7
$\begingroup$

By a theorem of Specker, there’s only the zero map since any map out of $ \prod_{n \in \mathbb{N}} \mathbb{Z} $ is determined by the values of the unit vectors, which all lie in $ \bigoplus_{n \in \mathbb{N}} \mathbb{Z} $, but the original proof is more general, uses a bunch of machinery, and in German. Isn’t there an easier way?

  • 0
    Could you link the original one in Germa? I'd be very grateful2017-02-06
  • 0
    @TryingToGraduate It's Specker's Additive Gruppen von Folgen ganzer Zahlen2017-02-06

1 Answers 1

5

There is a nice quick proof. I'm not sure who the proof is due to.

The statement is equivalent to

If $P$ is the group of sequences ${\bf a}=(a_0,a_1,\dots)$ of integers, and $f:P\to\mathbb{Z}$ is a homomorphism that vanishes on finite sequences (so that $f({\bf a})=f({\bf b})$ whenever ${\bf a}$ and ${\bf b}$ differ in only finitely many places), then $f=0$.

Suppose $f:P\to\mathbb{Z}$ is a homomorphism that vanishes on finite sequences.

For any ${\bf a}\in P$, we can write $a_n=b_n+c_n$, where $b_n$ is divisible by $2^n$ and $c_n$ is divisible by $3^n$.

Then for each $n$, ${\bf b}$ differs in only finitely many places from a sequence divisible by $2^n$, so $f({\bf b})$ is divisible by $2^n$ for all $n$, and so $f({\bf b})=0$. Similarly $f({\bf c})=0$, and so $f({\bf a})=f({\bf b}+{\bf c})=0$.

  • 2
    Dear Jeremy: this beautiful proof can be found in Lam's ''Lectures on Modules and Rings'' as Lemma 2.8' on page 23. After presenting this argument, Lam ''thank[s] I. Emmanouil, P Farbman, and D. Shapiro for their help in formulating the preceding proof''. Apologies if you already knew this.2017-02-08