4

(Rewritten following earlier feedback...)

(1) What are the ideals of ${\mathbb Z}^{\mathbb N}$? That is, take the ring which is the product of a countable infinity of copies of $\mathbb Z$; is there a simple explicit way to describe its ideals?

More generally, suppose $R = \prod_{i \in I} R_i$ is a product of rings (unital, not necessarily commutative). We'd like to describe the left ideals of $R$ in terms of $R_i$.

I see that in a finite product all the ideals are the products of ideals in the factors. For example, if $J$ is an ideal of $R_0 \times R_1$, define $J_i = \pi_i(J)$; then $J \subseteq J_0 \times J_1$, but also if $r_i \in J_i$, then we can get $s_i \in J \cap \pi_i^{-1}(r_i)$ and we have $(r_0, r_1) = (1_{R_0}, 0)s_0 + (0, 1_{R_1})s_1 \in J$.

So only the infinite products are interesting. There we have ideals such as the elements which are zero except at finitely many coordinates. Here are the ones I can see. Suppose each $R_i$ has left ideals $J_{i0} \subseteq J_{i1}$, and $K \subseteq 2^I$ is an ideal in the power set, in the sense that it's closed under subset and finite union. Then take $J$ to be the set of elements whose coordinates are in the small ideals and occasionally in the large ideals: $J = \{r \in R: r_i \in J_{i1}, \{i: r_i \notin J_{i0}\} \in K\}$. This seems to be a left ideal in $R$.

(2) Are there any other left ideals in $R$?

user26857
  • 52,094
Hew Wolff
  • 4,074
  • 6
    The better way to look at this problem is to find all ideals of $\prod \mathbb Z$. In this context your question is equivalent to: Is every ideal in $\prod \mathbb Z$ product of ideals in $\mathbb Z$? This is not true for infinite products. – user52045 Dec 19 '13 at 13:12
  • 2
    For example, every filter on $\mathbb{N}$ gives rise to an ideal. – Martin Brandenburg Dec 19 '13 at 13:49
  • @MartinBrandenburg: Is it true tho that every ideal comes from filter on index set? – user52045 Dec 19 '13 at 13:59
  • 2
    No, $\prod_i n_i \mathbb{Z}$ doesn't come from a filter. In a product of fields, ideals correspond to filters on the index set. The situation of $\mathbb{Z}$ is much more complicated and probably not solvable. – Martin Brandenburg Dec 19 '13 at 14:00
  • @Hew Wolff: Is your index set finite or infinite? You cannot simply write $\prod_i$ without specifying $i$. – Martin Brandenburg Dec 19 '13 at 21:43
  • @Martin, actually I think I can! :-) But the answer is that I'm interested in both cases. Maybe someone would like to post an answer for the finite case. – Hew Wolff Dec 21 '13 at 16:11
  • Well ideals of $R_1 \times \dotsc \times R_n$ are of the form $I_1 \times \dotsc \times I_n$ for ideals $I_l \subseteq R_l$. The proof is straightforward, no ideas are needed. Interesting and mysterious things happen when you take infinite products. – Martin Brandenburg Dec 21 '13 at 17:04
  • Intriguing, @MartinBrandenburg. Where can I learn more about the infinite products? For example, how does one build an ideal from a filter as you mention above? – Hew Wolff Dec 27 '13 at 03:56
  • 1
    See http://math.stackexchange.com/questions/1049867 – Martin Brandenburg Nov 17 '15 at 10:01
  • Actually there is a quite nicely written paper on exactly that question by Levy, Loustaunau an Shapiro. Follow https://eudml.org/doc/211861. – Daniel W. Jan 10 '20 at 17:55
  • Thanks @DanielW. I will look at that. It looks like they describe prime and maximal ideals of $\Bbb{Z}^\Bbb{N}$; it's not clear whether this is also a description of all ideals. If it is, feel free to supply that as an answer. – Hew Wolff Jan 12 '20 at 20:31

0 Answers0