ottaviani).

Original languageEnglish
Pages (from-to)457-481
Number of pages25
JournalFoundations of Computational Mathematics
Volume14
Issue number3
DOIs
StatePublished - 1 Jan 2014
Externally publishedYes

Keywords

  • Matching polynomial
  • Schur ring
  • Tensor rank

ASJC Scopus subject areas

  • Analysis
  • Computational Mathematics
  • Computational Theory and Mathematics
  • Applied Mathematics

Fingerprint

Dive into the research topics of 'Tensor Rank: Matching Polynomials and Schur Rings'. Together they form a unique fingerprint.

Cite this