Let $K$ be a number field and $O_K$ be its ring of integers. Take $\alpha \in K$ and consider the principal fractional ideal $(\alpha)$. I am sure the following is true, but I couldn't quite prove it. I was wondering if someone could give me a hint on how I can find $\beta, \gamma \in O_K$ so that $\alpha = \beta/ \gamma$ and $(\beta) + (\gamma) = 1$. Thank you very much!

  • $\begingroup$ Can you show that every element of $K$ is the quotient of two elements of $O_K$? Can you show that that implies what you want? $\endgroup$ – Greg Martin Oct 16 '15 at 21:14
  • $\begingroup$ @GregMartin Yes, I can show that every element in $K$ is a quotient of two elements of $O_K$. But I am not seeing how this implies what I want yet. Could I get a hint? Thank you! :) $\endgroup$ – Johnny T. Oct 17 '15 at 21:46

This may not always be possible, take, for example $\mathcal{O}_k=\Bbb Z[\sqrt{-5}]$. Then consider the ideal generated by $\alpha={1+\sqrt {-5}\over 2}$. Both numerator and denominator are irreducibles, but due to lack of unique factorization we find both divide $6$, i.e.

$$\begin{cases} (1+\sqrt{-5})(1-\sqrt{-5})=6 \\ 2\cdot 3 = 6\end{cases}.$$

To see irreducibility it is sufficient to show there are no ideals of norm $2$ or $3$, which is readily verified by noting that the norm $N(a+b\sqrt{-5})=a^2+5b^2$ which clearly is never equal to $2$ or $3$ for integer $a,b$. But then it must be that $(1+\sqrt{-5})\supseteq (6)$ and therefore is contained in some maximal ideal containing $6$, but such a maximal ideal, $\mathfrak{p}$, must lie above $(2)$ or $(3)$ In fact it lies in one for each of them, $\mathfrak{p}_2=(2,1+\sqrt{-5}),\mathfrak{p}_3=(3,1+\sqrt{-5})$. This can also be seen by reducing the minimal polynomial for $1+\sqrt{-5}$ modulo $2$ and $3$. Knowing that $2$ is ramified, this means $\mathfrak{p}_2$ is the unique ideal above $2$ in $k=\Bbb Q(\sqrt{-5})$. We can see now that $(1\pm\sqrt{-5})+(2)\subseteq\mathfrak{p}_2$ so that they are not coprime. Even multiplying by conjugates we get

$${(1+\sqrt{-5})(1-\sqrt{-5})\over 2(1-\sqrt{-5})}={3\over 1-\sqrt{-5}}$$

which gives an expression where we have $(3)+(1-\sqrt{-5})=\mathfrak{p}_3'$, the other maximal ideal containing $3$ which still has them non-coprime integer ideals.

In general, say you have

$$\alpha = {\beta\over\gamma}\cdot{\delta\over\delta}$$

for some $\delta\in K$ such that $\delta\beta, \delta\gamma\in\mathcal{O}_K$. Then we ask if we may choose a $\delta$ so that $(\delta\beta)+(\delta\gamma)=\mathcal{O}_K$.

We know that $(\beta)=\mathfrak{p}\mathfrak{q}_1,(\gamma) =\mathfrak{p}\mathfrak{q}_2$. Then $\delta\in \mathfrak{p}^{-1}$ by definition, i.e.


and, as usual, this expression is unique with all $e_i>0$. Then we have


Then clearly for this to be equal to $\mathcal{O}_K$ we must have $r=0$, i.e. there are no other primes, and we already know that $\mathfrak{q}_1+\mathfrak{q}_2=\mathcal{O}_K$. But then this implies $(\delta)=\mathfrak{p}^{-1}$, which would imply that $\mathfrak{p}^{-1}$ is a principal fractional ideal in the ideal class group. However we know that $\mathfrak{p}$ is a non-principal ideal, and since the ideal class group is cyclic of order $2$, we have that the ideal class for $\mathfrak{p}$ is the same as that for $\mathfrak{p}^{-1}$. However, $\mathfrak{p}^{-1}=(\delta)$ is a contradiction, since the latter is a principal ideal. Hence no such expression exists.

  • $\begingroup$ Perhaps I am missing something... How does this show that $\alpha$ you have given me can not be expressed as $\beta / \gamma$ with $(\beta)+(\gamma) = (1)$? $\endgroup$ – Johnny T. Oct 22 '15 at 22:12
  • $\begingroup$ @JohnnyT. I have expanded my answer to include the full details. Let me know if anything else needs explanation. If you don't already know what the class groups is, that's a necessary part of the proof, so I can edit to give a quick primer enough to give you the conceptual picture as to how it implies the lack of existence in this case. Cheers. $\endgroup$ – Adam Hughes Oct 26 '15 at 23:46
  • $\begingroup$ Yes, thank you! I am just going through the details at the moment! $\endgroup$ – Johnny T. Oct 27 '15 at 18:59

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.