Kernel of the following Substitution Homomorphism over $\mathbb{C}[x]$

340 Views Asked by At

I'm hoping someone could review my proof for accuracy, thanks! There is another proof of this on stack exchange, but it uses quotient rings, which we haven't learned yet.

Problem:

Let $\phi: \mathbb{C}[x,y]$ -> $\mathbb{C}[t]$, a the homomorphism that sends x -> $t + 1$ and $ y$ -> $t^3 - 1 $. Determine the kernel $K$ of $\phi$, and prove that every ideal of $\mathbb{C}[x,y]$ that contains $K$ can be generated by two elements.

Proof:

Note that $f = (x-1)^3 - 1 - y \in K$. Suppose there is some $g \in K$ such that $f$ does not divide $g$ in $\mathbb{C}[x,y]$. Then we have that $f$ is monic of degree $1$ when considered as a polynomial in the variable $y$.

Then we have $g = fq + r$, where deg(r) $\lt$ 1 (with respect to the variable $y$).

This implies that $r(x)$ is either a polynomial in $\mathbb{C}[x]$ that is non constant or is the $0$ polynomial. If it is nonconstant in $x$ then since r$\in$ K we must have that $x+1$ is a root of r.

Part I think is wrong: But this implies that the polynomial $r$ has infinite roots, as $x$ is a variable that can range over all of $\mathbb{C}$ for example.

Part I think may work better: Alternatively, plugging in x -> $x + 1$ will not send a non zero polynomial to zero as we are working in an integral domain and the substitution will maintain the degree of the polynomial as distributing $x+1$ will just produce extra terms of lower degree.

Hence we have r(x) must be the zero polynomial and so f generates all of K and so K is principle and $K$ = $ ( (x-1)^3 - 1 - y )$.

Then Let $I$ be an ideal that contains K. If I = (1), then I contains K and is generated by 1 element. Suppose $I$ is a proper ideal then.

New attempt:

Then for any $g$ $\in$ $I$ we have $g = fq + r$ when dividing over $y$. Then deg(r) $\lt$ $1$ and so it is either a constant or a polynomial over $x$, call it $r(x)$. Since $r(x)$ $\in$ $\mathbb{C}[x]$ then the set of all remainders is an ideal and is principle as it is polynomials in one variable over a field.

Though I’m not sure if we can claim the set of remainders is an ideal. It is certainly a subset of $I$, but I’m not sure if it is an ideal itself.

______________below this is wrong__________

Then let $g$ $\in$ I and suppose $f$ does not divide $g$. Then we can similarly divide $g$ by $f$ and conclude the remainder, $r(x)$, must be degree 0 in the variable y which implies it may be of non zero degree of $x$. Indeed it may be as considering $f$ as a function in the variable $x$ now we have that it is monic of degree 3. Then $g = fq + r$ is valid as an equation in the variable $y$, and the variable $x$. So we are dealing with the same $r(x)$ value as previously, and it must satisfy the additional constraint that the degree of $r(x)$ is less than $3$.

Hence any g $\notin$ K for g$\in$ I is of the form $g = fq + r$, where $r(x)$ $\in$ $\mathbb{C}[x]$. Then $r(x)$ $\in$ I also, as $I$ is closed under addition. Then the the lowest degree of $r$ is $0$, $1$ or $2$. $0$ implies $I$ is the whole ring. $1$ or $2$ implies we can generate $I$ with either $x$ or $x$$^2$ as $\mathbb{C}[x]$ is a ring with only ideals that are principle that are generated by lowest degree monic polynomials, and so $x$ or x$^2$ would suffice in that case.

In any case we have I = or , or <1>.

Hence the claim is shown.

1

There are 1 best solutions below

11
On BEST ANSWER

My review:

Overall, this is a good proof. The idea is mostly correct, though there are a couple problems.

Part you think is wrong vs. Part you think is better:

The part you think is wrong is actually correct, but could be phrased better. You have that $r(x)\in \newcommand\CC{\mathbb{C}}\CC[x]$ has the property that $r(t+1)=0\in\CC[t]$. Therefore, for all $\alpha\in\CC$, $r(\alpha+1)=0$, so $r$ has infinitely many roots in $\CC$, and must be the zero polynomial.

The part you think is better is also correct though, except you should say $x\mapsto t+1$. Don't mess up the variables, that will confuse your readers.

Your proof that $K=(f)$ is good.

Proving that if $K\subseteq I$, then $I$ is generated by at most two elements.

The idea for this part starts well, but quickly becomes confusing.

The first problem is how you conclude $r$ has degree less than $3$ in $x$. You appear to be dividing by $f$ twice. However the second division by $f$ might cause the degree of $y$ to go up. Thus you can't conclude $\deg r < 3$.

In addition, the degree of $r$ being $1$ or $2$ would not imply that $I = (f,x)$ or $(f,x^2)$. The ideals $(f,x)$ and $(f,x+1)$ are different!

Also it's not clear how you're concluding that there cannot be a third generator. You don't explicitly say that this is not possible. You implicitly conclude this from your attempted characterization of the ideals containing $K$, but you haven't proved that these are the only possible ideals.

How to fix this part

Let $I$ be an ideal containing $K$. $I$ is finitely generated. Let $g_1,g_2,\ldots,g_n$ be the fewest possible elements of $I$ such that $I$ is generated by $f$, $g_1$, $\ldots$, $g_n$. We will prove that $n=1$.

If $n\ge 2$, then it suffices to prove that there exists a polynomial $g_1'$ such that $(f,g_1,g_2)=(f,g_1')$. Then $I=(f,g_1,\ldots,g_n)=(f,g_1',g_3,g_4,\ldots,g_n)$, contradicting the minimality of the collection of generators.

Now by division, $g_1=fq_1+r_1(x)$ and $g_2=fq_2+r_2(x)$. Therefore $(f,g_1,g_2)=(f,r_1,r_2)$. Now $r_1$ and $r_2$ are single variable polynomials, so if $d=\gcd(r_1,r_2)$, we know $(r_1,r_2)=(d)$. Thus $(f,g_1,g_2)=(f,d)$. Taking $g_1'=d$, this completes the proof.

Indeed, from the proof, we can see that every ideal $I$ containing $K$ is either $(f)=K$, or of the form $(f,g)$, where $g$ has zero $y$-degree.