Minimal polynomials of CM coefficients of the mirror quartic K3 family

edit · history · discussion · files · long url · K3 surfaces modular function algebraic special values polynomial
Polynomials
$r$ 
$M_r(x)$
1:
x^4 - 81/32
2:
x^8 - 457/16*x^4 - 2401/1024
3:
x^8 - 3357/16*x^4 + 1185921/4096
4:
x^8 - 17901/16*x^4 - 15752961/8192
5:
x^16 - 39537/8*x^12 + 12691307/512*x^8 - 43470553/8192*x^4 + 1908029761/1048576
6:
x^16 - 150921/8*x^12 - 67788495/512*x^8 + 1485551097/8192*x^4 + 1836036801/1048576
7:
x^8 - 1036557/16*x^4 + 69257922561/65536
8:
x^16 - 1632557/8*x^12 - 23848812649/4096*x^8 - 280880583419/65536*x^4 - 620512135426561/67108864
9:
x^16 - 4798929/8*x^12 + 15221689155/512*x^8 - 181546911441/8192*x^4 + 6836598566721/1048576
10:
x^16 - 13301577/8*x^12 - 67544071263/512*x^8 + 931370015817/8192*x^4 + 39754550824641/1048576
Definition
For each positive integer $r$, let $\tau_r=i\sqrt r$, with the positive square root and $i^2=-1$, and set $R_r=(\eta(\tau_r)/\eta(2\tau_r))^{24}$ and $t_r=((R_r+128+4096/R_r)/256)^{1/4}>0$. The entry $M_r(x)\in\mathbb Q[x]$ is the monic minimal polynomial of the mirror-quartic coefficient $t_r$ defined in (6); Dedekind's eta function $\eta$ is defined in (1).
Parameters
$r$
—   rectangular CM index ($r>0$)
Formulas
(1)
For $\operatorname{Im}\tau>0$, $\eta(\tau)=q^{1/24}\prod_{n\geq1}(1-q^n)$, where $q=\exp(2\pi i\tau)$. Thus $R_r=q_r^{-1}\prod_{n\geq1}(1+q_r^n)^{-24}$, with $q_r=\exp(-2\pi\sqrt r)$. Equation (2.5) of [2] gives $t_r^{-4}=256R_r/(R_r+64)^2$.
(2)
Let $H_D(y)$ be the Hilbert class polynomial of the imaginary quadratic order of discriminant $D=-4r$, and let $h_D=\deg H_D$. With $m_r=4096/R_r$, $j(\tau_r)=(m_r+16)^3/m_r$ and $256t_r^4m_r=m_r^2+128m_r+4096$. Hence $m_r^{h_D}H_D((m_r+16)^3/m_r)=0$; these exact equations give an annihilating polynomial for $t_r$ by a resultant. The level-two equation uses the canonical normalization of the modular-polynomial table [4]. Here $j$ is the classical elliptic modular invariant.
(3)
In the zero-constant convention of the McKay-Thompson table, $T_{2A}(\tau_r)=R_r+24+4096/R_r=256t_r^4-104$. In particular $R_1=512$ and $M_1(x)=x^4-81/32$.
(4)
$\gamma_r^2=-4r$ and $\gamma_r\cdot(P(t_r)h-e+2P(t_r)^2f)=2(P(t_r)^2+r)=0$.
(5)
Let $\lambda_r=(\theta_2(\tau_r)/\theta_3(\tau_r))^4$, where $\theta_2(\tau)=\sum_{n\in\mathbb Z}e^{\pi i\tau(n+1/2)^2}$ and $\theta_3(\tau)=\sum_{n\in\mathbb Z}e^{\pi i\tau n^2}$. Then $t_r^{-4}=16(1-\lambda_r)\lambda_r^2/(\lambda_r-2)^4$ [3], equation (5.3). Yang's series variable $t=\psi^{-4}$ equals the present $t_r^{-4}$; its theta nome is $\exp(\pi i\tau)$.
Comments
(6)
Let $F_t:X_0^4+X_1^4+X_2^4+X_3^4-4tX_0X_1X_2X_3=0$ in $\mathbb P^3$, and let $X_t$ be the minimal resolution of $F_t/G$, where $G=\{(a_0,a_1,a_2,a_3)\in\mu_4^4:\prod_j a_j=1\}/\mu_4$. Here $\mu_4$ denotes the fourth roots of unity, with the denominator acting diagonally. The base coefficient is $t$, rather than the monomial coefficient $-4t$. The selected positive roots satisfy $t_r>1$, so they avoid $t^4=1$ and infinity [1] [2].
(7)
Use the marked period coordinate $P(t)$ on $t>1$. In the ordered integral basis $(h,e,f)$ of $T_0=\langle4\rangle\oplus U$, the only nonzero pairings are $h^2=4$ and $e\cdot f=f\cdot e=1$, and $[\Omega_t]=[P(t)h-e+2P(t)^2f]$. Here $U$ is the hyperbolic plane and $[\Omega_t]$ is the holomorphic two-form line. Take the positive imaginary lift with $P(t)\to i\infty$ as $t\to+\infty$, without modular reduction. Then $P(t_r)=i\sqrt r$. This projective ratio cancels a common differential scale [1] [3].
(8)
The lattice $\mathbb Z+\mathbb Z\tau_r$ has endomorphism order $\mathbb Z[\sqrt{-r}]$ of discriminant $D=-4r$. Square factors in $r$ are retained, so different conductors are distinct inputs. These are principal rectangular CM points, rather than all ideal classes or all CM fibres.
(9)
The normalization is monic over $\mathbb Q$, not primitive integral. The polynomial describes the selected positive coefficient exactly; its other conjugate roots are not asserted to satisfy the same marked-period condition. Elimination polynomials are factored and the appropriate irreducible factor is selected using the positive embedding.
(10)
The marked integral vector $\gamma_r=e-2rf$ is primitive, has square $-4r$ and is orthogonal to $\Omega_{t_r}$ (4). Hartmann's primitive embedding and the Lefschetz $(1,1)$ theorem give an additional divisor class outside the generic algebraic lattice. The attached lattice certificate gives the exact pairings. It supplies neither effective curve equations, a polarization degree bound nor the saturated Neron-Severi lattice.
Programs
(P1)
Sage
import numberdb.sage as numberdb
from sage.rings.rational_field import QQ
from sage.rings.real_arb import RealBallField
from sage.rings.polynomial.polynomial_ring_constructor import PolynomialRing
from sage.libs.pari import pari
r = 1  # any positive integer; the stored reference slice is 1 through 10
S = PolynomialRing(QQ, 'x'); x = S.gen()
A = PolynomialRing(S, 'm'); m = A.gen()
H = S(pari(-4*r).polclass()); h = H.degree()
E = sum((H[i]*(m+16)**(3*i)*m**(h-i) for i in range(h+1)), A(0))
F = S(E.resultant(256*x**4*m-m**2-128*m-4096))
B = RealBallField(600); q = (-2*B.pi()*B(r).sqrt()).exp()
N = 100
v = sum(((1+q**n).log() for n in range(1,N+1)), B(0))
v = v.add_error(q**(N+1)/(1-q))
R = (-24*v).exp()/q
t = ((R+128+4096/R)/256).sqrt().sqrt()
factors = [f.monic() for f,e in F.factor()]
chosen = [f for f in factors if f(t).contains_zero()]
assert t.is_finite() and t > 1 and t.accuracy() > 450 and len(chosen) == 1
chosen[0]  # x^4 - 81/32 for r=1
References
[1]
Heinrich Hartmann, Period- and Mirror-maps for the Quartic K3 (2011), Sections 4.1, 4.3 and 4.7, Theorems 4.8, 4.11 and 4.29 and equation (10). (arXiv)
[2]
Lassina Dembele, Alexei Panchishkin, John Voight and Wadim Zudilin, Special hypergeometric motives and their L-functions: Asai recognition (2020), equation (2.5). (arXiv)
[3]
Wenzhe Yang, K3 mirror symmetry, Legendre family and Deligne's conjecture for Fermat quartic, Nuclear Physics B 959 (2020), 115303, equations (4.29), (4.33), (5.3), (5.15)-(5.17). (arXiv) (doi)
Links
Similar tables
CM coefficients of the mirror quartic K3 family —   gives the positive real roots whose monic minimal polynomials are listed here, with the same rectangular CM index
Period map of the Dwork quartic K3 family —   gives the marked coordinate at rational coefficients; the selected coefficients here have prescribed periods i*sqrt(r)
Minimal polynomials of monstrous McKay-Thompson CM values —   its 2A polynomials give annihilating polynomials after substituting 256*x^4-104; irreducible factor selection remains necessary
Hilbert class polynomials —   supplies the classical CM j polynomials used in exact elimination
Canonical modular polynomials —   its level-two eta quotient relates the elliptic j invariant to the quartic coefficient
Data properties
Entries are of type: rational polynomial
Table is complete: no (it holds every integer 1 <= r <= 10 in the principal rectangular order family of discriminant -4r, retaining square factors and conductors; ten is a noncanonical arithmetic reference cutoff, and larger r, nonrectangular CM points and other ideal classes are not covered)
How they were obtained:

The generator uses PARI polclass with its default j invariant [5], exact rational resultants and irreducible factorization. A bounded eta product at 600 bits identifies the positive embedding and excludes every other factor; exact rational endpoint signs and a derivative enclosure certify a unique simple root.

more

The attached checks use a different elimination through the modular lambda function and independently bounded theta series for all ten entries, together with inverse-period AGM and exact lattice pairings. Exactness concerns the polynomial coefficients, while the root enclosures are auxiliary certificates. These build checks share PARI class polynomials and Sage/Arb; they are not an independent review verdict.