2
$\begingroup$

Let given ring $R$ of finite rank. Is it true that for all primes $p$ large enough modules $Der_{\mathbb{Z}}(R/pR) = \{0\}$?

For every ring we define $Der_{\mathbb{Z}}(R)$ as set of linear operators $D : R\to R$, where $\forall a, b\in R, D(ab) = aD(b) + bD(a)$.

I can prove this in case $R/pR$ without nilpotent elements:

In this case for any $r\in R\setminus\{0\}$, $D\in Der_{\mathbb{Z}}(R/pR)$ and any prime $p > n = rk(R)$, it is well known that $$\exists f = a_0 +...+ a_qx^q\in \mathbb{Z}[x] : f(r) = 0$$ with $$\deg f\leq n$$ Moreover if $p|a_0,...,a_q$ then $p|f$, so we can think that $f\not= 0\in\mathbb{Z}_p[x]$.

Let $\{P_i\}_{1\leq i\leq k}$ be different primes in $\mathbb{Z}_p[x]$, such that $f = P_1^{w_1}...P_k^{w_k}\in\mathbb{Z}_p[x]$.

Let $R/pR$ be without nilpotents so that $g = P_1...P_k\implies g(r) = 0$.

So $D(g(r)) = g'(r)D(r) = g(r)D(r) = 0$, $gcd(g, g') = 1$, so $D(r) = 0$. $\Box$

$\endgroup$
1
  • 1
    $\begingroup$ Could you - say what you mean by ring: is it meant to be associative (certainly)? commutative? and what is "finite rank"? Somewhere and for commutative rings, it is defined as "noetherian with a universal upper bound on the number of generators of ideals". If you mean commutative, the tag "commutative algebra" would be more appropriate. If you consider non-commutative rings, "linear operator" is somewhat unclear. $\endgroup$
    – YCor
    Dec 26, 2015 at 16:37

1 Answer 1

-1
$\begingroup$

I found helpfull lemma

Lemma 1

Let given ring of finite rank $n$, let $p>(2n)!$ be a prime prove that next conditions are equivalent:

a) $Der_{\mathbb{Z}}(R/pR) = \{0\}$.

b) $Nil(R/pR)=\{0\}$, where $Nil(R)$ is nilradical of $R$.

Proof:

b) => a) is proved in previous post. So we need to prove a) => b).

Lemma 2

Let given ring $R$, number $n$ and prime number $p>(2n)!$, where $pR = \{0\}$ and for every element $a\in Nil(R), a^n = 0$. Prove that for some natural $k$, $(Nil(R))^k = \{0\}$.

Proof:

Induction by $n$. Let $I = <\{a^{n-1}|a\in Nil(R)\}>R$. For every $x, y\in Nil(R)$, $x+y\in Nil(R)$, so $0 = (x+y)^{2n-2} = \frac{(2n-2)!}{(n-1)!^2}x^{n-1}y^{n-1}$, so $x^{n-1}y^{n-1}=0$, $I^2 =\{0\}$. For every $a\in Nil(R), a^{n-1}\in I$, so $\forall r\in Nil(R/I) = Nil(R)/I$, $r^{n-1} = 0$. So by induction for some natural $k$, $(Nil(R/I))^k = (Nil(R)/I)^k = \{0\}$ and $Nil(R)^k\subseteq I$. So $Nil(R)^{2k} = \{0\}$. done

Lemma 3

Let given ring $R$ of finite rank $n$, let $p > (2n)!$ is prime. Prove that for some natural $k$, $(Nil(R/pR))^k = \{0\}$.

Proof:

Let $a = r + pR$ is any element from $Nil(R/pR)$, let $\lambda : a^{\lambda} =0\not= a^{\lambda - 1}$. Rank of $R$ is $n$, so exists $f = a_0+ ... +a_nx^n\in\mathbb{Z}[x] : f(r) = 0$. If $p|a_i, i=1,..., n$, then we can factor $f$ by $p$, so we can think that $f\not= 0\in \mathbb{Z}_p[x]$. Let $f = a_i'x^i + ... + a_n'x^n\in \mathbb{Z}_p[x]$, where $a_i'\not= 0\in \mathbb{Z}_p$. If $i < \lambda$, then $0=a^{\lambda - i - 1}f(a) = a_i'a^{\lambda - 1}\not= 0$, so $\lambda\leq i \leq n$. Now from lemma 2 we finish proof. done

Lemma 4

Let given ring $R$ and prime $p$, such that $pR =\{0\}$ and for some natural $k$, $(Nil(R))^k =\{0\}$. Prove that there exists ring $Q\subseteq R: Q\oplus Nil(R) = R$.

Now return to main lemma 1. From lemmas 3 and 4 we get that there exists ring $Q\subseteq R/pR: Q\oplus Nil(R/pR) = R/pR$. Consider linear projector $P : R/pR \to Nil(R/pR)$, where $\ker P = Q$, $Im P = Nil(R/pR)$. Let $Nil(R/pR)\not=\{0\}$ one can find element $a\in R/pR : (Nil(R/pR))^2\subseteq Ann(a), Nil(R/pR)\notin Ann(a)$, so we can define linear operator $D : R/pR\to aR/pR\subseteq R/pR$, $D(x) = aP(x)$, one can check that $D\in Der_{\mathbb{Z}}(R/pR)$, $Nil(R/pR)\notin Ann(a)$, so $D\not= 0$, but $Der_{\mathbb{Z}}(R/pR) = \{0\}$, so $Nil(R/pR)=\{0\}$. done

$\endgroup$
1
  • $\begingroup$ This appears to have been copied and pasted from another source withou explaining the context adequately $\endgroup$
    – Yemon Choi
    Feb 24, 2016 at 15:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.