The integers of $ \mathbb{Q}(\sqrt{2})$

Let $ \alpha=\sqrt{2}$. The quadratic number field $ \mathbb{Q}(\sqrt{2})$ is the field

$\displaystyle {\mathbb{Q}}(\alpha)=\{x+\alpha y\ \vert\ x,y\in\mathbb{Q}\},
$

which we denote by $ K$ for short. The integers of $ K$ is the ring

$\displaystyle {\mathbb{Z}}[\alpha]=\{a+\alpha b\ \vert\ a,b\in {\mathbb{Z}}\}
={\mathbb{Z}}+\alpha {\mathbb{Z}}.
$

Definition 2.1.11   An element $ p=a+\alpha b\not= 0$ is a prime (element of $ \mathbb{Z}[\alpha ]$) if it has the following property: for all $ r,s\in \mathbb{Z}[\alpha ]$, $ p\vert rs$ then $ p\vert r$ or $ p\vert s$.

Is there an analog of Theorem 2.1.9 for primes in this field?

Yes. Before describing them, we need to introduce the notion of an associate.

Definition 2.1.12   If $ R$ is an commutative ring with unit $ 1$ and $ u\in R$ has the property that $ uv=1$, for some $ v\in R$ then we call $ u$ a unit of $ R$ and call $ v$ the inverse of $ u$. . The set of units of $ R$ is denoted $ R^\times$.

Let $ x,y\in R$. We say $ x$ is an associate of $ y$ if $ x=uy$, for some $ u\in R^\times$. The set of associates of $ x$ is denoted $ xR^\times$.

There is an analog of the fundmental theorem of arithmetic for $ \mathbb{Z}[\alpha ]$. In other words, each element in $ \mathbb{Z}[\alpha ]$ can be uniquely (up to order and ``unit'' factors of the form $ \pm (1+\sqrt{2})^k$, $ k\in \mathbb{Z}$) factored into a product of prime elements. In other words, $ \mathbb{Z}[\sqrt{2}]$ is a unique factorization domain.

Example 2.1.13 (a)   If $ R=\mathbb{Z}$ then $ R^\times =\{1,-1\}$.

(b) If $ R=\mathbb{Z}[i]$ then $ R^\times =\{1,-1,i,-i\}$.

(c) if $ R=\mathbb{Z}[\alpha ]$ then

$\displaystyle R^\times =\{ \pm (1+\sqrt{2})^k\ \vert\ k\in \mathbb{Z}\}.
$

The proof of this goes beyond the scope of this book. See for example, Hardy and Wright, [HW], §14.5. (See the exercises below for further examples.)

The following characterization of primes is known.

Theorem 2.1.14   An element $ a+\alpha b\not= 0$ is a prime element of $ \mathbb{Z}[\alpha ]$ if and only if either

(a) $ a=0$ and $ b=\pm 1$,

(b) $ b=0$ and $ \vert a\vert$ is a prime integer which is $ \equiv \pm 3\, ({\rm mod}\, 8)$, or

(c) $ ab\not= 0$ and $ a^2-2b^2$ is a prime integer which is $ \equiv 1\, ({\rm mod}\, 8)$,

(d) $ a+\alpha b$ is an associate of an element as in (a), (b) or (c).

Remark 2.1.15   If we replace $ \alpha=\sqrt{2}$ by the root of some higher degree polynomial in $ \mathbb{Z}[x]$ then there is, in general, no analog of Theorem 2.1.14 for $ \mathbb{Q}(\alpha)$.

In fact, the search for an analog is one of the motivations for the very deep Langlands philosophy [Art], §1.

If we replace $ \alpha=\sqrt{2}$ by the root of some higher degree polynomial in $ \mathbb{Z}[x]$ then there is, in general, no analog of the fundamental theorem of arithmetic. The fundmental theorem of arithmetic fails for the integers in $ \mathbb{Q}(\sqrt{-5})$ and for the integers in $ \mathbb{Q}(\sqrt{10})$, for example. (See Hardy and Wright, §§14.6 for more details.)



David Joyner 2007-09-03