English

Rank probabilities for real random $N\times N\times 2$ tensors

Probability 2012-01-11 v1 Algebraic Geometry

Abstract

We prove that the probability PNP_N for a real random Gaussian N×N×2N\times N\times 2 tensor to be of real rank NN is PN=(Γ((N+1)/2))N/G(N+1)P_N=(\Gamma((N+1)/2))^N/G(N+1), where Γ(x)\Gamma(x), G(x)G(x) denote the gamma and Barnes GG-functions respectively. This is a rational number for NN odd and a rational number multiplied by πN/2\pi^{N/2} for NN even. The probability to be of rank N+1N+1 is 1PN1-P_N. The proof makes use of recent results on the probability of having kk real generalized eigenvalues for real random Gaussian N×NN\times N matrices. We also prove that logPN=(N2/4)log(e/4)+(logN1)/12ζ(1)+O(1/N)\log P_N= (N^2/4)\log (e/4)+(\log N-1)/12-\zeta '(-1)+{\rm O}(1/N) for large NN, where ζ\zeta is the Riemann zeta function.

Keywords

Cite

@article{arxiv.1106.5581,
  title  = {Rank probabilities for real random $N\times N\times 2$ tensors},
  author = {G. Bergqvist and P. J. Forrester},
  journal= {arXiv preprint arXiv:1106.5581},
  year   = {2012}
}

Comments

8 pages