Classification of flat pencils of foliations on compact complex surfaces

Liliana Puchuri

REVIEW 3 major objections 4 minor 13 references

Classification of flat pencils of foliations on compact complex surfaces

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A flat pencil of foliations with a rational first integral can only live on a rational surface, and the paper classifies which members admit first integrals.

desk verdict A genuine completion of Lins Neto's flat-pencil classification, held together by one unpublished lemma that needs a proof. read the letter →

arxiv 1908.08197 v2 pith:UJXJ7FAH submitted 2019-08-22 math.CV math.AG

Liliana Puchuri This is my paper · ORCID

classification math.CVmath.AG MSC 34C0714J2714D0632S65

keywords compactcomplexsurfacesflatpencilsoffoliationsinvarianttangencysetmeromorphicfirstintegralsholonomyrationalHopftori

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 completes a corner of the classification of compact complex surfaces that carry flat pencils of foliations: it treats pencils whose tangency set is invariant and whose member $F_\infty$ has a holomorphic first integral of genus zero. It proves that any such surface must be rational, and that the parameter set $I_p(\mathcal{P})$ of members with meromorphic first integrals is tightly constrained: either finite, all parameters with isolated singularities, or a rational coset $(\lambda\mathbb{Q}+\beta)$ intersected with the isolated-singularity set. For pencils with empty tangency set, it proves the surface is either a complex torus or a Hopf surface, with first integrals absent on Hopf surfaces and forced to be $\mathbb{Q}$ or $\mathbb{Q}(\tau)$ on split elliptic tori. These results settle a rigid geometric question, reducing it to a short list of explicit surfaces and parameter sets with direct bearing on the classical problem of bounding invariant algebraic curves.

What carries the argument

The engine of the proof is the local flat normal form for a flat pencil: outside the tangency set, coordinates can be chosen so that every member $F_\alpha$ is defined by $dy+\alpha\,dx=0$ (Lemma 3.5, quoted from earlier work). Around an invariant critical fiber of the rational first integral, $F_\alpha$ is a Riccati foliation, and the holonomy representation along loops around those fibers yields explicit generators: either affine maps $f_{j,\alpha}(z)=\lambda_j z+a_j\alpha+b_j$ or multiplicative maps $f_{j,\alpha}(z)=\exp(2\pi i(\mu_j\alpha+\nu_j))z$. The classification then runs on a finiteness criterion: if the global holonomy group is finite, each tangency component has positive GSV index, and the foliation has a local meromorphic first integral at its singularities, then the foliation admits a global meromorphic first integral (Lemma 5.4).

What would settle it

Find a flat pencil with invariant tangency set on a compact non-rational surface whose member $F_\infty$ has a genus-zero holomorphic first integral; Theorem 5.8 predicts no such pencil exists. A less global check is to compute the holonomy generators of the four explicit example pencils directly and look for any generator with a functional dependence on $\alpha$ other than the two stated forms.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central assertion is Theorem 5.8: if a compact complex surface $X$ carries a flat pencil $\mathcal{P}$ with invariant tangency set and $F_\infty$ has a holomorphic first integral $f:X\to\mathbb{P}^1$ of genus zero, then $X$ is rational and, after a reparametrization of the pencil, $I_p(\mathcal{P})\cap I_S(\mathcal{P})$ is either finite, equals $I_S(\mathcal{P})$, or equals $(\lambda\mathbb{Q}+\beta)\cap I_S(\mathcal{P})$. The companion classification for empty tangency set says that $X$ is either a torus or a Hopf surface and the pencil is generated by linear foliations; on Hopf surfaces $I_p(\mathcal{P})=\varnothing$, while on a torus with at least three first-integral members, $X=E\times E$ and $I_p(\mathcal{P})\setminus\{\infty\}$ is $\mathbb{Q}$ or $\mathbb{Q}(\tau)$, up to reparametrization. The paper thereby complements the earlier genus-one classification by settling the genus-zero case.

Load-bearing premise

The whole computation of holonomy generators rests on the quoted local normal form lemma asserting that a flat pencil is conjugate to $dy+\alpha\,dx=0$ outside its tangency set; if that lemma fails, the two generator forms and the classification built on them do not follow.

Editorial extensions

If this is right

A flat pencil with invariant tangency and a genus-zero holomorphic first integral can only live on a rational surface; no torus, Hopf, K3, or other non-rational compact surface can carry one.
The set of parameters with first integrals in such a pencil is either finite, all isolated-singularity parameters, or a rational coset intersected with the isolated-singularity set, so no other configurations occur.
An empty tangency set pins the surface down to a torus or a Hopf surface, and the pencil is generated by linear foliations.
On Hopf surfaces no member of such a pencil admits a meromorphic first integral; on tori, three such members force the torus to split as $E\times E$ and force the parameter set to be $\mathbb{Q}$ or $\mathbb{Q}(\tau)$.
For rational genus-zero pencils, finiteness of the holonomy group together with local first integrals is enough to conclude that a member has a global meromorphic first integral, which is the mechanism behind the classification.

Reading between the lines

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

If the theorem extends to meromorphic first integrals after resolution, the rational-surface conclusion should survive for any first integral whose resolved fibration has genus zero; this would broaden the classification beyond holomorphic first integrals.
The affine-versus-multiplicative dichotomy in the holonomy generators suggests a normal form statement: flat pencils with rational first integrals are locally pullbacks of constant-coefficient pencils on $\mathbb{P}^1\times\mathbb{P}^1$, which could be checked directly on the four explicit example pencils.
The $\mathbb{Q}$ or $\mathbb{Q}(\tau)$ parameter sets in the torus case tie the classification to endomorphism rings of elliptic curves, so a computational search on a generic torus with fewer than three linear foliations admitting first integrals should confirm that no pencil has three first-integral members.
A concrete next step would be to classify which rational surfaces actually arise in each of the three cases of Theorem 5.8, since the theorem proves rationality but does not list the surfaces.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Z — quick take on 1908.08197. This is a real classification result: for flat pencils with invariant tangency set, it pins down the genus-zero case (Theorems 5.6 and 5.8) and gives the empty-tangency classification (Theorems 4.5 and 4.6). It genuinely complements Lins Neto's genus-one work, and the key technique — using flatness to force holonomy generators into two explicit normal forms — is appropriate. I believe the main theorem is true, conditional on one quoted lemma.

Where it's solid: the empty-tangency classification is well argued. The proof of Theorem 4.5 uses Brunella's regular-foliation classification, Kodaira, and a neat contradiction with Lemma 4.4 to exclude elliptic fibrations. The torus case via Ghys and Pereira-Pirio gives a clean characterization of Ip(P). The gen(f)=0 section is structurally sound: Proposition 5.5 reduces to Riccati foliations, Theorem 5.6 derives the two normal forms, Lemma 5.7 handles finiteness, and Theorem 5.8's case split is valid once the normal forms are granted.

Soft spots, in order:

1. Lemma 5.1 is quoted from Lins Neto's unpublished preprint [10] and is load-bearing. Proposition 5.2 uses it to conclude gen(f)≥1 when ∞∈IS(P); drop it and Proposition 5.5(1) no longer excludes ∞∈IS(P), which is needed for the Δ(P) decomposition in Theorem 5.6. The lemma — a rigidity statement about a genus-zero fibration foliation sharing its tangent bundle with another foliation — may well be true, but the manuscript depends on it without proof or a published reference.

2. Lemma 4.2 has a sign error: for a non-invariant curve, NG·C = χ(C)+Tang(G,C), not minus. The lemma is repairable from TF=NF together with NF=NG forcing TF=TG, so this is a write-up flaw, not a broken result.

3. Minor notational slips: the proof of Proposition 5.5 says items (1) and (2) follow from transversality, but (1) really follows from Prop 5.2; a line in Lemma 4.2 mixing TF and TG is garbled. Cosmetic.

The citation pattern is healthy: it builds on Brunella, Ghys, Lins Neto, and Pereira-Pirio without pretending to prove imported results. If Lemma 5.1 is supplied, I think the classification stands.

Bottom line: this deserves serious refereeing. Send it out, ask the author to give a proof or published source for Lemma 5.1, and fix the Lemma 4.2 sign issue. A specialist in holomorphic foliations will want to cite the main theorem once it is through.

Referee Report

3 major / 4 minor

Summary. The paper studies flat pencils of holomorphic foliations on compact complex surfaces, focusing on two complementary problems. First, it classifies surfaces admitting a pencil with empty tangency set, proving that such surfaces are either complex tori or Hopf surfaces and that the pencil is generated by linear foliations; it also characterizes the parameter set Ip(P) of foliations in the pencil admitting a holomorphic first integral. Second, for flat pencils with invariant, non-empty tangency set, under the hypothesis that F∞ admits a holomorphic first integral f of genus zero, it proves that X is rational, gives explicit normal forms for the generators of the global holonomy group of each Fα, and concludes that Ip(P) is either finite, contains IS(P), or is of the form (λQ+β)∩IS(P) up to reparametrization. The proofs use Lins Neto's machinery of pencils, Brunella's classification of regular foliations, Ghys's results on torus foliations, and holonomy computations for Riccati foliations.

Significance. If the main results hold, they provide a substantial complement to Lins Neto's classification of flat pencils with first integrals of genus one, and they give a complete description of the surfaces and of the parameter set Ip(P) in the genus-zero invariant-tangency case. The explicit holonomy normal forms in Theorem 5.6 are a concrete and useful contribution. The paper is built on a coherent network of established results (Brunella, Ghys, Lins Neto's published work), and the arguments are detailed rather than merely sketched. However, one load-bearing ingredient, Lemma 5.1, is quoted from an unpublished preprint, and a second ingredient, Lemma 3.5, is quoted without proof from the published literature; the latter is acceptable, while the former leaves a gap that must be closed before the main theorem can be regarded as fully proved.

major comments (3)

[§4, Lemma 4.2, equation (9)] The proof of Lemma 4.2 contains a sign error in the intersection formula for a non-invariant curve. In equation (9) the manuscript writes NG·C = χ(C) − Tang(G,C), but the formula stated earlier in §2 (display before §3.3) and in Brunella's Lemme 2 is NF·C = χ(C) + Tang(F,C). With the correct formula, equation (9) becomes C·C − Tang(F,C) = NF·C = χ(C) + Tang(G,C), and since TF = NG one gets Tang(F,C) = Tang(G,C) = (C·C − χ(C))/2 = (−1−2)/2 = −3/2, an impossibility because tangency indices are nonnegative integers. Thus the contradiction still follows, and the lemma is true, but the displayed equation must be corrected.
[§5, Lemma 5.1 and Proposition 5.2] Lemma 5.1 is stated as '[10, Lemma 3.2.1]' and is not proved in the manuscript. This lemma is load-bearing: Proposition 5.2 begins with 'As F0 ≠ F∞, by Lemma 5.1, we obtain that gen(f) ≥ 1', and Proposition 5.5(1) uses Proposition 5.2 to exclude ∞ ∈ IS(P) when gen(f)=0. Without Lemma 5.1(1), the decomposition Δ(P)=Σ n_j[f^{-1}(c_j)]+Σ C'_s in Theorem 5.6 and the two normal forms for the holonomy generators would lack their premise. Because [10] is an unpublished IMPA preprint, this is a genuine gap. The authors should either include a complete proof of Lemma 5.1 in the paper or replace it with a published, verifiable reference.
[§5, proof of Proposition 5.5] The argument ruling out gen(f) ≥ 2 is too terse and leaves a central step to an external reference. The assertion that Fγ(α,·) does not depend on α because Aut(Tc) is finite is plausible, but the jump from 'the holonomy maps are locally constant in α' to 'there exists a neighborhood V of β such that Fα = Fβ for all α ∈ V' needs a precise justification: one must show that coinciding global holonomy representations for a Riccati foliation relative to f force the foliations themselves to coincide. Since this is used to contradict F0 ≠ F∞, please expand this step or state and prove it as a lemma rather than citing '[10, p. 34]'.

minor comments (4)

[§5, opening paragraph] The sentence 'From now on P = {F∞}α∈C is a flat pencil' appears to contain a typo: it should read P = {Fα}α∈C.
[§5, proof of Theorem 5.6] The symbol P is used both for the pencil and for the polynomial P(x,y) in the local expression ωα = dy + αP(x,y)dx. Please rename the polynomial (for example, Q(x,y)) to avoid confusion, especially in the phrase 'Since P is flat'.
[§3.4, equation (7)] In equation (7), the constants λ, a, b are not defined locally; it would help the reader to specify that λ ∈ C*, a,b ∈ C depend on the generator and on the chosen cross-section, and that this is the form obtained from Lins Neto's computation.
[§4, Theorem 4.5 proof] The sentence 'Inoue surfaces just admit at most two regular foliations, so they cannot contain a pencil of foliations with empty tangency set' would benefit from a specific citation to the part of [3] where this is proved, since it is used to exclude Inoue surfaces.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the gen(f)=0 classification is derived from flatness, holonomy computations, and external classification results, not from its own conclusion.

full rationale

The paper's central claim, Theorem 5.8, is not assumed as an input. It is obtained by first showing, via Proposition 5.5 and Lemma 5.1, that an invariant non-empty tangency set forces gen(f) ≤ 1 and, in the gen(f)=0 case, that the foliations are Riccati with ∞ ∉ IS(P). Theorem 5.6 then computes the holonomy generators from flatness and the local normal form Lemma 3.5, and Theorem 5.8 converts finiteness or arithmetic conditions on those holonomy generators into membership in Ip(P) using Lemma 5.4 and Ghys's theorem. The use of Lins Neto's Theorem 1.2 is confined to the already-treated genus-one case and is not used to prove the genus-zero classification. Lemma 5.1 is load-bearing for ruling out gen(f) ≥ 2, but it is an external result quoted from Lins Neto's preprint, not a self-citation, and it does not contain the paper's claimed classification. Its unpublished status is a verification or correctness concern, not evidence of circularity. No parameter is fitted and later renamed as a prediction: Ip(P) is defined independently, and the final descriptions are explicit sets such as (λQ+β) ∩ IS(P). The empty-tangency classification in Section 4 imports torus and Hopf surface results from Ghys and Pereira-Pirio, but those are external benchmarks rather than the paper's own conclusions. Consequently, the derivation chain is self-contained relative to its cited external machinery, and no step reduces by construction to its own inputs.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

The paper introduces no new free parameters or invented entities. The constants λ, β, μ_j, ν_j appearing in the holonomy generators and in the characterization of Ip(P) are structural outputs of the flatness condition, not fitted values. The external theorems used are standard or established domain results, quoted from the literature.

assumptions (7)

domain assumption Brunella's classification of regular foliations on compact complex surfaces with Kod(X)<2
Used in Theorem 4.5 to narrow the possible surfaces after Kodaira classification gives Kod(X)<2.
standard math Kodaira classification of compact complex surfaces
Used in Theorems 4.5 and 5.8 to convert c1^2=c2=0 and minimality into Kod(X)<2, and to identify rational surfaces.
domain assumption Lins Neto's Lemma 3.5 on local coordinates for flat pencils
Quoted from [12, Lemma 2.1.4]; basis for the holonomy generator normal forms in Theorem 5.6 and for Lemma 4.3.
domain assumption Lemma 5.1 from Lins Neto's preprint [10, Lemma 3.2.1]
Used in Proposition 5.2 to show that two distinct foliations with common holomorphic first integral imply gen(f)≥1; forces gen(f)≤1 in the invariant-tangency case.
domain assumption Lins Neto's Theorem 1.2 for the genus-one case
Used in Proposition 5.9 to describe Ip(P) when ∞∈IS(P); the paper's new results cover gen(f)=0 and empty tangency.
domain assumption Pereira-Pirio Proposition 2.2 on tori with many linear foliations
Used in Corollary 4.7 to identify X=E×E and Ip(P)\ {∞} as Q or Q(τ).
domain assumption Ghys's classification of regular foliations on complex tori
Used in Theorem 4.6 to prove that a pencil with empty tangency on a torus is generated by linear foliations.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Classification of flat pencils of foliations on compact complex surfaces." pith.science (2026). https://pith.science/paper/UJXJ7FAH

@misc{pith2026190808197,
  author       = {Pith},
  title        = {Pith review of: Classification of flat pencils of foliations on compact complex surfaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UJXJ7FAH}},
  note         = {Machine review of arXiv:1908.08197}
}

read the original abstract

Related to the classification of regular foliations in a complex algebraic surface, we address the problem of classifying the complex surfaces which admit a flat pencil of foliations. On this matter, a classification of flat pencils which admit foliations with a first integral of genus one and isolated singularities was done by Lins Neto. In this work, we complement Lins Neto's work, by obtaining the classification of compact complex surfaces which have a pencil with an invariant tangency set.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

[10]

Lins Neto

A. Lins Neto. Exceptional families of foliations and th e Poincar´ e Problem. Preprint IMP A serie A183/2002, www.preprint.impa.br, 2002

work page 2002
[1]

Barth, C

W . Barth, C. Peters, and A. V an de V en. Compact complex surfaces, volume 4 of Ergebnisse der Mathematik und ihrer Grenzgebiete (3) [Resu lts in Mathematics and Related Areas (3)] . Springer-V erlag, Berlin, 1984. 20

work page 1984
[2]

P . F. Baum and R. Bott. On the zeros of meromorphic vector- ﬁelds. In Essays on Topology and Related Topics (M ´emoires d ´edi´es `a Georges de Rham) , pages 29–47. Springer, New Y ork, 1970

work page 1970
[3]

Brunella

M. Brunella. Feuilletages holomorphes sur les surfaces complexes compactes. Ann. Sci. ´Ecole Norm. Sup. (4) , 30(5):569–594, 1997

work page 1997
[4]

Brunella

M. Brunella. Some remarks on indices of holomorphic vect or ﬁelds. Publ. Mat., 41(2):527–544, 1997

work page 1997
[5]

Brunella

M. Brunella. Birational geometry of foliations . Publicac ¸ ˜ oes Matem´ aticas do IMP A. [IMP A Mathematical Publications]. Instituto de Matem´ atica Pura e Apli- cada (IMP A), Rio de Janeiro, 2004

work page 2004
[6]

Ehresmann

C. Ehresmann. Les connexions inﬁnit´ esimales dans un es pace ﬁbr´ e diff´ erentiable. In Colloque de topologie (espaces ﬁbr ´es), Bruxelles, 1950, pages 29–55. Georges Thone, Li` ege; Masson et Cie., Paris, 1951

work page 1950
[7]

´E. Ghys. Feuilletages holomorphes de codimension un sur les espaces homog` enes complexes. Ann. Fac. Sci. T oulouse Math. (6), 5(3):493–519, 1996

work page 1996

Show all 13 references

[8]

´E. Ghys. `A propos d’un th´ eor` eme de J.-P. Jouanolou concernant les f euilles ferm´ ees des feuilletages holomorphes.Rend. Circ. Mat. Palermo (2) , 49(1):175– 180, 2000

2000
[9]

Kobayashi and T

S. Kobayashi and T. Ochiai. Holomorphic projective stru ctures on compact com- plex surfaces. Math. Ann., 249(1):75–94, 1980

1980
[11]

Lins Neto

A. Lins Neto. Some Examples for the Poincar´ e and Painle v´ e problems.Ann. Sci. ´Ecole Norm, 4(35):231–266, 2002

2002
[12]

Lins Neto

A. Lins Neto. Curvature of pencil of foliations. Ast´erisque, (296):167–190, 2004

2004
[13]

J. V . Pereira and L. Pirio. The classiﬁcation of excepti onal CDQL webs on com- pact complex surfaces. Int. Math. Res. Not. IMRN , (12):2169–2282, 2010. 21

2010

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [10]

Lins Neto

A. Lins Neto. Exceptional families of foliations and th e Poincar´ e Problem. Preprint IMP A serie A183/2002, www.preprint.impa.br, 2002

work page 2002

[2] [1]

Barth, C

W . Barth, C. Peters, and A. V an de V en. Compact complex surfaces, volume 4 of Ergebnisse der Mathematik und ihrer Grenzgebiete (3) [Resu lts in Mathematics and Related Areas (3)] . Springer-V erlag, Berlin, 1984. 20

work page 1984

[3] [2]

P . F. Baum and R. Bott. On the zeros of meromorphic vector- ﬁelds. In Essays on Topology and Related Topics (M ´emoires d ´edi´es `a Georges de Rham) , pages 29–47. Springer, New Y ork, 1970

work page 1970

[4] [3]

Brunella

M. Brunella. Feuilletages holomorphes sur les surfaces complexes compactes. Ann. Sci. ´Ecole Norm. Sup. (4) , 30(5):569–594, 1997

work page 1997

[5] [4]

Brunella

M. Brunella. Some remarks on indices of holomorphic vect or ﬁelds. Publ. Mat., 41(2):527–544, 1997

work page 1997

[6] [5]

Brunella

M. Brunella. Birational geometry of foliations . Publicac ¸ ˜ oes Matem´ aticas do IMP A. [IMP A Mathematical Publications]. Instituto de Matem´ atica Pura e Apli- cada (IMP A), Rio de Janeiro, 2004

work page 2004

[7] [6]

Ehresmann

C. Ehresmann. Les connexions inﬁnit´ esimales dans un es pace ﬁbr´ e diff´ erentiable. In Colloque de topologie (espaces ﬁbr ´es), Bruxelles, 1950, pages 29–55. Georges Thone, Li` ege; Masson et Cie., Paris, 1951

work page 1950

[8] [7]

´E. Ghys. Feuilletages holomorphes de codimension un sur les espaces homog` enes complexes. Ann. Fac. Sci. T oulouse Math. (6), 5(3):493–519, 1996

work page 1996