Let $\mathbb{Z}[\sqrt{3}]$ be the ring $\{a + b\sqrt{3} | a, b \in \mathbb{Z}\}$ with a natural structure. We consider it's norm:

$$a + b\sqrt{3} \mapsto a^2 - 3b^2$$

Then, finding the solution to the Pell's diophantine equation $x^2 - 3y^2 = 1$ amounts to find the multiplicative kernel of the norm map $\mathbb{Z}[\sqrt{3}] \to \mathbb{Z}$. This kernel is easily found as it is contained in $\mathbb{Z}[3]^\times$, the group of units of this ring. I'm stuck at trying to prove the following result, which amounts to the existence of a fundamental solution to the equation:

There is an isomorphism $\mathbb{Z}[\sqrt{3}]^\times \simeq \mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}$

Where the torsion part corresponds to the roots of unity in $\mathbb{Z}[\sqrt{3}]^\times$, which are only $\{ \pm 1\}$, and the torsion-free part is given by the powers of the fundamental solution.

Here is what I've done so far: a friend of mine gave me a partial proof of the Dirichlet Unit Theorem (from which the proposition above is a special case). I tried to adapt it, and make it work for this ring using simpler arguments. Let us consider the map: $$f\colon \mathbb{Z}[\sqrt{3}]^\times \to \mathbb{R}$$ $$z \mapsto \log|z|$$ where $| \cdot |$ here represents the absolute value function on the real numbers. This application is a group homomorphism, from the multiplicative group of the ring to the additive group of the reals. It's kernel is precisely the torsion part of the group of units, so I just need to show that it's image is cyclic. Now, let $z_1$ and $z_2$ be two units in $\mathbb{Z}[\sqrt{3}]^\times$. Then, it is enough to show that there exists a rational number $\frac{p}{q}$ such as that $z_1 = \pm(z_2)^{\frac{p}{q}}$. Now, because we have $\dim_\mathbb{R}\mathbb{R} = 1$ as a real vector space, we know that there exists real numbers $k_1$ and $k_2$ that satisfy: $$k_1\log |z_1| + k_2\log |z_2| = 0$$ This implies that $z_1 = \pm (z_2)^{\frac{k_1}{k_2}}$ inside $\mathbb{R}$. Now here comes the tricky part: though $z_1$ and $z_2$ are (possibly) irrational numbers, they are very controlled irrational numbers. Seems intuitive that for a real power of $z_2$ to still lie in $\mathbb{Z}[\sqrt{3}]$, this power must actually be a rational power (and a very special one at that, given that roots are scarce in this ring). This however, seems very non-trivial and outside the realm of the number theory. Is this true? Is there some "elementary" proof of this fact? Otherwise, how can we characterize the units of $\mathbb{Z}[\sqrt{3}]$ without resorting to the full power of the Dirichlet Unit Theorem?


I think it is worth pointing out how close you were to finishing off your own argument!

The key insight is that is any non-trivial subgroup of $\mathbb R$ is either dense or isomorphic to $\mathbb Z$. (See here and here.) So we must prove that the image of your homomorphism $f : \mathbb Z[\sqrt 3]^\times \to \mathbb R$ is not dense in $\mathbb R$.

To prove this, pick any $\delta > 0$. It is enough to prove that the set $f^{-1}(-\delta, \delta)$ is a finite set.

If $z = a + b\sqrt {3}$ is a unit such that $\log |z| \in (-\delta, \delta)$, then $$ | a + b \sqrt{3}|<e^\delta, \ \ \ \ \ \ \ \ | a - b \sqrt 3 | < e^\delta.$$ [Note that $z$ is a unit, so $|a^2 - 3b^2| = |N(z)| = 1$, hence $-\log |z| = \log|z^{-1}| = \log|a - b \sqrt{3}|$.]

Therefore, $ | a | < e^\delta$ and $|b| < \frac{e^\delta}{\sqrt 3},$ by the triangle inequality. Since $a, b \in \mathbb Z$, there are only finitely many possible $a$ and $b$ satistfying these inequalities, and we're done.


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.