0

Let $A \subseteq B$ be two GCD domains with $1$. Assume that the extension $A \subseteq B$ is algebraic, not necessarily integral.

Notation: For $w \in B$ denote its minimal polynmial over $A$ by $m_w$ and denote the leading coefficient of $m_w$ by $c_w$.

Question: Let $u,v \in B$. Assume we have $c_u$ and $c_v$. What can we say about $c_{u+v}$ in terms of $c_u$ and $c_v$? Especially, when $\gcd(c_u,c_{u+v})=1$ and $\gcd(c_v,c_{u+v})=1$? Or more strongly, when $(c_u,c_{u+v})=A$ and $(c_v,c_{u+v})=A$ (unit ideals).

Remarks:

(1) In the special case when the extension is integral, clearly all coefficients $c_u,c_v,c_{u+v}$ are units, so trivially $(c_u,c_{u+v})=A$ and $(c_v,c_{u+v})=A$.

(2) I do not mind to restrict to the case where $A$ and $B$ are Noetherian UFD's, if this simplifies my question.

(3) Please see this related question, which hints that considering resultants may help.


Edit: Perhaps it is better to change my question to the following one: Let $k$ be a characteristic zero base field $k \subsetneq A \subseteq B$. Instead of considering $u$, $v$ and $u+v$, let us consider $u$, $v$ and $u+\lambda v$, where $\lambda \in k$.

Therefore, I would like to ask:

New Question: Let $u,v \in B$. Assume we have $c_u$ and $c_v$. Can we fine $\lambda \in k$ such that $c_{u+\lambda v}$ satisfies $\gcd(c_u,c_{u+\lambda v})=1$ and $\gcd(c_v,c_{u+\lambda v})=1$? Or more strongly, $(c_u,c_{u+\lambda v})=A$ and $(c_v,c_{u+\lambda v})=A$ (unit ideals).

Thank you!

user237522
  • 7,263
  • Note, minimal polynomial is not unique. The lead coefficients of all polynomials of degree $d$ satisfied by a fixed element are an ideal (if you include $0.$) So the lead coefficient is really only defined if that ideal is principal. – Thomas Andrews Jun 04 '25 at 17:06
  • @ThomasAndrews, thank you for your important coment. what if we take m_w that has no common factor for all its coefficients. Are now m_w and c_w defined uniquely? – user237522 Jun 04 '25 at 17:09
  • @ThomasAndrews, when that ideal is principal? – user237522 Jun 04 '25 at 17:11
  • Well, in your case, your gcd of the polynomials is enough to show that the ideal is principal, I think. – Thomas Andrews Jun 04 '25 at 17:13
  • @ThomasAndrews, thank you. Isn't it "automatic" that for a given element $w$, all its minimal polynomials (polynomials $m$ of minimal degree such that $m(w)=0$) have the property that they are multiples of one another by an element of $A[T]$ (of degree zero, hence an element of $A$), hence their leading coefficients belong to some principal ideal $(a)$, for some $a \in A$? – user237522 Jun 04 '25 at 17:18
  • I was concerned for the non-Noetherian case, where you might have a sequence of minimal polynomials with lead coefficients $c_1,c_2,\dots$ where each $c_{i+1}$ is not a unit multiple of $c_i$ and $c_{i+1}\mid c_i.$ Then all of the minimal polynomials could be multiples , but there would be no "minimal" lead ing coefficient. But the $\gcd$ of the coefficients, I think, makes that impossible. – Thomas Andrews Jun 04 '25 at 17:26
  • @ThomasAndrews, thank you for the clarification. Do you have any ideas for the Noetherian UFD case? – user237522 Jun 04 '25 at 18:08
  • If I had any idea, I might have contribute more. As a rule, don't request from the one person helping you to help you more, just because the system lets you tag them. I'm apparently the one person on this site who is engaging with your question so far. Reminding me you still want an answer is slightly rude, as if I don't know. – Thomas Andrews Jun 04 '25 at 18:14
  • @ThomasAndrews, I really apologize, I did not mean to be rude and I am thankful for your comments and interest in my question. I will try to remember your advice and apply it next times. – user237522 Jun 04 '25 at 18:23
  • Yeah, I've seen far worse along this lines, of people getting suggestions of problems with their question with comments that were, approximately, "So what? Can you answer my question." Your response is a very mild, by comparison, and you seemed polite, and it is an error I can imagine making. – Thomas Andrews Jun 04 '25 at 18:27
  • @ThomasAndrews, thank you. (I am trying to learn and improve from any comment, mathematical or not, including diminish my errors, mathematical or not). – user237522 Jun 04 '25 at 18:38

0 Answers0