A continuous model for systems of complexity 2 on simple abelian groups

From MaRDI portal
Publication:1661514

DOI10.1007/S11854-018-0037-8zbMATH Open1442.11026arXiv1509.04485OpenAlexW2962729391MaRDI QIDQ1661514FDOQ1661514


Authors: Pablo Candela, Balázs Szegedy Edit this on Wikidata


Publication date: 16 August 2018

Published in: Journal d'Analyse Mathématique (Search for Journal in Brave)

Abstract: It is known that if p is a sufficiently large prime then for every function f:mathbbZpo[0,1] there exists a continuous function on the circle f:mathbbTo[0,1] such that the averages of f and f across any prescribed system of linear forms of complexity 1 differ by at most epsilon. This result follows from work of Sisask, building on Fourier-analytic arguments of Croot that answered a question of Green. We generalize this result to systems of complexity at most 2, replacing mathbbT with the torus mathbbT2 equipped with a specific filtration. To this end we use a notion of modelling for filtered nilmanifolds, that we define in terms of equidistributed maps, and we combine this with tools of quadratic Fourier analysis. Our results yield expressions on the torus for limits of combinatorial quantities involving systems of complexity 2 on mathbbZp. For instance, let m4(alpha,mathbbZp) denote the minimum, over all sets AsubsetmathbbZp of cardinality at least alphap, of the density of 4-term arithmetic progressions inside A. We show that limpoinftym4(alpha,mathbbZp) is equal to the infimum, over all measurable functions f:mathbbT2o[0,1] with intmathbbT2fgeqalpha, of the following integral: int_{mathbb{T}^5} f�inom{x_1}{y_1}; f�inom{x_1+x_2}{y_1+y_2}; f�inom{x_1+2x_2}{y_1+2y_2+y_3}; f�inom{x_1+3 x_2}{y_1+3y_2+3y_3} ,dmu_{mathbb{T}^5}(x_1,x_2,y_1,y_2,y_3).


Full work available at URL: https://arxiv.org/abs/1509.04485




Recommendations




Cites Work






This page was built for publication: A continuous model for systems of complexity 2 on simple abelian groups

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1661514)