How come only irreducible polynomials are considered when talking about field extensions?

83 Views Asked by At

I'm reading the section about field extensions in Dummit and Foote. It says that if you have a field $\mathbb{F}$ and a polynomial $p(x) \in \mathbb{F}[X]$ we can ask the question whether there exists an extension $\mathbb{K}$ of $\mathbb{F}$ containing a solution to $p(x) = 0$. Then it says that "we may assume here that the polynomial $p(x)$ is irreducible in $\mathbb{F}[X]$ since a root of any factor any factor $p(x)$ is certainly a root of $p(x)$ itself.

I'm having trouble understanding why you can assume that $p(x)$ is always irreducible because I think you can have reducible polynomials that don't have any roots over your field.

For example if we have $p(x) = (x^2 +1)(x^2 +3)$ and our polynomial ring is $\mathbb{R}[X]$ then $p(x)$ doesn't have any roots over $\mathbb{R}$ but it is reducible.

1

There are 1 best solutions below

0
On BEST ANSWER

As pointed out in the comments, if $\alpha$ is a root of $p(x)$, then it must be a root of one of its irreducible factors since we are working over a field. If $p(x) = \pi_1(x)^{e_1} \cdots \pi_t(x)^{e_t}$ is the prime factorization of $p(x)$ and $\alpha$ is a root of $p(x)$, then $$ 0 = p(\alpha) = \pi_1(\alpha)^{e_1} \cdots \pi_t(\alpha)^{e_t} \, . $$ Since we are working over a field (which is in particular an integral domain), then we must have $\pi_i(\alpha) = 0$ for some $i$. For instance in your example, if $0 = p(\alpha) = (\alpha^2 + 1)(\alpha^2 + 3)$, then either $\alpha^2 + 1 = 0$ or $\alpha^2 + 3 = 0$, i.e., $\alpha$ is either a root of the irreducible factor $x^2 + 1$ or the irreducible factor $x^2 + 3$.

The usual strategy for constructing a field extension containing a root of $p(x)$ also requires $p$ to be irreducible. If you look just below at the proof of Theorem 3 in $\S13.1$ (p. 512), you'll see that this field extension is defined as $F[x]/(p(x))$. Since $F[x]$ is a PID and $p(x)$ is irreducible, then $(p(x))$ is a maximal ideal so this quotient is indeed a field. On the other hand, quotienting out by the ideal generated by a reducible polynomial will not produce a field since it will contain zero-divisors.

One certainly can quotient by a reducible polynomial and still obtain an interesting object. If the polynomial is at least separable, the resulting quotient is called an étale algebra.

Edit: To respond to your question in the comments, it is certainly possible that adjoining the roots of one irreducible factor of $p(x)$ will not produce a field containing all roots of $p(x)$. However, we can adjoin the roots of the irreducible factors successively.

For example, consider the polynomial $p(x) = (x^2 - 2)(x^2 - 3) \in \mathbb{Q}[x]$. Adjoining the roots of the first irreducible factor produces the field $F := \mathbb{Q}(\sqrt{2}) = \frac{\mathbb{Q}[x]}{(x^2 - 2)}$, but one can show that $F$ does not contain the roots of $x^2 - 3$. So we repeat the process and adjoin these roots to $F$, obtaining $K := F(\sqrt{3}) = \frac{F[x]}{(x^2 - 3)} = \mathbb{Q}(\sqrt{2},\sqrt{3})$.