A determinant-line and degree obstruction to foliation transversality

Basic Information

• Paper ID: 2509.10799
• Title: A determinant-line and degree obstruction to foliation transversality
• Author: Mostafa Khosravi Farsani (University of Alberta)
• Classification: math.GT (Geometric Topology)
• Publication Date: November 4, 2025 (arXiv v3)
• Paper Link: https://arxiv.org/abs/2509.10799v3

Abstract

This paper investigates the transversality problem for foliations, proposing two concise obstruction criteria for when a closed embedded complementary submanifold $S^n \subset M^{\ell+n}$ is everywhere transverse to a $C^1$ foliation $\mathcal{F}$:

(A) Determinant-line bundle obstruction: For the line bundle $L := \det(TS)^* \otimes \det(\nu\mathcal{F})|_S$, after $C^1$ small isotopic perturbation, there exists a smooth section transverse to the zero section whose zero set $Z$ is a closed codimension-1 submanifold representing the Poincaré dual $PD(w_1(L))$; in particular, when $n=1$, the parity of the tangency number is determined by $\langle w_1(L), [S]\rangle$.

(B) Twisted degree criterion: If $\mathcal{F}$ is given by a $C^1$ submersion $\pi: M \to B^n$ and $f := \pi|_S$ satisfies $f_*[S]_{f^*\mathcal{O}_B} = 0 \in H_n(B; \mathcal{O}_B)$ (equivalent to $\deg(f)=0$ when $B$ is orientable), then $S$ must be tangent to the foliation somewhere.

These two criteria apply to situations with holonomy and lack of transverse orientability, with applications to periodic directions on translation surfaces and rational polygonal billiard systems.

Research Background and Motivation

1. Core Research Problem

The central question addressed is: given a rank-$\ell$ $C^1$ foliation $\mathcal{F}$ on a manifold $M^{\ell+n}$ and a closed embedded submanifold $S^n \subset M$, when can $S$ be everywhere transverse to the leaves of $\mathcal{F}$?

2. Problem Significance

• Foundational problem in geometric topology: Transversality is a core concept in differential topology; understanding foliation transversality is crucial for studying global geometric and topological properties of manifolds
• Applications in dynamical systems: In dynamical systems, the transversality relationship between closed orbits and invariant foliations relates to system stability and existence of periodic orbits
• Concrete geometric applications: Periodic direction problems in translation surfaces and billiard systems directly correspond to transversality problems

3. Limitations of Existing Methods

Classical transversality theory primarily addresses:

• High smoothness requirements: Traditional methods require $C^2$ or higher regularity
• Orientability assumptions: Classical degree theory depends on orientability of manifolds and foliations
• Simple foliations: Existing results are mostly limited to simple cases without holonomy

4. Research Motivation

The innovation of this paper lies in:

• Extending transversality theory to merely $C^1$ foliations
• Handling non-orientable cases through Stiefel-Whitney classes
• Providing precise characterization of topological obstructions (via $w_1(L)$ and twisted cohomology)
• Giving computable discrimination criteria applicable to concrete geometric problems

Core Contributions

1. Determinant-line bundle obstruction theorem (Theorem 1):
   • Proves that even for $C^1$ foliations, after small perturbation the tangency set $Z$ is a smooth hypersurface
   • Establishes precise relationship between the homology class of the tangency set and the first Stiefel-Whitney class: $[Z] = PD(w_1(L))$
   • Provides a parity formula for the tangency number when $n=1$
2. Twisted degree forcing tangency criterion (Proposition 3):
   • Extends classical degree theory to homology with local orientation coefficients
   • Proves that zero twisted degree forces tangency
   • Applies to non-orientable base spaces
3. Smoothing technique (Lemma 14):
   • Develops local smoothing methods preserving Stiefel-Whitney classes
   • Via frame bundles and homotopy arguments, locally replaces $C^1$ foliations with smooth subbundles
4. Geometric applications:
   • Proves that on translation surfaces with cone singularities, any closed curve must be tangent to the foliation in periodic directions (Corollary 10)
   • Extends to rational polygonal billiard systems (Corollary 12)
   • Provides multiple non-trivial examples illustrating complementarity of the two criteria

Detailed Methodology

Problem Setup

Setting:

• $M^{\ell+n}$: smooth $(\ell+n)$-dimensional manifold
• $\mathcal{F}$: rank-$\ell$ $C^1$ foliation with tangent bundle $T\mathcal{F} \subset TM$
• $\nu\mathcal{F} := TM/T\mathcal{F}$: rank-$n$ normal bundle
• $S^n \subset M$: closed connected embedded submanifold with inclusion $\iota: S \hookrightarrow M$

Core map: $q: TM \to \nu\mathcal{F}, \quad d^\perp := q \circ \iota_* : TS \to \nu\mathcal{F}|_S$

Tangency criterion: $S$ is tangent to $\mathcal{F}$ at point $p$ $\Leftrightarrow$ $\text{rank}(d^\perp_p) \leq n-1$ $\Leftrightarrow$ $\det(d^\perp)(p) = 0$

Model Architecture

1. Determinant-line bundle method

Core construction: Define the line bundle $L := \det(TS)^* \otimes \det(\nu\mathcal{F})|_S \to S$

Determinant section: $\det(d^\perp) \in \Gamma(L)$

whose zero set precisely corresponds to the tangency set.

Stiefel-Whitney class computation (Formula (1)): $w_1(L) = w_1(TS) + w_1(\nu\mathcal{F})|_S \in H^1(S; \mathbb{Z}_2)$

utilizing properties: $w_1(\det E) = w_1(E)$ and $w_1(L^*) = w_1(L)$.

2. Smoothing technique (Core of Lemma 14)

Problem: The $C^1$ foliation $T\mathcal{F}$ is only a continuous subbundle; the determinant section $\det(d^\perp)$ lacks sufficient smoothness for direct application of classical transversality theorems.

Solution:

1. Local smooth approximation: On a neighborhood $U$ of $S$, choose a finite trivialization cover $\{U_i\}$
2. Frame approximation: On each $U_i$, use continuous frames $E_i = (e_1^{(i)}, \ldots, e_\ell^{(i)})$ spanning $T\mathcal{F}$, then uniformly approximate with smooth frames $\tilde{E}_i$
3. Gluing: Via smooth approximations $\tilde{g}_{ij}$ of transition functions $g_{ij}: U_i \cap U_j \to GL(\ell)$, construct a global smooth subbundle $\tilde{T} \subset TM|_U$
4. Homotopy connection: Construct linear homotopy $E_i^t = (1-t)E_i + t\tilde{E}_i$; for sufficiently close approximations, $E_i^t$ maintains full rank, yielding a subbundle homotopy $T^t$

Key property: The homotopy preserves Stiefel-Whitney classes: $w_1(\tilde{L}) = w_1(L)$ where $\tilde{L} := \det(TS)^* \otimes \det(\tilde{\nu})|_S$ and $\tilde{\nu} := TM/\tilde{T}$.

3. Transversality realization (Lemma 15)

Theorem statement: Let $\xi \to S$ be a real line bundle and $Z \subset S$ a closed embedded hypersurface. Then $Z$ is the zero set of some smooth section $s \in \Gamma(\xi)$ transverse to the zero section if and only if $[Z] = PD(w_1(\xi)) \in H_{n-1}(S; \mathbb{Z}_2)$

Proof strategy:

• $(\Rightarrow)$: Classical result; transverse zero sets represent Poincaré duals
• $(\Leftarrow)$: Constructive proof
  1. On $S \setminus Z$, $j^*w_1(\xi) = 0$, so $\xi|_{S\setminus Z}$ is trivial; take a nowhere-zero section $\sigma_0$
  2. Choose tubular neighborhood $\tau: (-\varepsilon, \varepsilon) \times Z \to S$ with normal bundle $\nu_{Z/S}$ satisfying $w_1(\nu_{Z/S}) = i^*PD([Z]) = i^*w_1(\xi) = w_1(\xi|_Z)$
  3. Construct bundle isomorphism $\Phi: \xi|_{\text{tube}} \xrightarrow{\sim} \tau^*(\nu_{Z/S})$
  4. Define section in tube $s_{\text{in}} := \Phi^{-1}(t \cdot n)$, where $t$ is the normal coordinate and $n$ is the unit normal vector
  5. Via smooth extension and cobordism techniques, glue $s_{\text{in}}$ with $\sigma_0$ into a global transverse section

Technical Innovations

1. Avoiding determinant singularities (Remark 13):
   • Rather than directly handling the determinant variety $\Sigma_{\leq r} = \{A: \text{rank } A \leq r\}$ (singular)
   • Use the determinant map $\det: \text{Hom}(TS, \nu\mathcal{F}|_S) \to L$, where the zero section is a smooth submanifold
2. Stiefel-Whitney classes as obstructions:
   • Convert geometric tangency problems to topological computations of characteristic classes
   • $w_1(L) \neq 0$ provides necessary obstruction; when $w_1(L) = 0$ an even number of tangencies may still occur
3. Degree theory in twisted homology:
   • Classical degree theory: orientable case, $\deg(f) \neq 0 \Rightarrow f$ surjective $\Rightarrow S$ transverse
   • This paper's extension: use orientation local coefficients $\mathcal{O}_B$, condition becomes $f_*[S]_{f^*\mathcal{O}_B} = 0$
   • Key observation: if $S$ is everywhere transverse, then $f$ is a finite covering; transfer maps give $f_*[S] = d[B_0] \neq 0$
4. Complementarity of the two criteria:
   • Example 4 (twisted Reeb torus): $w_1(L) \neq 0$ but not a simple foliation; Proposition 3 inapplicable
   • Example 5 (torus with transverse covering): $w_1(L) = 0$ and $\deg(f) \neq 0$; neither criterion obstructs
   • Example 6 (non-orientable base): $w_1(TB) \neq 0$ leads to $w_1(L) \neq 0$

Experimental Setup

Application scenario: Translation surfaces

Definition: A compact translation surface $(M, u)$ is obtained by gluing a parallelogram with a Riemann surface structure, possessing a flat metric and finite set of cone singularities $\Sigma$.

Periodic directions: A direction $\theta$ is called periodic if $M \setminus \Sigma$ decomposes into finitely many open cylinders $C_j \cong S^1_{L_j} \times (0, h_j), \quad j = 1, \ldots, k$ whose leaves are horizontal circles $S^1_{L_j} \times \{y\}$.

Base space construction (Corollary 7 & Remark 8):

• Cylindrical part: $B_\theta^\circ := \bigsqcup_{j=1}^k I_j$, $I_j = (0, h_j)$
• Saddle connections: $M \setminus \Sigma$ also contains finitely many open saddle connections (degenerate leaves, open intervals)
• Augmented base space: $B_\theta := B_\theta^\circ \sqcup \{p_1, \ldots, p_r\}$, each $p_m$ corresponding to a saddle connection
• Projection map: $\pi_\theta: M \setminus \Sigma \to B_\theta$ is a $C^1$ submersion on $M_\theta^\circ$, locally constant on saddle connections

Homological properties (Remark 9): $H_1(B_\theta; \mathbb{Z}) = 0, \quad \mathcal{O}_{B_\theta}\text{ trivial on each component}$

Application scenario: Rational polygonal billiards

Zemlyakov-Katok unfolding: The billiard flow in a rational polygon $P \subset \mathbb{R}^2$ can be unfolded via reflections to a translation surface $(M, u)$.

Local isometry: $U: P^\circ \setminus V \to M \setminus \Sigma$ ($V$ is the vertex set)

Conclusion: If direction $\theta$ is periodic on $(M, u)$, then any closed $C^1$ loop in $P^\circ \setminus V$ must be tangent to the billiard trajectory in direction $\theta$ somewhere.

Experimental Results

Main result: Tangency theorem on translation surfaces

Corollary 10: On a compact translation surface $(M, u)$ with cone singularities, for any periodic direction $\theta$, every closed $C^1$ loop $S$ in $M \setminus \Sigma$ is tangent to the linear foliation $\mathcal{F}_\theta$ somewhere.

Proof outline:

1. If $S$ intersects an open saddle connection, then $f = \pi_\theta|_S$ is locally constant on that segment, so $df = 0$, giving a tangency
2. Otherwise $S \subset M_\theta^\circ$, and $f := \pi_\theta^\circ|_S$ is a proper local diffeomorphism
3. From $H_1(B_\theta^\circ; \mathbb{Z}) = 0$, we get $f_*[S]_{f^*\mathcal{O}} = 0$
4. Apply Proposition 3; $S$ must have a tangency

Case studies

Example 4 (Twisted Reeb torus):

• Construction: Starting from Reeb component $(V = S^1 \times D^2)$, construct a codimension-1 foliation on 3-manifold $M$ by reversing transverse orientation via boundary diffeomorphism
• Suspended loop: $\gamma$ has orientation-reversing holonomy, $\langle w_1(\nu\mathcal{F}), [\gamma]\rangle = 1$
• Submanifold: Take $S = \Sigma_{g-1} \times \gamma$ ($\Sigma$ orientable)
• Computation: $w_1(L) = \iota^*w_1(\nu\mathcal{F}) \neq 0$
• Conclusion: Theorem 1 forces tangency; Proposition 3 inapplicable (non-simple foliation)

Example 5 (Transverse covering on torus):

• Foliation: Vertical circle foliation on $T^2$: $\pi(\theta, \phi) = \theta$
• Curve: $S_{p/q} = \{(\theta, \phi) = (qt, pt)\}$ ($\gcd(p,q)=1, q \neq 0$) everywhere transverse
• Computation:
  • $w_1(\nu\mathcal{F}) = 0$, $w_1(\det TS) = 0 \Rightarrow w_1(L) = 0$
  • $f = \pi|_{S_{p/q}}$ is $|q|$-fold covering, $\deg(f) = \pm q \neq 0$
• Conclusion: Neither criterion obstructs transversality (indeed transverse)

Example 6 (Non-orientable base):

• Construction: $M = F^\ell \times B^n$, projection $\pi$, $w_1(TB) \neq 0$ (e.g., $B = \mathbb{RP}^n$, $n$ even)
• Submanifold: $\Sigma^{n-1} \subset F$ orientable, $\gamma \subset B$ one-sided loop ($\langle w_1(TB), [\gamma]\rangle = 1$)
• Computation: $S = \Sigma \times \gamma$ has $w_1(\det TS) = 0, \quad f^*w_1(\det TB)|_\gamma \neq 0 \Rightarrow w_1(L) \neq 0$
• Conclusion: Theorem 1 forces tangency (odd number when $n=1$)

Experimental findings

1. Comparison: flat torus vs. cone singularity surfaces (Remark 11):
   • Flat torus $\Sigma = \emptyset$: periodic directions give closed cylinders, leaf space $\cong S^1$, admits circle-valued first integral, $\deg(\pi_\theta|_S)$ can be nonzero
   • Cone singularity surfaces: periodic cylinders are open, boundaries on saddle connections; base space $B_\theta$ is open intervals plus isolated points, $H_1(B_\theta) = 0$, Corollary 10 applies
2. Independence of the two criteria:
   • Exist cases where $w_1(L) = 0$ but twisted degree obstructs transversality
   • Exist cases where twisted degree is nonzero but $w_1(L) \neq 0$ obstructs transversality
   • Exist cases where neither obstructs and transversality indeed holds
3. Precision of parity information: When $n=1$, $\#Z \pmod{2}$ is completely determined by $\langle w_1(L), [S]\rangle$, providing a computable discrimination criterion

Related Work

1. Classical transversality theory

• Hirsch 2: $C^1$ transversality theorem, requiring sufficient smoothness of maps and target submanifolds
• This paper's contribution: Extends transversality theory to $C^1$ foliations via local smoothing preserving topological invariants

2. Characteristic class theory

• Milnor-Stasheff 4: Fundamental properties of Stiefel-Whitney classes and Poincaré duality of zero sets
• This paper's application: Uses $w_1(L)$ as topological obstruction to transversality, handling non-orientable cases

3. Degree theory and covering spaces

• Hatcher 1: Homology with local coefficients and transfer maps
• Lee 3: Proper local diffeomorphisms are finite coverings
• This paper's extension: Degree conditions in twisted homology, applicable to non-orientable base spaces

4. Translation surfaces and billiards

• Masur-Tabachnikov 6: Survey on rational billiards and flat structures
• Zemlyakov-Katok 5: Unfolding techniques for polygonal billiards
• This paper's application: Closed loops in periodic directions must have tangencies, supplementing topological constraints on translation surface dynamics

5. Foliation theory

• Husemoller 7: Classification theory of fiber bundles and homotopy properties
• This paper's technique: Via frame bundle homotopy, realizes smooth approximation of $C^1$ subbundles

Conclusions and Discussion

Main conclusions

1. Determinant-line bundle obstruction: Even for $C^1$ foliations, the tangency set is a smooth hypersurface after small perturbation, with homology class determined by $w_1(L)$; when $n=1$ gives parity of tangency number
2. Twisted degree criterion: When foliation is presented by a submersion and twisted degree is zero, tangency must exist; extends classical degree theory to non-orientable cases
3. Geometric applications: Proves that on translation surfaces with cone singularities and rational polygonal billiards, closed loops cannot be everywhere transverse in periodic directions

Limitations

1. Smoothness requirements:
   • Though extended to $C^1$ foliations, still requires $S$ to be smoothly embedded
   • Smoothing technique requires local trivializations; may be complex for globally non-trivial bundles
2. Sufficiency issues:
   • Both criteria are necessary conditions (obstructing transversality), not sufficient
   • When $w_1(L) = 0$ and twisted degree is nonzero, tangencies may still exist (Example 5 happens to have none)
3. Computational complexity:
   • Computing $w_1(L)$ requires knowing $w_1(\nu\mathcal{F})$
   • Twisted homology computation may be difficult on complex base spaces
4. Higher codimension cases:
   • When $n > 1$, only gives homology class of tangency hypersurface, not fine structure
   • Topological type of tangency set (connectivity, genus, etc.) not discussed

Future directions

1. Higher-order characteristic classes:
   • Study effects of $w_2, w_3$, etc. on transversality
   • Consider Pontryagin classes in orientable cases
2. Quantitative transversality:
   • Provide upper and lower bounds on tangency numbers
   • Study distribution and stability of tangencies
3. Dynamical systems applications:
   • Apply results to Anosov flows, pseudo-Anosov maps, etc.
   • Study transversality of periodic orbits with stable/unstable foliations
4. Algorithmic implementation:
   • Develop effective algorithms for computing $w_1(L)$ and twisted degree
   • Apply to classification of concrete geometric objects (polygonal billiards, translation surfaces)

In-depth Evaluation

Strengths

1. Theoretical innovation:
   • First systematic extension of transversality theory to $C^1$ foliations, filling the gap in smoothness requirements
   • Clever use of determinant line bundles to avoid handling singular determinant varieties
   • Natural and profound use of Stiefel-Whitney classes as obstructions
2. Technical contributions:
   • Lemma 14's smoothing technique has independent value; homotopy method preserving Stiefel-Whitney classes applicable to other problems
   • Lemma 15's constructive proof provides concrete method for realizing given homology classes as zero sets
3. Application value:
   • Corollary 10 is new for translation surface theory, revealing essential difference between cone singularity and flat tori
   • Billiard application (Corollary 12) connects dynamical systems and topology
4. Writing clarity:
   • Clear structure, hierarchical from general theory to concrete applications
   • Three examples (Examples 4-6) effectively illustrate complementarity and independence of the two criteria
   • Remarks provide rich background and intuitive explanations

Weaknesses

1. Proof conciseness trade-off:
   • Some technical details (e.g., specific frame bundle operations) are sketchy
   • Quantitative control of "sufficiently close approximation" in Lemma 14 not explicit
2. Example diversity:
   • Main examples concentrate on low dimensions ($n=1,2$) and special geometry (translation surfaces)
   • Lack applications to higher dimensions or more general geometries
3. Comparison with existing results:
   • Lacks detailed discussion of precise relationship between this paper's results and classical results in $C^2$ or orientable cases
   • Question of "when are the two criteria equivalent" not deeply explored
4. Algorithmic and computational aspects:
   • No algorithms or software for computing $w_1(L)$
   • For given translation surface or billiard system, no practical method for efficiently determining periodic directions

Impact

1. Contribution to field:
   • Foliation theory: Provides new topological obstruction tools applicable to low-regularity cases
   • Translation surface dynamics: Corollary 10 likely becomes standard lemma for periodic direction research
   • Billiard theory: Offers new perspective for studying rational polygonal billiard trajectories
2. Practical value:
   • Both criteria are computable topological invariants, in principle implementable
   • For specific geometric objects (particular translation surfaces or polygons), can directly apply discrimination
3. Reproducibility:
   • Complete theoretical proofs with sufficient references
   • Concrete, unambiguous examples independently verifiable
   • Lacks numerical experiments or code, acceptable for pure mathematics

Applicable scenarios

1. Theoretical research:
   • Study global properties of foliations (compactness, transversality, stability)
   • Explore topological constraints on manifolds (via characteristic classes)
   • Analyze invariant foliations in dynamical systems
2. Concrete geometry:
   • Moduli space research for translation surfaces
   • Periodic orbit classification in rational polygonal billiards
   • Foliation study in Teichmüller dynamics
3. Extension directions:
   • Lagrangian foliations in symplectic geometry
   • Characteristic foliations in contact geometry
   • Holomorphic foliations in complex geometry
4. Inapplicable scenarios:
   • Problems requiring quantitative estimates (exact tangency numbers)
   • Fine structure research of high-codimension tangency sets
   • Non-compact or manifolds with boundary (theory requires modification)

References

The paper cites the following key references:

1. Hatcher (2002): Standard algebraic topology text, local coefficient homology theory
2. Hirsch (1976): Classical differential topology work, transversality theorems
3. Lee (2013): Introduction to smooth manifolds, covering space theory
4. Milnor-Stasheff (1974): Foundational work on characteristic class theory
5. Zemlyakov-Katok (1975): Topological transitivity of polygonal billiards
6. Masur-Tabachnikov (2002): Survey on rational billiards and flat structures
7. Husemoller (1994): Fiber bundle theory

Overall assessment: This is a high-quality pure mathematics paper making substantial contributions to foliation transversality theory. Through clever techniques (determinant line bundles, local smoothing, twisted homology), it extends classical theory to low-regularity and non-orientable cases, with non-trivial applications to translation surfaces and billiard systems. The paper is logically rigorous with complete proofs, providing important reference value for researchers in geometric topology and dynamical systems. Main limitations are lack of examples in higher dimensions or more general settings, and absence of algorithmic details for practical computation.