Euler Class in Mathematics

• Euler class is a characteristic invariant defined via obstruction theory and Thom’s construction, measuring the failure of an oriented vector bundle to admit a nowhere-vanishing section.
• It extends to diverse settings—from plane fields and foliations to surface bundles and Bloch bundles in band topology—guiding rigidity and realizability phenomena.
• Recent developments provide explicit combinatorial formulas, bounded cohomology techniques, and algebraic-motivic approaches that link classical topology with modern applications.

Searching arXiv for the cited papers and related Euler class sources. The Euler class is a characteristic class attached to an oriented real vector bundle, an oriented plane field, or closely related geometric structures. In its obstruction-theoretic form, it is the primary obstruction to the existence of a nowhere-vanishing section; in its Thom-theoretic form, it is the pullback of the Thom class along the zero section. For tangent bundles it recovers the Euler characteristic, while in low-dimensional topology, foliation theory, algebraic geometry, bounded cohomology, and recent band-topological settings it becomes a refined invariant controlling existence, rigidity, and realizability phenomena (Panina, 2022, Yazdi, 9 Apr 2026, Schlichting, 2015, Jankowski et al., 2023).

1. Classical definition and basic geometric meaning

For an oriented rank-$(n+1)$ real vector bundle $\pi:E\to B$, the Euler class $e(E)\in H^{n+1}(B;\mathbb{Z})$ is the primary obstruction to the existence of a nowhere-vanishing section. Equivalently, if $u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$ is the Thom class and $s_0:B\to E$ is the zero section, then

$e(E)=s_0^*(u).$

This obstruction-theoretic and Thom-theoretic equivalence is a recurring template in both topology and algebraic geometry (Panina, 2022, Asok et al., 2013).

For oriented rank-$2$ bundles, the same description appears in especially concrete form. If $E\to M$ is an oriented rank-$2$ real vector bundle, then $e(E)\in H^2(M;\mathbb{Z})$ measures the obstruction to a nowhere-vanishing section. On a closed oriented surface $\pi:E\to B$0, the tangent bundle satisfies

$\pi:E\to B$1

and Gauss–Bonnet gives

$\pi:E\to B$2

In the notation of orientation classes, if $\pi:E\to B$3 is the orientation class of a connected, closed, oriented $\pi:E\to B$4-manifold, then

$\pi:E\to B$5

These formulas place the Euler class at the interface of characteristic classes, index theory, and global curvature (Kim et al., 18 Aug 2025, Menichi, 2013).

For oriented circle bundles over closed oriented surfaces, the Euler class is the degree-$\pi:E\to B$6 characteristic class classifying the bundle. This low-dimensional case underlies several later developments, including group-cohomological central extensions, local singularity formulas, and flux–Euler transgression phenomena (Panina et al., 2024, Kim et al., 18 Aug 2025).

2. Plane fields, foliations, contact structures, and the Thurston norm

On an orientable $\pi:E\to B$7-manifold, an oriented $\pi:E\to B$8-plane field $\pi:E\to B$9 has an Euler class $e(E)\in H^{n+1}(B;\mathbb{Z})$0, defined as the primary obstruction to trivializing $e(E)\in H^{n+1}(B;\mathbb{Z})$1 over the $e(E)\in H^{n+1}(B;\mathbb{Z})$2-skeleton. For manifolds with boundary there is a relative Euler class in $e(E)\in H^{n+1}(B;\mathbb{Z})$3 once a boundary trivialization is fixed. If $e(E)\in H^{n+1}(B;\mathbb{Z})$4 is transversely oriented on a closed orientable $e(E)\in H^{n+1}(B;\mathbb{Z})$5-manifold, then the parity condition holds:

$e(E)\in H^{n+1}(B;\mathbb{Z})$6

For cooriented contact structures one has

$e(E)\in H^{n+1}(B;\mathbb{Z})$7

after viewing $e(E)\in H^{n+1}(B;\mathbb{Z})$8 as a complex line bundle via a compatible almost complex structure (Yazdi, 9 Apr 2026).

Thurston’s norm on $e(E)\in H^{n+1}(B;\mathbb{Z})$9 is defined by

$u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$0

with dual norm on cohomology

$u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$1

Its dual unit ball

$u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$2

is a compact convex polytope with integral vertices (Yazdi, 9 Apr 2026).

For a transversely oriented taut foliation $u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$3, the tangent plane field $u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$4 has Euler class $u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$5, and Thurston’s index-sum argument yields

$u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$6

for incompressible $u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$7 with $u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$8, equivalently

$u\in H^{n+1}(E,E\setminus 0;\mathbb{Z})$9

Hence $s_0:B\to E$0. If $s_0:B\to E$1 is a compact leaf whose transverse orientation agrees, then

$s_0:B\to E$2

so equality occurs. Tight contact structures satisfy the analogous Eliashberg inequalities, and therefore their Euler classes also satisfy $s_0:B\to E$3 (Yazdi, 9 Apr 2026).

These inequalities motivated Thurston’s Euler class one conjecture: on a closed, orientable, irreducible, atoroidal $s_0:B\to E$4-manifold with $s_0:B\to E$5, every integral class of dual norm one satisfying the parity condition should be realized as the Euler class of a taut foliation. Gabai proved that every vertex of the dual unit ball is realized. Yazdi constructed counterexamples using the Fully Marked Surface Theorem, and Liu later proved that every closed hyperbolic $s_0:B\to E$6-manifold has a finite cover where an even lattice point of dual norm one is not the real Euler class of any weakly symplectically fillable contact structure, hence not of any transversely oriented taut foliation (Yazdi, 9 Apr 2026, Yazdi, 2016, Liu, 2024).

The same circle of ideas extends to pseudo-Anosov flows, quasigeodesic flows, universal circle actions, and circular orders on $s_0:B\to E$7, all of which produce integral points in the dual Thurston norm ball. In that setting the Euler class is simultaneously geometric, dynamical, and order-theoretic (Yazdi, 9 Apr 2026).

3. Surface bundles, circle actions, and bounded or unbounded Euler classes

For a smooth oriented surface bundle $s_0:B\to E$8 with fiber $s_0:B\to E$9 and section $e(E)=s_0^*(u).$0, the vertical tangent bundle $e(E)=s_0^*(u).$1 pulls back to an oriented rank-$e(E)=s_0^*(u).$2 bundle $e(E)=s_0^*(u).$3. The Euler class of the bundle with marked point is

$e(E)=s_0^*(u).$4

where $e(E)=s_0^*(u).$5 is induced by the derivative at the marked point. For Nielsen-convex hyperbolic surfaces, the Nielsen action $e(E)=s_0^*(u).$6 gives the same class:

$e(E)=s_0^*(u).$7

For many infinite-type surfaces this class is nontrivial, its powers are often nontrivial, and their order depends on genus and end structure (Bustamante et al., 25 Sep 2025).

For infinite-genus surfaces, one has an injective ring homomorphism

$e(E)=s_0^*(u).$8

while for finite-genus infinite-type surfaces $e(E)=s_0^*(u).$9 the classes $2$0 obtained from Freudenthal compactification combine with the marked-point Euler classes in an injective map up to degree $2$1:

$2$2

There are also uncountable genus-zero families for which $2$3 and all powers $2$4 have infinite order, and these results feed directly into extensions of Morita’s non-lifting theorem to infinite-type surfaces (Bustamante et al., 25 Sep 2025).

A different but related perspective comes from bounded cohomology. For flat oriented real rank-$2$5 bundles, Bucher and Monod proved that the norm of the Euler class is

$2$6

for even $2$7, while the class vanishes in odd dimension. They also constructed a cocycle representative taking only the values $2$8 and proved uniqueness of the antisymmetric bounded representative (Bucher et al., 2010). This sharpens the Sullivan–Smillie upper bound and gives universal Milnor–Wood-type estimates for Euler numbers of flat bundles.

The behavior can be sharply different for homeomorphism groups of higher-dimensional fibers. For closed Seifert fibered $2$9-manifolds $E\to M$0 satisfying the stated $E\to M$1 hypothesis, the degree-$E\to M$2 Euler class in $E\to M$3 is unbounded. More strongly, for any integer $E\to M$4 there exists

$E\to M$5

with

$E\to M$6

This contrasts with the bounded Euler class for circle actions and shows that “higher Euler classes” for flat topological $E\to M$7-bundles can behave fundamentally differently (Mann, 2017).

4. Algebraic, motivic, and commutative-algebraic forms of the Euler class

In algebraic geometry, Schlichting constructed a cohomological Euler class for an oriented rank-$E\to M$8 projective module $E\to M$9 over a commutative noetherian ring $2$0 of dimension $2$1 with infinite residue fields. Writing $2$2, the class lies in

$2$3

and satisfies the splitting criterion

$2$4

The same holds for orientations in an arbitrary line bundle $2$5, with

$2$6

Here Milnor–Witt $2$7-theory appears as the exact obstruction to “one more step” of homology stability for special linear groups (Schlichting, 2015).

Asok and Fasel compared two algebraic Euler classes: the Chow–Witt characteristic class and the $2$8-obstruction class. For a rank-$2$9 vector bundle $e(E)\in H^2(M;\mathbb{Z})$0 over a smooth $e(E)\in H^2(M;\mathbb{Z})$1-scheme, they define

$e(E)\in H^2(M;\mathbb{Z})$2

and show that under the identification

$e(E)\in H^2(M;\mathbb{Z})$3

the obstruction-theoretic Euler class and the Chow–Witt Euler class agree up to a unit in $e(E)\in H^2(M;\mathbb{Z})$4 (Asok et al., 2013).

Bachmann, Déglise, Jin, and Khan further unified several motivic Euler constructions. For a vector bundle $e(E)\in H^2(M;\mathbb{Z})$5 and a motivic ring spectrum $e(E)\in H^2(M;\mathbb{Z})$6, the tautological Euler class is

$e(E)\in H^2(M;\mathbb{Z})$7

and for a section $e(E)\in H^2(M;\mathbb{Z})$8 with zero scheme $e(E)\in H^2(M;\mathbb{Z})$9 the refined class is

$\pi:E\to B$00

They proved the section-pullback identity

$\pi:E\to B$01

and identified the Barge–Morel, Kass–Wickelgren, Déglise–Jin–Khan, and Hopkins–Raksit–Serre constructions. In the presence of isolated zeros, the global Euler number decomposes into local indices computed by the Scheja–Storch bilinear form

$\pi:E\to B$02

and over $\pi:E\to B$03 the resulting Grothendieck–Witt-valued Euler number is determined by the complex and real topological Euler numbers (Bachmann et al., 2020).

Hu and Li introduced yet another extension: for a perfect derived object $\pi:E\to B$04 on an integral Deligne–Mumford stack, a birational derived resolution makes $\pi:E\to B$05 locally free, and one defines

$\pi:E\to B$06

Applied to $\pi:E\to B$07 on the primary component of the moduli stack of stable maps, this yields Euler numbers conjectured to be the reduced Gromov–Witten invariants of the smooth quintic (Hu et al., 2010).

In commutative algebra, Euler class groups $\pi:E\to B$08 encode the obstruction to splitting off a free rank-one summand from a rank-$\pi:E\to B$09 projective module. For a stably free $\pi:E\to B$10-module $\pi:E\to B$11 of rank $\pi:E\to B$12 in the stable range $\pi:E\to B$13, one has

$\pi:E\to B$14

and the weak Euler class group $\pi:E\to B$15 controls analogous unoriented and even-dimensional splitting phenomena (Keshari, 2010, Keshari, 2014).

5. Explicit formulas and combinatorial representatives

Although the Euler class is intrinsically obstruction-theoretic, several recent works give local or combinatorial formulas that make it computable.

For a fiber-oriented triangulated spherical bundle $\pi:E\to B$16, Panina constructs a rational simplicial cocycle representing the Euler class by averaging partial sections on the $\pi:E\to B$17-skeleton and extending them face-by-face by harmonic chains or, in the final step, winding numbers. The resulting cochain $\pi:E\to B$18 satisfies

$\pi:E\to B$19

In the circle-bundle case the formula reduces to

$\pi:E\to B$20

recovering the local combinatorial formula of Igusa–Mnëv–Sharygin (Panina, 2022).

For an oriented circle bundle $\pi:E\to B$21 over an oriented closed surface, a quasisection is a smooth surface mapped generically to $\pi:E\to B$22 and surjecting onto $\pi:E\to B$23 after projection. Chernyshev and Panina proved that the Euler number is the sum of local weights of three essential singular-vertex types. The weights are

$\pi:E\to B$24

$\pi:E\to B$25

and

$\pi:E\to B$26

They also proved uniqueness: any local formula with the natural axioms must coincide with this one (Panina et al., 2024).

For foliations carried by cooriented branched surfaces, a similarly explicit formula exists. If $\pi:E\to B$27 is a cooriented branched surface with product-ball exterior and $\pi:E\to B$28 is a fully carried foliation, then each sector $\pi:E\to B$29 has maw Euler characteristic

$\pi:E\to B$30

where $\pi:E\to B$31 counts double corners. Writing

$\pi:E\to B$32

with $\pi:E\to B$33 the coorientation-oriented dual edge, one obtains

$\pi:E\to B$34

This formula also quantifies the change under reversing the orientation of a final decomposing disk in a sutured hierarchy:

$\pi:E\to B$35

It recovers and generalizes earlier formulas of Lackenby and Dunfield (Cigna, 16 Feb 2026).

These constructions show that the Euler class is not merely a formal obstruction. In many settings it admits local weights, dual cycles, or harmonic representatives that make its variation under surgeries, triangulations, or singular maps effectively trackable.

6. Loop spaces, non-orientable dynamics, and modern band topology

The Euler class also appears in string topology. For a connected, closed, oriented manifold $\pi:E\to B$36 with free loop fibration

$\pi:E\to B$37

Menichi proved that for every class $\pi:E\to B$38 of positive degree,

$\pi:E\to B$39

where $\pi:E\to B$40 is the orientation class. In particular, if

$\pi:E\to B$41

is surjective, then either $\pi:E\to B$42 is a point or $\pi:E\to B$43 divides $\pi:E\to B$44 (Menichi, 2013). Here the Euler class of the tangent bundle enters via the diagonal shriek map and the identity $\pi:E\to B$45.

For non-orientable compact surfaces with one boundary component, the correct framework uses the orientation line bundle $\pi:E\to B$46. If $\pi:E\to B$47 is an area density with $\pi:E\to B$48, the flux homomorphism is

$\pi:E\to B$49

Pairing with a closed ordinary $\pi:E\to B$50-form $\pi:E\to B$51 gives

$\pi:E\to B$52

The transgression of $\pi:E\to B$53 along the boundary restriction exact sequence yields

$\pi:E\to B$54

where $\pi:E\to B$55 is the Euler class of circle diffeomorphisms. In the same non-orientable setting, the kernel of the flux homomorphism is simple, excluding Calabi-type homomorphisms analogous to the orientable case (Kim et al., 18 Aug 2025).

A distinct but structurally related usage appears in real band topology. For a real two-band subspace in a $\pi:E\to B$56-dimensional Brillouin zone with $\pi:E\to B$57 or $\pi:E\to B$58 symmetry, the Euler class is a multigap non-Abelian invariant measuring the obstruction to an everywhere-orientable globally smooth real frame. In the notation of recent condensed-matter papers, it is written $\pi:E\to B$59 or $\pi:E\to B$60. One representation is

$\pi:E\to B$61

and in a three-band model it can also be expressed as a skyrmion number. This invariant controls optical response, Landau levels, and quench dynamics (Jankowski et al., 2023, Guan et al., 2021, Ünal et al., 2020).

Optically, the many-band quantum metric obeys

$\pi:E\to B$62

which implies the high-frequency optical-weight bound

$\pi:E\to B$63

Near Euler nodes, the optical conductivity and jerk photoconductivity exhibit universal signatures, and momentum-resolved optical measurements reconstruct the Euler connection and curvature directly (Jankowski et al., 2023).

In magnetic field, Euler phases have robustly gapless Hofstadter spectra in the flat-band limit, and the Euler class gives a lower bound for magnetic subgap Chern numbers:

$\pi:E\to B$64

For quenches in optical lattices, the first Hopf map produces a signed linking invariant $\pi:E\to B$65, making the Euler class dynamically observable through momentum-time tomography (Guan et al., 2021, Ünal et al., 2020).

Across these settings, the Euler class retains a common core: it measures an obstruction to global trivialization of an oriented structure. What changes is the ambient category—topological bundles, foliated plane fields, projective modules, perfect derived objects, or real Bloch bundles—and with it the most natural language for realizing the obstruction: homotopy, cohomology, bounded norms, local singularity weights, or observable response functions.