AFTD Auto-Formalizing Theoretical Domains

subfield_nonzero_mem_cubes

theoremverified

Let E be a field, m an odd natural number, and t a nonzero element of E. If t^(2^m) = t, then t belongs to nonzero_cubes E.

Statement

theorem subfield_nonzero_mem_cubes {E : Type*} [Field E] (m : ℕ) (hm : Odd m) (t : E) (ht : t ≠ 0) (htq : t ^ (2 ^ m) = t) : t ∈ nonzero_cubes E
Source
arXiv:2611.10269 Sec. 7
Verified
11 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 1
  • nonzero_cubesThe set of nonzero cubes in a field E, consisting of all elements of the form x^3 for nonzero x in E.
Used by 1
  • spread_line_contained_in_non_bentLet E be a finite field equipped with an algebra structure over ZMod 2 such that the cardinality of E is 2^(2m), where m is an odd natural…

Read back from the Lean

Let $E$ be a field, and let $m$ be a natural number such that $m$ is odd. For any element $t \in E$, if $t \ne 0$ and $t^{2^m} = t$, then $t \in \text{nonzero\_cubes } E$.

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

The proof, explained

Because m is odd, write m = 2k + 1 and let N = 2^m - 1. From t^(2^m) = t and t ≠ 0, cancelling t gives t^N = 1. Since 2 ≡ -1 (mod 3) and m is odd, 2^m ≡ -1 ≡ 2 (mod 3), which implies 3 divides 2N + 1; let c = (2N + 1) / 3. Then (t^c)^3 = t^(3c) = t^(2N+1) = (t^N)^2 * t = t, and since t ≠ 0, t^c is nonzero, showing that t is a cube of a nonzero element.

  1. m can be written as 2k + 1, and setting N = 2^(2k+1) - 1, we have 2^(2k+1) = N + 1.

    Since m is odd, we express it as 2k + 1 for some k. Defining N = 2^(2k+1) - 1, the fact that 2^(2k+1) > 0 (by Nat.pow_pos) allows linear arithmetic (omega) to deduce 2^(2k+1) = N + 1.

    lines 3–8

    uses Nat.pow_pos

      rcases hm with ⟨k, rfl⟩
      let N := 2 ^ (2 * k + 1) - 1
      have h2m : 2 ^ (2 * k + 1) = N + 1 := by
        dsimp [N]
        have hpos : 0 < 2 ^ (2 * k + 1) := Nat.pow_pos (by norm_num)
        omega
  2. t^N = 1.

    From the hypothesis t^(2^m) = t and the identity 2^m = N + 1, we rewrite t^(N+1) as t^N * t using pow_succ. Since t ≠ 0, cancelling t on the right via mul_right_cancel₀ gives t^N = 1.

    lines 9–12

    uses pow_succ, mul_right_cancel₀

      have hpowN : t ^ N = 1 := by
        have h := htq
        rw [h2m, pow_succ] at h
        exact mul_right_cancel₀ ht (by simpa using h)
  3. 3 divides 2N + 1.

    Writing 2^(2k+1) as 2 * 4^k via pow_succ and pow_mul, 2N + 1 simplifies to 4 * 4^k - 1. Modulo 3, 4 ≡ 1, so 4^k ≡ 1 and 4 * 4^k ≡ 1 (using Nat.ModEq.refl). Since 1 ≤ 4 * 4^k (via Nat.one_le_iff_ne_zero.mpr), Nat.modEq_iff_dvd' translates this congruence into 3 ∣ (4 * 4^k - 1). Finally, since 1 ≤ 4^k (by Nat.one_le_pow), omega equates 2(2 * 4^k - 1) + 1 with 4 * 4^k - 1, yielding 3 ∣ 2N + 1.

    lines 13–29

    uses pow_succ, pow_mul, Nat.mul_comm, Nat.ModEq.refl, Nat.one_le_iff_ne_zero.mpr, Nat.modEq_iff_dvd', Nat.one_le_pow

      have hdiv : 3 ∣ 2 * N + 1 := by
        dsimp [N]
        have hpow2 : 2 ^ (2 * k + 1) = 2 * 4 ^ k := by
          rw [pow_succ, pow_mul]
          norm_num
          exact Nat.mul_comm _ _
        have hd4 : 3 ∣ 4 * 4 ^ k - 1 := by
          have h4 : (4 : ℕ) ≡ 1 [MOD 3] := by norm_num [Nat.ModEq]
          have hpow : (4 : ℕ) ^ k ≡ 1 [MOD 3] := by simpa using h4.pow k
          have hmul : 4 * (4 : ℕ) ^ k ≡ 4 * 1 [MOD 3] := (Nat.ModEq.refl 4).mul hpow
          have hmul' : 4 * (4 : ℕ) ^ k ≡ 1 [MOD 3] := hmul.trans h4
          have h : 1 ≤ 4 * 4 ^ k := Nat.one_le_iff_ne_zero.mpr (by positivity)
          exact (Nat.modEq_iff_dvd' h).mp hmul'.symm
        convert hd4 using 1
        rw [hpow2]
        have ha : 1 ≤ 4 ^ k := by exact Nat.one_le_pow k 4 (by norm_num)
        omega
  4. Setting c = (2N + 1) / 3, we have 3c = 2N + 1.

    Because 3 divides 2N + 1, multiplication by 3 inverts the integer division by Nat.mul_div_cancel'.

    lines 30–33

    uses Nat.mul_div_cancel'

      let c := (2 * N + 1) / 3
      have hc : 3 * c = 2 * N + 1 := by
        dsimp [c]
        exact Nat.mul_div_cancel' hdiv
  5. t^(3c) = t.

    First, t^(2N) = (t^N)^2 = 1^2 = 1 using Nat.mul_comm, pow_mul, and the earlier result t^N = 1. Then, replacing 3c with 2N + 1, we compute t^(3c) = t^(2N+1) = t^(2N) * t^1 = 1 * t = t using pow_add, one_mul, and pow_one.

    lines 34–40

    uses Nat.mul_comm, pow_mul, pow_add, one_mul, pow_one

      have htrip : t ^ (3 * c) = t := by
        have h2N : t ^ (2 * N) = 1 := by
          calc
            t ^ (2 * N) = t ^ (N * 2) := by rw [Nat.mul_comm]
            _ = (t ^ N) ^ 2 := by rw [pow_mul]
            _ = 1 := by rw [hpowN]; norm_num
        rw [hc, pow_add, h2N, one_mul, pow_one]
  6. t is a nonzero cube in E, with cube root t^c.

    We propose t^c as the cube root. Since t ≠ 0, pow_ne_zero ensures t^c ≠ 0. Furthermore, (t^c)^3 = t^(3c) = t by pow_mul, Nat.mul_comm, and the previous equality, confirming that t belongs to nonzero_cubes E.

    lines 41–46

    uses pow_ne_zero, pow_mul, Nat.mul_comm

      refine ⟨t ^ c, ?_, ?_⟩
      · exact pow_ne_zero c ht
      · calc
          (t ^ c) ^ 3 = t ^ (c * 3) := (pow_mul t c 3).symm
          _ = t ^ (3 * c) := by rw [Nat.mul_comm]
          _ = t := htrip
Nat.pow_pos
A positive natural number raised to any power is positive.
pow_succ
In a monoid, a^(n + 1) = a^n * a.
mul_right_cancel₀
In a structure with cancellation away from zero (such as a field), a * b = c * b with b ≠ 0 implies a = c.
pow_mul
In a monoid, a^(m * n) = (a^m)^n.
Nat.mul_comm
Multiplication of natural numbers is commutative: n * m = m * n.
Nat.ModEq.refl
Modular congruence is reflexive: a ≡ a [MOD n].
Nat.one_le_iff_ne_zero.mpr
A natural number is at least 1 if it is nonzero.
Nat.modEq_iff_dvd'
For natural numbers a ≤ b, a ≡ b [MOD n] if and only if n divides b - a.
Nat.one_le_pow
Any power of a positive natural number is at least 1.
Nat.mul_div_cancel'
If n divides m, then n * (m / n) = m.
pow_add
In a monoid, a^(m + n) = a^m * a^n.
one_mul
In a monoid with identity 1, 1 * a = a.
pow_one
In a monoid, a^1 = a.
pow_ne_zero
In a reduced monoid with zero (such as a field), any power of a nonzero element is nonzero.

Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.

Lean source view module on GitHub

/-- For odd m, every nonzero element of the subfield F_{2^m} is a cube. -/theorem subfield_nonzero_mem_cubes {E : Type*} [Field E] (m : ℕ) (hm : Odd m) (t : E) (ht : t ≠ 0) (htq : t ^ (2 ^ m) = t) : t ∈ nonzero_cubes E := by  rcases hm with ⟨k, rfl⟩  let N := 2 ^ (2 * k + 1) - 1  have h2m : 2 ^ (2 * k + 1) = N + 1 := by    dsimp [N]    have hpos : 0 < 2 ^ (2 * k + 1) := Nat.pow_pos (by norm_num)    omega  have hpowN : t ^ N = 1 := by    have h := htq    rw [h2m, pow_succ] at h    exact mul_right_cancel₀ ht (by simpa using h)  have hdiv : 3 ∣ 2 * N + 1 := by    dsimp [N]    have hpow2 : 2 ^ (2 * k + 1) = 2 * 4 ^ k := by      rw [pow_succ, pow_mul]      norm_num      exact Nat.mul_comm _ _    have hd4 : 3 ∣ 4 * 4 ^ k - 1 := by      have h4 : (4 : ℕ) ≡ 1 [MOD 3] := by norm_num [Nat.ModEq]      have hpow : (4 : ℕ) ^ k ≡ 1 [MOD 3] := by simpa using h4.pow k      have hmul : 4 * (4 : ℕ) ^ k ≡ 4 * 1 [MOD 3] := (Nat.ModEq.refl 4).mul hpow      have hmul' : 4 * (4 : ℕ) ^ k ≡ 1 [MOD 3] := hmul.trans h4      have h : 1 ≤ 4 * 4 ^ k := Nat.one_le_iff_ne_zero.mpr (by positivity)      exact (Nat.modEq_iff_dvd' h).mp hmul'.symm    convert hd4 using 1    rw [hpow2]    have ha : 1 ≤ 4 ^ k := by exact Nat.one_le_pow k 4 (by norm_num)    omega  let c := (2 * N + 1) / 3  have hc : 3 * c = 2 * N + 1 := by    dsimp [c]    exact Nat.mul_div_cancel' hdiv  have htrip : t ^ (3 * c) = t := by    have h2N : t ^ (2 * N) = 1 := by      calc        t ^ (2 * N) = t ^ (N * 2) := by rw [Nat.mul_comm]        _ = (t ^ N) ^ 2 := by rw [pow_mul]        _ = 1 := by rw [hpowN]; norm_num    rw [hc, pow_add, h2N, one_mul, pow_one]  refine ⟨t ^ c, ?_, ?_⟩  · exact pow_ne_zero c ht  · calc      (t ^ c) ^ 3 = t ^ (c * 3) := (pow_mul t c 3).symm      _ = t ^ (3 * c) := by rw [Nat.mul_comm]      _ = t := htrip
See it in the graphDiscuss this resultChallenge it