Equivalence between lax algebras for parametrised endofunctors and iterated monads

Let (\cM,×,1)(\cM,\times,1) be a cartesian monoidal category, let (\cC,)(\cC,\blacktriangleright) be an \cM\cM-actegory, and let FF be a strong endofunctor on \cC\cC such that FF and \cC\cC satisfy the hypotheses of the stated Kelly theorem. For a lax algebra (X,(P,f))(X,(P,f)) for Para(F)\mathsf{Para}(F), let PP carry the comonoid structure

!P:P1,ΔP:PP×P.!_P:P\to 1,\qquad \Delta_P:P\to P\times P.

Equivalence conjecture. The assignment sending (X,(P,f))(X,(P,f)) to the corresponding lax algebra for FκF^\kappa,

(X,(P,f))(X,((P,!P,ΔP),laxlim((P,f)Para(F)(P,f)Para(F)α(P,f)):FκXX)),(X,(P,f))\longmapsto (X,((P,!_P,\Delta_P),\underrightarrow{\mathsf{laxlim}}((P,f)\circ\mathsf{Para}(F)(P,f)\circ\cdots\circ\mathsf{Para}(F)^\alpha(P,f)):F^\kappa X\to X)),

defines an equivalence of categories

Lax-AlgEndo(Para(F))Lax-AlgMnd(Fκ).\mathsf{Lax}\textbf{-}\mathsf{Alg}_{\mathsf{Endo}}(\mathsf{Para}(F))\longrightarrow \mathsf{Lax}\textbf{-}\mathsf{Alg}_{\mathsf{Mnd}}(F^\kappa).

The claim identifies lax algebras for parametrised endofunctors with lax algebras for the associated iterated monad, using the induced comonoid of parameters. Its resolution depends on the hypotheses invoked from the Kelly theorem and the construction of the displayed lax limit; the supplied text gives no status evidence beyond the statement itself.

Sources & referencesView supporting material

Primary source

Bruno Gavranović, Paul Lessard, Andrew Dudzik, Tamara von Glehn, João G. M. Araújo and Petar Veličković, “Position: Categorical Deep Learning is an Algebraic Theory of All Architectures”, arXiv:2402.15332 (2024).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.