On the equivalence of quasirandomness and exchangeable representations independent from lower-order variables
Leonardo N. Coregliano, Henry P. Towsner

TL;DR
This paper explores when quasirandomness properties in exchangeable graph representations imply simpler, independent representations, introducing a new $ ext{ extasterisk}$-representation concept that extends the classical theon framework.
Contribution
It introduces $ ext{ extasterisk}$-representations with added random permutations and proves their equivalence to quasirandomness conditions, resolving a question by Coregliano and Razborov.
Findings
UCouple[$ ext{ extasterisk}$] implies $ ext{ extasterisk}$-$ ext{ extasterisk}$-independence
$ ext{ extasterisk}$-representations are necessary for certain quasirandomness properties
The results generalize the Aldous--Hoover theorem to more complex structures.
Abstract
It is often convenient to represent a process for randomly generating a graph as a graphon. (More precisely, these give \emph{vertex exchangeable} processes -- those processes in which each vertex is treated the same way.) Other structures can be treated by generalizations like hypergraphons, permutatons, and, for a very general class, theons. These representations are not unique: different representations can lead to the same probability distribution on graphs. This naturally leads to questions (going back at least to Hoover's proof of the Aldous--Hoover Theorem on the existence of such representations) that ask when quasirandomness properties on the distribution guarantee the existence of particularly simple representations. We extend the usual theon representation by adding an additional datum of a random permutation to each tuple, which we call a -representation. We show…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsMatrix Theory and Algorithms
