Suppose we are working with a commutative ring $R$. I am reading a proof where the author writes

"If we are in $R[X]$ and we know that $f\mid X$. Thus either $f$ is a unit multiple of $X$ or it is a unit, so $(f)=(X)$ or $(f)=(1)$."

This seems to assume that $X$ is irreducible in $R[X]$. However I don't know how we would prove this unless we assume $R$ is an integral domain. Solving by hand$$(a_0+a_1X+\cdots+a_nX^n)(b_0+b_1X+\cdots+b_mX^m)=X$$ does not seem to require that one of these polynomials is a unit. Am I missing something here?

  • 13
    $\begingroup$ $(2X+3)(3X+2)=X$ in $\mathbb Z_6[X]$. $\endgroup$ – user26857 May 30 '17 at 9:51
  • 3
    $\begingroup$ Technically the question is not well-posed since there is not a unique definition of irreducible in use for general commutative rings, e.g. see the links in this answer. $\endgroup$ – Bill Dubuque May 30 '17 at 14:13

Let me summarize and give another answer, which the question has deserved.

  1. Yes, this is correct. The claim does not follow unless we assume that $R$ is an integral domain. We can easily give examples, where $R$ is a commutative ring with zero divisors, and the claim is false. Let $n=ab$ be a composite number and take $R=\mathbb{Z}/n\mathbb{Z}$. Then $a,b$ are zero divisors in $R$, and we have $$ (ax+b)(bx+a)=(a^2+b^2)x. $$ If we can choose $a,b$ such that $a^2+b^2=1$ in $R$, then we obtain a counterexample. For example, take $(a,b)=(3,4)$ with $n=12$, or $(a,b)=(2,3)$ with $n=6$ (see the comment).

  2. If $R$ is an integral domain, we have $\deg(fg)= \deg(f)+\deg(g)$ for $f,g\neq 0$. Then $x$ is irreducible.


In fact, one can prove the following theorem:

Thm. Let $A$ be a commutative ring with 1. Then $X$ is irreducible in $A[X]$ if and only if $A$ has no nontrivial idempotent elements (nontrivial meaning: different from $0$ and $1$)

Proof. I will prove in fact that $X$ is reducible if and only if $A$ has a nontrivial idempotent.

Assume that $e\in A$ is an idempotent different from $0$ and $1$. Then the polynomials $(1-e)X+e$ and $eX+(1-e)$ are nonzero, non invertible (otherwise their constant terms would be invertible, which would imply that $e=0$ or $1$, since $e(1-e)=0$). However, $((1-e)X+e)(eX+(1-e))=X$, so $X$ is reducible.

Assume now that $X$ is reducible. Notice that it is nonzero, and non invertible (since its constant term is not invertible). Hence $X=RS,$ where $R$ and $S$ are not invertible

Write $R=a+XP,S=b+XQ.$ We then have $X=ab+(aQ+bP)X+X^2PQ$.

Evaluation at $0$ yields $ab=0$. Therefore $X=(aQ+bP)X+X^2PQ$. Sicne multiplication by $X$ is injective, we get $$1=aQ+bP+XPQ.$$ Set $v=P(0)$ et $u=Q(0)$. Evaluation at $0$ yields $au+bv=1$. Set $e=au$. We have $$e^2=au (1-bv)=au-abuv=au=e,$$ hence $e$ is an idempotent.Assume that $e=0$. Then, $bv=1$, so $b$ is invertible and consequently $a=0.$

We then have $X=XPS$. Since multiplication by $X$ isinjective, we obtain $1=PS$, so $S$ is invertible, contradiction. Hence, $e\neq 0$. Similarly if $e=1$, then $au=1$ and we obtain this time that $R$ is invertible. Conclusion: $e$ is a nontrivial idempotent of $A$.


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.