3

Is there an easy way to prove that any surjective homomorphism $\mathbb{Z} \oplus \mathbb{Z} \rightarrow \mathbb{Z} \oplus \mathbb{Z}$ is an isomorphism?

I was told that this holds more generally for any direct sum of $\mathbb{Z}$. This is easy in the case of $\mathbb{Z} \rightarrow \mathbb{Z}$ since any homomorphism has $1 \mapsto n$ for some $n$.

Tuo
  • 4,882

2 Answers2

5

Given any Noetherian module $M$, a surjection $f:M\to M$ is always an isomorphism. To see why, consider $$ ker(f)\subset ker(f^2)\dots ker(f^n)\subset\dots $$ As $M$ is Noetherian, this chain stabilizes at, say $ker(f^k)$. Let us examine $ker(f^k)\cap im(f^k)$. Assume $x$ is in this intersection; then $f^k(x)=0$, and we can write $f^k(y)=x$ for some $y\in M$. Then $f^k(f^k(y))=f^k(x)=0$, so $y\in ker(f^{2k})=ker(f^k)$. Thus $x$ is in fact zero.

But now $f$ is surjective, so $f^k$ is also surjective; hence $ker(f^k)\cap im(f^k)=ker(f^k)=\{0\}$, showing that $f^k$ (and hence $f$ as well) is injective.

Now you just have to show that $\mathbb{Z}\oplus\mathbb{Z}$ is Noetherian!

TomGrubb
  • 13,177
1

I'm not sure what your background is so I'll offer a most elementary way to reason it specific to $\mathbb{Z}$, although as ThomasGrubb explained there are far more general results (see Nakayama's Lemma).

If $1 \oplus 0 \rightarrow a \oplus b$ and $0 \oplus 1 \rightarrow c \oplus d$ then the homomorphism is completely defined by $x \oplus y \rightarrow (ax + cy) \oplus (bx + dy)$

In matrix notation,

$\begin{bmatrix} x \\ y \end{bmatrix} \rightarrow \begin{bmatrix} a & c \\ b & d \end{bmatrix} \begin{bmatrix} x \\ y \end{bmatrix} $

Clearly the matrix is invertible iff our original homomorphism is.

The problem thus boils down to showing that a ($2 \times 2$) integer matrix has full column rank iff it is invertible. This is easy, because if it has full column rank then we can find $x_1,y_1$ such that $$\begin{bmatrix} 1 \\ 0 \end{bmatrix} = \begin{bmatrix} a & c \\ b & d \end{bmatrix} \begin{bmatrix} x_1 \\ y_1 \end{bmatrix} $$ and similarly $x_2, y_2$ such that $$\begin{bmatrix} 0 \\ 1 \end{bmatrix} = \begin{bmatrix} a & c \\ b & d \end{bmatrix} \begin{bmatrix} x_2 \\ y_2 \end{bmatrix} $$

Thus $$\begin{bmatrix} 1 & 0 \\ 0 & 1 \end{bmatrix} = \begin{bmatrix} a & c \\ b & d \end{bmatrix} \begin{bmatrix} x_1 & x_2 \\ x_2 & y_2 \end{bmatrix} $$

The above argument immediately generalizes to $\bigoplus_n \mathbb{Z}$.

Badam Baplan
  • 9,123
  • You show that the integer matrix has a right inverse, does it imply that it also has a left inverse? This is true for matrices over a field, is it also true in this case? – ImHackingXD Jun 08 '24 at 21:40
  • 1
    Yes for square matrices over commutative rings, a right inverse is automatically also a left inverse. I think the best way to see it is with the universal formula for the adjoint. See also here. https://math.stackexchange.com/questions/298791/right-invertible-and-left-zero-divisor-in-matrix-rings-over-a-commutative-ring – Badam Baplan Jun 08 '24 at 22:53