2
$\begingroup$

Let $A,B$ be two rings and $f : A[X] \to B[X]$ is a ring homomorphism. The main question is:

What should be assumed on $f$ (or on $A$ and $B$) so that the degree of a polynomial $P(X) \in A[X]$ equals the degree of the polynomial $f(P(X)) \in B[X]$?

If I take $f=\text{ev}_0 : P \longmapsto P(0)$ then my condition is clearly not satisfied.

If I assume $f(1)=1$, $A = \Bbb Z$ (or $A=\Bbb Q$ I think) and if $f$ is bijective then $f(a_nX^n+...+a_1X+a_0)=a_nf(X)^n+...+a_1f(X)+a_0$ implies that the degree is preserved, I think.

But in general, does bijectivity of $f$ (and $f(1)=1$) is a sufficient condition? I only know that $$f(a_nX^n+...+a_1X+a_0)=f(a_n)f(X)^n+...+ f(a_1) f(X)+ f(a_0)$$ If I could show that $f(a) \in B$ for all $a\in A$ then I think that the degree is preserved.

Moreover, is bijectivity a necessary condition (this is just out of curiosity... I didn't found a counterexample)?

Thank you very much!

$\endgroup$
  • 1
    $\begingroup$ Bijectivity is not necessary, consider the inclusion $R[X] \hookrightarrow S[X]$ for a ring $S$ and a proper subring $R \subsetneq S$. If both $A$ and $B$ have no zero-divisors then $\deg f(P) = \deg P$ for every $P \in A[X]$ should be equivalent to $\deg f(X) = 1$, i.e. that $f(X)$ is of the form $f(X) = b X + c$. $\endgroup$ – Jendrik Stelzner Feb 28 '16 at 13:31
5
$\begingroup$

I first point that you should always assume that $f(1) = 1$, that is part of the definition of a ring morphism. I will also assume that $A$ and $B$ are commutative (otherwise it gets a little trickier).

Then an obvious necessary condition for $f$ to preserve degree is that $f$ sends constants to constants, meaning that the restriction of $f$ to $A$ is a ring homomorphism from $A$ to $B$ (it is not automatic !). It should also be injective since otherwise you have $a\in A$ such that $f(a)=0$, hence $f(aX)=0$ so degree is not preserved.

Then the second condition is that $f(X)$ has degree $1$, hence $f(X)=aX + b$, with $a,b\in B$ and $a\neq 0$. Then if $P\in A[X]$ has highest-degree term $a_nX^n$, the highest-degree term of $f(P)$ is $f(a_n)a^nX^n$. Since you want this term to be non-zero, it follows that $a$ should not satisfy $xa^n=0$ for any non-zero $x\in A$ and $n\in \mathbb{N}$ (this is clearly satisfied if $a$ is not a zero divisor, and in particular when $B$ is a domain, but this needs not be the case).

On the other hand, if you want $f$ to be invertible, then first obviously you need its restriction to $A$ to be invertible, so $f$ is an isomorphism between $A$ and $B$. You also need $a$ to be invertible, and this is enough, since then you can define its inverse by the inverse of $f_{|A}$ on the elements of $B$, and $f^{-1}(X) = a^{-1}(X-b)$.

In conclusion : for $f$ to preserve degree is equivalent to $f_{ |A}$ being an injective homomorphism from $A$ to $B$, and $f(X) = aX+b$ with $a,b\in B$ and no power of $a$ being annihilated by an element of $f(A)\setminus \{0\}$ ; on the other hand, for $f$ to be bijective is equivalent to $f_{|A}$ being an isomorphism from $A$ to $B$ and $f(X) = aX + b$ with $a,b\in B$ and $a$ invertible. So clearly the second condition is stronger, and they are almost never equivalent (you would need huge restrictions on $A$ and $B$, though it does work for $A=B=\mathbb{Q}$).

$\endgroup$
  • $\begingroup$ I have two questions: 1) when you say at the end "for $f$ to be bijective is equivalent to $f \vert_A$ being an isomorphism from $A$ to $B$", are you also supposing that $f$ preserves degree? Because I don't understand why we should have $f(A) \subset B$ only assuming that $f$ is an isomorphism. $\endgroup$ – Alphonse Mar 2 '16 at 18:41
  • $\begingroup$ 2) Do you have an example of an isomorphism $f : A[X] \to B[X]$ which does not preserve degree? According to your last sentence, it is never possible. But wouldn't this question give a counterexample? $\endgroup$ – Alphonse Mar 2 '16 at 18:43
  • $\begingroup$ (There is just a typo at the end, it should be $f(X)=aX+b$, not $aX+B$). $\endgroup$ – Alphonse Mar 2 '16 at 18:46
  • $\begingroup$ Sure, thanks, I edited it. $\endgroup$ – Captain Lama Mar 2 '16 at 18:59
  • 1
    $\begingroup$ Yes, I assume throughout my answer that $f$ preserves idegree, otherwise things get quite complicated, as you point out yourself. For an easy example of ring isomorphism that does not preserve degree, consider $A=B=k[T]$, and the isomorphism $k[T,X]\rightarrow k[T,X]$ exchanging $X$ and $T$ (sorry, I didn't see your questions before, I just read the last comment). $\endgroup$ – Captain Lama Mar 4 '16 at 19:33
1
$\begingroup$

One necessary condition is that the induced ring homomorphism in degree $0$. $f_0\colon A\to B$ is injective. This allows us to consider $A$ as a subring of $B$ from now on. Another necessary condition is that $f(X)=b_1X+b_0$ with $b_1\ne 0$. As an implication we have $f(a_nX^n+\ldots+a_0)=a_nb_1^nX^n+\ldots$, so what we additionally need is that multiplication with $b_1^n$ is injective on $A$. This is most easily achieved if $B$ has no divisors of zero at all ...

$\endgroup$

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.