Enomoto–Satoh Obstruction in Johnson Cokernel

• The Enomoto–Satoh obstruction is a canonical symplectic trace that detects the top layer of the Johnson cokernel using cyclic and dihedral coinvariants.
• It translates the geometric non-surjectivity of the Johnson homomorphism into explicit Sp-representation computations and dihedral invariant analysis.
• Graphical and topological reformulations extend this approach, underpinning modern methods in graph complexes, regular-homotopy Lie bialgebras, and non-commutative torsion.

Searching arXiv for papers on the Enomoto–Satoh obstruction and related Johnson cokernel work. The Enomoto–Satoh obstruction is the obstruction to surjectivity of the Johnson homomorphism furnished by the Enomoto–Satoh trace, a degree-zero $Sp$-equivariant trace from symplectic derivations of a free Lie algebra, or equivalently from the Johnson cokernel, to a space of cyclic or dihedral coinvariants. In the setting of the Torelli group of a surface with one boundary component, it detects classes in the cokernel of the Johnson homomorphism that cannot lie in its image. In its modern form, the obstruction is realized as the rank-$1$ or $1$-loop part of a broader graphical trace formalism, and in the stable range it identifies the top layer of the Johnson cokernel with a dihedral coinvariant quotient $[V(d)]_{D_{2d}}$ (Conant, 2013, Conant, 2016). Subsequent work has placed it in several parallel frameworks: hairy graph complexes, top cohomology of $\mathrm{Out}(F_n)$, regular-homotopy refinements of the Goldman–Turaev Lie bialgebra, non-commutative torsion, and uniqueness theorems for degree-zero cocycles (Conant et al., 2015, Kawazumi, 2014, Nozaki et al., 2022, Baudat, 10 Nov 2025).

1. Johnson homomorphisms and the cokernel problem

Let $\Sigma_g$ or $\Sigma_{g,1}$ be an oriented surface, and let $V = H_1(\Sigma_g;\mathbb{Q})$ or $H = H_1(\Sigma_{g,1};\mathbb{Q})$, equipped with its symplectic form. The Torelli group is the subgroup of the mapping class group acting trivially on first homology. The Johnson homomorphism packages the induced action on the lower central series of $\pi_1$ into graded maps

$1$0

or, equivalently, into the Lie algebra of symplectic derivations of the free Lie algebra (Conant, 2016).

The Johnson cokernel in degree $1$1 is the quotient of this target by the image of $1$2. In the notation of Conant, one writes

$1$3

or, in related notation,

$1$4

where $1$5 is the Lie subalgebra generated by the degree-$1$6 part and Hain’s theorem identifies it with the Johnson image in the stable range (Conant, 2013, Kuno et al., 26 Aug 2025). The central problem is that the Johnson homomorphism is injective but not surjective, and the obstruction theory concerns explicit invariants on this cokernel.

As an $1$7-module, the degree $1$8 Johnson cokernel decomposes into layers indexed by partition sizes descending by even steps: $1$9 The summand $1$0 is the top layer, or the piece of “top-level partitions,” indexed by partitions of $1$1 itself (Conant, 2016). The Enomoto–Satoh obstruction is, in the first instance, the mechanism detecting this top layer.

2. Definition of the Enomoto–Satoh trace

In the symplectic Lie-algebra formulation, let $1$2 be the free Lie algebra on

$1$3

and let $1$4 denote the Lie algebra of derivations preserving the standard symplectic form encoded by $1$5. For $1$6, the Enomoto–Satoh trace is defined by

$1$7

where $1$8, $1$9 contracts the first tensor slot, $[V(d)]_{D_{2d}}$0 embeds Lie words into the tensor algebra by expanding brackets, and $[V(d)]_{D_{2d}}$1 projects to cyclic words (Baudat, 10 Nov 2025).

Concretely, one starts with an element of $[V(d)]_{D_{2d}}$2, expands the Lie word into a signed sum in $[V(d)]_{D_{2d}}$3, contracts the first slot against the dual element, and then passes to the cyclic class of the remaining $[V(d)]_{D_{2d}}$4 letters. The result is a degree-zero $[V(d)]_{D_{2d}}$5-cocycle on $[V(d)]_{D_{2d}}$6 with values in $[V(d)]_{D_{2d}}$7 (Baudat, 10 Nov 2025).

A basic degree-$[V(d)]_{D_{2d}}$8 example is given by

$[V(d)]_{D_{2d}}$9

Expanding $\mathrm{Out}(F_n)$0, contracting via $\mathrm{Out}(F_n)$1, and projecting cyclically yields

$\mathrm{Out}(F_n)$2

This exemplifies the defining feature of the trace: the contracted index disappears, and the remaining letters are recorded only up to cyclic rotation (Baudat, 10 Nov 2025).

In the graphical formulation, the same trace is defined on rank-$\mathrm{Out}(F_n)$3 hairy graphs. A loop with $\mathrm{Out}(F_n)$4 hairs labeled cyclically by $\mathrm{Out}(F_n)$5 is sent to the class

$\mathrm{Out}(F_n)$6

where cyclic rotation corresponds to moving the basepoint around the loop and reversal corresponds to reflection of the loop orientation (Conant, 2016). This is the form in which the obstruction is most visibly tied to dihedral symmetry.

3. Dihedral coinvariants and the top layer of the Johnson cokernel

The decisive structural result is that the Enomoto–Satoh trace does not merely detect some ad hoc quotient: in the stable range it identifies the entire top layer of the Johnson cokernel with a dihedral coinvariant space. Let

$\mathrm{Out}(F_n)$7

act on $\mathrm{Out}(F_n)$8 by literal permutation of tensor factors:

• $\mathrm{Out}(F_n)$9 acts by cyclic rotation,
• $\Sigma_g$0 acts by reversal, with no sign inserted in the action (Conant, 2016).

Let $\Sigma_g$1 be the intersection of the kernels of all pairwise symplectic contractions, and let

$\Sigma_g$2

be the natural projection removing lower-weight pieces obtained by contractions. Then the addendum proves the main theorem

$\Sigma_g$3

establishing Conjecture 7.2 of Conant’s earlier paper (Conant, 2016).

This statement has several immediate consequences. First, the top layer $\Sigma_g$4 is completely accounted for by the Enomoto–Satoh obstruction: there are no additional top-level obstructions beyond those seen by the ES trace (Conant, 2016). Second, the topological problem of Johnson-surjectivity is converted into a representation-theoretic computation involving the restriction of symmetric-group representations to the embedded dihedral subgroup $\Sigma_g$5.

Using Schur–Weyl duality,

$\Sigma_g$6

and hence

$\Sigma_g$7

Therefore the multiplicity of $\Sigma_g$8 in the top layer is

$\Sigma_g$9

which in characteristic zero also equals the dimension of the corresponding invariant space by semisimplicity (Conant, 2016).

This is the precise sense in which the ES obstruction “detects all top-level partitions.” The surviving irreducibles are exactly those Specht modules with nontrivial dihedral coinvariants, and the trace realizes the corresponding $\Sigma_{g,1}$0-isotypic components explicitly (Conant, 2016).

4. Graphical and topological realizations

Conant’s 2013 construction introduced a graphical trace

$\Sigma_{g,1}$1

defined from the Conant–Kassabov–Vogtmann hairy graph trace by quotienting out relations that force vanishing on the Johnson image. Its rank-$\Sigma_{g,1}$2 part is precisely the Enomoto–Satoh trace (Conant, 2013). In that framework, the quotient $\Sigma_{g,1}$3 is identified with dihedral coinvariants of tensor powers: $\Sigma_{g,1}$4 for $\Sigma_{g,1}$5, with the reflection action twisted by the nontrivial $\Sigma_{g,1}$6-character when $\Sigma_{g,1}$7 is even (Conant, 2013).

The graphical description is not merely auxiliary. In the stable range,

$\Sigma_{g,1}$8

where $\Sigma_{g,1}$9 is the degree-$V = H_1(\Sigma_g;\mathbb{Q})$0 rank-$V = H_1(\Sigma_g;\mathbb{Q})$1 hairy Lie graph complex and $V = H_1(\Sigma_g;\mathbb{Q})$2 is the iterated contraction operator encoding iterated brackets of tripods (Conant, 2016). The crucial rank-$V = H_1(\Sigma_g;\mathbb{Q})$3 quotient

$V = H_1(\Sigma_g;\mathbb{Q})$4

arises because the nontrivial relations among loop graphs are exactly the slide relations corresponding to cyclic permutations and reversals (Conant, 2016).

A distinct but related topological interpretation appears in Kawazumi’s regular-homotopy version of the Goldman–Turaev Lie bialgebra. There, the Enomoto–Satoh traces are interpreted as part of a regular-homotopy Turaev cobracket $V = H_1(\Sigma_g;\mathbb{Q})$5. The classical Turaev cobracket is insensitive to monogon birth–death, but the regular-homotopy refinement retains the monogon action and rotation-number data. This refinement permits the definition of maps $V = H_1(\Sigma_g;\mathbb{Q})$6 and $V = H_1(\Sigma_g;\mathbb{Q})$7 whose graded version recovers the Enomoto–Satoh traces; in genus zero it identifies with the divergence cocycle in the Kashiwara–Vergne problem, up to a low-degree correction (Kawazumi, 2014).

Another topological realization is provided by the non-commutative Reidemeister–Turaev torsion of homology cylinders. In that setting, the degree-$V = H_1(\Sigma_g;\mathbb{Q})$8 leading term

$V = H_1(\Sigma_g;\mathbb{Q})$9

splits into a $H = H_1(\Sigma_{g,1};\mathbb{Q})$0-eigenspace and a $H = H_1(\Sigma_{g,1};\mathbb{Q})$1-eigenspace under reversal. The main theorem identifies the $H = H_1(\Sigma_{g,1};\mathbb{Q})$2-part with the Enomoto–Satoh trace: $H = H_1(\Sigma_{g,1};\mathbb{Q})$3 while the $H = H_1(\Sigma_{g,1};\mathbb{Q})$4-part is $H = H_1(\Sigma_{g,1};\mathbb{Q})$5 times the $H = H_1(\Sigma_{g,1};\mathbb{Q})$6-loop part of the LMO homomorphism (Nozaki et al., 2022). This places the ES obstruction and the $H = H_1(\Sigma_{g,1};\mathbb{Q})$7-loop LMO term inside a single $H = H_1(\Sigma_{g,1};\mathbb{Q})$8-valued torsion invariant.

5. Representation-theoretic content and low-degree behavior

The ES obstruction has an explicit representation-theoretic profile. In Conant’s 2013 formulation, there is an epimorphism

$H = H_1(\Sigma_{g,1};\mathbb{Q})$9

where $\pi_1$0 is the intersection of the kernels of all pairwise contractions (Conant, 2013). This yields concrete series of $\pi_1$1-representations in the Johnson cokernel.

Two classical families are singled out. The irreducible $\pi_1$2 occurs in $\pi_1$3 only when $\pi_1$4, with multiplicity $\pi_1$5; this is the Enomoto–Satoh series. The irreducible $\pi_1$6 occurs only when $\pi_1$7, again with multiplicity $\pi_1$8; this is Morita’s series (Conant, 2013). In contrast, $\pi_1$9 and $1$00 do not occur in the dihedral-coinvariant target (Conant, 2013).

The low-degree calculations in the addendum illustrate the mechanism. For $1$01,

$1$02

so $1$03. For $1$04,

$1$05

so $1$06 (Conant, 2016). The paper also states analogous expectations for $1$07 and $1$08, obtained by character computations on $1$09-classes (Conant, 2016).

From the graph-complex viewpoint of embedding calculus, the $1$10-loop target $1$11 is concentrated in cohomological degree $1$12 and identified with dihedral words in $1$13. Its stable $1$14-decomposition in weights $1$15 is given explicitly, and up to weight $1$16 the Johnson cokernel coincides with this $1$17-loop cohomology, hence with the image of the classical Enomoto–Satoh trace (Naef et al., 10 Feb 2026). This viewpoint makes the $1$18-loop nature of the obstruction literal: trees encode derivations, and gluing hairs to form one loop yields the classical ES trace.

6. Extensions, uniqueness, and limits of the classical obstruction

The Enomoto–Satoh trace is not only effective; within its formal class it is canonical. A 2025 uniqueness theorem proves that there is a unique, up to scalar multiple, degree-zero $1$19-cocycle on $1$20 with values in $1$21. In other words, among degree-zero traces to cyclic words on symplectic derivations, the Enomoto–Satoh trace is the only one up to scale (Baudat, 10 Nov 2025). The proof uses Hain’s theorem that the stable Lie algebra is generated by $1$22, together with the $1$23-irreducibility of $1$24 and Schur’s lemma (Baudat, 10 Nov 2025).

The same paper places ES alongside the non-commutative divergence cocycle on derivations of the free associative algebra. In the associative setting, degree-zero $1$25-cocycles on finite-degree quotients are linear combinations of $1$26 and its switch; in the symplectic Lie setting, ES is the unique degree-zero cocycle (Baudat, 10 Nov 2025). This comparison clarifies the role of ES as the symplectic Lie analogue of the associative divergence.

At the same time, the classical ES obstruction has a sharp limitation: it captures only the $1$27-loop part. Conant’s later loop decomposition and the 2025 study of the $1$28-loop Johnson cokernel show that the first degree in which the $1$29-loop trace fails to be injective is $1$30. In that degree,

$1$31

so there are stable $1$32-types invisible to the Enomoto–Satoh trace (Kuno et al., 26 Aug 2025). The refined $1$33-loop trace $1$34 remedies this: for sufficiently large genus,

$1$35

is injective, and $1$36 detects precisely the missing $1$37, $1$38, and $1$39 components (Kuno et al., 26 Aug 2025).

This shows that the classical Enomoto–Satoh obstruction is complete for the top layer and for the $1$40-loop part, but not for the entire Johnson cokernel in higher degree. The embedding-calculus formulation sharpens this point further: the Johnson image is the joint kernel of all higher Enomoto–Satoh traces $1$41, with the classical trace appearing as the $1$42 case

$1$43

on tree cocycles (Naef et al., 10 Feb 2026). A plausible implication is that “the Enomoto–Satoh obstruction” now names both a specific classical $1$44-loop obstruction and the first stage of a hierarchy of higher-loop obstructions.

7. Conceptual significance and related frameworks

The obstruction is significant because it translates a geometric non-surjectivity problem for the mapping class group into explicit representation theory. In the addendum’s formulation, the amount by which the Johnson homomorphism fails to be surjective in top degree is exactly measured by the dihedral coinvariants $1$45 (Conant, 2016). In Conant’s 2013 work, the rank-$1$46 trace is the first piece of a larger hierarchy of graphical obstructions, and the rank-$1$47 piece already detects modular-form-related families beyond the classical ES target (Conant, 2013).

A further conceptual generalization is the Hopf-algebraic and $1$48-cohomological framework of Conant–Kassabov. For a cocommutative Hopf algebra $1$49, they construct an $1$50-module $1$51; for $1$52, the generalized trace $1$53 on the Johnson cokernel projects to

$1$54

The $1$55 case recovers the Enomoto–Satoh trace, while higher $1$56 produce new obstruction families in top cohomology (Conant et al., 2015). This situates ES as the first instance of a larger cohomological theory rather than an isolated construction.

In summary, the Enomoto–Satoh obstruction is the canonical degree-zero cyclic-word trace obstructing surjectivity of the Johnson homomorphism, the $1$57-loop or rank-$1$58 piece of the Johnson cokernel, and—by the stable theorem of the 2016 addendum—the complete detector of all top-level partitions of that cokernel (Conant, 2016). Its later reformulations in graph complexes, regular-homotopy Lie bialgebras, non-commutative torsion, and uniqueness theory have clarified both its exact range and its limitations: it is exhaustive at the top layer, foundational in the $1$59-loop theory, and the starting point for higher-loop refinements of Johnson-cokernel obstruction theory (Kawazumi, 2014, Nozaki et al., 2022, Kuno et al., 26 Aug 2025, Naef et al., 10 Feb 2026).