A quantitative version of the non-Abelian idempotent theorem (Q633235)

Let \(G\) be a finite (not necessarily abelian) group with Haar measure \(\mu\). Any complex-valued function \(f\) on \(G\) induces a convolution operator \(L_f\) from \(L^2(G)\) to \(L^2(G)\) by \(g\mapsto f*g\), where \[ f*g(x):=\int f(y)g(y^{-1}x)\,d\mu(y). \] It is not difficult to see that the algebra norm defined by \[ \|f\|_{A(G)}:= \sup\{|\langle f,g\rangle_{L^{2}(\mu)} : \|L_g\|\leq 1\}, \] where \(\|\cdot\|\) denotes the operator norm, reduces to the well known Wiener algebra in the case when \(G\) is abelian, i.e. \(\|f\|_{A(G)}=\|\hat{f}\|_{\ell^1}\). It is also relatively straightforward to verify, even in the non-abelian case, that the indicator function of a coset of a subgroup has \(A(G)\) norm equal to 1. The triangle inequality implies therefore that small plus/minus combinations of such indicator functions also have small \(A(G)\) norm. The paper under review establishes the following converse of this observation, stating that all functions with small \(A(G)\) norm are essentially of this form. Theorem. Let \(A\subseteq G\) be such that \(\|1_A\|_{A(G)} \leq M\). Then there is an integer \(L=L(M)\), subgroups \(H_1, \dots,H_L \leqslant G\), elements \(x_1, \dots, x_L \in G\) and signs \(\sigma_1, \dots, \sigma_L \in \{-1,0,1\}\) such that \[ 1_A=\sum_{i=1}^L \sigma_i 1_{x_i H_i}, \] where \(L\) may be taken to be at most triply tower in \(O(M)\). Inspired by a result of \textit{P. Cohen} [Am. J. Math. 82, 191--212 (1960; Zbl 0099.25504)], the abelian case of this theorem had been proved by \textit{B. Green} and \textit{T. Sanders} in [Ann. Math. (2) 168, No. 3, 1025--1054 (2008; Zbl 1170.43003)], and a qualitative version of the non-abelian theorem is due to \textit{M. Lefranc} [C. R. Acad. Sci., Paris, Sér. A 274, 1882--1883 (1972; Zbl 0247.43014)]. Proving the theorem above, however, is not the only aim of the paper under review. Perhaps even more importantly, the author develops a version of the non-abelian Fourier transform relative to a suitable notion of ``approximate group'', widely used in the abelian context since the work of \textit{J. Bourgain} [Geom. Funct. Anal. 9, No. 5, 968--984 (1999; Zbl 0959.11004)]. This allows him to transfer several results from additive combinatorics, including a Freiman-type result, to the non-abelian setting. Although the paper contains an excellent introduction to guide the reader, he or she is well advised, given its conceptual as well as methodological complexity, to be comfortable with the general strategy and technical tools of [Green and Sanders, loc. cit.] before embarking on the present paper.