On the 2-loop part of the Johnson cokernel
Yusuke Kuno, Masatoshi Sato
TL;DR
This work advances the understanding of the Johnson cokernel by isolating and explicitly presenting its 2-loop component via the hairy Lie graph complex. It provides a concrete presentation of the 2-loop space $\widetilde{\Omega}_2$ through generators $\Theta(t,u,v,w)$ and the generalized Conant third relation, and uses this to analyze the degree-6 Johnson cokernel, identifying specific $\mathrm{GL}$- and $\mathrm{Sp}$-modules that appear. The paper also clarifies how two different 2-loop trace constructions relate, showing that any 2-loop information captured beyond the Enomoto–Satoh trace is already detected by ES, up to a proved scalar relation between traces. Overall, the results connect graph-complex models of the Johnson cokernel with explicit module decompositions, deepening the link between low-dimensional topology, representation theory, and number-theoretic phenomena.
Abstract
We study stable Sp-decompositions of the cokernel of the Johnson homomorphism. Continuing the work of Conant in 2016, which identified the 1-loop part of the Johnson cokernel as the Enomoto-Satoh obstruction, we study the 2-loop part. Using the corresponding 2-loop trace map, we capture all the components of the Johnson cokernels in degree 6 that cannot be detected by the Enomoto-Satoh trace.
