Additivization of functors in an abelian monoidal category

68 Views Asked by At

Crossposted on MathOverflow here.


I'm having trouble with the proof of Lemma 2.9 in "Cohomology of Monoids in Monoidal Categories" by Baues, Jibladze, and Tonks, and I was wondering if someone could clarify a detail. I'll try to summarize the context of the lemma.

Context

Let $(\Bbb A,\circ,I)$ be an abelian monoidal category where $\circ$ is left distributive, i.e. the natural transformation $$(X_1\circ Y)\oplus(X_2\circ Y)\rightarrow (X_1\oplus X_2)\circ Y$$ is an isomorphism. For example, $\Bbb A$ could be the category of linear operads (this is a motivating example of the article). Given an endofunctor $F$ of $\Bbb A$, we define its cross-effect $$F(A|B):=\ker(F(A\oplus B)\rightarrow F(A)\oplus F(B)).$$ The additivization of $F$ is then the functor $F^\text{add}$ defined by $$F^\text{add}(A):=\text{coker}\left(F(A|A)\rightarrow F(A\oplus A)\xrightarrow{F(+)}F(A)\right).$$ The idea is that $F^\text{add}$ is the additive part of $F$.

Let $(M,\mu,\eta)$ be an internal monoid in $\Bbb A$, and let $L_0$ be the endofunctor $L_0:A\mapsto M\circ(M\oplus A)$. Let $L:=L_0^\text{add}$ be the additivization of $L_0$. (In the case of operads, represented as planar trees, I see $L(A)$ as the space of trees whose nodes are all labeled by elements of $M$ except for one leaf, which is labeled by an element of $A$.)

Suppose now that $\Bbb A$ is right compatible with cokernels, i.e. that

for each $A\in\Bbb A$, the additive functor $A\circ-:\Bbb A\rightarrow\Bbb A$ given by $B\mapsto A\circ B$ preserves cokernels.

Then, in the proof of Lemma 2.9, the authors claim the following:

By the assumption that $\Bbb A$ is right compatible with cokernels it follows that $L(L(X))$ is the additivisation of $L_0(L_0(X))$ in $X$ [...].

Remarks

If anyone could provide an explanation of the last claim, I would be very grateful. However, my inability to understand how to show this might be related to two other issues I have:

1) Elsewhere in the literature, cross-effects are only defined when $F$ is reduced, i.e. $F(0)=0$ (e.g. here, section 2). But we can always reduce a functor by taking the cokernel of $F(0)\rightarrow F(X)$, so I don't think it's much of a problem.

2) In the first quote, the authors state that $A\circ -$ is additive, which is quite the opposite of the initial hypothesis that $\circ$ be left distributive, and not necessarily right distributive. How to resolve this apparent conflict?