A geometric criterion for the boundedness of characteristic classes (Q652234)

The authors characterize connected Lie groups \(G\) whose Borel cohomology classes are bounded. Recall that a Borel \(n\)-cochain of \(G\) is a map \(G^n\to \mathbb Z\) such that the inverse image of any subset of \(\mathbb Z\) is a Borel set. The set of Borel \(n\)-cochains being stable with the differential, it gives rise to cohomology groups, \(H^*_B(G;\mathbb Z)\), called Borel cohomology groups of \(G\). A class in this cohomology is said to be bounded if it admits a representative Borel cocycle whose image is finite. Finally, recall that the radical of \(G\) is the biggest connected normal solvable subgroup of \(G\). The main result of this work is the equivalence of the two following conditions. -1- The radical of \(G\) is linear. -2- Each Borel cohomology class of \(G\) can be represented by a Borel bounded cocycle. The authors prove also the equivalence with a third condition involving distortion. Let \(G\) be a virtually connected Lie group and \(G^{\delta}\) be the same group endowed with the discrete topology. As a corollary of the quoted theorem, the authors prove: if the radical of \(G\) is linear, then the image of the natural map \(H^*(BG;\mathbb R)\to H^*(BG^{\delta};{\mathbb R})\) consists of bounded classes. This result is a generalization of a theorem of \textit{M. Gromov} [``Volume and bounded cohomology'', Inst. Hautes Études Sci. Publ. Math. 56, 5--99 (1982; Zbl 0516.53046)], whose original statement was concerning real Lie groups associated to linear algebraic groups defined over \(R\). Finally, the authors observe that the hypothesis of linearity of the radical is still too strong. It is worth underlining that the paper is very clearly written with many examples, remarks and comments.