Pith. sign in

REVIEW 4 major objections 5 minor 20 references

A Real K3 Automorphism with Most of Its Entropy in the Real Part

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A K3 automorphism with most of its entropy in the real part

desk verdict First serious candidate for a real K3 automorphism with more than half its entropy in the real locus; the main inequality is convincing but one computer step is not fully certified. read the letter →

arxiv 2506.03479 v1 pith:YOY2OVEG submitted 2025-06-04 math.DS math.AG

Ethan Cohen This is my paper
classification math.DSmath.AG MSC 14J2837B4037E30
keywords K3surfacestopologicalentropyrealdynamicspseudo-AnosovmapsshadowinglemmamappingclassgroupSalempolynomials(222)
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper constructs a real projective K3 surface $X_{10}$ and an automorphism $f_{10}$ whose action on the real points carries more than half of the total entropy. It proves that the complex topological entropy is $\ln 6.1393$ for every nonzero parameter $A$, and that for $A=10$ there is a period-10 point on $X_{10}(\mathbb{R})$; deleting that orbit leaves a ten-punctured sphere on which $f_{10}^2$ acts with stretch factor about $8.1998$. Because the real entropy is at least half of $\ln 8.1998$, the ratio $\rho = h_{top}(f_{10},X_{10}(\mathbb{R}))/h_{top}(f_{10},X_{10}(\mathbb{C}))$ is about $0.58$, strictly above $1/2$. The construction therefore supplies the first example in this setting where the real dynamics contribute more than half of the entropy, and it settles an open comparison problem in the survey literature.

What carries the argument

The argument has three load-bearing pieces. First, the complex entropy is identified with the log spectral radius of $f^*$ on $H^2$, and that spectral radius is computed exactly as the dominant real root of the Salem polynomial coming from $f$'s action on a Minkowski subspace $W$ spanned by twelve algebraic $(-2)$-curves. Second, a quantitative shadowing lemma for $C^2$ surface diffeomorphisms converts a highly recurrent, uniformly hyperbolic pseudo-orbit into a true period-10 point of $f_{10}$ on $X_{10}(\mathbb{R})$, with all constants explicit enough that interval arithmetic certifies the required $10^{-29}$ precision. Third, the punctured-sphere mapping class of $f_{10}^2$ is reconstructed from plotted arc images as a product of Dehn half-twists, and the stretch factor of that product is computed as approximately $8.1998$.

What would settle it

Re-run the arc extraction with a certified computational-topology pipeline that checks each drawn arc isotopy and over/under crossing exactly, then recompute the dilatation of the resulting half-twist word; any change from the word displayed in Section 3.6.2 would disprove the $\lambda\approx 8.1998$ bound. Alternatively, compute the full real entropy of $f_{10}$ by a high-resolution numerical scheme and check whether it is close to $\frac{1}{2}\ln 8.1998$; a value well above that number would show the punctured-sphere model leaves out additional real entropy.

Watch

Extended reading notes

Core claim

For the $(2,2,2)$ surface in $\mathbb{P}^1 \times \mathbb{P}^1 \times \mathbb{P}^1$ defined by $(1+x^2)(1+y^2)(1+z^2)+10xyz-2=0$, the automorphism $f=\sigma_3\circ\sigma_2\circ\sigma_1$ has $h_{top}(f,X(\mathbb{C}))=\ln 6.1393$. On the real locus there is a genuine period-10 point whose orbit, when removed, turns $X_{10}(\mathbb{R})$ into a sphere with ten punctures; the square of $f$ on this punctured sphere is a pseudo-Anosov mapping class with dilatation approximately $8.1998$. Consequently $h_{top}(f,X(\mathbb{R})) \ge \frac{1}{2}\ln 8.1998$, so the real-to-complex entropy ratio is about $0.58$ and in particular exceeds $1/2$. The complex side is computed exactly from the characteristic polynomial of $f$ acting on a twelve-dimensional invariant subspace of the algebraic curve lattice; the real side is certified by an explicit shadowing argument that upgrades a numerical pseudo-orbit to a true periodic point and by an explicit half-twist product for the mapping class.

Load-bearing premise

The real-entropy lower bound rests on reading the mapping class of $f_{10}^2$ from floating-point plots of arc images: if any over/under crossing in that arc-data is wrong, the product of half-twists changes, the stretch factor $8.1998$ is no longer the right number, and the inequality $\rho>1/2$ may fail.

Editorial extensions

If this is right

  • The ratio $\rho(f_{10})$ is bounded below by about $0.58$, so the real part of this automorphism carries more than half of the complex entropy.
  • The equality $h_{top}(f_A,X_A(\mathbb{C}))=\ln 6.1393$ holds for every nonzero $A$, so the one-parameter family provides many surfaces with identical complex entropy and potentially different real dynamics.
  • The certified period-10 point makes the real-dynamics computation fully explicit: one can work with a ten-punctured sphere and a concrete pseudo-Anosov representative rather than an abstract limit.
  • Because $f_{10}$ reverses orientation on $X_{10}(\mathbb{R})$, the square $f_{10}^2$ is the natural pseudo-Anosov representative, which is why the real-entropy lower bound appears with a factor of $1/2$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same explicit shadowing certificate could be reused to certify other periodic orbits of real surface automorphisms, turning numerical sightings into rigorous existence proofs whenever a sufficiently recurrent and hyperbolic pseudo-orbit is available.
  • Since the complex entropy does not depend on $A$, varying $A$ in the family may tune the real-to-complex ratio continuously, with the $A=10$ example possibly sitting on one side of a transition where the ratio crosses $1/2$.
  • If the period-10 orbit and the drawn arcs describe the whole real dynamics, then $h_{top}(f_{10},X_{10}(\mathbb{R}))$ may equal $\frac{1}{2}\ln 8.1998$ exactly; a direct numerical estimate of the full real entropy could test whether all real entropy is carried by this one pseudo-Anosov factor.
  • The construction does not address the companion open cases $0<h_{top}(f,X(\mathbb{R}))=h_{top}(f,X(\mathbb{C}))$ and $0=h_{top}(f,X(\mathbb{R}))<h_{top}(f,X(\mathbb{C}))$; the mechanism here produces only the middle regime.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper constructs an explicit real projective K3 surface X_A, for nonzero A in R, with automorphism f_A = σ_3 σ_2 σ_1, and proves two assertions. First, by Gromov-Yomdin and an explicit 12-dimensional invariant subspace W spanned by twelve (-2)-curves, it computes h_top(f_A, X_A(C)) = ln(6.1393...) independent of A, with a Salem polynomial whose largest root is approximately 6.1393. Second, for A=10, a quantitative shadowing lemma for C^2 surface diffeomorphisms is proved and applied to a computer-generated pseudo-orbit to obtain a real f_10-periodic point of order 10; the mapping class of f_10^2 on the ten-punctured sphere is then computed as an explicit product of half-twists, and Flipper yields a stretch factor approximately 8.1998. These facts give rho(f_10) ⪆ (1/2 ln 8.1998)/(ln 6.1393) > 1/2, providing the first example of a real K3 automorphism for which twice the real entropy exceeds the complex entropy.

Significance. If the main theorem is correct, this resolves an open question highlighted in Cantat's ICM survey: it gives the first real K3 automorphism with h_top(f, X(R)) > h_top(f, X(C))/2. The paper has substantial strengths: the cohomology computation is explicit and checkable, with the full intersection matrix and matrices for σ_i^* displayed; the shadowing lemma is proved in detail; and the numerical parts are accompanied by reproducible code and an interval-arithmetic error analysis in Appendix C. The use of Flipper to compute the dilatation of a pseudo-Anosov mapping class is standard and appropriate. The main risk is the numerical extraction of the mapping class from arc-data, which is not yet certified; that is the principal reason this is not an immediate accept.

major comments (4)
  1. [§3.6, §3.6.1, Figure 4] The lower bound on h_top(f_10, X_10(R)) depends entirely on the claimed mapping class [f_10^2] on the ten-punctured sphere. That mapping class is obtained from the arc-data produced by 'arcs.c', which evaluates the iterates with ordinary double-precision (or similar floating-point) arithmetic and records over/under crossings. The paper asserts that the output is verifiable by hand, but no interval-arithmetic certificate or topological stability analysis is supplied to show that the crossing pattern is unchanged under the 10^-26 perturbation allowed by the shadowing lemma. A single erroneous over/under datum would replace [f^2] by a different product of half-twists, and the resulting stretch factor could be far below 8.1998, invalidating the inequality ρ(f_10) > 1/2. This is a load-bearing computational step; the author should either provide a rigorous certificate that the recorded crossing pattern is correct or replace this step with a certified computation.
  2. [Lemma B.2] Lemma B.2 is stated with 'We omit the proof; it can be done by hand with Lagrange multipliers.' This lemma bounds max_{(x,y,z) ∈ X(R)} |x| ≈ 2.3 and is used in Lemma B.3, Lemma B.5, and the interval error analysis of Appendix C. As written, this is an unproved supporting statement in a part of the paper that is supposed to be fully checkable. A proof, or at least a precise computer-assisted verification with an interval-arithmetic certificate, should be included.
  3. [Claim 2.3, §2.3] The proof of Claim 2.3 uses the notation ∥·∥^2 for the intersection form, which has signature (1,11) on W and is not a norm. The displayed computation can at best show q(σ_1^*(w) − \tilde σ_1(w)) = 0, where q is the intersection form, and the deduction that σ_1^*(w) = \tilde σ_1(w) requires the additional fact that the difference lies in W^⊥ together with negative definiteness of W^⊥. Those ingredients are present in the following paragraph, but the proof as written is not valid and should be rewritten. Since this claim underlies the computation of h_top(f, X(C)), the revision should make the argument fully rigorous.
  4. [§3.5.2, C<21 bound] The bound C < 21 is obtained from a Mathematica computation of the Frobenius norm of (\tilde L − I)^{-1}, together with an a priori perturbation estimate with r = 10^{-2}. This bound is used to conclude that the shadowing lemma applies and that the periodic point is within 10^{-26} of the pseudo-orbit. The matrix L is 20×20 and the claimed bound is very loose, so a rigorous rational or interval-arithmetic certificate should be easy to provide; at present, however, this is another numerical step that is not formally certified and it is load-bearing for the existence and location of the periodic point.
minor comments (5)
  1. [§2.3] The text says that σ_1^* should fix [c_2] = p_1 + p_2 + p_11 + p_12 and also [c_3] = p_1 + p_2 + p_11 + p_12; the second expression should presumably be p_3 + p_4 + p_7 + p_8.
  2. [Lemma 3.2] The notation 'L−I20' is nowhere defined; it should be L − I_{2n} (or I_2 in the equivalence statement).
  3. [§3.6] The sentence 'Since f doesn't fix γ (although γ does have order 2)' is unclear; the intended meaning is likely that the point γ is fixed by f^2 or that f(γ) has order two, but as written it is confusing.
  4. [Appendix C] The constants in Claims C.4–C.9 refer to a bound c on |x| and |y|, but c is not explicitly set to the value supplied by Lemma B.2; defining c = 2.5 at the start of the appendix would make the estimates easier to verify.
  5. [Figure 4] In the entry for g_3, the symbol ~~s_3~ appears with a tilde that is not defined; this is likely a typo and should be corrected to s_3.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the complex entropy is computed from explicit linear algebra and the real stretch factor from the mapping class; the ratio is an output, not an input.

full rationale

The paper's two main quantities are derived from independent data. The complex entropy h_top(f, X_A(C)) ≈ ln 6.1393 is computed in Section 2 by writing explicit intersection matrices for twelve curves, restricting f* to an invariant subspace W, and taking the spectral radius of a 12×12 integer matrix; the Salem polynomial is an output of that matrix computation, not an assumed value. The real lower bound is obtained separately in Section 3: a pseudo-orbit is exhibited, the ad hoc shadowing lemma (Lemma 3.2, proved in Appendix A) produces a genuine 10-periodic point, and the mapping class of f^2 on the 10-punctured sphere is represented as a product of half-twists using arc-data from arcs.c; Flipper then computes the dilatation λ ≈ 8.1998 from that mapping class. No parameter is fitted to make the final entropy ratio exceed 1/2, and the ratio is never used as an input to any earlier step. There are no load-bearing self-citations: the cited external results and software (Cantat, Moncet, McMullen's programs, Flipper, MPFR) are not invoked as substitutes for the computations, and no uniqueness theorem from the author's own prior work is imported. The reliability concerns about Section 3.6—uncertified floating-point over/under crossings in the arc-data and the omitted proof of Lemma B.2—are correctness or verification risks, not circularity: an error there would change the input mapping class, but the stretch factor would still be derived from that mapping class rather than being defined by it. Therefore no circular step is present.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The central claim does not depend on fitted constants. It relies on standard K3-surface theory, on one unproved calculus bound (Lemma B.2), on the assertion that X10(R) is a sphere, and on the correctness of the external program Flipper.

assumptions (6)
  • standard math Gromov-Yomdin theorem: h_top(f,X(C)) = ln R(f*)
    Invoked in Section 2.1 to convert topological entropy to the spectral radius of the induced map on cohomology.
  • standard math Hodge theory and Hodge index theorem for K3 surfaces
    Used in Section 2.1 to restrict attention to H^{1,1} and to the Neron-Severi group, and to assert the signature of the intersection form.
  • domain assumption The surface X_A is a smooth projective K3 surface
    Derived in Section 1.1 from adjunction and Lefschetz hyperplane; standard but asserted rather than fully detailed.
  • domain assumption X10(R) is homeomorphic to a sphere
    Stated in Section 1 and supported by Lemma B.1 (star-shaped component) and Figure 1, but no complete proof is given.
  • domain assumption Lemma B.2: max_{X(R)} |x| approximately 2.3
    Stated without proof ('we omit the proof; it can be done by hand with lagrange multipliers'); used in derivative bounds (Lemma B.3) and in the MPFR error estimates (Appendix C).
  • domain assumption Flipper computes the stretch factor lambda([f^2]) approximately 8.1998 correctly
    The paper relies on the external program Flipper for the dilatation of the mapping class, without providing interval-certified error bounds.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Real K3 Automorphism with Most of Its Entropy in the Real Part." pith.science (2026). https://pith.science/paper/YOY2OVEG

@misc{pith2026250603479,
  author       = {Pith},
  title        = {Pith review of: A Real K3 Automorphism with Most of Its Entropy in the Real Part},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YOY2OVEG}},
  note         = {Machine review of arXiv:2506.03479}
}
abstract

This article describes an example of a real projective K3 surface admitting a real automorphism $f$ satisfying $h_{top}(f, X(\mathbb{C})) < 2 h_{top}(f, X(\mathbb{R}))$. The example presented is a $(2,2,2)$-surface in $\mathbb{P}^1 \times \mathbb{P}^1 \times \mathbb{P}^1$ given by the vanishing set of $(1 + x^2)(1 + y^2)(1 + z^2) + 10xyz - 2$, first considered by McMullen. Along the way, we develop an ad hoc shadowing lemma for $C^2$ (real) surface diffeomorphisms, and apply it to estimate the location of a periodic point in $X(\mathbb{R})$. This result uses the GNU MPFR arbitrary precision arithmetic library in C and the Flipper computer program.

Figures

Figures reproduced from arXiv: 2506.03479 by the authors.

Figure 1
Figure 1. X10(R) plotted in Mathematica 1.2. Outline of the paper. Our computation of htop(f, XA(C)) in section 2 relies on the Gromov￾Yomdin theorem (see [10], [11], [21]) and working with explicit algebraic curves. In section 3.2, we prove a checkable shadowing lemma for an arbitrary C 2 -diffeomorphism of a (real) surface. In sections 3.4 and 3.5, we apply the shadowing lemma to a pseudo-orbit of f10 on X10(R), with the he… view at source ↗
Figure 2
Figure 2. The pseodo-orbit xi = ˜φi(ai , bi) plotted on X(R), and projected into the z-x plane. Blue dots denote points on the top sheet and orange the bottom sheet. The curve C := {x = −z}∩X can be seen intersecting f 5 (C) near x0. The figure is obtained using a program developed by C. McMullen. where z 7→ z 1 2 is taken to be the principle branch of the square root. Then the maps Ψ ± 1 (x, y) := ([1 : p±(y, z)], [1 : y], [… view at source ↗
Figure 3
Figure 3. The algorithm described in Section 3.6.1 applied in a simple case. We obtain g = g2 ◦ g1 ◦ g0 = s2.s1.s0.s0.s1.s2.s−1 2 s −1 2 .s1. As we saw in 3.13, [f 2 ] is readily obtained from g. For completeness, we include its mapping class below: f 2 ∼= [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The algorithm described in Section 3.6.1, applied to the 10-periodic point approximated in Section 3.4. See 3.6.1 for notation [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

20 extracted references · 20 canonical work pages

  1. [1]

    Eric Bedford and Kyounghee Kim, Periodicities in linear fractional recurrences: degree growth of birational surface maps, Michigan Math. J. 54 (2006), no. 3, 647–670. MR 2280499

    work page 2006

  2. [3]

    187 (2001), no

    Serge Cantat, Dynamique des automorphismes des surfaces K3, Acta Math. 187 (2001), no. 1, 1–57. MR 1864630

    work page 2001

  3. [4]

    Serge Cantat, Dynamics of automorphisms of compact complex surfaces , 2011, Last checked: 2025-06-03

    work page 2011

  4. [5]

    619–634, 2018

    , Automorphisms and dynamics: A list of open problems , pp. 619–634, 2018

    work page 2018

  5. [6]

    Serge Cantat and Christophe Dupont, Automorphisms of surfaces: Kummer rigidity and measure of maximal entropy , J. Eur. Math. Soc. (JEMS) 22 (2020), no. 4, 1289–1351. MR 4071328

    work page 2020

  6. [7]

    Devaney and Z

    R. Devaney and Z. Nitecki, Shift automorphisms in the H´ enon mapping, Comm. Math. Phys. 67 (1979), no. 2, 137–146. MR 539548

    work page 1979

  7. [8]

    Fathi, Fran¸ cois Laudenbach, and Valentin

    Albert. Fathi, Fran¸ cois Laudenbach, and Valentin. Poenaru, Travaux de Thurston sur les surfaces , Ast´ erisque, vol. 66-67, Soci´ et´ e Math´ ematique de France, Paris, 1979, S´ eminaire Orsay, With an English summary. MR 568308

    work page 1979

  8. [9]

    Laurent Fousse, Guillaume Hanrot, Vincent Lef` evre, Patrick P´ elissier, and Paul Zimmermann, MPFR: a multiple- precision binary floating-point library with correct rounding , ACM Trans. Math. Software 33 (2007), no. 2, Art. 13,

    work page 2007

Show all 20 references
  1. [10]

    Gromov, Entropy, homology and semialgebraic geometry , no

    M. Gromov, Entropy, homology and semialgebraic geometry , no. 145-146, 1987, S´ eminaire Bourbaki, Vol. 1985/86, pp. 5, 225–240. MR 880035

    1987

  2. [11]

    Mikha ¨ ıl Gromov,On the entropy of holomorphic maps , Enseign. Math. (2) 49 (2003), no. 3-4, 217–235. MR 2026895

    2003

  3. [12]

    158, Cambridge University Press, Cambridge, 2016

    Daniel Huybrechts, Lectures on K3 surfaces , Cambridge Studies in Advanced Mathematics, vol. 158, Cambridge University Press, Cambridge, 2016. MR 3586372

    2016

  4. [13]

    247, Springer, New York, 2008, With the graphical assistance of Olivier Dodane

    Christian Kassel and Vladimir Turaev, Braid groups, Graduate Texts in Mathematics, vol. 247, Springer, New York, 2008, With the graphical assistance of Olivier Dodane. MR 2435235

    2008

  5. [14]

    Katok, Lyapunov exponents, entropy and periodic orbits for diffeomorphisms , Inst

    A. Katok, Lyapunov exponents, entropy and periodic orbits for diffeomorphisms , Inst. Hautes ´Etudes Sci. Publ. Math. (1980), no. 51, 137–173. MR 573822

    1980

  6. [15]

    54, Cambridge University Press, Cambridge, 1995, With a supplementary chapter by Katok and Leonardo Mendoza

    Anatole Katok and Boris Hasselblatt, Introduction to the modern theory of dynamical systems , Encyclopedia of Math- ematics and its Applications, vol. 54, Cambridge University Press, Cambridge, 1995, With a supplementary chapter by Katok and Leonardo Mendoza. MR 1326374

    1995

  7. [16]

    html, 2006, Version 3.2, Software, March 2006

    Curt McMullen, Orbits of automorphisms of k3 surfaces , https://people.math.harvard.edu/~ctm/programs/index. html, 2006, Version 3.2, Software, March 2006

    2006

  8. [17]

    Arnaud Moncet, Real versus complex volumes on real algebraic surfaces , Int. Math. Res. Not. IMRN (2012), no. 16, 3723–3762. MR 2959025 27

    2012

  9. [18]

    Paul Reschke and Bar Roytman, Lower semi-continuity of entropy in a family of K3 surface automorphisms , Rocky Mountain J. Math. 47 (2017), no. 7, 2323–2349. MR 3748233

    2017

  10. [19]

    Rowe, Structures on a k3 surface , Master’s thesis, UNL V Theses, Dissertations, Professional Papers, and Capstones

    Nathan P. Rowe, Structures on a k3 surface , Master’s thesis, UNL V Theses, Dissertations, Professional Papers, and Capstones. 737., 2010

    2010

  11. [20]

    The Sage Developers, Sagemath, the Sage Mathematics Software System (Version x.y.z) , YYYY, https://www.sagemath.org

  12. [21]

    Yomdin, Volume growth and entropy , Israel J

    Y. Yomdin, Volume growth and entropy , Israel J. Math. 57 (1987), no. 3, 285–300. MR 889979 Department of Mathematics, Yale University Email address : ethan.cohen@yale.edu

    1987

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.