Diophantine properties of elements of \(SO(3)\) (Q5956636)

A real number \(\alpha\) is said to be of Diophantine type \((K, \sigma)\) if \[ \Biggl|\alpha-{p\over q}\Biggr|\geq {K\over q^{2+\sigma}} \] for all rationals \(p/q\). It is well known in the theory of Diophantine approximation that given \(\varepsilon> 0\), almost all real numbers \(\alpha\) are of type \((K, 2 +\varepsilon)\) for some \(K= K(\alpha)\), or more simply are Diophantine. The notion is important in dynamical systems; in KAM theory it is used to control small denominators. The theory of Diophantine approximation has been extended to a variety of settings, from Euclidean and hyperbolic space to manifolds, but in this ground breaking paper the authors address the Diophantine property for a non-Abelian group. They show that for almost all pairs \(A\), \(B\) in \(\text{SO}(3)\), nontrivial words generated by \(A\), \(B\) of length \(n\) are at least \(c^{-n^2}\) from the identity (the most interesting case). The question stems from a conjectured analogue of the Diophantine property for SO(3)'s double cover \(\text{SU}(2)\), raised in [\textit{A. Gamburd}, \textit{D. Jakobson} and \textit{P. Sarnak}, J. Eur. Math. Soc. (JEMS) 1, No. 1, 51--85 (1999; Zbl 0916.22009)]. While the authors' result is a long way from the conjectured bound of \(c^{-n}\), the proof is not at all easy, and the authors draw on a wide range of mathematics, including the crucial fact that an integer polynomial cannot be small on a set of large measure [see \textit{D. Y. Kleinbock} and \textit{G. A. Margulis}, Ann. Math. (2) 148, No. 1, 339--361 (1998; Zbl 0922.11061)] and algebraic geometry techniques. They discuss obstructions to the exponential bound and provide informative details of related work.