Quotients of higher-dimensional Cremona groups (Q2044121)

The Cremona group \(\mathrm{Cr}_n(\mathbb{C})\) is the group of birational transformations of the complex projective space \(\mathbb{P}^n\) of dimension \(n\). It was a long-standing question whether \(\mathrm{Cr}_n(\mathbb{C})\) is simple group. Several years ago Serge Cantat and Stéphane Lamy made a breakthrough in [\textit{S. Cantat} et al., Acta Math. 210, No. 1, 31--94 (2013; Zbl 1278.14017)] by proving that \(\mathrm{Cr}_2(\mathbb{C})\) is not simple. In the paper under review, the authors proved the remarkable result that \(\mathrm{Cr}_n(\mathbb{C})\) is not simple for any \(n\geq 3\). Note that the previous proof for \(n=2\) cannot be generalized to \(n\geq 3\) and the proof for \(n\geq 3\) in this paper fails when \(n=2\). The authors prove the non-simplicity of \(\mathrm{Cr}_n(\mathbb{C})\) by constructing a surjective homomorphism of groups from \(\mathrm{Cr}_n(\mathbb{C})\) to a group \(A\) isomorphic to an infinite free sum of \(\mathbb{Z}/2\mathbb{Z}\). For this purpose they consider the groupoid \(\mathrm{BirMori}(n)\) of birational maps between rational Mori fiber spaces which contains \(\mathrm{Cr}_n(\mathbb{C})\) as a subgroupoid. They first construct a homomorphism of groupoids from \(\mathrm{BirMori}(n)\) to \(A\). Then they restrict it to \(\mathrm{Cr}_n(\mathbb{C})\) to get a group homomorphism. The construction relies on the Sarkisov Program of factorization of birational maps. The Sarkisov Program says that \(\mathrm{BirMori}(n)\) is generated by some birational maps of particular forms called Sarkisov links. The authors define first a map on these generators and then prove that the map is trivial on the ralations between these generators. The bulk of the construction is to understand the relations between Sarkisov links. The authors introduced a new concept called rank-\(r\) fibrations, which are Mori dream spaces with relative Picard rank \(r\) satisfying some nice conditions on singularities. The idea is that rank-\(1\) fibrations are Mori fiber spaces, rank-\(2\) fibrations correspond to Sarkisov links, rank-\(3\) fibrations correspond to relations between Sarkisov links, etc. The authors proved that the relations given by rank-\(3\) fibrations generate all relations between Sarkisov links. Roughly speaking, the authors consider some special types of Sarkisov links and their groupoid homomorphism \(\mathrm{BirMori}(n)\rightarrow A\) counts how many Sarkisov links of these types there are in the factorization of a birational map. By their description of relations between Sarkisov links and by BAB Conjecture recently proved by Caucher Birkar, they manage to prove that if such a special Sarkisov link appears in a relation, then it must appear in a pair. Therefore the counting leads to a homomorphism towards a free sum of \(\mathbb{Z}/2\mathbb{Z}\). The paper is well written and is an important piece of work towards the understanding of birational maps. It also contains many consequences of the above construction.