← All problems

Functoriality of the Becker–Gottlieb transfer

Given composable fibrations F→gE→fBF\xrightarrow{g}E\xrightarrow{f}B with homotopy-finite, or more generally finitely dominated, fibers, does the Becker–Gottlieb transfer satisfy tr⁡f∘g≃tr⁡g∘tr⁡f:Σ∞B+→Σ∞F+\operatorname{tr}_{f\circ g}\simeq \operatorname{tr}_g\circ\operatorname{tr}_f:\Sigma^\infty B_+\to\Sigma^\infty F_+? In particular, does this equality hold after rationalization, equivalently on rational cohomology?

Equivalent formulations
  1. For every composable pair of fibrations F→gE→fBF\xrightarrow{g}E\xrightarrow{f}B in the stated class, do the induced maps on rational cohomology satisfy (tr⁡f∘g)∗=(tr⁡f)∗∘(tr⁡g)∗:H∗(F;Q)→H∗(B;Q)(\operatorname{tr}_{f\circ g})^*=(\operatorname{tr}_f)^*\circ(\operatorname{tr}_g)^*:H^*(F;\mathbb{Q})\to H^*(B;\mathbb{Q})?

    source: Naef, Stoll, The Becker–Gottlieb transfer is not functorial: An example using $C_\infty$-algebras

References

Progress summary

Refreshed
Open · solution reported

Two new preprints claim explicit counterexamples, so functoriality appears false, but the result has not been independently verified.

The problem asks whether Becker–Gottlieb transfers compose functorially for suitable composable fibrations, including after rationalization. Earlier positive claims were weakened or left with a naturality gap.

Known results

  • Haugseng’s 2013 functoriality proof was withdrawn after an error in Lemma 4.114.11.
  • Klein–Malkiewich (2018) claimed functoriality up to homotopy for fibrations with finitely dominated fibers; a 2022 corrigendum restricted this to partial cases.
  • A 2019 result established functoriality in the stable homotopy category for homotopy-finite fibers, while leaving the canonical natural homotopy unresolved.

October 13, 2026 counterexamples

Lehner, Malkiewich, and Naef, and concurrently Naef and Stoll, report explicit counterexamples. The latter constructs, for every even n≥4n\ge 4, a fibration with fiber S2∨SnS^2\vee S^n and an oriented S1S^1-bundle whose composite transfer is nontrivial on rational cohomology although one factor transfer vanishes. Thus the claimed functoriality fails; both reports are primary preprints and remain unverified.

Current status (as of October 2026): two concurrent preprints claim to settle the question negatively, but their counterexamples lack independent verification, so the resolution remains formally unverified.

Sources

Solutions 0

No solutions have been posted yet.