Non-Orientable Lefschetz Fibration

• Non-orientable Lefschetz fibrations are smooth surjective maps from compact nonorientable 4-manifolds to orientable 2-manifolds with finitely many singular points modeled on the local complex product z₁z₂.
• They integrate techniques from mapping class groups, open book decompositions, and trisections, providing explicit monodromy factorizations and constructive models for nonorientable 3- and 4-manifolds.
• The study reveals deep connections with Pin±-structures and orientation double covers while highlighting unique features such as two-sided vanishing cycles and commutator length phenomena.

A non-orientable Lefschetz fibration is, in the now-standard sense, a smooth surjective map

$\pi:X\to B$

from a compact, connected, nonorientable $4$-manifold $X$ to a compact, connected, orientable $2$-manifold $B$, with finitely many critical points in the interior of $X$, such that near each critical point the map is locally modeled by

$\pi(z_1,z_2)=z_1z_2$

in local complex coordinates. Because the total space is nonorientable while the base is orientable, the regular fiber is necessarily a nonorientable surface. The subject emerged as a nonorientable analogue of the classical Lefschetz-fibration technology in smooth $4$-manifold topology, and it now interacts with monodromy factorizations, open books, trisections, Pin${}^\pm$-structures, and orientation double covers (Miller et al., 2020, Onaran et al., 2021, Bais, 3 Jan 2025, Yoshikawa, 8 Sep 2025).

1. Definition and local geometric structure

The basic definition differs from the orientable case precisely in how orientation enters. Since $X$ is nonorientable, one cannot require the local complex coordinates near a singularity to agree with a global orientation on $4$0. Nevertheless, the singularity model is the same Lefschetz model used in the orientable theory, and singular fibers may be arranged, after perturbation, to contain a single critical point (Miller et al., 2020).

For fibrations over $4$1, the handle description is parallel to the orientable case. One starts with the trivial fibration $4$2, where $4$3 is a nonorientable surface with nonempty boundary, and attaches $4$4-handles along two-sided simple closed curves in the fiber, one for each critical point, with framing $4$5 relative to the surface framing. The local monodromy around a critical value is the Dehn twist about the corresponding vanishing cycle (Miller et al., 2020). The condition that vanishing cycles be two-sided recurs throughout the subject; in the closed-fiber setting, an admissible nonorientable Lefschetz fibration is required to have at least one singular fiber and all vanishing cycles nontrivial, meaning that no vanishing cycle bounds a disk or a Möbius band (Onaran et al., 2021).

A distinctive subtlety is the status of twist signs. In the orientable case, a Dehn twist is canonically labeled right-handed or left-handed once an orientation is fixed. In the nonorientable case, a two-sided curve has an annular neighborhood but no canonical orientation on that annulus, so the sign of a twist is only locally meaningful (Miller et al., 2020). This affects both monodromy notation and classification statements.

Relative minimality is defined in the expected way: a nonorientable Lefschetz fibration is relatively minimal if no fiber contains an exceptional sphere, namely a sphere with self-intersection $4$6 (Miller et al., 2020). When $4$7 has boundary and the regular fiber has nonempty boundary, the fibration induces an open book decomposition on $4$8, providing a direct bridge to $4$9-manifold topology (Miller et al., 2020).

2. Existence results and the nonorientable Harer theorem

A foundational existence theorem states that every nonorientable $X$0-dimensional handlebody without $X$1- and $X$2-handles admits an explicit Lefschetz fibration over $X$3 whose regular fiber is a nonorientable surface with nonempty boundary (Miller et al., 2020). This is the nonorientable analogue of Harer’s theorem in the orientable setting.

The construction begins with the $X$4-handlebody

$X$5

where $X$6 is obtained from $X$7 by attaching $X$8 nonorientable $X$9-handles. This already gives a trivial Lefschetz fibration over $2$0. One then projects the attaching link $2$1 of the $2$2-handles to $2$3, resolves double points by stabilizing the fiber, isotopes each component of $2$4 into a distinct page of the induced open book, and performs further stabilizations so that each attaching circle has framing $2$5 relative to the page framing. The $2$6-handles then become Lefschetz singularities, with vanishing cycles given by the resulting two-sided curves (Miller et al., 2020).

The first major application is an open-book existence theorem: every nonorientable closed $2$7-manifold admits an open book decomposition whose monodromy can be expressed as a product of Dehn twists (Miller et al., 2020). The construction supplies more information than the earlier branched-covering proof of Berstein and Edmonds, because the monodromy factorization is explicit. Moreover, since Dehn twists generate the index-two twist subgroup of the mapping class group of a nonorientable surface, the resulting monodromies land in that twist subgroup (Miller et al., 2020).

The same framework yields concrete examples. There is an explicit genus-one open book for $2$8 whose page is a projective plane with two holes and whose monodromy is the product of Dehn twists about two boundary-parallel curves $2$9. By Murasugi sum, this extends to open books on

$B$0

with genus-one page $B$1 with $B$2 holes. The associated minimality statement is sharp: for any $B$3, a genus-one open book for $B$4 must have at least $B$5 binding components (Miller et al., 2020).

3. Monodromy, mapping class groups, and orientation-sensitive notation

The monodromy theory of nonorientable Lefschetz fibrations parallels the orientable case but is phrased in the mapping class group $B$6 of a nonorientable fiber $B$7. For a fibration over a closed orientable base of genus $B$8, the existence problem is equivalent to a factorization of the identity in $B$9 as a product of commutators and Dehn twists. In particular, if a Dehn twist $X$0 about a two-sided curve $X$1 is equal to a product of $X$2 commutators in $X$3, then there is an admissible genus-$X$4 Lefschetz fibration with one singular fiber over a closed orientable surface of genus $X$5 (Onaran et al., 2021).

Because a Dehn twist on a nonorientable surface depends on a choice of orientation of the tubular neighborhood, later work writes such a twist as

$X$6

where $X$7 is a two-sided curve and $X$8 is an orientation of its annular neighborhood; reversing $X$9 inverts the twist (Yoshikawa, 8 Sep 2025). This notation becomes essential when comparing nonorientable monodromy with the monodromy of the orientation double cover.

If

$\pi(z_1,z_2)=z_1z_2$0

is a genus-$\pi(z_1,z_2)=z_1z_2$1 non-orientable Lefschetz fibration and

$\pi(z_1,z_2)=z_1z_2$2

is the standard orientation double covering map, then the composite

$\pi(z_1,z_2)=z_1z_2$3

is an achiral Lefschetz fibration of genus $\pi(z_1,z_2)=z_1z_2$4, with the same number of positive and negative critical points (Yoshikawa, 8 Sep 2025). On mapping class groups, there is an injective homomorphism

$\pi(z_1,z_2)=z_1z_2$5

for $\pi(z_1,z_2)=z_1z_2$6, sending a nonorientable mapping class to its unique orientation-preserving lift. If the preimage of $\pi(z_1,z_2)=z_1z_2$7 in the orientation double cover is $\pi(z_1,z_2)=z_1z_2$8, then

$\pi(z_1,z_2)=z_1z_2$9

Accordingly, the monodromy representation satisfies

$4$0

so each nonorientable Dehn twist factor lifts to a pair of inverse twists upstairs (Yoshikawa, 8 Sep 2025).

This viewpoint produces a nonorientable analogue of Hurwitz-equivalence classification. For $4$1, two genus-$4$2 non-orientable Lefschetz fibrations over $4$3 or $4$4 are isomorphic if and only if their positive monodromy factorizations are related by two elementary Hurwitz-type moves together with simultaneous conjugation by an auto-diffeomorphism of $4$5 (Yoshikawa, 8 Sep 2025).

4. Singular fibers, admissibility, and genus thresholds

A central quantitative invariant is

$4$6

the minimal number of singular fibers in an admissible genus-$4$7 nonorientable Lefschetz fibration over a closed orientable surface of genus $4$8. The sharp theorem is

$4$9

equivalently, there exists an admissible nonorientable genus-${}^\pm$0 Lefschetz fibration with exactly one singular fiber over a closed orientable surface of genus ${}^\pm$1 if and only if ${}^\pm$2 and ${}^\pm$3 (Onaran et al., 2021).

The proof is entirely group-theoretic. For genus ${}^\pm$4, if ${}^\pm$5 is a closed nonorientable surface and ${}^\pm$6 is any nontrivial separating curve, then

${}^\pm$7

for every ${}^\pm$8. By contrast, for genus ${}^\pm$9 or $X$0, no Dehn twist along a nontrivial curve belongs to the commutator subgroup $X$1 (Onaran et al., 2021). For $X$2, the theorem uses Szepietowski’s result that every power of every Dehn twist on a closed nonorientable surface of genus at least $X$3 has commutator length $X$4. The low-genus obstructions are equally explicit: when $X$5, $X$6, so its commutator subgroup is trivial; when $X$7, $X$8, so nontrivial twists survive in abelianization (Onaran et al., 2021).

Independent low-genus rigidity appears already in the existence theory over closed bases. If

$X$9

is a relatively minimal genus-one Lefschetz fibration on a closed nonorientable $4$00-manifold over a closed orientable surface $4$01, then $4$02 is an $4$03-bundle over $4$04 (Miller et al., 2020). The reason is that on a genus-one nonorientable surface the only homotopically nontrivial simple closed curve is one-sided, whereas vanishing cycles must be two-sided; hence a relatively minimal genus-one fibration has no singular fibers. By contrast, for every $4$05 there are relatively minimal nonorientable genus-$4$06 Lefschetz fibrations over closed orientable surfaces with arbitrarily many singular fibers (Miller et al., 2020).

Genus two already exhibits specifically nonorientable phenomena. For the Klein bottle $4$07,

$4$08

one generator being a Dehn twist about the unique essential two-sided curve and the other a $4$09-homeomorphism. Since $4$10, this gives genus-two examples with two singular fibers. There is also a non-generic genus-two fibration over $4$11 whose vanishing cycle bounds a Möbius band, and the associated Dehn twist is isotopic to the identity (Miller et al., 2020).

5. Open books, trisections, and closed nonorientable $4$12-manifolds

Nonorientable Lefschetz fibrations over $4$13 induce open books on boundary $4$14-manifolds, and these open books are compatible with relative trisections. Given any Lefschetz fibration

$4$15

on a nonorientable $4$16-dimensional handlebody without $4$17- and $4$18-handles, there is an explicit algorithm producing a relative trisection diagram

$4$19

such that the open book induced by the Lefschetz fibration coincides with that induced by the trisection (Miller et al., 2020). In the construction, each vanishing cycle is replaced by a local red-blue-green configuration, the fiber is stabilized by adding tubes, and one obtains curve systems

$4$20

The pairs $4$21 and $4$22 are standard, while $4$23 becomes standard after handleslides (Miller et al., 2020).

Doubling the relative picture gives closed-manifold trisections. If

$4$24

is the double of $4$25, then gluing the identical relative trisection diagrams of the two copies yields a trisection diagram for $4$26. The construction is illustrated in the case

$4$27

(Miller et al., 2020). More generally, if $4$28 is a closed nonorientable $4$29-manifold admitting a Lefschetz fibration over $4$30 equipped with a section of square $4$31, then there is an explicit trisection diagram of $4$32 determined by the vanishing cycles of the fibration (Miller et al., 2020).

A related but distinct boundary phenomenon appears in the theory of real structures. There exists a real open book on a boundary $4$33-manifold that cannot be filled by any real Lefschetz fibration with the same fiber topology, even though it is filled by non-real Lefschetz fibrations (Ozturk et al., 2013). This does not concern nonorientable total spaces, but it shows that compatibility conditions visible on the boundary need not extend to the filling fibration.

6. Pin$4$34-structures, orientation double covers, and neighboring notions

Recent work has made the topology of nonorientable Lefschetz fibrations computable in terms of vanishing cycles. For a Lefschetz fibration

$4$35

with regular fiber $4$36, Pin$4$37-structures on $4$38 correspond to quadratic enhancements

$4$39

satisfying

$4$40

for all $4$41, together with the conditions

$4$42

for every vanishing cycle $4$43 (Bais, 3 Jan 2025). There is no Pin$4$44-structure on $4$45 if and only if there exist vanishing cycles $4$46 such that

$4$47

and

$4$48

Pin$4$49-structures are governed by a different quadratic enhancement,

$4$50

with

$4$51

for each vanishing cycle. The total space supports a Pin$4$52-structure if and only if $4$53 supports one and

$4$54

where $4$55 is the $4$56-reduction of the matrix of vanishing-cycle classes in $4$57, and $4$58 (Bais, 3 Jan 2025). These criteria extend Stipsicz’s orientable Spin-theoretic results to the nonorientable Pin setting.

The same paper uses the Miller–Ozbagci decomposition

$4$59

where $4$60 is a nonorientable Lefschetz fibration over $4$61 and $4$62 a nonorientable $4$63-handlebody, to reduce Pin questions for closed nonorientable smooth $4$64-manifolds to the Lefschetz-fibration piece (Bais, 3 Jan 2025). This suggests that nonorientable Lefschetz fibrations are not merely examples but organizing structures for broader obstruction theory.

Several nearby notions should be separated from the nonorientable case. The paper on non-holomorphic surface bundles and Lefschetz fibrations constructs closed oriented $4$65-manifolds admitting oriented genus-$4$66 Lefschetz fibrations over oriented genus-$4$67 bases, with nonzero signature and no complex structure in either orientation; it does not construct non-orientable Lefschetz fibrations (Baykur, 2011). Likewise, achiral Lefschetz fibrations over $4$68 with orientable fibers and no sections, although orientation-sensitive, are not fibrations with nonorientable fibers (Gompf, 22 Jun 2025). A plausible implication is that the terminology surrounding orientation in the Lefschetz-fibration literature must be used with care: “non-orientable,” “non-holomorphic,” and “achiral” encode different failures of compatibility with orientation or complex structure.

At a structural level, the orientation double-cover construction furnishes the cleanest bridge between the genuinely nonorientable and orientable theories. Starting from a non-orientable Lefschetz fibration, one obtains an achiral orientable fibration of genus one lower, with monodromy obtained by lifting each nonorientable Dehn twist to a pair of inverse Dehn twists on the double cover (Yoshikawa, 8 Sep 2025). This places nonorientable Lefschetz fibrations within the classical Lefschetz-fibration framework while preserving the distinctive mapping-class-group and vanishing-cycle features of the nonorientable category.