Frobenius polynomials of rigid hypergeometric Calabi–Yau motives

edit · history · discussion · files · long url · hypergeometric motive L-function polynomial number theory
Polynomials
$\mathrm{template}$
$p$ 
$P^{\mathrm{rig}}_{A,p}(T)$
2.2.2.2
7:
343*T^2 - 24*T + 1
2.2.2.2
11:
1331*T^2 + 44*T + 1
2.2.2.2
13:
2197*T^2 - 22*T + 1
2.2.2.2
17:
4913*T^2 - 50*T + 1
2.2.2.2
19:
6859*T^2 - 44*T + 1
2.2.2.2
23:
12167*T^2 + 56*T + 1
2.2.2.2
29:
24389*T^2 - 198*T + 1
2.2.2.2
31:
29791*T^2 + 160*T + 1
2.2.3
7:
343*T^2 - 8*T + 1
2.2.3
11:
1331*T^2 + 36*T + 1
2.2.3
13:
2197*T^2 + 10*T + 1
2.2.3
17:
4913*T^2 + 18*T + 1
2.2.3
19:
6859*T^2 + 100*T + 1
2.2.3
23:
12167*T^2 + 72*T + 1
2.2.3
29:
24389*T^2 - 234*T + 1
2.2.3
31:
29791*T^2 + 16*T + 1
2.2.4
7:
343*T^2 + 24*T + 1
2.2.4
11:
1331*T^2 - 44*T + 1
2.2.4
13:
2197*T^2 - 22*T + 1
2.2.4
17:
4913*T^2 - 50*T + 1
2.2.4
19:
6859*T^2 + 44*T + 1
2.2.4
23:
12167*T^2 - 56*T + 1
2.2.4
29:
24389*T^2 - 198*T + 1
2.2.4
31:
29791*T^2 - 160*T + 1
2.2.6
7:
343*T^2 + 24*T + 1
2.2.6
11:
1331*T^2 - 28*T + 1
2.2.6
13:
2197*T^2 + 74*T + 1
2.2.6
17:
4913*T^2 + 82*T + 1
2.2.6
19:
6859*T^2 - 92*T + 1
2.2.6
23:
12167*T^2 + 8*T + 1
2.2.6
29:
24389*T^2 - 138*T + 1
2.2.6
31:
29791*T^2 - 80*T + 1
3.3
7:
343*T^2 + 25*T + 1
3.3
11:
1331*T^2 - 15*T + 1
3.3
13:
2197*T^2 - 20*T + 1
3.3
17:
4913*T^2 + 72*T + 1
3.3
19:
6859*T^2 - 2*T + 1
3.3
23:
12167*T^2 + 114*T + 1
3.3
29:
24389*T^2 + 30*T + 1
3.3
31:
29791*T^2 - 101*T + 1
3.4
7:
343*T^2 - 20*T + 1
3.4
11:
1331*T^2 + 1
3.4
13:
2197*T^2 + 70*T + 1
3.4
17:
4913*T^2 + 1
3.4
19:
6859*T^2 - 56*T + 1
3.4
23:
12167*T^2 + 1
3.4
29:
24389*T^2 + 1
3.4
31:
29791*T^2 - 308*T + 1
3.6
7:
343*T^2 + 1*T + 1
3.6
11:
1331*T^2 + 63*T + 1
3.6
13:
2197*T^2 + 28*T + 1
3.6
17:
4913*T^2 + 72*T + 1
3.6
19:
6859*T^2 - 98*T + 1
3.6
23:
12167*T^2 + 126*T + 1
3.6
29:
24389*T^2 - 126*T + 1
3.6
31:
29791*T^2 + 259*T + 1
4.4
7:
343*T^2 + 16*T + 1
4.4
11:
1331*T^2 - 40*T + 1
4.4
13:
2197*T^2 + 50*T + 1
4.4
17:
4913*T^2 + 30*T + 1
4.4
19:
6859*T^2 + 40*T + 1
4.4
23:
12167*T^2 + 48*T + 1
4.4
29:
24389*T^2 + 34*T + 1
4.4
31:
29791*T^2 + 320*T + 1
4.6
7:
343*T^2 - 12*T + 1
4.6
11:
1331*T^2 + 64*T + 1
4.6
13:
2197*T^2 - 58*T + 1
4.6
17:
4913*T^2 - 32*T + 1
4.6
19:
6859*T^2 - 136*T + 1
4.6
23:
12167*T^2 - 128*T + 1
4.6
29:
24389*T^2 + 144*T + 1
4.6
31:
29791*T^2 + 20*T + 1
6.6
7:
343*T^2 + 9*T + 1
6.6
11:
1331*T^2 + 17*T + 1
6.6
13:
2197*T^2 + 44*T + 1
6.6
17:
4913*T^2 - 56*T + 1
6.6
19:
6859*T^2 + 94*T + 1
6.6
23:
12167*T^2 + 50*T + 1
6.6
29:
24389*T^2 + 30*T + 1
6.6
31:
29791*T^2 + 139*T + 1
5
7:
343*T^2 - 6*T + 1
5
11:
1331*T^2 + 43*T + 1
5
13:
2197*T^2 + 28*T + 1
5
17:
4913*T^2 - 91*T + 1
5
19:
6859*T^2 + 35*T + 1
5
23:
12167*T^2 - 162*T + 1
5
29:
24389*T^2 - 160*T + 1
5
31:
29791*T^2 - 42*T + 1
8
7:
343*T^2 + 20*T + 1
8
11:
1331*T^2 + 14*T + 1
8
13:
2197*T^2 - 54*T + 1
8
17:
4913*T^2 + 66*T + 1
8
19:
6859*T^2 + 162*T + 1
8
23:
12167*T^2 + 172*T + 1
8
29:
24389*T^2 + 2*T + 1
8
31:
29791*T^2 - 128*T + 1
10
7:
343*T^2 + 6*T + 1
10
11:
1331*T^2 + 19*T + 1
10
13:
2197*T^2 - 12*T + 1
10
17:
4913*T^2 + 75*T + 1
10
19:
6859*T^2 + 91*T + 1
10
23:
12167*T^2 - 174*T + 1
10
29:
24389*T^2 + 272*T + 1
10
31:
29791*T^2 + 230*T + 1
12
7:
343*T^2 + 13*T + 1
12
11:
1331*T^2 + 65*T + 1
12
13:
2197*T^2 + 56*T + 1
12
17:
4913*T^2 + 108*T + 1
12
19:
6859*T^2 + 58*T + 1
12
23:
12167*T^2 - 66*T + 1
12
29:
24389*T^2 + 118*T + 1
12
31:
29791*T^2 - 145*T + 1
Definition
The polynomial $P^{\mathrm{rig}}_{A,p}(T)=\det(1-T\operatorname{Frob}_p\mid R_A)$ of geometric Frobenius on the canonically normalized pure rank-two, weight-three constituent $R_A$ at the classical coordinate $z=1$ of a classified hypergeometric Calabi–Yau family over $\mathbb Q$ [1]. The constituent is realized in the third cohomology of the resolved rigid mirror, has Hodge types $(3,0)$ and $(0,3)$, and is taken with no additional quadratic or Tate twist.
Parameters
$\mathrm{template}$
—   cyclotomic numerator multiset ($A$ is one of the fourteen cyclotomic numerator multisets in (3); the key lists its indices separated by dots)
$p$
—   unramified rational prime ($p$ is prime and $p\nmid N_A$, where $N_A$ is the conductor in (4))
Formulas
(1)
$P^{\mathrm{rig}}_{A,p}(T)=1-a_{A,p}T+p^3T^2$, where $a_{A,p}=\operatorname{Tr}(\operatorname{Frob}_p\mid R_A)$ is the $p$-th coefficient of the corresponding normalized weight-four newform in (4).
(2)
For $p>5$, $a_{A,p}\equiv\displaystyle\sum_{n=0}^{p-1}\frac{(\alpha_1)_n(\alpha_2)_n(\alpha_3)_n(\alpha_4)_n}{(n!)^4}\pmod{p^3}$ [1], Theorem 1. Here $(x)_n=x(x+1)\cdots(x+n-1)$ and $(x)_0=1$. The weight bound $a_{A,p}^2\leq4p^3$ determines the signed integral trace uniquely from this congruence.
Comments
(3)
A key such as 2.2.4 denotes the multiset $A=(2,2,4)$. Each index $d$ contributes every reduced fraction $j/d$ with $1\leq j<d$ and $\gcd(j,d)=1$, preserving multiplicity, to the four upper parameters $\alpha_1,\ldots,\alpha_4$. The four arithmetic lower residues are zero modulo integers, corresponding to the three analytic lower parameters $1,1,1$. Equivalently, the upper multiset is $\{r_1,1-r_1,r_2,1-r_2\}$ with the pairs in (4). These are the complete fourteen systems classified in [1].
(4)
The source-model correspondence uses the notation and equations of Tables 3–4 of [1]. The rigid motive is obtained from the resolved mirror at $z=1$, rather than from the entire cohomology of the displayed source variety. Here $X$ denotes the indicated hypersurface or complete intersection, $\mathbb P^n(w_0,\ldots,w_n)$ is weighted projective space, $N_A$ is the conductor, and $\chi_d(p)=(d/p)$ is the quadratic character at an odd prime, with $\chi_1$ trivial. The weight-four normalized newforms (listed in [3]) and extra characters are those of Table 1 of [1].

- 2.2.2.2: $(r_1,r_2)=(1/2,1/2)$; $X(2,2,2,2)\subset\mathbb P^7$, Table 4; $N_A=8$; newform 8.4.a.a; $\chi_A=\chi_{1}$.
- 2.2.3: $(r_1,r_2)=(1/2,1/3)$; $X(2,2,3)\subset\mathbb P^6$, Table 4; $N_A=36$; newform 36.4.a.a; $\chi_A=\chi_{3}$.
- 2.2.4: $(r_1,r_2)=(1/2,1/4)$; $X(2,4)\subset\mathbb P^5$, Table 4; $N_A=16$; newform 16.4.a.a; $\chi_A=\chi_{2}$.
- 2.2.6: $(r_1,r_2)=(1/2,1/6)$; $X(2,6)\subset\mathbb P^5(1,1,1,1,1,3)$, Table 4; $N_A=72$; newform 72.4.a.b; $\chi_A=\chi_{1}$.
- 3.3: $(r_1,r_2)=(1/3,1/3)$; $X(3,3)\subset\mathbb P^5$, Table 4; $N_A=27$; newform 27.4.a.a; $\chi_A=\chi_{1}$.
- 3.4: $(r_1,r_2)=(1/3,1/4)$; $X(3,4)\subset\mathbb P^5(1,1,1,1,1,2)$, Table 4; $N_A=9$; newform 9.4.a.a; $\chi_A=\chi_{6}$.
- 3.6: $(r_1,r_2)=(1/3,1/6)$; $X(6)\subset\mathbb P^4(1,1,1,1,2)$, Table 3; $N_A=108$; newform 108.4.a.a; $\chi_A=\chi_{3}$.
- 4.4: $(r_1,r_2)=(1/4,1/4)$; $X(4,4)\subset\mathbb P^5(1,1,2,1,1,2)$, Table 4; $N_A=32$; newform 32.4.a.a; $\chi_A=\chi_{1}$.
- 4.6: $(r_1,r_2)=(1/4,1/6)$; $X(4,6)\subset\mathbb P^5(1,1,2,1,2,3)$, Table 4; $N_A=144$; newform 144.4.a.f; $\chi_A=\chi_{2}$.
- 6.6: $(r_1,r_2)=(1/6,1/6)$; $X(6,6)\subset\mathbb P^5(1,2,3,1,2,3)$, Table 4; $N_A=216$; newform 216.4.a.c; $\chi_A=\chi_{1}$.
- 5: $(r_1,r_2)=(1/5,2/5)$; $X(5)\subset\mathbb P^4(1,1,1,1,1)$, Table 3; $N_A=25$; newform 25.4.a.b; $\chi_A=\chi_{5}$.
- 8: $(r_1,r_2)=(1/8,3/8)$; $X(8)\subset\mathbb P^4(1,1,1,1,4)$, Table 3; $N_A=128$; newform 128.4.a.b; $\chi_A=\chi_{2}$.
- 10: $(r_1,r_2)=(1/10,3/10)$; $X(10)\subset\mathbb P^4(1,1,1,2,5)$, Table 3; $N_A=200$; newform 200.4.a.f; $\chi_A=\chi_{1}$.
- 12: $(r_1,r_2)=(1/12,5/12)$; $X(12,12)\subset\mathbb P^5(1,1,4,6,6,6)$, Table 4; $N_A=864$; newform 864.4.a.a; $\chi_A=\chi_{1}$.
(5)
The normalization is cohomological/geometric Frobenius: its two eigenvalues have complex absolute value $p^{3/2}$ and determinant $p^3$. Thus the zeros of $P^{\mathrm{rig}}_{A,p}(T)$ have absolute value $p^{-3/2}$, and its Euler-product contribution is $P^{\mathrm{rig}}_{A,p}(p^{-s})^{-1}$. This is the constant-term-one convention, distinct from $\det(T-\operatorname{Frob}_p)$.
(6)
The specialization $z=1$ has a rank-three singular stalk. In the canonical rational hypergeometric normalization of [2], its finite-field trace $H_p(A;1)$ includes a one-dimensional character term: $H_p(A;1)=a_{A,p}+\chi_A(p)p$ for $p>5$ [1]. The table records the pure rank-two constituent, not the full stalk and not a $p$-adic unit root.
(7)
The geometric singularity of the fibre does not make every reduction prime arithmetically bad for $R_A$. Every selected prime $p>5$ is unramified for these constituents. Primes $2,3,5$ are outside the computed selection; some are individually good, as determined by $N_A$. Bad local factors are omitted, and no omitted factor is represented by $0$ or $1$.
Programs
(P1)
Sage
import numberdb.sage as numberdb
from sage.rings.integer_ring import ZZ
from sage.rings.polynomial.polynomial_ring_constructor import PolynomialRing
p = ZZ(37)
m = p**3
term = total = ZZ(1)
for n in range(int(p)-1):
    for j in (1, 2, 3, 4):  # upper parameters j/5, template 5
        term = (term*(j+5*n)*ZZ(5).inverse_mod(m)) % m
    term = (term*ZZ(n+1).inverse_mod(m)**4) % m
    total = (total+term) % m
a = total if 2*total <= m else total-m
assert a*a <= 4*p**3
R = PolynomialRing(ZZ, 'T')
R([1, -a, p**3])
References
[1]
L. Long, F.-T. Tu, N. Yui and W. Zudilin, Supercongruences for rigid hypergeometric Calabi–Yau threefolds, Theorems 1–2, Table 1 and Section 5.2; source models in Tables 3–4 (version 4). (arXiv)
[2]
F. Beukers, H. Cohen and A. Mellit, Finite hypergeometric functions, Definition 1.1, Theorem 1.3 and Assumption 1.4 (version 2). (arXiv)
Links
Similar tables
Frobenius polynomials of hypergeometric Calabi–Yau motives —   gives rank-four local factors of the same templates at smooth rational fibres $z=1/2,-1,2$; those fibres exclude the rigid specialization tabulated here
Normalized holomorphic periods of hypergeometric Calabi–Yau families —   gives the normalized analytic periods of the same templates, including their limits at $z=1$; these are different quantities from the local arithmetic factors
Canonical coordinates (mirror maps) of hypergeometric Calabi–Yau families —   gives canonical analytic coordinates of the same templates, including conifold limits; it does not identify an integral cohomology basis
Critical L-values of higher-level newforms —   includes critical values of the associated weight-four newforms at conductors $8,9,16,25,27$; its current level range does not include the other nine forms
Data properties
Entries are of type: integral polynomial
Table is complete: no (it holds all fourteen classified templates at every prime $7\leq p\leq31$, namely $7,11,13,17,19,23,29,31$, giving $112$ local quadratics. The method uses the $p>5$ hypothesis of the trace-congruence theorem. The upper prime cutoff is a noncanonical practical reference bound, providing several reductions and different splitting classes for every template before increasing prime depth. All signed traces, including zeros, are retained)
How they were obtained:

Exact modular rational arithmetic computes the truncated series in Theorem 1 of [1] modulo $p^3$. Every denominator is invertible for the selected primes. The residue is recovered as a signed integer and checked against $a_{A,p}^2\leq4p^3$; the inequality $p^6>16p^3$ makes recovery unique. No floating-point rounding is used.

more

Independent exact PARI newform computations reproduce all $112$ signed traces and complete polynomials. A coefficient prefix through index $16$ selects each source form, and its coefficients through index $99$ agree with the inspected exact LMFDB data. The unused primes $37,41$ provide further recognition controls.

A separately implemented finite-character-sum formula, Theorem 1.3 of [2], checks every trace, including nonsplit prime classes, after subtraction of $\chi_A(p)p$. The $384$-bit complex-ball computation requires finite real and imaginary radii below $10^{-80}$ and isolates a unique integer using the integrality of the geometric trace. Exact all-half quadratic-character convolutions check all eight selected primes. Known nonzero factors, all four zero traces, and deliberately reversed trace and incorrect character corrections control the comparisons. The attached evidence records software versions, actual bounds, source hashes and checks of values read from the draft.