1. If $D$ is an integral domain containing an irreducible element, then $D[x]$ is not a principal ideal domain.

  2. If $F$ is a field and $n \ge 2$, then $F[x_1,...,x_n]$ is not a principal ideal domain. [Hint: show that $x_1$ is irreducible in $F[x_1,...,x_{n-1}]$.

I have already worked on part 1, but I am having trouble proving what the hint in part 2. suggests. I imagine the statement is to be proven through induction. Here is my proof of the base case. Suppose that $x =pq$ for $p,q \in F[x]$. Then $1 =\deg(pq) = \deg(p) + \deg (q)$, which implies, WLOG, that $p = ax + b$ and $q = c$, where $a,b,c \in F$ and $c \neq 0$. Then $x = pq = acx + bc$ implies $ac=1$ and $bc=0$. Since $c \neq 0$, it must be that $b=0$, and $a$ and $c = q$ are units. Hence, $x$ is irreducible.

The inductive step is giving me a little more trouble. Suppose that $x_1$ is irreducible in $F[x_1,...,x_{n-1}]$, and suppose that $x_1 = pq$ for $p,q \in F[x_1,..,x_n]=(F[x_1,...,x_{n-1}])[x_n]= D[x_n]$ (strictly speaking, I think this is an isomorphism). Note that $x_1$ as a constant in $D[x_n]$, so that $0=\deg(x_1) = \deg(pq)= \deg(p)+\deg(q)$ implies $p = f(x_1,...,x_{n-1})$ and $q=g(x_1,...,x_{n-1})$ are constants. In particular, they satisfy $1 = fg$ and so are units in $D=F[x_1,...,x_{n-1}]$......

Isn't it true that the units of $D[x_n]$ will be the units of $D$? If so, can I conclude that $p$ and $q$ are units in $D[x_n]$ and then conclude that $x_1$ is irreducible in $F[x_1,...,x_n]$? Is that right? Obviously proving 2. is trivial at this point: $F[x_1,...,x_{n-1}]$ contains the irreducible element $x_1$, and so by part 1. $F[x_1,...,x_n]$ cannot be a PID.

  • $\begingroup$ There is something strange about the claim 1, is it really talking about $c$ or $x$ being irreducible in $D$? Also in that case wouldn't $\mathbb{Z}[p] = \mathbb{Z}$ be a counterexample for any prime $p$ (and thus irreducible in $\mathbb{Z}$)? $\endgroup$ – Tob Ernack Jan 2 '18 at 17:17
  • $\begingroup$ @TobErnack The statement is not taking $x$ to be irreducible in $D$. $\endgroup$ – user193319 Jan 2 '18 at 17:19
  • $\begingroup$ @TobErnack $x$ is a variable. The statement says that if there is an irreducible element in $D$, then $D[x]$ is not a PID. For example, there is no irreducible element in the field $F$, so $F[x]$ being a PID does not contradict statement 1. $\endgroup$ – Pedro Jan 2 '18 at 17:21
  • $\begingroup$ Statement 1 should be formulated as ‘If $D$ is an integral domain which contains an irreducible element, then …’. $\endgroup$ – Bernard Jan 2 '18 at 17:28


For step 2, you don't need really induction. You just have to prove this:

If $D$ is an integral domain, an irreducible element in $D$ is also an irreducible element in $D[X]$.

  • $\begingroup$ Okay. But, taking $D=F[x_1,...,x_{n-1}]$, I still need to show that $D$ has an irreducible element--namely, $x_1$--which I was trying to prove using induction. $\endgroup$ – user193319 Jan 2 '18 at 23:30
  • $\begingroup$ I'll complete the hint: since $x_1$is irreducible in $F[x_1]$, it is irreducible in $F[x_1,x_2]$, hence also in $F[x_1,x_2, x_3]$, &c. $\endgroup$ – Bernard Jan 2 '18 at 23:33

The argument you give for the base case, combined with similar arguments as $deg_{x_{i}}(p)=deg_{x_{i}}(q)=0$ for all $i>1$, shows at once that $x_{1}$ is irreducible in $F[x_{1},\ldots,x_{n-1}]$. Once you know this, then the statement follows immediately from part 1, since $$ F[x_{1},\ldots,x_{n}]\cong F[x_{1},\ldots,x_{n-1}][x] $$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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