Stationary measures and invariant subsets of homogeneous spaces. III. (Q375297)

\noindent This important paper is the third in a series of papers dealing with the ergodic theory of homogeneous spaces under real and \(p\)-adic Lie groups. For Parts I and II see [C. R., Math., Acad. Sci. Paris 347, No. 1--2, 9--13 (2009; Zbl 1155.22014); ibid. 349, No. 5--6, 341--345 (2011; Zbl 1211.22006)]. The authors have obtained random analogues of equidistribution properties of unipotent flows on homogeneous spaces of Lie groups obtained by Dani, Margulis and Ratner. \noindent One of the main theorems of the paper asserts: Let \(G\) be a real Lie group, \(\Lambda\) a lattice in \(G\) and \(X = G/\Lambda\). Let \(\Gamma\) be a compactly generated closed subsemigroup of \(G\). Further, assume that the Zariski closure of the semigroup \(Ad(\Gamma)\) is semisimple and without compact factors. Then, for every \(x \in G/\Lambda\), there exists a closed subgroup \(H\) of \(G\) containing \(\Gamma\) such that the closure of the orbit \(\Gamma x\) is the orbit \(Hx\). Moreover, \(Hx\) carries an \(H\)-invariant probability measure. \noindent Under the assumption that \(G\) is simple and \(\Gamma\) is already Zariski dense in \(G\), the authors had proved this theorem in the first of the three papers referred to above. The above theorem is already new for the striking cases such as: (i) \(G =\mathrm{SL}_2(\mathbb{R}) \times\mathrm{SL}_2(\mathbb{R})\) and \(\Gamma\) is Zariski dense in \(G\) and; (ii) \(G =\mathrm{SO}(3,1)\) and \(\Gamma\) is Zariski dense in \(\mathrm{SO}(2,1)\). \noindent The authors work more generally with the so-called weakly regular \(S\)-adic Lie groups. Such a topological group is isomorphic to a closed subgroup of products of finitely many real and \(p\)-adic Lie groups where weak regularity in a \(p\)-adic Lie group \(G_p\) is the property that two one-parameter subgroups \(\alpha, \beta : \mathbb{Q}_p \rightarrow G_p\) are equal if their derivatives at 1 are equal. The authors prove a general version of the above theorem in the context of weakly regular \(S\)-adic Lie groups. \noindent We don't state this general result but point out to two striking cases when it applies: (i) \(G =\mathrm{SL}_2(\mathbb{Q}_p)\) and \(\Gamma\) is Zariski dense and unbounded in \(G\) and; (ii) \(G =\mathrm{SL}_2(\mathbb{Q}_p) \times\mathrm{SL}_2(\mathbb{R})\) and the projection of \(\Gamma\) on each factor is Zariski dense and unbounded. \noindent Several subsidiary results are proved of independent interest. For instance, an interesting byproduct of the main theorem is: Let \(\{g_n \}\) be a sequence of independent identically distributed random elements of \(\mathrm{SL}_2(\mathbb{Z})\) whose law has finite support and generates a non-solvable group. Then, starting from any irrational point \(x\) in the torus \(\mathbb{T}^2\), almost surely, the trajectory \(g_n \cdots g_1x\) equidistributes toward the Haar probability on \(\mathbb{T}^2\). \noindent The powerful techniques introduced by the authors have already led to new, concrete, striking results in the subject and are expected to continue to do so.