On the dynamical degree of surjective endomorphisms
Ilya Karzhemanov · 2026
Read On the dynamical degree of surjective endomorphisms by Ilya Karzhemanov (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-24173.txt.
Introduction
Karzhemanov studies surjective rational maps on smooth projective surfaces. He proves a couple of dynamical properties of such maps. He also gives a numerical condition that ensures the map is regular. Worked examples and calculations in the study relate to the topological entropy of the map. The results add basic structure facts for surface dynamics. His test gives a concrete check that can be applied to given examples.
Catalog record
| CREATOR | Ilya Karzhemanov |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 15450 |
| LICENSE | http://creativecommons.org/publicdomain/zero/1.0/ |
| RIGHTS | CC0 |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.24173
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the dynamical degree of surjective endomorphisms Thanks: MS 2010 classification : 14E05, 37F10, 14J26 Thanks: Key words : surjective rational map, del Pezzo surface, dynamical degree Ilya Karzhemanov Abstract . We establish a couple of dynamical properties of surjective rational maps f : X ⇢ X f:X\dashrightarrow X for smooth projective surfaces X X . We also give a numerical characterization of regular f f in the case when X X is a del Pezzo surface. Some explicit constructions and calculations, related to the topological entropy of f f , are provided. 1. Introduction 1.1 . Let X be a smooth projective complex surface. In the paper [ 11 ] , we have initiated the study of the geometry of surjective endomorphisms f:X\dashrightarrow X , that is rational maps satisfying the property f(X\setminus Z)=X for some finite subset Z\subset X (cf. [ 9 ] , [ 10 ] and [ 8 ] for statistical, physical and computational aspects of the story). The aim of the present note is to highlight some dynamical sides of such maps f — again in the case when the divisor -K_{X} is ample (i.e. when X is a del Pezzo surface ). 1) 1) 1) We refer to [ 5 , Ch. 8] for the basic properties of such X . Consider the Néron–Severi group \text{Pic}(X)\otimes_{\mathbb{Z}}\mathbb{R}=N^{1}(X) . It follows from [ 11 , 2.1] that f induces a linear map f^{*}:N^{1}(X)\to N^{1}(X) with the property (f^{n})^{*}=(f^{*})^{n}\ \forall\ n\in\mathbb{N}\quad\text{for the corresponding iterates}. The maps with this algebraic stability behavior had acquired a great deal of studies in rational dynamics (see [ 3 ] , [ 16 ] , [ 15 ] ). In this regard, one important numerical characteristic of f is the (first) dynamical degree \lambda(f) , defined as follows. Let H\in\text{Pic}(X) be an ample divisor. Then \lambda(f):=\lim_{n\to\infty}\sqrt[n]{((f^{n})^{*}(H)\cdot H)}. This definition does not actually depend on the choice of H . Here is a (classical) relation between \lambda(f) and the spectral radius \rho(f^{*}) of the linear operator f^{*} : Proposition 1.2 . \lambda(f)=\rho(f^{*}) and there is a ( Perron ) vector v\in N^{1}(X) , which is a nef class on X , satisfying f^{*}(v)=\lambda(f)v . This claim is very well-known to specialists of course. Yet we include its proof for transparency and in order to adapt to our present setting (see Section 2 below). Remark 1.3 . Proposition 1.2 implies that \lambda(f) is an algebraic integer because f^{*} preserves the (full) lattice \text{Pic}(X) . However, in the case of an arbitrary rational dominant map X\dashrightarrow X , already when X=\mathbb{P}^{2} , its dynamical degree can be transcendental (see [ 2 ] ). Note also that Proposition 1.2 holds for any surjective map f and any surface X with irregularity q(X)=0 (see the arguments in Section 2 ). 1.4 . In Section 3 , we establish a relation between \lambda(f) and the (topological) degree \deg f of f , namely that \lambda(f)^{2}\geq\deg f . Again, this is a classical log-concavity property of the dynamical degrees, which we prove via algebro-geometric means. Next, following the steps of the proof, we obtain a numerical characterization of regularity (in the current setting): Theorem 1.5 . The map f is regular iff \lambda(f)^{2}=\deg f . Note that the implication “ f is regular \Longrightarrow\lambda(f)^{2}=\deg f ” is evident via the projection formula f_{*}f^{*}=(\deg f)\text{Id} . In turn, the opposite direction requires some extra reasoning, which seems to has not been done previously. Remark 1.6 . The case of monomial endomorphisms of \mathbb{P}^{2} shows that Theorem 1.5 does not hold for arbitrary rational dominant maps X\dashrightarrow X (see [ 6 , Section 3] ). The paper closes with Section 4 , where we discuss some basic properties of the entropy of f , which we further support by explicit calculations for a wide class of surjective endomorphisms. We also consider a seemingly interesting class of rank 1 maps (see Definition 4.5 ) and discuss their structure. Acknowledgments. I am grateful to I. Zhdanovskiy for fruitful conversations. The work was carried out at the Center for Pure Mathematics (MIPT) and was supported by the Russian Science Foundation under grant № 25-21-00083 (https://rscf.ru/project/25-21-00083/). 2. Spectral interpretation (revisited) 2.1 . Let V be a finite-dimensional real normed space and K\subset V a closed cone with nonempty interior. Suppose T:V\longrightarrow V is a linear map such that T(K)\subseteq K . Denote by \rho(T) the spectral radius of T . Let us recall the following general result (see [ 13 ] ): Theorem 2.2 (Krein–Rutman) . In the previous setting, \rho(T) is an eigenvalue of T and there exists an eigenvector v\in K,v\neq 0 , such that T(v)=\rho(T)v . 2.3 . We retain the notations from Introduction. Take V:=N^{1}(X) , T:=f^{*} and K:= the nef cone of X (cf. [ 11 , Lemma 2.5] ), so that Theorem 2.2 provides a nef \mathbb{R} -class v on X with f^{*}(v)=\rho(f^{*})v . Choose the divisor H from 1.1 above in the form v+H^{\prime} for some ample H^{\prime}\in N^{1}(X) . Lemma 2.4 . There exists an absolute constant C such that C^{-1}\rho(f^{*})^{n}\leq((f^{*})^{n}(H)\cdot H)\leq C\rho(f^{*})^{n}. Proof. The upper bound follows from the operator norm growth. In turn, the lower bound follows from the fact that v is an eigenvector of f^{*} and that the intersections H\cdot v,H\cdot H^{\prime} are positive. ∎ In order to prove Proposition 1.2 it suffices to take the n^{\text{th}} roots of the inequalities in Lemma 2.4 and then pass to the limit n\to\infty . Remark 2.5 . Note that the class v is not necessarily big (i.e. it need not satisfy (v^{2})>0 ). As an example we consider X:=\mathbb{P}^{1}_{\bf t}\times\mathbb{P}^{1}_{\bf w} and its (regular) endomorphism f([t_{0}:t_{1}]\times[w_{0}:w_{1}]):=[t^{d}_{0}:t^{d}_{1}]\times[w_{0}:w_{1}] for some d\geq 2 . In the basis of N^{1}(X) , corresponding to two rulings on X , the operator f^{*} is represented by the matrix \begin{pmatrix}d&0\\ 0&1\end{pmatrix} . Then we have \rho(f^{*})=d and (v^{2})=0 . 3. Relation between \lambda and \deg 3.1 . Let us see first that \lambda(f)^{2}\geq\deg f in our setting. Consider a resolution diagram for f : \textstyle{Y\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{g} \scriptstyle{h} \textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces} \scriptstyle{f} \textstyle{X.} Here g is a composition of blow-ups and h is a regular morphism. Then for H from 1.1 we have f^{*}(H)=g_{*}(h^{*}(H)) and so h^{*}(H)=g^{*}(f^{*}(H))-E, where E:=\displaystyle\sum_{i}a_{i}E_{i} , E_{i} are g -exceptional curves and a_{i}\geq 0 are some coefficients. It follows that (f^{*}(H)^{2})=(h^{*}(H)^{2})-E^{2}=(\deg f)(H^{2})-E^{2}\geq(\deg f)(H^{2}) because the intersection matrix \{(E_{i}\cdot E_{j})\}_{i,j} is negative definite. Further, since both classes f^{*}(H) and H are nef, the Hodge Index Theorem gives (f^{*}(H)\cdot H)^{2}\geq(f^{*}(H)^{2})(H^{2}). Thus we get (f^{*}(H)\cdot H)^{2}\geq(\deg f)(H^{2})^{2}. Finally, replacing f with f^{n} and using \deg f^{n}=(\deg f)^{n} , we get \sqrt[n]{((f^{n})^{*}(H)\cdot H)^{2}}\geq(\deg f)\sqrt[n]{(H^{2})^{2}}, which for n\to\infty becomes \lambda(f)^{2}\geq\deg f , as claimed. 3.2 . We now turn to the proof of Theorem 1.5 . Let \lambda(f)^{2}=\deg f and suppose the map f is non-regular . The latter means that the resolution g:Y\longrightarrow X in 3.1 is a non-trivial birational contraction. Consider the Perron class v from Proposition 1.2 . We have (v^{2})\geq 0 because v is nef. Lemma 3.3 . The class v can not be big ( \Longleftrightarrow can not satisfy (v^{2})>0 ). Proof. Suppose the contrary. Then v equals an \mathbb{R}_{\geq 0} -linear combination of big divisors v_{j} on X (see [