1
$\begingroup$

Let $F$ be a field, and let $P(X_1,\dots,X_m)$, $Q(X_1,\dots,X_m) \in F[X_1,\dots,X_m]$ be two coprime polynomials. Consider $n$ new polynomials $R_1(Y_{1,1},\dots,Y_{1,n}) \in F[Y_{1,1},\dots,Y_{1,n}]$, $\dots$, $R_m(Y_{m,1},\dots,Y_{m,n}) \in F[Y_{m,1},\dots,Y_{m,n}]$ in the new $mn$ indeterminates $Y_{1,1},\dots,Y_{1,n},\dots,Y_{m,1},\dots,Y_{m,n}$, and assume that $R_1,\dots,R_m$ all have positive degree. My question is: are the polynomials $P(R_1(Y_{1,1},\dots,Y_{1,n}),\dots,R_m(Y_{m,1},\dots,Y_{m,n}) )$ and $Q(R_1(Y_{1, 1},\dots,Y_{1, n}),\dots,R_m(Y_{m,1},\dots,Y_{m,n}))$ coprime?

I don't know the answer in general, neither I can say whether it is a trivial question or a deep one. Does someone have some idea about it? Thank you very very very much in advance for your kind attention.

NOTE 1 The answer is positive for $m=1$. This is an easy consequence of Bézout's Identity, by which there exist $S(X),T(X) \in F[X]$ such that \begin{equation} S(X)P(X)+T(X)Q(X)=1, \end{equation} so that for any $R(Y_1,\dots,Y_n) \in F[Y_1,\dots,Y_n]$ we have \begin{equation} S(R(Y_1,\dots,Y_n))P(R(Y_1,\dots,Y_n))+T(R(Y_1,\dots,Y_n))Q(R(Y_1,\dots,Y_n))=1, \end{equation} which implies that $P(R(Y_1,\dots,Y_n))$ and $Q(R(Y_1,\dots,Y_n))$ are coprime. Note that this results holds true even if $R(Y_1,\dots,Y_n)$ is a constant polynomial. For $m > 1$ we cannot allow $R_1,\dots,R_m$ to be constant polynomials, otherwise the answer is for sure in general negative: take e.g. $P(X_1,X_2)=X_1+X_2$, $Q(X_1,X_2)=X_1$, $R_1(Y_{1,1},Y_{1,2})=Y_{1,1}$,$R_2(Y_{2,1},Y_{2,2})=0$.

NOTE 2 This question has been posted on math.stackechange.com as Substituting Polynomials into Two Relatively Prime Polynomials, but it has received no attention there, whether it has been deemed too difficult or of little interest.

$\endgroup$
3
  • $\begingroup$ Perhap this remark at inhttps://math.stackexchange.com/questions/412980/kx-1-x-2-dots-is-a-ufd might be useful: Paul Cohn introduced the idea of inert extensions when studying Bezout rings. Cohn proved that every gcd domain can be inertly embedded in a Bezout domain, and every UFD can be inertly embedded in a PID. $\endgroup$ Commented Aug 30 at 22:55
  • $\begingroup$ @W.EdwinClark Thank you very much Edwin for having kindly considered my post. I would be very grateful to you, if you could give me the precise references for these results of Paul Cohn. $\endgroup$ Commented Aug 31 at 13:54
  • $\begingroup$ @W.EdwinClark Dear Edwin, I have found the article in which Cohn established the results you quoted: Cohn P.M, Bezout rings and their subrings. Mathematical Proceedings of the Cambridge Philosophical Society (1968) 64(2):251-264. doi:10.1017/S0305004100042791. Thank you very much again for your suggestion. $\endgroup$ Commented Sep 1 at 11:43

1 Answer 1

2
$\begingroup$

The answer is positive. Since $P,Q$ are coprime the variety $V=\{P=Q=0\}\subset\mathbb A^m$ has dimension $\le m-2$. Consider the variety $W=\{P(R_1,\ldots,R_m)=Q(R_1,\ldots,R_m)=0\}\subset\mathbb A^{nm}$ and the map $f=(R_1,\ldots,R_m):W\to V$. The fibers of $f$ have dimension $m(n-1)$ (because $R_i$ are non-constant and each is applied to a separate set of variables) and therefore $\dim W=\dim V+m(n-1)\le m(n-1)+m-2=mn-2$, so $P(R_1,\ldots,R_m),Q(R_1,\ldots,R_m)$ are coprime.

$\endgroup$
3
  • $\begingroup$ Should $f$ be the map $(R_1,\dots,R_m),$ rather than $(P,Q)$? $\endgroup$
    – Tom WIlde
    Commented Sep 6 at 20:28
  • $\begingroup$ Yes, fixed. Thanks. $\endgroup$ Commented Sep 7 at 0:48
  • $\begingroup$ @AlexeiEntin Wow, great! I don't have the notions required to understand your proof, but I trust you! $\endgroup$ Commented Sep 8 at 20:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .