# Example of ring “quasi-homomorphism” not conserving multiplicative identity?

Example of ring "quasi-homomorphism" not conserving multiplicative identity?

I'm using a definition of homomorphism that requires f(1)=1 and I'm trying understand the reasons and consequence of this point.

I mean: if we're connecting two rings with identity, does this not force f(1)=1?

-
– Martin Jan 17 '13 at 8:57

Take any idempotent $e\notin\{0,1\}$ in any ring $R$.

The inclusion of $eRe\rightarrow R$ does not preserve identity.

To add some details, it is easy to check that $eRe$ is a subring of $R$ with identity $e$. The inclusion $x\mapsto x$ would not send $e$ to 1 in $R$, and so it does not preserve identity. There is no need for $e$ to be central.

-
You should ask for $e$ to be central, or else this need not be a map of rings. – David Speyer Jan 17 '13 at 14:16
@DavidSpeyer No, that's superfluous. It is an inclusion map of a subring into a ring... I don't see why centrality of $e$ matters to map $ere\mapsto ere$. – rschwieb Jan 17 '13 at 14:57
I'm guessing @David was thinking of the projection $R \to eRe : x \mapsto exe$. – Hurkyl Jan 17 '13 at 15:16
My bad, you're right. – David Speyer Jan 17 '13 at 23:04

You can have multiplicative and additive maps between rings that do not preserve identity. For instance $i: \mathbb Z \rightarrow \mathbb Z \times \mathbb Z$ given by $i(n)=(0,n)$. One might call such a map a rng homomorphism.

-

consider $f:\mathbb Z \to \mathbb Z$ given by $f(a)=2\cdot a$.

-
This is not a multiplicative map, because $4=f(1)f(1)\neq f(1)$. – JSchlather Jan 17 '13 at 8:50
I apologize (going to sleep now). – Ittay Weiss Jan 17 '13 at 8:53