3
$\begingroup$

Let $\mathscr{T}$ be atriangulated category.

The third axiom for triangulated categories, namely,

if in the diagram

$$\begin{array} 0X &\stackrel{u}{\longrightarrow}&Y&\stackrel{v}{\longrightarrow}&Z&\stackrel{w}{\longrightarrow}&\Sigma X\\ \downarrow{f}&&\downarrow{g}&&\downarrow{\exists h}&&\downarrow{\Sigma f}\\ X'&\stackrel{u'}{\longrightarrow}&Y'&\stackrel{v'}{\longrightarrow}&Z'&\stackrel{w'}{\longrightarrow}&\Sigma X'\\ \end{array} $$

both rows are distinguished triangles and the left square is commutative, then there is a (not necessarily unique) map $h$ such that all the squares commute,

has the following

$\mathbf{Corollary:}$

If $f$ and $g$ are isomorphisms, so is $h$.

Now my question is:

Is there a way to prove this corollary, without using Yoneda's lemma?

Thanks for the help.

$\endgroup$
2
  • 10
    $\begingroup$ May I ask if there is a particular reason why one should like to prove things without using the Yoneda Lemma? $\endgroup$
    – Rasmus
    Jun 20, 2014 at 14:34
  • $\begingroup$ I did not want to involve Grothendieck universes, and was afraid to do so in some cases, but was note quite sure... $\endgroup$ Jun 21, 2014 at 1:39

1 Answer 1

10
$\begingroup$

Yes, it is possible (compare Balmer, Triangular Witt groups. Part I: The 12-term localization exact sequence, page 5).

One easily reduces your question to the following situation: \begin{array} 0 X &\stackrel{u}{\longrightarrow} & Y & \stackrel{v}{\longrightarrow} & Z & \stackrel{w}{\longrightarrow} & \Sigma X \\ \downarrow{1} && \downarrow{1} && \downarrow{h} && \downarrow{1} \\ X&\stackrel{u}{\longrightarrow} & Y & \stackrel{v}{\longrightarrow} & Z & \stackrel{w}{\longrightarrow} & \Sigma X\\ \end{array} and the claim is that $h$ is an isomorphism.

Two observations:

  1. If in an endomorphism $(f,g,h)$ of a distinguished triangle $f$ and $g$ are zero then $h^2 = 0$.

    Proof. Since $hv = 0$, we can write $h = kw$ and thus $h^2 = kwh = 0$.

  2. In the above diagram, $\varepsilon = h-1$ satisfies $\varepsilon^2 = 0$ by point 1. Thus $h = 1+\varepsilon$ has the inverse $1-\varepsilon$.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.