Surfaces et cohomologie bornée. (Surfaces and bounded cohomology) (Q1100769)

Let \(\Gamma\) be a discrete group. For \(x\in H_ 2(\Gamma;{\mathbb{Z}})=H_ 2(K(\Gamma,1),{\mathbb{Z}})\) let g(x) be the minimal genus of a singular surface representing x. (A singular surface is a map \(f: F\to K(\Gamma,1)\); it represents x if \(x=f_ x[F]\); its genus is the genus of F). The authors relate g(x) and the Thurston-Gromov norm \(\| x\|\) of x, considered as an element in \(H_ 2(\Gamma;{\mathbb{R}})\) by \[ \| x\| =\lim_{n\to \infty}4g(nx)/n \] thus showing that g contains more information than \(\|\|\). They substantiate this claim by showing that for each k there is a finitely presented nilpotent group \(\Gamma\) such that \(H_ 2(\Gamma;{\mathbb{Z}})\) is not generated by classes x with g(x)\(\leq h\) (Théorème 2.1), while \(\| \|\) vanishes on all amenable groups (in particular on solvable groups). In this context they ask: Given an amenable group \(\Gamma\). Is for any \(x\in H_ 2(\Gamma;{\mathbb{Z}})\) the sequence g(nx) bounded? The answer is yes for polycyclic groups (Théorème 2.2). In the second part of the paper the authors consider fundamental groups of surfaces of genus \(>1\). They associate with each 2-form on the surface a bounded cocycle and show that the induced map \(\Omega^ 2(surface)\to H^ 2_{bounded}(surface)\) is injective, thus giving another proof that \(H^ 2_{bounded}(surface)\) is a huge space (Théorème 3.2). From this they deduce that a surface of negative curvature all of whose ideal triangles in the universal covering have the same area is already a surface of constant negative curvature (Théorème 3.11).