Infinite permutation groups. I: Products of conjugacy classes (Q1117026)

\textit{R. D. Anderson} [Am. J. Math. 80, 955-963 (1958; Zbl 0090.388)] showed that the automorphism group Aut(\(\Omega)\) of a topological space \(\Omega\) is simple when \(\Omega\) satisfies certain homogeneity conditions (``Anderson's criterion''). More precisely he showed that under these conditions: whenever \(\sigma\),\(\tau\in Aut(\Omega)\) and \(\sigma\neq 1\), then \(\tau\) is a product of six conjugates of \(\sigma\) or \(\sigma^{- 1}\). The first part of the present paper shows that if \(\Omega\) is either the set of rationals \({\mathbb{Q}}\) or the irrationals with the usual topology, then the number ``six'' in the statement above can be reduced to ``three'' (Theorem 2.1). The second part of the paper deals with point stabilizers of Aut(\(\Omega)\). In the case that \(\Omega\) is either the rationals, the irrationals or the Cantor set, the author finds the smallest number of conjugates (never more than six) which suffice when \(\sigma\) and \(\tau\) lie in a one-point stabilizer of Aut(\(\Omega)\) and \(\tau\) is in the normal closure of \(\sigma\) (Corollary 3.14). The final section of the paper considers the group Aut(\(\Omega\),\(\leq)\) where \(\Omega\) is the rationals or the reals with the usual order relation. The author examines the conjugacy classes K of Aut(\(\Omega\),\(\leq)\) which have the property that \(K^ 2\) is a union of a finite number of conjugacy classes, and shows how this gives a proof that Aut(\({\mathbb{Q}},\leq)\) and Aut(\({\mathbb{R}},\leq)\) are not elementarily equivalent. This last result was already known from the work of \textit{Yu. Gurevich} and \textit{W. C. Holland} [Trans. Am. Math. Soc. 265, 527-534 (1981; Zbl 0471.03033)], but the proof here may be simpler.