Rational endomorphisms of codimension one holomorphic foliations (Q2164376)

Rational endomorphisms with interesting dynamics exist only on quite special complex projective manifolds (Abelian varities, projectives spaces\(\dots\) ). This article aims at generalizing this observation to the foliated setting: the typical situation studied here is a (smooth) projective variety \(X\), endowed with a codimension one (singular) foliation \(\mathcal{F}\) and a rational endomorphism \(f:X\dashrightarrow X\) preserving \(\mathcal{F}\) (\emph{i.e.} \(f^*\mathcal{F}=\mathcal{F}\)). One of the guide lines of this article is a conjecture formulated by Cerveau and Lins-Neto that predicts that codimension one foliations should be either obtained as the pull-back of a foliation on a projective surface or should admit a \emph{projective transverse structure}. The notion of transverse structures of codimension foliation is a central theme of this article and the authors use the following by now classical hierarchy of transverse structure: \begin{itemize} \item (virtually) transversely additive: \(\mathcal{F}\) is defined by a closed rational \(1\)-form (up to a generically finite cover); \item transversely affine: \(\mathcal{F}\) is defined by \(\omega\) such that \(\nabla(\omega)=0\) for a rational flat connection on \(N_\mathcal{F}\) the normal bundle of \(\mathcal{F}\); \item transversely projective: there exists a triple \((\omega_0,\omega_1,\omega_2)\) of rational \(1\)-forms such that \(d\omega_0=\omega_0\wedge\omega_1\), \(d\omega_1=2\omega_0\wedge\omega_2\) and \(d\omega_2=\omega_1\wedge\omega_2\). \end{itemize} This last condition has a geometric interpretation in terms of \(\mathbb{P}^1\)-bundle (explaining by the way the terminology). The rational endomorphisms \(f\) as above form a semi-group \(\mathrm{End}(X,\mathcal{F})\) and the complexity of the dynamics is measured in terms of the transverse action: two endomorphisms are identified if they acts in the same way on the ``space of leaves''. This is an equivalence relation on \(\mathrm{End}(X,\mathcal{F})\) and the quotient should be thought of as the transverse action. The surface case was studied in [\textit{C. Favre} and \textit{J. V. Pereira}, Math. Z. 268, No. 3--4, 753--770 (2011; Zbl 1237.37039)] where it was proven that a foliation \(\mathcal{F}\) with \(\mathrm{End}(X,\mathcal{F})\) infinite is automatically virtually transerversely additive (Theorem~2.3 of this article). It is thus natural to consider the transversely projective case in view of the above conjecture (Cerveau-Lins Neto). This is Theorem~A of this article: if \(\mathcal{F}\) is transversely projective and purely transcendental (meaning that the leaf through a general point of \(X\) does not contain any positive dimensional algebraic subvarities) and if the transverse action of \(\mathrm{End}(X,\mathcal{F})\) is infinite, then \(\mathcal{F}\) is virtually transversely additive. Among transversely projective foliations, there is a prominent class: the tranversely hyperbolic ones (such an \(\mathcal{F}\) is given by local maps to the disk \(\mathbb{D}\) and the changes of foliated coordinates are given by elements of \(\mathrm{PSL}_2(\mathbb{R})\)). This class has been intensively studied by \textit{F. Touzet} [in: Foliation theory in algebraic geometry. Proceedings of the conference, New York, NY, USA, September 3--7, 2013. Cham: Springer. 157--216 (2016; Zbl 1353.37099)] and \textit{J. V. Pereira} et al. [Compos. Math. 158, No. 6, 1428--1447 (2022; Zbl 1498.32007)]. Here they prove a striking fact (Theorem~5.13): if \(\mathcal{F}\) is transversely hyperbolic and \(f:X\dashrightarrow X\) is a rational endomorphism such that \(f^*\) fixes the monodromy representation associated with \(\mathcal{F}\), then \(f\in \mathrm{End}(X,\mathcal{F})\) (``the monodromy determines the hyperbolic structure''). This is an important tool in the above stated theorem. As an illustration of the first sentence of this review, the authors prove the following result (Theorem~B): if \(f\in \mathrm{End}(X,\mathcal{F})\) with \(\mathcal{F}\) transversely projective and purely transcendental has a Zariski-dense orbit, then \(f\) is conjugated to a Lattes-like exmaple. It exactly means that \(f\) comes from an affine map on an Abelian varity; in particular \(X\) is the image of such an Abelian variety by a generically finite map and is thus rather special. This article contains a lot of very interesting results concerning codimension one foliations, their symmetries, the geometry of \(1\)-forms\(\dots\) the statements presented above are just a small sample of the content of this very well written article.