7
$\begingroup$

Let $R=\mathbb C\{x_1,...,x_n\}\subset S=\mathbb C [[x_1,...,x_n]]$ denote the ring of convergent, respectively formal, power series over $\mathbb C$.
Suppose $f\in R$ is irreducible in $R$. Does it remain irreducible in $S$?

$\endgroup$
9
  • 3
    $\begingroup$ When crossposting, it's important to link all versions together to avoid duplication of work. Here is the post on MSE. $\endgroup$
    – KReiser
    Commented Sep 1, 2020 at 19:21
  • 1
    $\begingroup$ What partial results do you know of? I think Nagata proved this for prime instead of irreducible. $\endgroup$
    – Arno Fehm
    Commented Sep 1, 2020 at 20:38
  • $\begingroup$ The ring of germs of holomorphic functions is a UFD (see section 2.2 of people.math.harvard.edu/~yifei/Weierstrass_theorems.pdf . I think I remember this being in Gunning and Rossi, but I can't find my copy right now.) So prime = irreducible. Does this finish the question? $\endgroup$ Commented Sep 2, 2020 at 12:30
  • 3
    $\begingroup$ I do find the comments of the OP quite rude. That said, I think that prime elements stay prime because Nagata claims that prime ideals stay prime in the introduction to his 1953 paper "Some remarks on local rings, II", but I haven't actually looked into the proofs. $\endgroup$
    – Arno Fehm
    Commented Sep 2, 2020 at 13:30
  • 1
    $\begingroup$ @DavidESpeyer: You were right about Gunning and Rossi: books.google.co.vi/…. $\endgroup$
    – joriki
    Commented Feb 4, 2023 at 16:33

1 Answer 1

5
$\begingroup$

As Arno Fehm points out, this follows from results in Nagata's Some Remarks on Local Rings II. Both $R$ and $S$ are UFD's, so $f$ is irreducible, in $R$ or $S$ respectively, if and only if the ideal it generates, in $R$ or $S$ respectively, is prime. At the bottom of page 1 of Nagata's paper, he states that, if $\mathfrak{p}$ is a prime ideal of $R$, then $\mathfrak{p}S$ is prime as well.

I found it hard to absorb all of Nagata's vocabulary; here is a route to get the desired claim from his results while skipping some of the sophisticated language.

Let $R'$ and $S'$ denote the versions of $R$ and $S$ with $n-1$ variables. Suppose that $f=gh$ for $f \in R$ and $g$, $h \in S$ nonunits. Use the Weierstrass preparation theorem to factor $f = pu$, $g=qv$ and $h = rw$ where $p \in R'[x_n]$, $q \in S'[x_n]$ and $r \in S'[x_n]$ are Weierstrass polynomials and $u \in R^{\times}$, $v \in S^{\times}$ and $w \in S^{\times}$ are units. Then $qr$ is a Weierstrass polynomial of $S$, and $vw \in S^{\times}$, so $f = (qr) (vw)$ and $f = pu$ are both Weierstrass factorizations in $S$. Since such factorizations are unique, we have $p = qr$ and $u = vw$.

Write $p(x_n) = x_n^a + \sum_{i=0}^{a-1} p_i x_n^i$, $q(x_n) = x_n^b + \sum_{i=0}^{a-1} q_i x_n^i$, $r(x_n) = x_n^c + \sum_{i=0}^{c-1} r_i x_n^i$. Then the $q_i$ and $r_i$ are polynomial combinations of the roots of $p$, so the $q_i$ and $r_i$ are integral over $R'$. But Nagata, in his proof of Theorem 5, shows that $R'$ is integrally closed in $S'$, and this proof is extremely concrete. So that shows that the $q_i$ and $r_i$ land in $R'$. Thus $q$ and $r \in R$, and we deduce that $f$ factors in $R$ as well.

$\endgroup$
1
  • $\begingroup$ Thanks a lot for your remarkable work, David. Your extracting an answer to my question so quickly from Nagata's hard to read aricle (and idiosyncratic terminology) is a real *tour de force" . $\endgroup$
    – lefuneste
    Commented Sep 3, 2020 at 8:36

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .