Skip to content

About

Oppermann intervals in F_q[t]: characteristic-2 counterexamples with exact counts for d ≤ 3, an exact d = 2 formula for odd q, and a Hayes-FFT census

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

23 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Oppermann intervals in F_q[t]

Galois obstructions, exact counts for small degree, and the Oppermann subgroup in characteristic 2

Ruqing Chen. Version 2 (November 2026), DOI: 10.5281/zenodo.24211741 (version 1: 10.5281/zenodo.23353949).

New in version 2: level-3 L-functions in characteristic 2 (Theorem 7.7), coset sums on H_d (Proposition 7.8), and the mean drift on H_d for d = 6 and d = 7 for every q = 2^k (Theorems 7.10, 7.11), so the drift is known in closed form for all d ≤ 7.

Paper: paper/oppermann_function_fields.tex / .pdf (30 pages).

For monic f of degree d, the Legendre interval {f² + s : deg s ≤ d} splits into q Oppermann sub-intervals

O_{f,c} = {f² + c·f + s : deg s < d}, c ∈ F_q,

each of size q^d. For q = 2 these are the two halves, the analogue of (n², n²+n) and (n²+n, (n+1)²). Does every O_{f,c} contain an irreducible polynomial?

Main results

Characteristic 2: no. The failures are explained by three Galois obstructions (§3).

  • (A) d = 1: the c = 0 piece consists of squares.
  • (B) d = 2: Frobenius acts on the roots of an additive polynomial through AGL₂(F₂) ≅ S₄. The exact counts are q²/4, 0, q(q−2)/4 and q(q−1)/2. Empty sub-intervals occur exactly for q = 2 and q = 4^j.
  • (C) q = 2, d = 4: the Stickelberger–Swan sign of Frobenius obstructs.
  • Dead zone (§3.4): for r ≥ 3, AGL_r(F₂) contains no 2^r-cycle. Hence no affine polynomial of degree 2^r ≥ 8 is irreducible, and the mechanism of (B) does not extend to d ≥ 4.

Exact counts for small degree

  • Odd q, d = 2 (§4): a ternary trace form on F_{q⁴} gives π = (q² − 1 − (q − η(−1))η(2e))/4 > 0.
  • Characteristic 2, d = 3 (§5): a count in the field tower F_q ⊂ F_{q³} ⊂ F_{q⁶}, together with a cubic Gauss sum, gives every class. Let c be the t³-coefficient and ε = (−1)^{k/2}.
    • k odd: π = q(q² − 1)/6 for every class.
    • k even, c = 0: π = q(q − 1)²/6.
    • k even, c a cube: π = q(q² + 1 − 2ε√q)/6.
    • k even, c a non-cube: π = q(q² + 1 + ε√q)/6.

Large q and computation (§6)

  • Large q: for fixed d ≥ 3, Bank–Bary-Soroker–Rosenzweig together with the d = 3 count gives nonemptiness once q is large.
  • Odd q: exact Hayes-FFT counts find every sub-interval nonempty for q = 3, d ≤ 16; q = 5, d ≤ 10; q = 7, d ≤ 9; q = 9, d ≤ 8; q = 11, d ≤ 7; q = 13, d ≤ 6.
  • Characteristic 2: the same computation finds no empty sub-intervals beyond (A)–(C) for q = 2, d ≤ 26; q = 4, d ≤ 13; q = 8, d ≤ 7; q = 16, d ≤ 5.
  • Weil bound: it sits at the square-root barrier. At d = 2 every level-2 L-function in characteristic 2 is rigid, with γ⁴ = (−1)^k q².

The Oppermann subgroup in characteristic 2 (§2.1, §7)

  • Structure: H_d = ker(P ↦ S₁, S₃, …), so G_d/H_d ≅ F_q^{⌊d/2⌋}. The annihilator's explicit-formula coefficients are Artin–Schreier sums.
  • Mean drift: E_H[Ψ] − q^d is 0 for d = 2, 3 and for d = 4 with k even. It is −2q² + 2q for d = 4 with k odd, q² − q for d = 5, (−1)^(k+1) q²(q − 1) for d = 6 and q³ − q for d = 7, for every q = 2^k (Thms 7.10, 7.11). The d = 6 case rests on an elimination showing that the only irreducible cubics X³ + X² + c₁X + c₀ dividing X¹⁰ + AX⁶ + BX⁵ + B² are those with c₀ = c₁², A = B + 1, B = c₁⁴ + c₁³ a non-cube (k even).
  • Variance: it is a coherent sum over annihilator cosets.
    • d = 2: Var_H/q² = q − 1, against Var_G/q² = 1 − 1/q.
    • d = 3: Var_H/q³ = 1 − 1/q (k odd) or 3(1 − 1/q) (k even).
    • The coset sums of the characters (−1)^Tr(a·e₂) of H_d vanish for odd d; for d = 4 they are −q⁶ (k even), −q⁵(q − 2) (k odd), so Var_H/q⁴ ≥ q − 1 for k even (Prop 7.8).
    • For d ≥ 5 there is a numerical census, compared with the Keating–Rudnick value d − 1.
  • Level-3 L-functions (Thm 7.7): every character χ of exact level 3 of G₃ has L(u, χ) = (1 − γ₁u)(1 − γ₂u) with γ₁γ₂ = ±q or ±iq, (γ₁ + γ₂)²/γ₁γ₂ ∈ {0, 1, 2, 4} and γ⁶ ∈ q³{±1, ±i}; the normalised zeros are 16 of the 24th roots of unity. This holds for every q = 2^k. On each coset of q characters with a fixed restriction to H₃ the values of r and c₂ are distributed explicitly (r = 1; or r = 2, 0 for q/2 each; or r = 4 for q/4 and r = 0 for 3q/4), and exactly q(q²−1)/4 (k even), q(q−1)(q−2)/8 (k odd) characters have ψ₆ = −2q³.
  • Mean of π at d = 6: E_H[π] = (q⁶ − q³ + (−1)^(k+1) q²(q − 1))/12 (Thm 7.11).

Files

script (code/) output (data/, figures/, paper/) paper
hayes_oppermann.py hayes_oppermann_census.txt Thms 1.2, 1.3
verify_pointwise_oppermann.py, compare_bruteforce.py oppermann_census.txt, compare_bruteforce.txt cross-checks
char2_d2_blocks.py, char2_d2_theorem.py char2_d2_blocks.txt, char2_d2_theorem.txt Thm 1.1(B)
swan_parity.py swan_parity.txt Thm 1.1(C), Table 1
verify_additive_slices.py additive_slices.txt Thm 3.6, Cor 3.7
odd_d2_formula.py odd_d2_formula.txt Thm 1.4
char2_d3_p3zero.py, char2_d3_all.py char2_d3_p3zero.txt, char2_d3_all.txt Thm 5.2
cubic_gauss_m.py, make_d3_table.py cubic_gauss_m.txt, paper/tab_d3.tex Thm 5.3, Cor 5.4, Table 2
char2_d2_rigidity.py char2_d2_rigidity.txt Prop 6.1
verify_level3_cosets.py level3_cosets.txt Prop 7.6 (level-3 coset sums; γ⁶ = q³ on the non-cube cosets)
verify_level3_theorem.py level3_theorem.txt Thm 7.7: every character of exact level 3 of G_3 (all q²(q−1) of them, q = 2, 4, 8, 16)
verify_level2_cosets.py level2_cosets.txt Prop 7.8 (coset sums of the characters (−1)^Tr(a·e_2))
verify_mean_d67.py, verify_d67_factors.py verify_mean_d67.txt, verify_d67_factors.txt Lemma 7.9, Thm 7.10 (d = 7), Thm 7.11 (d = 6; all polynomial identities of the elimination, and the census of cubic factors for q ≤ 64)
subgroup_lib.py, verify_subgroup_mean.py verify_subgroup_mean.txt Thms 2.4, 2.5, 7.3
subgroup_variance_census.py, verify_variance23.py subgroup_variance.{txt,json}, verify_variance23.txt Thm 7.5
make_figure.py, make_figure2.py, make_tables.py figures/oppermann_min_ratio.pdf, figures/oppermann_char2_rigidity.pdf, paper/census_tables.tex Figs 1, 2
make_subgroup_figures.py, make_subgroup_tables.py figures/subgroup_mean_gauss.pdf, figures/subgroup_variance.pdf, paper/tab_subgroup_*.tex Figs 3, 4
hayes_fft_fq.py, fqlib.py, small_q_gf.py — from Part I. The only change is extra moduli in fqlib.py for q = 32, …, 256

notes/REPORT.md contains the working notes of the first round.

Run with:

cd code
python hayes_oppermann.py 3 1 16 <tmpdir>      # q d_min d_max [memmap dir]
python verify_subgroup_mean.py > ../data/verify_subgroup_mean.txt
python subgroup_variance_census.py <tmpdir>
python cubic_gauss_m.py 14 > ../data/cubic_gauss_m.txt
python make_figure.py; python make_figure2.py; python make_subgroup_figures.py
python make_tables.py; python make_d3_table.py; python make_subgroup_tables.py
cd ../paper && pdflatex oppermann_function_fields.tex

About

Oppermann intervals in F_q[t]: characteristic-2 counterexamples with exact counts for d ≤ 3, an exact d = 2 formula for odd q, and a Hayes-FFT census

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages