Proper affine actions on semisimple Lie algebras (Q332202)

In [Topology 3, 131--139 (1964; Zbl 0136.43102)], \textit{L. Auslander} conjectured that any crystallographic group (a discrete subgroup in the affine group \(\mathrm{GL}_n(\mathbf R)\cdot R^n\) which acts properly discontinuously and such that the quotient space is compact) contains a solvable subgroup of finite index. In fact, he gave a proof of this statement, but soon a flaw was discovered in his proof. So this statement is now known as Auslander's conjecture. In [Adv. Math. 25, 178--187 (1977; Zbl 0364.55001)], \textit{J. W. Milnor} asked whether this statement is actually true for any affine subgroup which acts properly discontinuously. The answer to this question turned out to be negative: \textit{G. A. Margulis} [Sov. Math., Dokl. 28, 435--439 (1983; Zbl 0578.57012); translation from Dokl. Akad. Nauk SSSR 272, 785--788 (1983)] constructed a nonabelian free group of affine transformations with Zariski-dense linear part in the Lie group \(\mathrm{SO}(2,1)\), acting properly discontinuously on \(\mathbf R ^3\). Later this counterexample was generalized. In the paper under review, the author constructs a wide class of counterexamples to Milnor's question. The main result is the follows: Let \(G\) be a noncompact semisimple real Lie group. Consider the ``affine group'' -- the semidirect product \(G\cdot g\) for the adjoint action of \(G\) on its Lie algebra \(g\). Then there is a discrete subgroup \(\Gamma \subset G \cdot g\) whose linear part is Zariski-dense in \(G\) and which is free, nonabelian and acts properly discontinuously on the affine space corresponding to \(g\). The general strategy of the proof comes from Margulis's original paper. But some new ideas are presented.