English

Permutations Unlabeled beyond Sampling Unknown

Information Theory 2019-06-26 v2 Signal Processing math.IT

Abstract

A recent unlabeled sampling result by Unnikrishnan, Haghighatshoar and Vetterli states that with probability one over iid Gaussian matrices AA, any xx can be uniquely recovered from an unknown permutation of y=Axy = A x as soon as AA has at least twice as many rows as columns. We show that this condition on AA implies something much stronger: that an unknown vector xx can be recovered from measurements y=TAxy = T A x, when the unknown TT belongs to an arbitrary set of invertible, diagonalizable linear transformations T\mathcal{T}. The set T\mathcal{T} can be finite or countably infinite. When it is the set of m×mm \times m permutation matrices, we have the classical unlabeled sampling problem. We show that for almost all AA with at least twice as many rows as columns, all xx can be recovered either uniquely, or up to a scale depending on T\mathcal{T}, and that the condition on the size of AA is necessary. Our proof is based on vector space geometry. Specializing to permutations we obtain a simplified proof of the uniqueness result of Unnikrishnan, Haghighatshoar and Vetterli. In this letter we are only concerned with uniqueness; stability and algorithms are left for future work.

Cite

@article{arxiv.1812.00498,
  title  = {Permutations Unlabeled beyond Sampling Unknown},
  author = {Ivan Dokmanić},
  journal= {arXiv preprint arXiv:1812.00498},
  year   = {2019}
}
R2 v1 2026-06-23T06:28:37.625Z