6
$\begingroup$

Let $\eta(\tau)$ be the Dedekind eta function and $f(\tau)=C\frac{\eta(\tau)\eta(4\tau)}{\eta(2\tau)^4}$, with the constant $C=2^{1/4}\eta(i)^2$. My questions concern the minimal polynomials of $f(\frac{i}{2p})$, $p$ an odd prime. The minimal polynomials seem to exhibit simple patterns (see below) and I would like to understand this better. Not being very familiar with algebraic number theory, my questions are: 1) are patterns like these expected? 2) if so, is there an intuitive understanding of them? 3) recommended reading?

Let $M_p(x)$ be the minimal polynomial of $f(\frac{i}{2p})$ when $p\equiv 1 \pmod 4$, and let it be the minimal polynomial of $f(\frac{i}{2p})^4$ when $p\equiv 3 \pmod 4$. A few examples of such minimal polynomials, so we are on the same page:

$$M_{13}(x)=x^3-9x^2+x-1,$$

$$M_{41}(x)=x^{10} - 168222x^9+194609x^8+41520x^7-47346x^6-25332x^5-4530x^4-48x^3+113x^2+18x+1,$$

$$M_{19}(x)=x^5-17041207x^4-236870x^3-6142x^2-11x-\frac{1}{19}.$$

The patterns I am referring to concern the degree 1 coefficient (most surprising to me), the constant term, and the degree of the minimal polynomials.

Degree

The degree of $M_p(x)$ is $\frac{p-1}{4}$ when $p\equiv 1 \pmod 4$ and $\frac{p+1}{4}$ when $p\equiv 3 \pmod 4$.

Constant Term

When $p\equiv 1 \pmod 4$, $M_p(0)=1$ if $p=x^2+32y^2$ for some natural numbers $x$ and $y$, otherwise $M_p(0)=-1$. So, the first $p$ with constant term $1$ are $41, 113, 137, 257, 313, 337, 353$.

When $p\equiv 3 \pmod 4$, $M_p(0)=-\frac{1}{p}$.

Degree 1 Coefficient

For $p\equiv 1\pmod 4$, let $p=even^2+odd^2$, where $even$ and $odd$ are even and odd natural numbers, respectively.

When $p\equiv 5\pmod 8$, the degree 1 coefficient of $M_p(x)$ is $\frac{even}{2}(-1)^{even/4-1/2}$.

When $p\equiv 1\pmod 8$, the degree 1 coefficient of $M_p(x)$ is $$2\left\lfloor \frac{odd}{4}+\frac{1}{4}\right\rfloor(-1)^{\lfloor odd/4-1/4\rfloor}+h(-4p^2)(-1)^{(p+7)/8},$$ where $h(d)$ is the class number for discriminant $d$. So, the degree 1 coefficients for $p=5,13,17,29,37,41,53,61,73,89,97,\ldots,353$ are $1,1,-8,1,-3,18,1,-3,38,42,-44,\ldots,-168$.

I haven't found a simple result yet for the degree 1 coefficient when $p\equiv 3 \pmod 4$, though I haven't looked much. The degree 2 coefficients for $p\equiv 1 \pmod 4$ are very suggestive, but I haven't found the pattern, yet, either.

Are these types of formulas typical for eta quotients? I haven't found many references to general properties of eta quotient minimal polynomials, mainly computations for specific values. Is there any intuition behind any of the details, such as the role of $x^2+32y^2$ in the constant term or the dependence of the degree 1 coefficient on the class number when $p\equiv 1 \pmod 8$, but not when $p \equiv 5 \pmod 8$?

It is important to note that I am not claiming to have proven any of the above, merely to have observed it after finding putative minimal polynomials for $p<200$ and $p\leq 353$ for $p\equiv 1 \pmod 4$ using PARI/GP's algdep function and high precision (e.g., 50,000 digits). (Don't know whether it's worthwhile to share all those minimal polynomials here, but it's straightforward to find them with PARI/GP.)

Thanks!

$\endgroup$

1 Answer 1

1
$\begingroup$

Observe that $$4\dfrac{\eta(2\tau)^4}{\eta(\tau)\eta(4\tau)}=\sum_{x,y}\genfrac(){}{0}{24}{x}\genfrac(){}{0}{12}yq^{(x^2+2y^2)/24}$$ where $\genfrac(){}{0}ab$ is the Kronecker-Legendre symbol. Your observations should then easily follow from the excellent book of Cox on primes of the form $x^2+ny^2$.

Edit: sorry, I was more thinking about the coefficients. Still, it should be a standard complex multiplication matter.

$\endgroup$
1
  • 1
    $\begingroup$ Thanks very much, will check Cox's book. And thank you for the wonderful PARI! $\endgroup$
    – ljk
    Oct 18, 2017 at 16:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.