I've tried looking around for an explanation to this problem, but I've been having trouble finding a clear solution that specifically focuses on this question:

How many homomorphisms are there of $\Bbb Z \times \Bbb Z\times \Bbb Z$ into $\Bbb Z$?

I'm studying abstract algebra and we have just begun discussing rings. Thank you in advance for your help. I would appreciate it if you could give me step by step help for this problem. This is not for homework- I just need to figure out how to do this problem to prepare for my exam. Thanks!

Edit: Here is a post that addressed the question, but did not answer it fully: Describe all ring homomorphisms

  • 4
    $\begingroup$ Are you regarding $\mathbb{Z}^3$ as a module, or as a ring? $\endgroup$ – goblin Apr 5 '16 at 3:24
  • $\begingroup$ See the above edit- the question comes from a series of questions about ring homomorphisms, but this part of the question is left kind of ambiguous. I guess that's part of my question, too. $\endgroup$ – asoiaf_obsession_91 Apr 5 '16 at 3:32
  • 1
    $\begingroup$ A ring, not a module. $\endgroup$ – asoiaf_obsession_91 Apr 5 '16 at 3:37
  • $\begingroup$ One should never forget that a $\Bbb Z$-linear map $f:\Bbb Z^n\to\Bbb Z^m$ is given by a matrix of size $m\times n$ with integer coordinates, and uniquely so. $\endgroup$ – Pedro Tamaroff Apr 5 '16 at 4:00

Assuming the question is about ring homomorphisms, there are $3$. Here is how to see this very explicitly.

Write $e_1 = (1, 0, 0), e_2 = (0, 1, 0), e_3 = (0, 0, 1)$. These three elements form a complete set of orthogonal idempotents. This means that they satisfy $e_i e_j = \delta_{ij} e_j$ and $\sum e_i = 1$, or more explicitly

$$e_i^2 = e_i, e_i e_j = 0 \text{ if } i \neq j, e_1 + e_2 + e_3 = 1.$$

The significance of these conditions for us is that they must be preserved by ring homomorphisms. Hence if $f : \mathbb{Z}^3 \to \mathbb{Z}$ is a ring homomorphism, $f(e_i)$ is a complete set of orthogonal idempotents in $\mathbb{Z}$. It's clear that $f$ is determined by the $f(e_i)$.

But $\mathbb{Z}$ only has two idempotents, namely $0$ and $1$. The condition that $\sum f(e_i) = 1$ means exactly one of the $f(e_i)$ is equal to $1$, and then orthogonality is automatic. Hence there are $3$ homomorphisms corresponding to whether $f(e_1), f(e_2)$, or $f(e_3)$ is $1$; these are the $3$ projection maps $\mathbb{Z}^3 \to \mathbb{Z}$.

More generally, homomorphisms $\mathbb{Z}^n \to \mathbb{Z}^m$ correspond to functions $m \to n$.

The algebraic significance of complete sets of orthogonal idempotents in a commutative ring $R$ is that they correspond to decompositions of $R$ as a finite product $\prod_i R e_i$ of rings. The geometric significance is that they correspond to decompositions of $\text{Spec } R$ into a finite disjoint union $\coprod_i \text{Spec } R e_i$ of affine schemes. Here $\text{Spec } \mathbb{Z}^n$ decomposes into $n$ copies of $\text{Spec } \mathbb{Z}$.

  • $\begingroup$ Good answer! $\;$ $\endgroup$ – goblin Apr 5 '16 at 3:44
  • $\begingroup$ Thank you! I think this is what I was looking for. $\endgroup$ – asoiaf_obsession_91 Apr 5 '16 at 3:54
  • $\begingroup$ The condition that $\sum f(e_i)=1$ alone actually tells you exactly one of the $f(e_i)$ is $1$. $\endgroup$ – Eric Wofsey Apr 5 '16 at 3:57
  • $\begingroup$ @Eric: whoops. Yes, of course. $\endgroup$ – Qiaochu Yuan Apr 5 '16 at 5:09
  • $\begingroup$ I don't quite understand the $Re_i$ thing in the last paragraph. Does this just mean $\prod_{i \in I} R$? $\endgroup$ – goblin Apr 5 '16 at 13:11

(I'm assuming that the question is about module homomorphisms, not ring homomorphisms.)

Adam is right. In more detail:


Suppose we're given

  • a commutative ring with unity, call it $R$,
  • an $R$-module, call it $X$,
  • a finite set $S$.

Then there's a natural bijective correspondence between:

  • homomorphisms $X \leftarrow R^S$ and
  • functions $X \leftarrow S.$

given as follows:

Firstly, assign to each $s \in S$ a vector $e_s \in R^S$ writing by $e_s(s') = [s=s']$, where $[]$ is the Iverson bracket.

$(\Rightarrow)$ Given a homomorphism $h : X \leftarrow R^S$, we get a function $F(h) : X \leftarrow S$ defined as follows: $$F(h)(s) = h(e_s).$$

$(\Leftarrow)$ Given a function $f : X \leftarrow S$, we get a homomorphism $H(f) : X\leftarrow R^S$ by writing: $$H(f)(a) = \sum_{s \in S} a(s)f(s)$$

(The above sum makes sense because $S$ is finite.)

To prove this:

Step 0. Check that $H(f)$ really is a homomorphism.

Step 1. Check that these processes are inverse to each other. That is:

$$H(F(h)) = h, \qquad F(H(f)) = f$$

Step 2. (Optional). If you know some category theory, you can also go ahead and try to prove that $H$ and $F$ are natural transformations.

  • 1
    $\begingroup$ Although there are moral reasons to write down arrows in functions the way you do, there are social reasons one would like to avoid it. Is it really a notation that carries a significant advantage? $\endgroup$ – Pedro Tamaroff Apr 5 '16 at 4:00
  • $\begingroup$ @PedroTamaroff, I really haven't decided yet. Sometimes I flip back and forth between them. The thing that made me use the backward $R \leftarrow S$ here is the agreement with the $R^S$ notation. But anyway, I don't really care about the social reasons, or tradition. As a general rule, I think we take these kinds of things altogether too seriously. What I want is to settle on a truly optimal system of conventions to hand to the next generation. $\endgroup$ – goblin Apr 5 '16 at 13:08

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.