6

Definition: A universal side divisor, is an element $s\in R\setminus R^\times$ such that for every $x\in R$ either $s\mid x$, or there is some unit $u\in R^\times$ such that $s\mid x+u$.

Fact: A Euclidean domain $R$ has universal side divisors.

Proof: Let $s$ be a non unit of minimal norm $N(s)$. Let $x\in R$ and write $$x=sq+r$$ Then either $r=0$, and so $s\mid x$, or $N(r)<N(s)$. But since $N(s)$ is minimal we must have $r$ a unit, so $s\mid x-r$. Thus $s$ is a universal side divisor.

I'm doing the following excercises:

Suppose $R=\Bbb Z[\alpha]$, where $\alpha=\frac{-1+\sqrt{-19}}{2}$. Prove that $R$ is not a Euclidean domain by showing that it has no universal side divisors.

a) Find the units of $R$.

b) Take the usual norm $N$ on $R$, $N(z)=z\bar z$. Show that the only elements of norm less than $5$ are $\{0,\pm 1,\pm 2\}$.

c) Taking $x=2$ in the definiton of a universal side divisor, show that any universal side divisor must be a non-unit divisor of $2$ or $3$.

d) Find all non-unit divisors of $2$ and $3$.

e) Now taking $x=\alpha$ in the definition of a universal side divisor, find a contradiction.

What I would like to know is:

  1. Is my proof correct?
  2. Why is the step b) needed?

Here we go:

a) Find the units of $R$.

Let $N(a+b\alpha)=a^2-ab+5b^2=1$, there are two cases to consider:

  1. $a\geq b$, then $1\geq 5b^2$, so $b=0$ and $a=1$
  2. $b\geq a$, then $1\geq a^2+4b^2$, so again $b=0$ and $a=-1$.

Thus the only units are $\pm 1$.

b) Take the usual norm $N$ on $R$, $N(z)=z\bar z$. Show that the only elements of norm less than $5$ are $\{0,\pm 1,\pm 2\}$.

Again, there are two cases to consider:

  1. $a\geq b$, then $5>a^2-ab+5b^2\geq 5b^2$, so $b=0$, and $a=0,1,2$
  2. $b\geq a$, then $5>a^2-ab+5b^2\geq a^2+4b^2$, so $b=0$, and $a=0,-1,-2$

Thus the elements $0,\pm 1, \pm 2$ is the only elements with norm less than $5$.

c) Taking $x=2$ in the definiton of a universal side divisor, show that any universal side divisor must be a non-unit divisor of $2$ or $3$.

Since $R^\times=\{\pm 1\}$ a universal side divisor $s$ must divide one of $2, (2+1), (2-1)$. Since $s$ is not a unit, we must have $s\mid 2$ or $s\mid 3$.

d) Find all non-unit divisors of $2$ and $3$.

Both $2,3$ are irreducible in $R$. To see this assume that $N(2)=N(x)N(y)=4$, and $N(3)=N(z)N(w)=9$. Is is then enough to show that $N(x),N(y)\neq 2$ and $N(z),N(w)\neq 3$. There are again two cases:

  1. If $a\geq b$, then $a^2+4b^2\geq 2,3\geq 5b^2$, and so $b=0$ and $a^2=2,3$, which is impossible.
  2. If $b\geq a$, then $5b^2\geq 2,3\geq a^2+4b^2$, and so $b=0$. Contradiction.

So both $2,3$ are irreducible.

e) Now taking $x=\alpha$ in the definition of a universal side divisor, find a contradiction.

If $R$ is a Euclidean domain, then there is a universal side divisor $s$ such that $s\mid 2$ or $s\mid 3$. As both $2,3$ are irreducible we have $s=\pm 2,\pm 3$. Therefore $N(s)=4,9$. Also $s$ divides one of $\alpha,\alpha-1,\alpha+1$. Computing norms we have:

  • $N(\alpha)=5$
  • $N(\alpha+1)=5$
  • $N(\alpha-1)=6$

As $4$ or $9$ divides none of them, we must have that $s$ doesn't divide any of $\alpha,\alpha-1,\alpha+1$. Contradiction. Therefore $R$ is not a Euclidean domain.

cansomeonehelpmeout
  • 12,782
  • 3
  • 22
  • 49
  • 1
    "can someone help me out"? Please do post such good questions, we will definitely help you out. +1 I am taking a look at the proof right now. – Sarvesh Ravichandran Iyer Mar 16 '21 at 17:34
  • It looks correct; as for (b), it simplifies your steps (c) and (d): a divisor of $2$ would have norm less than or equal to $N(2)=4$, so it is $\pm 1$ or $\pm 2$; same for $3$. – Arturo Magidin Mar 16 '21 at 17:35
  • 1
    @TeresaLisbon Thank you! My username reflects my profile picture: a man trapped inside a computer, therefore "can someone help me out". That's it, nothing more! I would feel bad if I chose a sketchy username in order to get help faster! – cansomeonehelpmeout Mar 16 '21 at 17:38
  • @ArturoMagidin Thank you very much! That makes sense! – cansomeonehelpmeout Mar 16 '21 at 17:41
  • @cansomeonehelpmeout Do not worry, it is the quality of your question and not your username that attracted me. The proof is correct as far as I see, and thanks because I actually didn't know about the universal divisor criteria. Where can I find the proof of the the "fact" that Euclidean domains have a universal side divisor? Or is it a nice exercise for students? – Sarvesh Ravichandran Iyer Mar 16 '21 at 17:44
  • Nice problem. It's correct the solution! (+1) –  Mar 16 '21 at 17:50
  • @TeresaLisbon I've updated the question to include a proof. Thank you again! – cansomeonehelpmeout Mar 16 '21 at 18:04
  • 1
    @cansomeonehelpmeout Seems nice for a first course in algebra. I will recommend it to who I can. Thanks once again! You are a role model, do help out new users as often as you can, they will not have someone better to guide them. – Sarvesh Ravichandran Iyer Mar 16 '21 at 18:08
  • Seems to be a duplicate of this (update of my comment 2hrs ago). – Bill Dubuque Mar 16 '21 at 20:10

0 Answers0