I am confused about a particular instance where adjoining an element of a field to itself makes it not equal to itself and I am asking for clarification. I can see the result is true, but I can not see why. We are not introducing any new element and we are not setting any new elements equal to zero.

In the finite field $\mathbb{F}_{11}$ we adjoin $\alpha$ where $\alpha^2 - 3 =0$. Because $(\pm 5)^2 -3 =0$ the two square roots of $3$ are already in $\mathbb{F}_{11}$, so we are either adjoining $5$ or $-5$. We do not know which, although both elements are invertible. However we cannot invert $\alpha +5$ because we don't know if $\alpha +5$ or $\alpha-5 = 0$ so $\mathbb{F}_{11}[\alpha]$ is not a field.

By adjoining an ambiguous element of the field to itself I thought maybe we were setting elements equal to zero, but it's not $5 = -5$ because then $10 = 0$ which makes every element $0$.

Source: This was an example in Algebra by Artin where Artin says $\mathbb{F}_{11}[\alpha] \simeq \mathbb{F}_{11}[x]/(x^2-3)$ is not a field on page 366.

Sorry for edits

Edit 2: If I am understanding what is being said, $\alpha$ must assume a specific value in $\mathbb{F}_{11}[\alpha]$. So it is $not$ true $\mathbb{F}_{11}[\alpha] \simeq \mathbb{F}_{11}[x]/(x^2-3)$ because the kernel of the evaluation homomorphism $\phi: \mathbb{F}_{11}[x] \rightarrow \mathbb{F}_{11}[\alpha]$ is not the ideal $(x^2-3)$, but one of the ideals $(x-5)$ or $(x+5)$.

I will rewrite a comment here hopefully to clarify.

In this image Artin describes $R'$ as "obtained by adjoining an element $\alpha$ to $\mathbb{F}_{11}$". A page earlier Artin defined "$R[\alpha] = \text{ring obtained by adjoining} \ \alpha \ \text{to} \ R$". There are also examples of using the evaluation homomorphism to show results such as $\mathbb{R}[x]/(x^2+1)\simeq \mathbb{Q}$ and $R[x,y] \simeq R[x][y]$.

Artin writes \begin{equation} R' = \mathbb{F}_{11}[x]/(x^2-3) \end{equation}

In the same paragraph he says "...procedure applied to $\mathbb{F}_{11}$ does not yield a field", and "But we haven't told $\alpha$ whether to be equal to $5$ or $-5$. We've only told that its square is $3$."

With this wording it sounds like the kernel of $\phi$ is not $(x-5)$ or $(x+5)$, but only $(x^2-3)$. Which again confuses me because then $\mathbb{F}_{11}[\alpha]$ is not a field.

  • 3
    $\begingroup$ I don't understand what your question is... Can you reformulate it in a precise way? $\endgroup$ Jun 15 '20 at 18:44
  • $\begingroup$ is this better? $\endgroup$
    – lamlame
    Jun 15 '20 at 18:52
  • $\begingroup$ Small remark: the fact that you cannot invert one of the two $(\alpha+5,\alpha-5)$, because one of them is $0$, doesn't mean that $\mathbb{F}_{11}[\alpha]$ is not a field. In order to conclude like that, you should find a non-zero element. In this particular case, $\mathbb{F}_{11}[\alpha]$ is indeed a field (which is isomorphic to $\mathbb{F}_{11}[X]/(X-\alpha)\sim\mathbb{F}_{11})$. $\endgroup$
    – quangtu123
    Jun 15 '20 at 18:55
  • $\begingroup$ And perhaps the title means "why isn't $\mathbb{F}_{11}[X]/(X^2-3)$ isomorphic to $\mathbb{F}_{11}$ eventhough $X^2-3$ has two solutions in $\mathbb{F}_{11}$?"? $\endgroup$
    – quangtu123
    Jun 15 '20 at 19:02

There are two different rings being discussed here:

  • $\mathbb{F}_{11}[\alpha]$, the ring obtained by adjoining some specific $\alpha\in \overline{\mathbb{F}}_{11}$ with $\alpha^2 = 3$;
  • $\mathbb{F}_{11}[X] / (X^2 - 3)$.

In the first case, $\mathbb{F}_{11}$ already contains $\alpha$; as you point out, $\alpha = \pm 5\in \mathbb{F}_{11}$. In the latter case, we no longer have a field: \begin{align*} \mathbb{F}_{11}[X]/(X^2 - 3) = \mathbb{F}_{11}[X]/(X - 5)\oplus \mathbb{F}_{11}[X]/(X + 5). \end{align*} If $f\in \mathbb{F}_{11}[X]$ is an irreducible nonconstant polynomial, then the map $\mathbb{F}_{11}[X]/(f) \to \mathbb{F}_{11}[\alpha]$ given by $X \to \alpha$, where $\alpha$ is a zero of $f$ in $\overline{\mathbb{F}}_{11}$, is an isomorphism; for then $\alpha$ is not a zero of any polynomial of degree less than $\deg f$, and comparing dimensions gives the result. That result doesn't hold without the irreducibility assumption, though.

  • $\begingroup$ Thank you for your answer. In the picture I linked Artin describes $R'$ as "obtained from adjoining an element $\alpha$ to $\mathbb{F}_{11}$ with the relation $\alpha^2-3 = 0$". He later says "we haven't told $\alpha$ whether to equal $5$ or $-5$. We've only told it it's square is $3$". With this wording it sounds like the substitution homomorphism $\phi : \mathbb{F}_{11}[x] \rightarrow \mathbb{F}_{11}[\alpha]$ does not have kernel $(x-5)$ or $(x+5)$, but only $(x^2 - 3)$. It seems in which case $F_{11}[\alpha] \simeq \mathbb{F}_{11}[x] / \ker \phi$ is not a field. $\endgroup$
    – lamlame
    Jun 15 '20 at 21:04
  • 1
    $\begingroup$ If Artin is describing adjoining a formal element satisfying $x^2 - 3$, he's presumably referring to $\mathbb{F}_{11}[X]/(X^2 - 3)$. Regardless, any ring $R\supset \mathbb{F}_{11}$ containing some $\alpha\not\in \mathbb{F}_{11}$ must not be a field; otherwise, the polynomial $X^2 - 3$ would have at most $2$ solutions. $\endgroup$
    – anomaly
    Jun 15 '20 at 21:13

The two square roots of $3$ in $F_{11}$ are $5$ and $6$. Adjoining an element of a field that's already in the field does not increase the size of the field, as you seem to understand.

I think perhaps you do not fully understand what $F[\alpha]$ means: it means "the smallest ring containing $F$ and $\alpha$." When $\alpha$ i algebraic over $F$, it conicides with the smallest field containing $F$ and $\alpha$ (and it is algebraic because it is a solution to $X^2-3$.)

Accordingly, $F_{11}[\alpha]=F_{11}[-\alpha]=F_{11}[\alpha,-\alpha]=F_{11}$.

By adjoining an ambiguous element of the field to itself I thought maybe we were setting elements equal to zero,...

I have a hard time deciphering that. When you adjoin an indeterminate (possibly what you mean by "ambiguous element"?) to $F$, you get $F[x]$, which is a domain but not a field. When $\alpha$ is a root of an irreducible polynomial $f(x)\in F[x]$, you can talk about the quotient $F[x]/(f(x))\cong F[\alpha]$ being a field extension of $F$.

Your case is no exception because $F_{11}[x]/(x-5)\cong F_{11}$ as we thought.

When $f(x)$ is reducible, $F[x]/(f(x))$ no longer yields a field extension of $F$, so it is useless to compare it to $F[\alpha]$ where $\alpha$ is a root of $f(x)$.

  • $\begingroup$ This is not what Artin means with adjoining an element to the field. If you check in the book, then the relevant chapter is about adjoining a universal solution of a polynomial to a ring, which by definition is the construction $R[x]/(f(x))$. The resulting extension is not a field extension, but it isn't intended to be - it's an extension of rings. $\endgroup$
    – Wojowu
    Jun 15 '20 at 22:16
  • $\begingroup$ @Wojowu ok: not having the resource, I am at a bit of a disadvantage. I’m a bit surprised he would still define that to be an extension, since it doesn’t correspond to a field extension. $\endgroup$
    – rschwieb
    Jun 15 '20 at 22:53

Here's why $\mathbb F_{11}[x]/(x^2-3)$ is not a field:

The notation $\mathbb F_{11}[x]$ denotes all polynomials with coefficients in $x$. The $/(x^2-3)$ part then means that you consider two polynomials as equivalent iff they differ by some multiple of $x^2-3$, and consider the equivalence classes of that equivalence relation.

Now every such class contains a polynomial of the form $ax+b$ where $a,b\in\mathbb F_{11}$, because in any higher-degree polynomial you can get rid of the highest power by using $ax^n \equiv ax^n - ax^{n-2}(x^2-3) = 3ax^{n-2}$, and by repeated application you can eliminate all terms of degree $2$ or higher.

What you cannot eliminate is the linear term, as there is no multiple of $x^2-3$ that is of degree $1$. That is, $[ax+b]=[cx+d]$ iff $a=c$ and $b=d$, where $[...]$ denote the equivalence classes of the relation above which contain the enclosed polynomial. In particular, $[ax+b]=[0]$ iff $a=b=0$.

So now you have a ring of $11^2=121$ elements, and we want to show that this is not a field. This is easily done by considering the product $(x+5)(x-5)$, which is a product of two non-zero terms: $$(x+5)(x-5) = x^2-5^2 = x^2-3 \equiv 0$$ Thus both $[x+5]$ and $[x-5]$ are zero divisors, which a field cannot have.

Basically what happened here is that we added a third square root of $3$ to our field (and also a fourth because also $[-x]^2=[3]$). But a field can only have two square roots of the same number (with the second one being the negative of the first), so adding yet another one leads to a non-field.


Your sequence

However we cannot invert $\alpha +5$ because we don't know if $\alpha +5$ or $\alpha-5 = 0$ so $\mathbb{F}_{11}[\alpha]$ is not a field.

is confusing.

Either $\alpha = 5$ and in this case $\alpha+5$ is invertible... And $\mathbb F_{11}[\alpha] = \mathbb F_{11}$ is a field. Or $\alpha = -5$ and in that case $\alpha+ 5=0$ is not invertible. But in that case still $\mathbb F_{11}[\alpha] = \mathbb F_{11}$ is a field.

  • $\begingroup$ But $\mathbb{F}[\alpha]$ isn't a field... $\endgroup$
    – lamlame
    Jun 15 '20 at 19:04
  • 1
    $\begingroup$ It is!!! How do you define $\mathbb{F}[\alpha]$? $\endgroup$ Jun 15 '20 at 19:04
  • $\begingroup$ If you check out Artin's book, you will find that $\alpha$ is defined as the residue of $X$ in the quotient $\mathbb F_{11}[X]/(X^2-3)$. So $\alpha$ is a square root of $3$ which is neither $5$ nor $-5$, in fact it doesn't like in (the image of) $\mathbb F_{11}$. $\endgroup$
    – Wojowu
    Jun 15 '20 at 22:22

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.