Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I encountered the following exact sequence a while ago, and wondered if it was a special case of the Snake lemma. It looks like it would be, but I don't quite see how...

The context is that $A,B$ are operators on a Hilbert space $H$ (but it holds in greater generality):

$$\DeclareMathOperator{\Im}{Im} 0 \rightarrow \ker(B) \rightarrow \ker(AB) \xrightarrow{B} \ker(A) \rightarrow H/\Im(B) \xrightarrow{A} H/\Im(AB) \rightarrow H/\Im(A) \rightarrow 0$$

share|cite|improve this question
up vote 2 down vote accepted

\begin{matrix} 0\to &H &\stackrel{(1,B)}{\longrightarrow} &H\oplus H &\stackrel{B\pi_1-\pi_2}{\longrightarrow} &H &\to0\\ &\downarrow\rlap{\small B} & &\downarrow\rlap{\small(AB,1)} & &\downarrow\rlap{\small A} \\ 0\to&H &\stackrel{(A,1)}{\longrightarrow} &H\oplus H &\stackrel{\pi_1-A\pi_2}{\longrightarrow} &H&\to0 \end{matrix} This should work.

share|cite|improve this answer

In general, let $\alpha: L \rightarrow M$ and $\beta: M \rightarrow N$ be morphisms in an abelian category. Then

$\begin{matrix} &L &\stackrel{\alpha}{\longrightarrow} &M &\stackrel{}{\longrightarrow} &\operatorname{coker} \alpha &\to0\\ &\downarrow\rlap{\beta \alpha} & &\downarrow\rlap{\beta} & &\downarrow\rlap{} \\ 0\to&N &\stackrel{\text{id}}{\longrightarrow} &N &\stackrel{}{\longrightarrow} &0 \end{matrix}$

yields an exact sequence $$\ker \beta \alpha \rightarrow \ker \beta \rightarrow \operatorname{coker} \alpha \rightarrow \operatorname{coker} \beta\alpha \rightarrow \operatorname{coker} \beta \rightarrow 0$$


$\begin{matrix} &0 &\stackrel{}{\longrightarrow} &L &\stackrel{\text{id}}{\longrightarrow} &L &\to0\\ &\downarrow\rlap{} & &\downarrow\rlap{\alpha} & &\downarrow\rlap{\beta \alpha} \\ 0\to& \ker \beta &\stackrel{}{\longrightarrow} &M &\stackrel{\beta}{\longrightarrow} &N \end{matrix}$

yields an exact sequence

$$0 \rightarrow \ker \alpha \rightarrow \ker \beta \alpha \rightarrow \ker \beta \rightarrow \operatorname{coker} \alpha \rightarrow \operatorname{coker} \beta\alpha \, .$$

It is not hard to check that the two sequences can be combined to give $$0 \rightarrow \ker \alpha \rightarrow \ker \beta \alpha \rightarrow \ker \beta \rightarrow \operatorname{coker} \alpha \rightarrow \operatorname{coker} \beta\alpha \rightarrow \operatorname{coker} \beta \rightarrow 0 \, .$$

share|cite|improve this answer
You don't need to apply the snake lemma twice. People rarely mention this, but it is true in general that the kernel of the first morphism in the ker-coker sequence is isomorphic to the kernel of the first morphism in the first exact sequence, so you can get the start $0 \to \ker{\alpha} \to \ker{\beta\alpha}$ from that. – Martin Jul 6 '13 at 9:26
@Martin: Sure, and dually, it is also true in general that $\operatorname{coker} \beta$ is a cokernel of $\operatorname{coker} \alpha \rightarrow \operatorname{coker} \beta \alpha$. I just wanted to mention that there are two diagrams for which the snake lemma gives at once almost all of the desired long exact sequence. – Cihan Jul 6 '13 at 10:53
I see. I guess I didn't make myself very clear in my first comment. What I was trying to say is this: In the snake sequence of $$ \begin{matrix} &A &\stackrel{f}{\longrightarrow} &B &\stackrel{g}{\longrightarrow} &C &\to0\\ &\downarrow\rlap{a} & &\downarrow\rlap{b} & &\downarrow\rlap{c} \\ 0\to&A' &\stackrel{f'}{\longrightarrow} &B' &\stackrel{g'}{\longrightarrow} &C' \end{matrix} $$ one has that $\ker{f}$ is isomorphic to the kernel of the map $\ker{a} \to \ker{b}$ (and dually $\operatorname{coker}{g'}$ is isomorphic to the cokernel of $\operatorname{coker}b \to \operatorname{coker}c$). – Martin Jul 6 '13 at 11:34
Oh now I see. In fact I didn't realize this situation in the general setup of the snake lemma before. Thanks! – Cihan Jul 6 '13 at 11:44

I think this should work: $$ \begin{matrix} &H &\stackrel{(-B,1)}{\longrightarrow} &H\oplus H &\stackrel{\pi_1+B\pi_2}{\longrightarrow} &H &\to0\\ &\downarrow\rlap{\small B} & &\downarrow\rlap{\small(AB,-AB)} & &\downarrow\rlap{\small A} \\ 0\to&H &\stackrel{(-AB,-A)}{\longrightarrow} &H\oplus H &\stackrel{}{\longrightarrow} &H \end{matrix} $$

Oops, after writing all down, it doesn't look that good anymore ... May have to sleep over it

share|cite|improve this answer
What is the map on the bottom right? – i707107 Jul 4 '13 at 16:23
Bottom left map should be injective, but it is not in yours. – i707107 Jul 4 '13 at 16:54
Bottom right could be $\pi_1-B\pi_2$. But then the right square is no more commutative. And indeed the bottom left is not injective without certain conditions. But there are verisimilarly some modifications which could solve the problem. – awllower Jul 5 '13 at 5:07

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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