1

Prove that if $\rm p(x)$ and $\rm q(x)$ are polynomials with rational coefficients whose product $\rm f(x)g(x)$ has integer coefficients , then the product of any coefficient of $\rm g(x)$ with any coefficient of $\rm f(x)$ is an integer.

I tried to prove this as follows:

Let $\rm G(x):=p(x)q(x)$. It is given that $\rm p(x),q(x)\in \mathbb Q[x]$.

If $\rm deg \, p(x)=0 $ or $\rm \, deg\, q(x)=0$, then the statement is true. So suppose that $\rm p(x), q(x)$ are of $\rm degree\gt 0$.

By Gauss' Lemma (stated below) $\rm G(x)$ is reducible in $\rm \mathbb Z[x]$ and there exist $\rm r,s\in \mathbb Q\setminus \{0\}$ such that $\rm rp(x), sq(x)\in \mathbb Z[x]$ and $\rm G(x)=(rp(x))(sq(x)).$ It follows that $\rm rs=1$.

Let coefficients of $\rm p(x), q(x)$ be denoted by $\rm p_i,q_i$'s respectively. For any $\rm i\in \{0,1,\cdots,\deg p(x)\}, j\in \{0,1,\cdots,\deg q(x)\} $, it follows that $\rm p_iq_j=1.p_iq_j=rs.p_iq_j=\overbrace {rp_i}^{\rm\in \mathbb Z}.\overbrace{sq_j}^{\rm\in \mathbb Z}\in \mathbb Z\quad\quad \square$

Is my proof correct? Thanks.

Gauss' Lemma denotes the following: If $\rm R$ is a Unique Factorization Domain with field of fractions $\rm F$, and $\rm p(x)\in R[x]$ is reducible in $\rm F[x]$ then $\rm p(x)$ is reducible in $\rm R[x]$. More precisely, if $\rm p(x)= A(x) B(x)$ for some non-constant polynomials in $\rm F[x]$ then there are non zero elements $\rm r,s\in R$ such that $\rm r A(x), sB(x)\in R[x]$ and $\rm p(x)=(r A(x))(sB(x)).$

Bill Dubuque
  • 272,048
Koro
  • 11,402
  • It looks extremely circular to me. It's like proving Euclid's lemma by invoking unique prime factorization. It would be far better if you find a proof of the desired theorem directly. – user21820 Apr 22 '22 at 15:57
  • 2
    Yes, that's a standard method of proof, e.g. see the linked dupe. – Bill Dubuque Apr 24 '22 at 06:03
  • @BillDubuque: Thank you so much for the review. :) – Koro Apr 24 '22 at 06:04
  • 1
    See also here and here for generalizations. – Bill Dubuque Apr 24 '22 at 06:06
  • Please edit the question to quote the entire statement of the GL theorem you apply (i.e. the part about $r$ and $s$ etc). – Bill Dubuque Apr 24 '22 at 09:36
  • @BillDubuque: done. – Koro Apr 24 '22 at 09:42
  • Is that really the wording of the entire lemma used in Dummit and Foote? Are you even aware that the first part ("If R is a UFD with field of fractions F, and p(x)∈R[x] is reducible in F[x] then p(x) is reducible in R[x].") is NOT saying the same as the second part ("if p(x)=A(x)B(x) for non-constant polynomials in F[x] then there are nonzero r,s∈R such that rA(x),sB(x)∈R[x] and p(x)=(rA(x))(sB(x)).")? – user21820 Apr 24 '22 at 09:45
  • 1
    @user21820 Yes, that's how D&F state it (which is not uncommon). Btw, that's not my downvote. – Bill Dubuque Apr 24 '22 at 10:09

0 Answers0