9
$\begingroup$

A special value: $$ J\big(i\sqrt{6}\;\big) = \frac{(14+9\sqrt{2}\;)^3\;(2-\sqrt{2}\;)}{4} \tag{1}$$ I wrote $J(\tau) = j(\tau)/1728$.

How up-to-date is the Wikipedia listing of known special values for the modular j invariant ? Value (1) is not on it. Alternatively, is there a compilation of special values elsewhere?

The value (1) is related to this hypergeometric value, too: $$ {}_2F_1\left(\frac{1}{6},\frac{1}{3};1;\frac{1}{2}\right) = \eta(i\sqrt{6}\;)^2 \,2^{1/2} \,3^{3/4} \,(1+\sqrt{2}\;)^{1/6} \tag{2}$$ Here, $\eta(\tau)$ is the Dedekind eta function

(I came to this while working on my unanswered question at math.SE)

$\endgroup$

4 Answers 4

10
$\begingroup$

I don't think that it's too hard, in principle, to compute many such examples. The Wikipedia article lists a bunch, but it's not clear why they chose those particular ones. The point, of course, is that $j(\tau)$ with $\tau$ imaginary quadratic generates a proper ideal in an order in a ring class field associated to $\tau$ and $\mathbb Q(\tau)$. As long as this field isn't too large, a computer algebra system will get you generators for the ideal, and thence express $j(\tau)$ in terms of the generators.

I'll also note in passing that your $J(i\sqrt 6)$ is actually in the ring of integers $\mathbb Z[\sqrt2]$ of $\mathbb Q(\sqrt2)$, since it is equal to $$ J(i\sqrt6) = (9+7\sqrt2)^3(-1+\sqrt2). $$ Not sure if that's helpful for the application you have in mind.

$\endgroup$
2
  • $\begingroup$ What do you mean by "generates a proper ideal in an order"? I would think that an element generates an ideal of every ring it belongs to, in this case even of the whole ring of integers. In any case, can you give a hint on how to prove it is never a unit? $\endgroup$ Commented Jan 29, 2017 at 1:35
  • $\begingroup$ @FilippoAlbertoEdoardo Sorry, what I said isn't quite right. When doing CM and class field theory, one has to take account of the fact that the ideals in an order $\mathfrak o$ in the ring of integers of $\mathbb Q(\tau)$ need not be locally principal. So one looks at the class group consisting of proper (= "locally principal") $\mathfrak o$-ideals, modulo principal ideals. So if I remember correctly, the ideal of the appropriate order in the appropriate ring class field that is generated by $1$ and $j(\tau)$ will be locally principal. (And the unit ideal is locally principal.) $\endgroup$ Commented Jan 29, 2017 at 2:18
5
$\begingroup$

I do not know enough about CM (complex multiplication?) or class field theory to tell whether Joe's answer is sensible or feasible. And (so far) we have received no references as I had hoped. So eventually I came up with a proof, perhaps more elementary.

Write $\tau_1 = i\sqrt{6}/6$. Then $$ \frac{-1}{\tau_1} = i\sqrt{6} = 6\tau_1 . $$ Now both $j(\tau)$ and $j(6\tau)$ are modular functions for the group $\Gamma_0(6)$. Thus, they are algebraically related to each other. And since $$ j(6\tau_1) = j\left(\frac{-1}{\tau_1}\right) = j(\tau_1), $$ if we put these into the algebraic relation we get an algebraic equation for $j(\tau_1)$. Evaluating $j(\tau_1)$ numerically, we can tell which zero of that equation is the right one.

To determine the algebraic relationship between $j(\tau)$ and $j(6\tau)$, I chose the Hauptmodul $$ j_{6E}(\tau) = \frac{\eta(2\tau)^3\eta(3\tau)^9} {\eta(\tau)^3\eta(6\tau)^9} = q^{-1} + 3 + 6q + 4q^2 - 3 q^3 +\dots $$ Then $j(\tau)$ and $j(6\tau)$ are both rational functions of it. My computer calculates—writing $x=j_{6E}(\tau)$: \begin{align*} j(\tau) &= \frac{(x+4)^3(x^3+228 x^2 + 48 x + 64)^3} {x^2(x-8)^6(x+1)^3} \\ j(6\tau) &=\frac{x^6(8/x^3+12/x^2+6/x-1)^3(1-2/x)^3} {8/x^3+15/x^2+6/x-1} \end{align*} When $\tau=\tau_1$, these are equal. Equating them, we get an equation to solve for $x$. Of degree $10$. (It factors somewhat.)

Maple doesn't numerically evaluate eta functions directly. But we can write them in terms of theta functions $$ j_{6E}(\tau) = {\frac {{{\rm e}^{-2\,i\pi \,\tau}} \theta_4 \left( \pi \,\tau,{{\rm e}^{6\,i\pi \,\tau}} \right) ^{3} \theta_4 \left( \frac{3}{2}\,\pi \,\tau,{{\rm e}^{9\,i\pi \,\tau}} \right) ^{9}}{ \theta_4 \left( \frac{1}{2}\,\pi \, \tau,{{\rm e}^{3\,i\pi \,\tau}} \right) ^{3} \theta_4 \left( 3\,\pi \,\tau,{{\rm e}^{18\,i\pi \,\tau}} \right) ^{9}}} $$ which Maple does numerically evaluate. We get $$ j_{6E}(\tau_1) \approx 16.48528137423857 . $$ The only zero of our polynomial that matches this is $x = 8 + 6\sqrt{2}$. Plugging it in, we get \begin{align*} j(i\sqrt{6}\,) &= 2417472+1707264\sqrt{2} =2^6 \;3^3 \;(1+\sqrt{2}\,)^5\;(3\sqrt{2}-1)^3 \\ J(i\sqrt{6}\,) &= 1399+988\sqrt{2}= (1+\sqrt{2}\,)^5\;(3\sqrt{2}-1)^3 . \end{align*}

$\endgroup$
1
  • $\begingroup$ If you want a reference, a good one is ch. 2 of Advanced Topics in the Arithmetic of Elliptic Curves by Joseph Silverman (So yes, you can trust his answer above is sensible and feasible). CM theory tells you that $j(i\sqrt{6})$ will be an algebraic integer, and its Galois conjugates are at evaluations of $j$ at other roots of integer binary quadratic forms of discriminant $-24$. There is only one other equivalence class of forms; the conjugate is $j(i\sqrt{6}/2)\approx 3036.09664...$ From this it is easy to see that $j(i\sqrt{6})$ is a root of the polynomial $x^2 - 4834944x + 14670139392$ $\endgroup$ Commented Mar 16, 2017 at 20:48
4
$\begingroup$

The value of $\eta(i\sqrt{6})$ and $\eta(i\sqrt{3/2})$ involves the use of gamma function values on a 24 basis, so we have:

$$\eta(i\sqrt{6})=\frac{1}{2^{3/2}3^{1/4}}\big(\sqrt{2}-1\big)^{1/12}\frac{\Big(\Gamma\big(\tfrac{1}{24}\big) \Gamma\big(\tfrac{5}{24}\big) \Gamma\big(\tfrac{7}{24}\big) \Gamma\big(\tfrac{11}{24}\big)\Big)^{1/4}}{\pi^{3/4}}$$

$$\eta(i\sqrt{3/2})=\frac{1}{2^{5/4}3^{1/4}}\big(\sqrt{2}+1\big)^{1/12}\frac{\Big(\Gamma\big(\tfrac{1}{24}\big) \Gamma\big(\tfrac{5}{24}\big) \Gamma\big(\tfrac{7}{24}\big) \Gamma\big(\tfrac{11}{24}\big)\Big)^{1/4}}{\pi^{3/4}}$$.

For completeness, we have:

$$\eta(i\sqrt{2/3})=\frac{1}{2^{3/2}}\big(\sqrt{2}+1\big)^{1/12}\frac{\Big(\Gamma\big(\tfrac{1}{24}\big) \Gamma\big(\tfrac{5}{24}\big) \Gamma\big(\tfrac{7}{24}\big) \Gamma\big(\tfrac{11}{24}\big)\Big)^{1/4}}{\pi^{3/4}}$$.

[added GEdgar] also
$$ \eta(i\sqrt{1/6})= \frac{1}{2^{5/2}}\big(\sqrt{2}-1\big)^{1/12} \frac{\Big(\Gamma\big(\tfrac{1}{24}\big) \Gamma\big(\tfrac{5}{24}\big) \Gamma\big(\tfrac{7}{24}\big) \Gamma\big(\tfrac{11}{24}\big)\Big)^{1/4}}{\pi^{3/4}} $$

$\endgroup$
3
  • $\begingroup$ +1 This is great! Do you have a reference explaining this? $\endgroup$ Commented Jul 17, 2020 at 14:24
  • $\begingroup$ @Gerald Edgar- It's too long to explain! Soon, I'm going to publish a lot of other values like $\eta(i\sqrt{m/n})$. $\endgroup$ Commented Jul 17, 2020 at 18:23
  • 1
    $\begingroup$ I hope you do not mind if I add one more consequence. $\endgroup$ Commented Jul 17, 2020 at 20:11
3
$\begingroup$

You are at $z=i\sqrt{6}$. Hence you can use singular modulus $k_r$. It holds in general (when $z=i\sqrt{r}$) $$ j_r=\frac{256(k_r^2+(k'_r)^4)^3}{(k_rk'_r)^4}\textrm{, }k'_r=\sqrt{1-k_r^2}\textrm{, }\forall r>0. $$ But $k_6=\lambda^*(6)=(2-\sqrt{3})(\sqrt{3}-\sqrt{2})$ (see here). Hence you get your value.

$\endgroup$
1
  • $\begingroup$ This one was the simplest approach +1. Finding singular moduli is tough for large $n$ but for small $n$ it is doable by hand. $\endgroup$ Commented Jan 22, 2021 at 7:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .