I am solving the following exercise:

Let $\phi : G_1 \rightarrow G_2$ be a homomorphism (where $G_1$ and $G_2$ are groups) and $\ker \phi := \{ g \in G_1 \mid \phi(g) = e \}$

now I have to prove that

a) $\ker \phi$ is a subgroup of $G_1$,

b) $\phi$ is injective if and only $\ker \phi = \{ e \}$

My Problem: Until yet we have not really covered the topic of homomorphism between groups in our lectures. Anyhow I looked it up on wikipedia and found the definition for a subgroup as the following: $(U, \circ)$ is a subgroup of $(G, \circ)$ if $U$ is not an empty set. Therefore:

  • $a,b \in U \Rightarrow a \circ b \in U$
  • $a \in U \Rightarrow a^{-1} \in U$
  • $a,b \in U \Rightarrow a \circ b^{-1} \in U$

so I began to work with these definitions. I somehow managed to prove what I'm supposed to but I'm not sure if I did it the right way. I would be very thankful about some additional words to my attempt and also corrections. Thank you a lot in advance.

My Attempt:

a) $\ker \phi$ is a subgroup of $G_1$. So we can take two elements $x,y \in \ker \phi$ which are $x := \phi(g_1) = e $ and $y := \phi(g_2) = e $ and show that $x^{-1}$ and $x \circ y$ $\in$ $\ker \phi$.

Since $x\circ x^{-1} \in \ker \phi$ we can say: $x\circ x^{-1} = e \ \Leftrightarrow \ \overbrace{\phi(g_1)}^{= \ e} \circ x^{-1} = e \ \Rightarrow \ x^{-1} = e \ \Rightarrow \ x^{-1} \in \ker \phi$.

It must also be true that $x \circ y \in \ker \phi$ this is easily shown by: $x \circ y \ \Leftrightarrow \ \overbrace{\phi(g_1)}^{= \ e} \circ \overbrace{\phi(g_2)}^{= \ e} = e \ \Rightarrow \ x \circ y \in \ker\phi$

b) To show that $\phi$ is injective when $ \ker\phi = \{ e_{G_1} \}$ we must show that $ \ker\phi = \{ e_{G_1} \}$ has only one fiber which then has to be $\phi^{-1}(e_{G_2})$.

So we can take two elements $g_1,g_2 \in G_1$ and if $\phi(g_1) = \phi(g_2) = e \ \Rightarrow \ g_1 = g_2$ we can state that $\phi$ with $\ker \phi = \{e\}$ is injective.

  • $\begingroup$ I think your proof that $x^{-1}\in\ker\phi$ is wrong. If $x^{-1}\in\ker\phi\subseteq G_1$ and $\phi(g_1)\in G_2$, then I don't think you can necessarily apply a common group operation. Also, you seem to miss the "only if" for part b. $\endgroup$ – Robin Goodfellow Oct 14 '14 at 10:31
  • $\begingroup$ Hey Robin could you be a little more precise or in detail because I don't understand what you mean. I mean since $e$ is the only element in our kernel the inverse $e$ again must also be in it? Or do I misunderstand something $\endgroup$ – Mainviel Oct 14 '14 at 10:33
  • $\begingroup$ There is definitely some confusion about which group things live in - for example, in your part a), you take $x\in\ker{\phi}$ such that $x=\phi(g_1)$ - but $\ker{\phi}\subseteq G_1$, and $\phi(g_1)\in G_2$ for any $g_1\in G_1$, so this is impossible. What you want is to notice that because $x\in\ker{\phi}$, then $\phi(x)=e_{G_2}$. For similar reasons, you can't write $\phi(g_1)\circ x^{-1}$, as the two elements live in different groups. $\endgroup$ – mdp Oct 14 '14 at 10:53
  • $\begingroup$ alright I understand my mistake, so if I would write $\phi^{-1}(g_1)$ instead of $\phi(g_1)$ my formulation would be correct? $\endgroup$ – Mainviel Oct 14 '14 at 10:58
  • $\begingroup$ @Mainviel Don't do that! If $\phi$ isn't injective, which might be the case at that stage, then $\phi^{-1}(g_1)$ is a set with more than one element, which makes things even worse! At the moment you haven't written the correct definition of $x\in\ker{\phi}$; if you fix this, the problem of elements living in the wrong group should go away. $\endgroup$ – mdp Oct 14 '14 at 11:01

Some notes, which hopefully should help you fix things.

1) If $x,y\in\ker{\phi}$, then $x,y\in G_1$, so $x$ and $y$ cannot be in the image of $\phi$, which is a subset of $G_2$. What you know is that $\phi(x)=\phi(y)=e_{G_2}$; so what is $\phi(x\circ y)$? Don't forget to show that the kernel is non-empty - it contains $e_{G_1}$.

2) $\phi(g_1)\circ x$ is undefined, because $\phi(g_1)$ is an element of $G_2$, whereas $x$ is an element of $G_1$ - always be careful not to do this.

3) For the second statement ("$\ker{\phi}=\{e_{G_1}\}$ if and only if $\phi$ is injective") it might help you to do each direction separately. Firstly, if $\phi$ is injective, and $\phi(g)=e_{G_2}$ then what must we be able to say about $g$? (Hint: what is $\phi(e_{G_1})$?). Secondly, assume $\ker{\phi}=\{e_{G_1}\}$; if $\phi(g_1)=\phi(g_2)$, what is $\phi(g_1g_2^{-1})$?

Please comment if anything is unclear!

  • $\begingroup$ Yes, that's much better! You should also check that $\phi(x^{-1})=e_{G_2}$ if $\phi(x)=e_{G_2}$ (and that $\ker{\phi}$ isn't empty, but this should be easier!). $\endgroup$ – mdp Oct 14 '14 at 11:30
  • $\begingroup$ For 1) $\phi(x \circ y) = e_{G_2}$ since $\phi$ is an homomorphism we can say: $\phi(x \circ y) \Leftrightarrow \overbrace{\phi(x)}^{= \ e_{G_2}} \circ \overbrace{\phi(y)}^{= \ e_{G_2}} = e_{G_2}$ For 3) 1st: $\phi$ injective $\phi(g) = e_{G_2}$ now we can say that $g = e_{G_1}$ since $\phi(e_{G_1}) = e_{G_2} = \phi(g)$ 2nd: Since $\phi$ is a homomorphism and $\phi(g_1) = \phi(g_2)$ so we know $g_1^{-1} = g_2^{-1} \Rightarrow \phi(g_1 \circ g_1^{-1} = \phi(e_{G_1}) = e_{G_2}$ is that somehow true? $\endgroup$ – Mainviel Oct 14 '14 at 11:32
  • $\begingroup$ So for $x^{-1} \in \ker\phi$ we can say $x \circ x^{-1} = e_{G_1}$ now we can apply $\phi$ so we get: $\phi(x \circ x{-1}) = \overbrace{\phi(e_{G_1})}^{= \ e_{G_2}} \Leftrightarrow \overbrace{\phi(x)}^{= \ e_{G_2}} \circ \phi(x^{-1}) = e_{G_2} \Rightarrow \phi(x^{-1})= e_{G_2}$. If the kernel was empty we would not have an injective function and therefore not an homomorphism. Or how do I prove this? Thank you a lot for your help! $\endgroup$ – Mainviel Oct 14 '14 at 11:52
  • $\begingroup$ The proof for the inverse is correct. Homomorphisms don't have to be injective, so that wasn't the right thing to say - the kernel is non-empty because it contains $e_{G_1}$; you even write this down in the proof that $\phi(x^{-1})=e_{G_2}$! $\endgroup$ – mdp Oct 14 '14 at 12:07
  • $\begingroup$ well then the proof is 'only' saying: that $e_{G_1}$ is in the kernel ? That somehow don't look like a proof to me. I mean the fact is given in the task setting. Thank you alot! $\endgroup$ – Mainviel Oct 14 '14 at 12:16

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.