Fourier and small ball estimates for word maps on unitary groups
Abstract
To a non-trivial word w(x1,...,xr) in a free group Fr on r elements and a group G, one can associate the word map wG:Gr→ G that takes an r-tuple (g1,...,gr) in Gr to w(g1,...,gr). If G is compact, we further associate the word measure τw,G, defined as the distribution of wG(X1,...,Xr), where X1,...,Xr are independent and Haar-random elements in G. In this paper we study word maps and word measures on the family of special unitary groups \ SUn\ n≥2. Our first result is a small ball estimate for wSUn. We show that for every w∈ Fr\ 1\ there are ε(w),δ(w)>0 such that if B⊂eqSUn is a ball of radius at most δ(w)diam(SUn) in the Hilbert-Schmidt metric, then τw,SUn(B)≤(μSUn(B))ε(w), where μSUn is the Haar probability measure. Our second main result is about the random walks generated by τw,SUn. We provide exponential upper bounds on the large Fourier coefficients of τw,SUn, and as a consequence we show there exists t(w)∈N, such that τw,SUn*t has bounded density for every t≥ t(w) and every n≥2, answering a conjecture by the first two authors. As a key step in the proof, we establish, for every large irreducible character of SUn, an exponential upper bound of the form |(g)|<(1)1-ε, for elements g in SUn whose eigenvalues are sufficiently spread out on the unit circle in C×.
Turn this paper into a lesson
ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.