9

Let $n\in \mathbb{Z}$, $n\geq3$ and $d=n^2-2$. I want to show that $n^2-1+n\sqrt{d}$ is the fundamental unit in $\mathbb{Z}[\sqrt{d}]$.

Substituting $n=3,4,5$ gives the elements $8+3\sqrt{7}$, $15+4\sqrt{14}$ and $24+5\sqrt{23}$ respectively, which, by inspection, are the fundamental units of $\mathbb{Z}[\sqrt{7}]$, $\mathbb{Z}[\sqrt{14}]$ and $\mathbb{Z}[\sqrt{23}]$ respectively. So by empirical observation, the statement seems to be true, at least for the first few values of $n$.

My ideas so far on how to go about proving this have been the following: We know that any unit in $\mathbb{Z}[\sqrt{d}]$ must be expressible as some power of the fundamental unit, or the additive inverse of some power of the fundamental unit, and furthermore, we know that only the fundamental unit has this property. So if we can show that any unit in $\mathbb{Z}[\sqrt{d}]$ must be expressible in the form $\pm(n^2-1+n\sqrt{d})^r$ for some $r\in \mathbb{Z}$, this would prove that $n^2-1+n\sqrt{d}$ is the fundamental unit. However, I seem to be at a loss as to how to prove this, and I suspect that there may be a simpler proof of the statement.

All help or input would, as always, be highly appreciated.

Update: In response to a highly relevant comment, I add the following: If $d$ is not square free, the statement seems to fail. $\textit{E.g.}$ if we take $n=10$. Then $d=10^2-2=98+2*7^2$, and we get the element $99+10\sqrt{98}$, but the fundamental unit in $\mathbb{Z}[\sqrt{98}]$ is demonstrably $1+\sqrt{2}$.

In fact, the fundamental unit is only defined for $\mathbb{Z}[\sqrt{d}]$, when $\mathbb{Z}[\sqrt{d}]$ is the ring of integers of some quadratic number field. This is only the case when $d$ is square-free. We observe that as $d \not\equiv 1\ (\textrm{mod}\ 4)$ for all $n\geq3$, $\mathbb{Z}[\sqrt{d}]$ does indeed constitute the ring of integers of some quadratic number field whenever $d$ is square-free.

But I do still suspect that the above statement is true for $d$ square-free.

Heinrich Wagner
  • 2,897
  • 11
  • 24
  • Are you assuming that $n^2-2$ is squarefree? – Servaes Feb 16 '19 at 19:13
  • You are right. The statement seems to fail for $n=10$, when $n^2-2=98=2*7^2$. This gives the element $99+10\sqrt{98}\in \mathbb{Z}[\sqrt{98}]$, but the fundamental unit in $\mathbb{Z}[7\sqrt{2}]$ is demonstrably $1+\sqrt{2}$. So the statement seems to be untrue in the general case. But can we prove it for $d$ square-free? – Heinrich Wagner Feb 16 '19 at 19:31
  • @WillJagy Thanks for the tip. – Heinrich Wagner Feb 16 '19 at 19:39
  • I disagree; it does not fail for $n=10$ because $1+\sqrt{2}\notin\Bbb{Z}[7\sqrt{2}]$. I asked because when $d$ is not squarefree, there are a few more subtleties to take care of, as your example shows. – Servaes Feb 16 '19 at 19:41
  • @Servaes I suppose the problem is that the question is poorly defined. Fundamental units should be expressed in terms of rings of integers of number fields. In this case the ring of integers of $\mathbb{Q}(\sqrt{98})=\mathbb{Q}(\sqrt{2})$ is $\mathbb{Z}[\sqrt{2}]$, which has fundamental unit $1+\sqrt{2}$. In fact, I believe that the fundamental unit is not defined for rings $\mathbb{Z}[\sqrt{d}]$ in general, but only when these are the rings of integers of a quadratic number field. – Heinrich Wagner Feb 16 '19 at 19:47
  • 1
    Well since $\Bbb{Z}[\sqrt{d}]$ is certainly of finite index in the ring of integers of $\Bbb{Q}(\sqrt{d})$, the unit group is also of finite index. So the fundamental unit still makes sense; it is some power of the fundamental unit of the ring of integers. – Servaes Feb 16 '19 at 19:49
  • @Servaes Does this mean that $\textit{e.g.}$ in the case of $\mathbb{Q}(\sqrt{98})$, we have the orthodox definition of a fundamental unit of its ring of integers $\mathbb{Z}[\sqrt{2}]$ which is $1+\sqrt{2}$, but the subring $\mathbb{Z}[\sqrt{98}]$ of $\mathbb{Q}(\sqrt{98})$ also has a well-defined fundamental unit in its own right. If so, how can we find this fundamental unit? – Heinrich Wagner Feb 16 '19 at 19:59
  • @Servaes The trouble is that all the sources I consult seem to define fundamental units in terms of the ring of integers of number fields. $\textit{E.g.}$ the Wikipedia article on fundamental units (number theory) begins with the following sentence: "In algebraic number theory, a fundamental unit is a generator (modulo the roots of unity) for the unit group of the ring of integers of a number field, when that group has rank 1." Are you certain that fundamental units are defined for rings that are not the ring of integers of some number field? If so, can you refer me to any sources? – Heinrich Wagner Feb 16 '19 at 20:06
  • 1
    I can highly recommend this reader. It defines a system of fundamental units for an order $R$ as a minimal subset ${\eta_1,\ldots,\eta_k}\subset R^{\times}$ satisfying $$R^{\times}=\mu_R\times\langle\eta_1\rangle\times\cdots\times\langle\eta_k\rangle,$$ where $\mu_R$ denotes the group of roots of unity of $R$. It does so just after Theorem 5.13 (Dirichlets unit theorem), though fundamental units occur earlier in the reader already. – Servaes Feb 16 '19 at 20:28
  • It is worth mentioning that an order is defined there as a subring of a number field that is finitely generated as an abelian group. – Servaes Feb 16 '19 at 20:33
  • @Servaes This looks like a highly valuable resource, and I shall consult it at some later date so that I may determine whether or not I can prove the statement in the general case, using your definition of a system of fundamental units. But as for now, I think I shall stick to the former definition of fundamental units in terms of the ring of integers of a number field and try to prove the statement in the case of $d$ square-free, in which case $\mathbb{Z}[\sqrt{d}]$ does indeed constitute the ring of integers of a quadratic number field. – Heinrich Wagner Feb 16 '19 at 20:35
  • 1
    In the general case, the fact I suggested earlier (which is proved here) is of great help. – Servaes Feb 16 '19 at 20:38
  • You might find the following in Wolfram Alpha very instructive: Table[NumberFieldFundamentalUnits[Sqrt[n^2 - 2]] * Boole[MoebiusMu[n^2 - 2] != 0], {n, 3, 25}] – Robert Soupe Feb 23 '19 at 02:06

1 Answers1

1

You wrote "the fundamental unit in $\mathbf Z[\sqrt{98}]$ is demonstrably $1 + \sqrt{2}$" but this is wrong because $1+\sqrt{2}$ is not even an element of $\mathbf Z[\sqrt{98}] = \mathbf Z[7\sqrt{2}]$. When $A$ is a ring that is a finite free $\mathbf Z$-module (such as $\mathbf Z[\sqrt{m}]$ for all integers $m$ that are not perfect squares, even $m$ that have square factors bigger than $1$ such as $98 = 2 \cdot 49$), the unit group $A^\times$ is finitely generated and a set of multiplicatively independent generators of $A^\times$ modulo its roots of unity is called a set of fundamental units for $A$. If $A^\times$ is finite, such as when $A = \mathbf Z[i]$, there are no fundamental units.

In any case, a real quadratic ring $\mathbf Z[\sqrt{m}]$ turns out to have a unit group of the form $\pm u^\mathbf Z$ for a unique unit $u > 1$ and we call $u$ a fundamental unit for $\mathbf Z[\sqrt{m}]$. To show $n^2 - 1 + n\sqrt{n^2-2}$ is the fundamental unit of $\mathbf Z[\sqrt{n^2-2}]$ when $n \geq 3$, show it is the smallest unit greater than $1$. Start by showing no unit $a+b\sqrt{n^2-2}$ in $\mathbf Z[\sqrt{n^2-2}]$ can have $|b| < n$.

KCd
  • 46,062
  • and what happens after you have proved that it's the smallest unit greater than 1? What do you do then? – kubo Sep 15 '23 at 19:00
  • @kubo I'm not sure what you mean. Have you ever computed the unit group of the real quadratic ring $\mathbf Z[\sqrt{d}]$ for some specific values of $d$, such as $\mathbf Z[\sqrt{2}]$, $\mathbf Z[\sqrt{3}]$, and $\mathbf Z[\sqrt{10}]$? – KCd Sep 15 '23 at 23:03
  • Yes, I have. I just want to prove that any unit of $\mathbb{Z}[\sqrt{m}]$ can be written as a power of that fundamental unit, and I do not know how to proceed. – kubo Sep 16 '23 at 09:15
  • This is discussed in many elementary number theory books. Have you looked in books that discuss solutions to the Pell equation $x^2 - my^2 = 1$? The particular details in the case $m = n^2-2$ are not important. – KCd Sep 16 '23 at 21:16