I'm working on a problem from Dummit and Foote's Abstract Algebra, 3rd Edition, reproduced below.

Suppose that this is a commutative diagram of groups and that the rows are exact.

enter image description here

Prove that if $\phi, \alpha,$ and $\gamma$ are surjective, then $\beta$ is surjective.

Here's what I've got so far:

If $b'\in B'$ then the surjectivity of $\gamma$ implies $\exists$ $ c\in C$ such that $\gamma(c)=\varphi'(b')$.

The surjectivity of $\varphi$ implies $\exists$ $ b\in B$ such that $\varphi(b)=c$.

Thus, $\varphi'(b')=\gamma(\varphi(b))$ and since the diagram commutes $\gamma(\varphi(b))=\varphi'(\beta(b))$.

This gives us that $\varphi'(b')=\varphi'(\beta(b))$

But we don't have that $\varphi'$ is injective, so we don't know that $b'=\beta(b)$, which would give us surjectivity of $\beta$.

Secondly, if $b'\in Ker(\varphi')$ then since the rows are exact we know that for some $a'\in A'$ it is true that $\psi'(a)=b'$.

Also, since $\alpha$ is surjective we have that $\exists$ $a\in A$ such that $ \alpha(a)=a'$.

Thus, $\psi'(\alpha(a))=b'$, and since the diagram commute we also have that $\beta(\psi(a))=\psi'(\alpha(a))=b'$.

Thus, we at least know that $\beta$ maps to the entire kernel of $\varphi'$.

But what about when $b'\notin Ker(\varphi')$?

Hope you guys can help me, because I really don't know what I'm missing.


I'll assume that these groups are all additive Abelian groups. If they aren't then you will have to amend accordingly

You have $\phi'(b')=\phi'(\beta(b))$. Rather than split cases according to whether $b'$ is in the kernel, instead note that $\phi'(b'-\beta(b))\in\text{Ker}(\phi')$ so $b'-\beta(b)=\psi'(a')$ for some $a'\in A'$. Now pick up the diagram chase again.

(By the way I do find the $\ni$ notation very hard to read.)

  • $\begingroup$ Edited the question to remove the $\ni$s, and I found the answer. Am I supposed to edit the question to have the answer I found, or just leave it as is? I'm new here and don't really know how etiquette works. $\endgroup$ – Dieterlan May 14 '17 at 3:56

Write $b'=\beta(b)\tilde{b}$ for some $\tilde{b}\in B'$ (namely $\tilde{b}=b'(\beta(b))^{-1}$). Note if $b'=\beta(b)$, then $\tilde{b}$ is just the identity. Now you know

$$\varphi'(b')=\varphi'(\beta(b)\tilde{b})=\varphi'(\beta(b))\varphi'(b) = \varphi'(b')\varphi'(\tilde{b}) \Longrightarrow 1=\varphi'(\tilde{b})$$

so $\tilde{b}\in \ker(\varphi')$. Now use the fact which you proved that $\beta$ maps to the entire kernel of $\varphi'$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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