Explicit tensors of border rank at least 2n-1

نویسنده

  • J. M. Landsberg
چکیده

For odd m, I write down tensors in C m ⊗C m ⊗C m of border rank at least 2m − 1, showing the non-triviality of the Young-flattening equations of [6] that vanish on the matrix multiplication tensor. I also study the border rank of the tensors of [1] and [3]. I show the tensors T 2 k ∈ C k ⊗C 2 k ⊗C 2 k , of [1], despite having rank equal to 2 k+1 − 1, have border rank equal to 2 k. I show the equations for border rank of [3] on C m ⊗C m ⊗C m are trivial in the case of border rank 2m − 1 and determine their precise non-vanishing on the matrix multiplication tensor. Let A, B, C be complex vector spaces of dimensions a, b, c. A tensor T ∈ A⊗B⊗C is said to have rank one if T = a⊗b⊗c for some a ∈ A, b ∈ B, c ∈ C. More generally the rank of a tensor T ∈ A⊗B⊗C is the smallest r such that T may be written as the sum of r rank one tensors. Letˆσ 0 r ⊂ A⊗B⊗C denote the set of tensors of rank at most r. This set is not closed (under taking limits or in the Zariski topology) so letˆσ r denote its closure (the closure is the same in the Euclidean or Zariski topology). The varietyˆσ r is familiar in algebraic geometry, it is cone over the r-th secant variety of the Segre variety, but we won't need that in what follows. The rank and border rank of a tensor are measures of its complexity. While rank is natural to complexity theory, border rank is more natural from the perspective of geometry, as one can obtain lower bounds on border rank via polynomials. Let R(T), R(T) respectively denote the rank and border rank of T. The maximum rank of a tensor in C m ⊗C m ⊗C m is at most m 2 , although it is not known in general if this actually occurs. The maximum border rank of a tensor in C m ⊗C m ⊗C m is ⌈ m 3 3m−2 ⌉ for all m = 3 and five when m = 3, see [9, 8]. It is an important problem to find explicit tensors of high rank and border rank, and to develop tests that …

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

An Improvement on Ranks of Explicit Tensors

We give constructions of n × n × n tensors of rank at least 2n − O(n). As a corollary we obtain an [n] shaped tensor with rank at least 2n − O(n) when r is odd. The tensors are constructed from a simple recursive pattern, and the lower bounds are proven using a partitioning theorem developed by Brockett and Dobkin. These two bounds are improvements over the previous best-known explicit tensors ...

متن کامل

Inner Rank and Lower Bounds for Matrix Multiplication

We develop a notion of inner rank as a tool for obtaining lower bounds on the rank of matrix multiplication tensors. We use it to give a short proof that the border rank (and therefore rank) of the tensor associated with n × n matrix multiplication over an arbitrary field is at least 2n − n + 1. While inner rank does not provide improvements to currently known lower bounds, we argue that this n...

متن کامل

On the geometry of border rank algorithms for matrix multiplication and other tensors with symmetry

We establish basic information about border rank algorithms for the matrix multiplication tensor and other tensors with symmetry. We prove that border rank algorithms for tensors with symmetry (such as matrix multiplication and the determinant polynomial) come in families that include representatives with normal forms. These normal forms will be useful both to develop new efficient algorithms a...

متن کامل

Tensor Rank : Some Lower and Upper

The results of Strassen [25] and Raz [19] show that good enough tensor rank lower bounds have implications for algebraic circuit/formula lower bounds. We explore tensor rank lower and upper bounds, focusing on explicit tensors. For odd d, we construct field-independent explicit 0/1 tensors T : [n]d -+ F with rank at least 2n td/ 2j + n 8(d Ig n). This improves the lower-order terms in known low...

متن کامل

Tensor Rank, Invariants, Inequalities, and Applications

Though algebraic geometry over C is often used to describe the closure of the tensors of a given size and complex rank, this variety includes tensors of both smaller and larger rank. Here we focus on the n × n × n tensors of rank n over C, which has as a dense subset the orbit of a single tensor under a natural group action. We construct polynomial invariants under this group action whose non-v...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:
  • CoRR

دوره abs/1209.1664  شماره 

صفحات  -

تاریخ انتشار 2012