Suppose we work in an algebraically closed field. Then, do the Waring rank (symmetric tensor rank) and tensor rank of a symmetric tensor coincide in general? Recall that tensor rank is rank with respect to the Segre variety and Waring rank is rank with respect to the Veronese variety.
Asked
Active
Viewed 876 times
1 Answers
10
A counterexample in $\mathbb{C}$ is given in A counterexample to Comon's conjecture:
We present an example of a symmetric tensor of size 800×800×800 which can be written a sum of 903 simple tensors with complex entries but not as a sum of 903 symmetric simple tensors.
It's a very recent result by Yaroslav Shitov, not yet published but it has survived some scrutiny by the tensor community.
Carlo Beenakker
- 177,695
-
Thank you @Carlo! Do you know if they coincide for monomials (special symmetric tensors)? I will look at the paper and some references in it though. – SMD Oct 22 '17 at 18:27
-
@SMD For Waring ranks of monomials, see Carlini, Catalisano, Geramita, The solution to the Waring problem for monomials and the sum of coprime monomials (J. Algebra, 2012) (also https://arxiv.org/abs/1112.3474). I don't know a reference for the tensor rank of a monomial; I'm not sure if it's known. – Zach Teitler Oct 23 '17 at 05:41
-
@SMD Some positive results are: here, Corollary 10; here; here. Each of these shows that the conclusion of Comon's conjecture (equality of tensor rank and symmetric tensor rank, aka Waring rank) holds in some special cases. I don't think that monomials are included in any of these special cases. – Zach Teitler Oct 30 '17 at 05:46
-
4Shitov's paper has been published today in SIAM J. Appl. Algebra Geometry: https://epubs.siam.org/doi/abs/10.1137/17M1131970 – Tadashi Sep 11 '18 at 23:39