Abstract
We classify invariant probability measures for non-elementary groups of automorphisms, on any compact Kähler surface X, under the assumption that the group contains a so-called “parabolic automorphism”. We also prove that except in certain rigid situations known as Kummer examples, there are only finitely many invariant, ergodic, probability measures with a Zariski dense support. If X is a K3 or Enriques surface, and the group does not preserve any algebraic subset, this leads to a complete description of orbit closures.
Similar content being viewed by others
Notes
The proof in [6] is written for K3 surfaces but extends to other projective surfaces
Indeed, if D ⊂ X is the graph of a section and F is a fiber, then F ⋅ D = 1, so F cannot be multiple; and if F is reducible, D intersects F along a component of multiplicity 1. Moreover, if a ∈Z+ is large enough, then aF + D is big and nef.
We must take the closure in Eq. (5.5) because \({L_{h}^{0}}(z^{\prime })\) is reduced to \(\{z^{\prime }\}\) when \(\pi _{h}(z^{\prime })\in \text {Tor}(B_{h}^{\circ })\).
References
Amerik, E., Verbitsky, M.: Parabolic automorphisms of hyperkähler manifolds. arXiv:2112.01951 (2021)
Barth, W.P., Hulek, K., Peters, C.A.M., Van de Ven, A.: Compact Complex Surfaces, 2nd edn. vol. 4 of Ergebnisse der Mathematik und ihrer Grenzgebiete. 3. Folge. A Series of Modern Surveys in Mathematics [Results in Mathematics and Related Areas. 3rd Series. A Series of Modern Surveys in Mathematics]. Springer-Verlag, Berlin (2004)
Bierstone, E., Milman, P.D.: Semianalytic and subanalytic sets. Inst. Hautes Études Sci. Publ. Math. 67, 5–42 (1988)
Blanc, J.: On the inertia group of elliptic curves in the Cremona group of the plane. Mich. Math. J. 56(2), 315–330 (2008)
Cantat, S.: Dynamique des automorphismes des surfaces K3. Acta Math. 187(1), 1–57 (2001)
Cantat, S.: Sur la dynamique du groupe d’automorphismes des surfaces K3. Transform. Groups 6(3), 201–214 (2001)
Cantat, S.: Bers and Hénon, Painlevé and Schrödinger. Duke Math. J. 149(3), 411–460 (2009)
Cantat, S.: Quelques aspects des systèmes dynamiques polynomiaux: existence, exemples, rigidité. In: Quelques Aspects Des SystÈMes Dynamiques Polynomiaux, vol. 30 of Panor. Synthèses, pp 13–95. Soc. Math., France, Paris (2010)
Cantat, S.: Dynamics of automorphisms of compact complex surfaces. In: Frontiers in Complex Dynamics, vol. 51 of Princeton Math. Ser, pp 463–514. Princeton Univ. Press, Princeton (2014)
Cantat, S., Dolgachev, I.: Rational surfaces with a large group of automorphisms. J. Amer. Math. Soc. 25(3), 863–905 (2012)
Cantat, S., Dujardin, R.: Random dynamics on real and complex projective surfaces. arXiv:2006.04394. (references to this article concern version v3 of this preprint on arXiv)
Cantat, S., Dujardin, R.: Finite orbits for large groups of automorphisms of projective surfaces. arXiv:2012.01762. (2020)
Cantat, S., Dujardin, R.: Hyperbolicity for large automorphism groups of projective surfaces. Preprint arXiv:2211.02418 (2022)
Cantat, S., Favre, C.: Symétries birationnelles des surfaces feuilletées. J. Reine Angew. Math. 561, 199–235 (2003)
Cantat, S., Gao, Z., Habegger, P., Xie, J.: The geometric Bogomolov conjecture. Duke Math. J. 170(2), 247–277 (2021)
Cantat, S., Oguiso, K.: Birational automorphism groups and the movable cone theorem for Calabi-Yau manifolds of Wehler type via universal Coxeter groups. Amer. J. Math. 137(4), 1013–1044 (2015)
Cartan, H.: Variétés analytiques réelles et variétés analytiques complexes. Bull. Soc. Math France 85, 77–99 (1957)
Cornulier, Y.: On the Chabauty space of locally compact Abelian groups. Algebr. Geom. Topol. 11(4), 2007–2035 (2011)
de la Harpe, P.: Topics in Geometric Group Theory. Chicago Lectures in Mathematics. University of Chicago Press, Chicago (2000)
Degtyarev, A., Itenberg, I., Kharlamov, V.: Real Enriques surfaces, vol. 1746 of Lecture Notes in Mathematics. Springer-Verlag, Berlin (2000)
Degtyarev, A., Kharlamov, V.: On the moduli space of real Enriques surfaces. C. R. Acad. Sci. Paris Sér. I Math. 324(3), 317–322 (1997)
Degtyarev, A.I., Kharlamov, V.M.: Topological properties of real algebraic varieties. Rokhlin’s way. Uspekhi Mat. Nauk 55,4(334), 129–212 (2000)
Diller, J., Favre, C.: Dynamics of bimeromorphic maps of surfaces. Amer. J. Math. 123(6), 1135–1169 (2001)
Diller, J., Jackson, D., Sommese, A.: Invariant curves for birational surface maps. Trans. Amer. Math. Soc. 359(6), 2793–2991 (2007)
Douady, A., Buff, X.: Le théorème d’intégrabilité des structures presque complexes. In: The Mandelbrot Set, Theme and Variations, vol. 274 of London Math. Soc. Lecture Note Ser, pp 307–324. Cambridge Univ. Press, Cambridge (2000)
Duistermaat, J.J.: Discrete integrable systems. Springer Monographs in Mathematics. Springer, New York (2010). QRT maps and elliptic surfaces
Dujardin, R.: Approximation des fonctions lisses sur certaines laminations. Indiana Univ. Math J. 55(2), 579–592 (2006)
Erëmenko, A.E.: Some functional equations connected with the iteration of rational functions. Algebra I Analiz 1(4), 102–116 (1989)
Furstenberg, H.: Noncommuting random products. Trans. Amer. Math Soc. 108, 377–428 (1963)
Furstenberg, H.: Recurrence in Ergodic theory and combinatorial number theory. Princeton University Press, Princeton (1981). M. B. Porter Lectures
Galbiati, M.: Stratifications et ensemble de non-cohérence d’un espace analytique réel. Invent. Math. 34(2), 113–128 (1976)
Gizatullin, M.H.: Rational G-surfaces. Izv. Nauk SSSR Ser. Mat. 44(1), 110–144,239 (1980)
Goldman, W.M.: Topological components of spaces of representations. Invent. Math. 93(3), 557–607 (1988)
Hatcher, A.: Algebraic Topology. Cambridge University Press, Cambridge (2002)
Łojasiewicz, S.: Sur la géométrie semi- et sous-analytique. Ann. Inst. Fourier (Grenoble) 43(5), 1575–1595 (1993)
Silhol, R.: Real algebraic surfaces, vol. 1392 of Lecture Notes in Mathematics. Springer-Verlag, Berlin (1989)
Stein, E.M.: Harmonic analysis real-variable methods, orthogonality, and oscillatory integrals, vol. 43 of Princeton Mathematical Series. Princeton University Press, Princeton (1993). With the assistance of Timothy S. Murphy, Monographs in Harmonic Analysis, III
Varadarajan, V.S.: Groups of automorphisms of Borel spaces. Trans. Amer. Math. Soc. 109, 191–220 (1963)
Wang, L.: Rational points and canonical heights on K3-surfaces in P1 ×P1 ×P1. In: Recent developments in the inverse Galois problem (Seattle, WA, 1993), vol. 186 of Contemp. Math. Amer. Math. Soc., pp 273–289, Providence (1995)
Whitney, H., Bruhat, F.: Quelques propriétés fondamentales des ensembles analytiques-réels. Comment Math. Helv. 33, 132–160 (1959)
Acknowledgements
We are grateful to Yves de Cornulier for useful discussions which led to Theorem 8.1, after we had obtained Corollary 8.2. We thank the referees for their detailed reading and their insightful remarks.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of Interest
The authors declare no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: . Abelian Surfaces
Appendix: . Abelian Surfaces
In this appendix, we consider the case when all parabolic automorphisms g of Γ induce an automorphism gB of infinite order on the base of their invariant fibration πg. In that case, we know from [14, Proposition 3.6] that X is a compact torus, and in fact an abelian surface since Γ is non-elementary. Thus, we assume that
-
(i)
X is an abelian surface, isomorphic to C2/Λ for some lattice Λ;
-
(ii)
Γ is a non-elementary group of automorphisms of X that contains a parabolic element g;
-
(iii)
every parabolic element g of Γ acts on the base of its invariant fibration πg: X → Bg by an automorphism gB: Bg → Bg of infinite order.
We provide an argument to complete the proof of Theorem A in that case; the strategy is the same as in Sections 4 and 5, but simpler since the dynamics is linear:
Proposition A.1
Under the above hypotheses (i), (ii), (iii), if μ is a Γ-invariant and ergodic measure, then either μ is the Haar measure on the abelian surface X, or there are finitely many subtori Sj ⊂ X of real dimension 2, and points aj ∈ X, j = 1,…,k, such that
-
(1)
\(\bigcup _{j} (a_{j}+S_{j})\) is Γ-invariant;
-
(2)
Γ permutes transitively the subsets aj + Sj, j = 1,…,k;
-
(3)
μ is supported on \(\bigcup _{j} (a_{j}+S_{j})\) and on each aj + Sj, μ is given by \(\frac {1}{k} m_{j}\) where mj is the Haar measure on aj + Sj.
Here, by Haar measure on aj + Sj, we mean the image of the Haar measure on Sj by the translation s ∈ Sj↦aj + s. With the results of Section 4, this proposition concludes the proof of Theorem A.
This proposition could be proved with techniques from homogeneous dynamics, but there are two subtleties: firstly the group Γ is not supposed to fix the origin of the torus so we have to deal with affine maps; secondly, the linear representation of Γ in GL2(C) given by the linear part of its elements is not supposed to be irreducible as a subgroup of GL4(R): totally real invariant subtori appear precisely in this case (see [15, Section 4] for this type of difficulty). Thus, we provide a proof in the spirit of this article.
1.1 A.1 Parabolic, Affine Transformations
The group Γ acts on X by affine transformations
where the linear part Af ∈GL2(C) preserves the lattice Λ ⊂C2 and the translation part Sf is an element of C2/Λ. Now, pick a parabolic element g ∈Γ; its linear part is given by
after a linear change of coordinates in C2. In these coordinates, the fibration πg is induced by the projection π1: (x,y)↦x, and a conjugation by a translation reduces g to the form
where s has infinite order in the elliptic curve Bg = C/π1(Λ).
Lemma A.2
If the orbits of gB: x↦x + s are dense in Bg, then g is uniquely ergodic: the unique g-invariant probability measure on X is the Haar measure.
This result is due to Furstenberg (see [30, Section 3.3]). Thus, in this case μ is the Haar measure on X and we are done. So in what follows, we assume that for every g ∈Hal(Γ) the orbits of the translation gB are not dense: they equidistribute along circles x + Tg, where Tg is the closure of the group Zs ⊂ Bg; changing g into a positive iterate we may assume that this closure is isomorphic to R/Z (as a real Lie group). We let ℓg be the quotient map C2/Λ → Bg/Tg.
Lemma A.3
Every fiber of the linear projection ℓg is a 3-dimensional g-invariant torus, and g is uniquely ergodic on almost every fiber.
To prove Lemma A.3, we think of C2 as a real vector space and fix a basis of Λ. Then C2 is identified with R4 and Λ with Z4 ⊂R4. The eigenspace of Ag for the eigenvalue 1 is defined over Z with respect to Λ = Z4. Moreover, Ag acts trivially on the quotient space R4/Fix(Ag). Thus adapting the basis of Z4 to g, we may assume that
for some irrational number s2 and some integers a, b, c, and d. The linear projection ℓg is now given by (x1,x2,x3,x4)↦x1 and Lemma A.3 boils down to the following statement.
Lemma A.4
If 1, s1 and s2 are linearly independent over Q, then g is uniquely ergodic on the level set \(\left \{x_{1} = s_{1}\right \}\).
Proof
Let us first observe that ad − bc≠ 0. Indeed, otherwise the linear part of g would have a fixed point set of dimension 3, which is impossible because g is holomorphic (see Eq. (A.2)). To prove unique ergodicity, we use the following criterion due to Furstenberg (see [30, Prop. 3.10]): let h be a homeomorphism on R/Z × (R2/Z2) of the form (x,y)↦(x + u,y + φ(x)), where u is irrational, then h is uniquely ergodic if and only if it is ergodic for the Haar measure. On the fiber x1 = s1, our map g is of the form
so we need to check that it is ergodic for the Haar measure. For this, we pick a measurable invariant subset A ⊂R3/Z3, we denote by \(\mathbf 1_{A}\in L^{2}(\mathbf {R}^{3}/\mathbf {Z}^{3})\) its indicator function, and we expand it into a Fourier series \(\mathbf 1_{A}(x_{2},x_{3},x_{4}) = {\sum }_{(k,\ell , m)\in \mathbf {Z}^{3}} c_{k,\ell , m} e^{2\textsf {i} \pi (kx_{2}+\ell x_{3}+ mx_{4})}\). Then
and, from the g-invariance of 1A and the uniqueness of the expansion, we get
for all (k,ℓ,m) ∈Z3. For ℓ = m = 0, the irrationality of s2 implies that ck,0,0 = 0 unless k = 0. If ℓb + md≠ 0, iterating the relation \(\left \vert c_{k,\ell , m}\right \vert = \left \vert c_{k- \ell b - md, \ell , m}\right \vert \) and using the fact that Fourier coefficients decay to zero at infinity, we infer that ck,ℓ,m = 0. Finally, if ℓb + md = 0 and one of ℓ or m is nonzero, since ad − bc≠ 0 we get that ℓa + mc≠ 0 and Eq. (A.7) gives \(c_{k,\ell , m} = e^{2{\mathsf {i}} \pi (ks_{2} + (\ell a+ mc)s_{1})} c_{k, \ell , m}\). Since 1, s1, and s2 are Q-linearly independent, we derive ck,ℓ,m = 0. Thus, 1A is a constant, which means that the Haar measure of A is 0 or 1. □
1.2 A.2 Using Distinct Parabolic Automorphisms
To complete the proof of Theorem A in the case of tori, one can now follow the same ideas as in Sections 4 and 5. We only need to replace the dimension \(\dim _{\mathbf {R}}(\mu )\) by the minimal dimension of a real subtorus Z ⊂ X such that μ(q + Z) > 0 for some q ∈ X, the invariant fibration πg by the R-linear projection ℓg, and the set \(\mathrm {R}_{g}(B_{g}^{\circ })\subset B_{g}^{\circ }\) by
Lemma A.4 shows that R(g) is countable.
Lemma A.5
If there is a parabolic element g ∈Γ for which ((ℓg)∗μ)(R(g)) < 1, then μ is the Haar measure on X.
Proof
The proof is the same as for Proposition 4.9. Pick another parabolic transformation h ∈Γ, such that ℓg and ℓh are linearly independent; such an h exist because Γ is non-elementary. The main tool is the disintegration of μ with respect to ℓg; for y in a subset \({\mathcal {Y}}_{g}\subset B_{g}/T_{g}\) of positive measure, the conditional measure λg,y is the Haar measure on the 3-dimensional torus \(\ell _{g}^{-1}(y)\). Hence \(\dim _{\mathbf {R}}(\mu )\geqslant 3\) and ((ℓg)∗μ)(R(g)) = 0, as in Step 2 of the proof of Proposition 4.9. As in Steps 3 and 4, we infer that (ℓh)∗μ does not charge R(h) and that (ℓh)∗μ is absolutely continuous with respect to the Lebesgue measure on R/Z. This, implies that μ itself is invariant by all translations along the fibers of ℓh, because \(\mu ={\int \limits }_{Y} \lambda _{h,y} d((\ell _{h})_{*}\mu )(y)\) and λh,y is the Haar measure for almost every y. Permuting the roles of g and h, μ is in fact invariant under all translations. Hence, μ is the Haar measure on X. □
Now, we are reduced to the case where (ℓg)∗μ(R(g)) = 1 for every parabolic automorphism g in Γ. Since R(g) is countable, \(d_{\mathbf {R}}(\mu )\leqslant 3\) and μ charges some fiber of ℓg. Using another parabolic automorphism h, we see that μ gives positive mass to a translate a0 + S0 of a 2-dimensional torus S0 ⊂ X whose projections ℓg(a0 + S0) = ℓg(a0) and ℓh(a0) are in the countable sets R(g) ⊂ Bg/Tg and R(h) ⊂ Bh/Th respectively. Thus, by ergodicity, we conclude that μ is supported on a finite union of translates of 2-dimensional tori aj + Sj ⊂ X, \(0\leqslant i\leqslant k-1\) for some \(k\geqslant 1\).
A subgroup Γ0 of index \(\leqslant k!\) in Γ preserves a0 + S0, and gk! and hk! act on \(a_{0}+ S_{0}\simeq \mathbf {R}^{2}/\mathbf {Z}^{2}\) as two linear parabolic transformations with respect to transverse linear fibrations. So, it follows that \(\mu _{\vert a_{0}+S_{0}}\) is proportional to the Haar measure of a0 + S0, and the proof of Proposition A.1 is complete.
1.3 A.3 No or Infinitely Many Invariant Real Tori
Consider a compact complex torus X = C2/Λ of dimension 2. Let Γ be a subgroup of Aut(X). As in Appendix Appendix, write the elements f of Aut(X) in the form f(x,y) = Af(x,y) + Sf, and denote by AΓ ⊂GL2(C) the image of Γ by the homomorphism f↦Af. The group Γ is non-elementary if and only if AΓ contains a free group, if and only if the Zariski closure of AΓ in the real algebraic group GL2(C) is semi-simple.
Now, assume that Γ is non-elementary and preserves at least one ergodic probability measure μ with \(\dim _{\mathbf {R}}(\mu )=2\). Equivalently, after conjugation by a translation, there is a finite index subgroup Γ0 ⊂Γ that preserves a real, two-dimensional subtorus Σ = π/Λπ, where π ⊂C2 is a real vector space of dimension 2 and ΛR := π ∩Λ is a lattice in π (the restriction of μ to π/Λπ is proportional to the Haar measure). The goal of this last section is to explain that, in fact, Γ preserves infinitely many ergodic measuresμj with \(\dim _{\mathbf {R}}(\mu _{j})=2\). Two mechanisms can be used to establish this fact.
The first one relies on the fact that Γ0 acts on the quotient \(Q=X/{\Sigma }\simeq \mathbf {R}^{2}/\mathbf {Z}^{2}\), fixing the origin. Moreover, the action of Γ0 on Q = R2/Z2 is induced by an injective homomorphism Γ0 →GL2(Z) (to see this, note that C2 = π ⊕Riπ and iπ surjects onto Q). This implies that Γ0 has arbitrarily large finite orbits in Q (coming from torsion points of Q). The preimages of these orbits in X provide surfaces Σj ⊂ X; they are “parallel” to Σ and have an arbitrarily large number of connected components; they are Γ0-invariant; and each of them supports a unique invariant, ergodic, probability measure μj with \(\dim _{\mathbf {R}}(\mu _{j})=2\).
For the second mechanism, we assume that Γ0 fixes the origin and, changing Γ0 in a finite index subgroup if necessary, we identify Γ0 with a subgroup of SL2(C). Identify (π,Λπ) to (R2,Z2) and the restriction Γ0|π to a subgroup of GL2(Z); since Γ0 is non-elementary, Γ0 is Zariski dense in SL2(C) and Γ0|π is Zariski dense in SL2(R) (resp. in SL2(C)). In particular, the Q-algebra generated by Γ0|π is the algebra of 2 × 2 matrices with rational coefficients. The decomposition C2 = π ⊕Riπ is Γ0-invariant, and the multiplication by i defines a Γ0-equivariant map from π to iπ. Thus, Γ0 preserves each of the real planes πη = {(x,y) + η i(x,y); for (x,y) ∈π}, with η ∈R. Now, consider the (real) projection q of C2 onto π parallel to iπ, and set \({\Lambda }^{\prime }=q({\Lambda })\). It is a Γ0-invariant subgroup of π of rank at most 4, and it contains \({\Lambda }_{\Pi }\simeq \mathbf {Z}^{2}\). Then, one checks easily that
-
(1)
\({\Lambda }^{\prime }\) is commensurable to Λπ ⊕ αΛπ, for some α ∈R ∖Q, or to Λπ, in which case we set α = 0;
-
(2)
Λ is commensurable to Λπ ⊕ Kα,β(Λπ) where Kα,β is the linear map from π to π ⊕iπ defined by Kα,β(u) = αu + βiu;
-
(3)
for m in Z, the real plane πmα/β is Γ-invariant and intersects Λ on a cocompact lattice \({\Lambda }_{{\Pi }_{m\alpha /\beta }}\).
Then, the surfaces \({\Sigma }_{m}={\Pi }_{m\alpha /\beta }/{\Lambda }_{{\Pi }_{m\alpha /\beta }}\) form an infinite family of Γ-invariant tori in X.
Remark A.6
This second argument does not apply in the following case. Let E = C/Z[i], Λ = Z[i] ×Z[i] ⊂C2, and X = C2/Λ = E × E. The group \({\Gamma }={\textsf {SL}}_{2}(\mathbf {Z})\ltimes \mathbf {R}^{2}/\mathbf {Z}^{2}\) is a subgroup of Aut(X) that preserves the torus π/Λπ for π = R2 ⊂C2, but has no fixed point (because Γ contains π/Λπ), and every Γ-invariant surface is a finite union of translates of this torus.
On the other hand, this second argument applies when X and Γ come from a genuine Kummer example, that is, a Kummer example defined on a surface that is not a compact torus. Indeed in that case Γ contains a finite index subgroup with a fixed point.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Cantat, S., Dujardin, R. Invariant Measures for Large Automorphism Groups of Projective Surfaces. Transformation Groups (2023). https://doi.org/10.1007/s00031-022-09782-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00031-022-09782-0