Specifying composites of morphisms in a localisation of a triangulated category

62 Views Asked by At

I am reading the book Triangulated Categories by Neeman. I have come across a sentence and I'm not really sure what it is trying to say. For those with access to the book, it is Remark 2.1.23.

Let $\mathcal{D}$ be a triangulated category with triangulated subcategory $\mathcal{C}$. Denote by $\text{Mor}_{\mathcal{C}}$ the class of morphisms whose mapping cone is in $\mathcal{C}$. In the usual way we can form the Verdier localisation which I will denote by $\mathcal{D}/\mathcal{C}$ with a quotient functor $Q : \mathcal{D} \rightarrow \mathcal{D}/\mathcal{C}$. The morphisms are given by roofs (or spans) and the composite of two composable spans is given by taking the homotopy pullback of their summits.

A composite of spans then looks something like this (diagram taken from the text) http://tinypic.com/r/2up360i/9

where $P$ is the homotopy pullback of $f_{2}$ and $g_{1}$. Here $v'$, $f_{1}$, and $f_{2}$ are in $\text{Mor}_{\mathcal{C}}$, and $$Q(g_{2})Q(f_{2})^{-1} Q(g_{1})Q(f_{1})^{-1} = Q(g_{1})Q(u')Q(v')^{-1}Q(f_{1})^{-1}. $$ The text then remarks that,

In other words, to give two composable morphisms $X \rightarrow Y$ and $Y \rightarrow Z$ in $\mathcal{D}/\mathcal{C}$, and their composite, is nothing other than to give $X' \rightarrow Y'$, and $Y' \rightarrow Z'$ in $\mathcal{D}$ and maps $X' \rightarrow X$, $Y' \rightarrow Y$, and $Z' \rightarrow Z$ in $\text{Mor}_{\mathcal{C}}$ expressing the isomorphism of hte composable pair in $\mathcal{D}/\mathcal{C}$ with a composable pair in $\mathcal{D}$.

I have no idea what is meant by this remark. How does the argument to that point imply that? I am struggling to see why it is true, and struggling even more to see how it is simply a rephrasing of the equality above. How is a composite of morphisms in the quotient category determined that way, and moreover, what precisely is meant by the "expressing the isomorphism of the composable pair..."?

Any help is appreciated.

1

There are 1 best solutions below

0
On

Comparing Neeman's notation to the picture, $X',Y',Z'$ become $ P, W_2, Z$ while $X,Y,Z$ are the same in both places. The vertical morphisms in the picture are the isos Neeman describes; in the picture, the isomorphism $Z'\to Z$ is trivial, but of course it could be nontrivial and we'd still have a well defined morphism. In particular $g_2$ is determined as a map in $D/C$ by the other data, so superfluous.