AFTD Auto-Formalizing Theoretical Domains

pell_log_shift_periodic

theoremverified

For any function F : ℝ → ℝ satisfying Function.Periodic F 1 and any positive real number Y, evaluating F at Real.log (pell_unit_squared * Y) / Real.log pell_unit_squared equals F (Real.log Y / Real.log pell_unit_squared).

Statement

theorem pell_log_shift_periodic (F : ℝ → ℝ) (hF : Function.Periodic F 1) (Y : ℝ) (hY : 0 < Y) :
    F (Real.log (pell_unit_squared * Y) / Real.log pell_unit_squared) =
      F (Real.log Y / Real.log pell_unit_squared)
Source
arXiv:2611.05898
Verified
06 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 1
  • pell_unit_squaredThe definition of pell_unit_squared as the real number 17 + 12 * Real.sqrt 2.

Read back from the Lean

For every function F : ℝ → ℝ that is periodic with period 1 (i.e., F(x + 1) = F(x) for all x : ℝ), and for every real number Y > 0, F(log(pell_unit_squared * Y) / log(pell_unit_squared)) = F(log(Y) / log(pell_unit_squared)), where pell_unit_squared is 17 + 12 * √2.

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

The argument of F expands via the logarithm product rule into (Real.log pell_unit_squared + Real.log Y) / Real.log pell_unit_squared. Simple algebra splits this quotient into (Real.log Y / Real.log pell_unit_squared) + 1, shifting the input by 1. Because F is assumed to be 1-periodic, this shift leaves the function value unchanged. Preliminary steps verify that pell_unit_squared > 1 and Y > 0 so that the logarithm product formula and division by the non-zero logarithm are valid.

  1. pell_unit_squared is strictly greater than 1, and in particular strictly positive.

    Unfolding the definition, pell_unit_squared equals 17 + 12 * √2. Since √2 ≥ 0 by Real.sqrt_nonneg, linear arithmetic (linarith) proves 17 + 12 * √2 > 1. Positivity of pell_unit_squared is confirmed directly by positivity.

    lines 5–9

    uses Real.sqrt_nonneg

      have hP : 0 < pell_unit_squared := by unfold pell_unit_squared; positivity
      have hP1 : 1 < pell_unit_squared := by
        unfold pell_unit_squared
        have := Real.sqrt_nonneg 2
        linarith
  2. pell_unit_squared ≠ 0, Y ≠ 0, and Real.log pell_unit_squared ≠ 0.

    Since 1 < pell_unit_squared, Real.log_pos shows Real.log pell_unit_squared > 0, which implies it is non-zero by ne_of_gt. Similarly, ne_of_gt applied to 0 < Y and 0 < pell_unit_squared establishes that Y ≠ 0 and pell_unit_squared ≠ 0.

    lines 10–12

    uses ne_of_gt, Real.log_pos

      have hlog : Real.log pell_unit_squared ≠ 0 := ne_of_gt (Real.log_pos hP1)
      have hYne : Y ≠ 0 := ne_of_gt hY
      have hPne : pell_unit_squared ≠ 0 := ne_of_gt hP
  3. The numerator inside F expands as Real.log pell_unit_squared + Real.log Y.

    Since pell_unit_squared and Y are both non-zero, Real.log_mul rewrites the logarithm of their product as the sum of their individual logarithms.

    line 13

    uses Real.log_mul

      rw [Real.log_mul hPne hYne]
  4. (Real.log pell_unit_squared + Real.log Y) / Real.log pell_unit_squared = Real.log Y / Real.log pell_unit_squared + 1.

    Because Real.log pell_unit_squared ≠ 0, clearing denominators with field_simp reduces this equality to Real.log pell_unit_squared + Real.log Y = Real.log Y + Real.log pell_unit_squared, which holds by commutativity of addition (ring).

    lines 14–17

      have hsplit : (Real.log pell_unit_squared + Real.log Y) / Real.log pell_unit_squared
          = Real.log Y / Real.log pell_unit_squared + 1 := by
        field_simp
        ring
  5. Evaluating F at the shifted argument equals F (Real.log Y / Real.log pell_unit_squared).

    Rewriting the argument of F with the fractional split yields F (Real.log Y / Real.log pell_unit_squared + 1). Because F is 1-periodic (hypothesis hF), shifting the input by 1 leaves the value of F unchanged, closing the goal.

    lines 18–19

      rw [hsplit]
      exact hF _
Real.sqrt_nonneg
The square root of any real number is non-negative.
ne_of_gt
In a preorder, if b < a, then a ≠ b.
Real.log_pos
If a real number x > 1, then its natural logarithm is strictly positive (log x > 0).
Real.log_mul
The natural logarithm of a product of two non-zero real numbers equals the sum of their natural logarithms.

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

/-- A 1-periodic function of the log-phase is invariant under multiplication of Y by the Pell unit P. -/theorem pell_log_shift_periodic (F : ℝ → ℝ) (hF : Function.Periodic F 1) (Y : ℝ) (hY : 0 < Y) :    F (Real.log (pell_unit_squared * Y) / Real.log pell_unit_squared) =      F (Real.log Y / Real.log pell_unit_squared) := by  have hP : 0 < pell_unit_squared := by unfold pell_unit_squared; positivity  have hP1 : 1 < pell_unit_squared := by    unfold pell_unit_squared    have := Real.sqrt_nonneg 2    linarith  have hlog : Real.log pell_unit_squared ≠ 0 := ne_of_gt (Real.log_pos hP1)  have hYne : Y ≠ 0 := ne_of_gt hY  have hPne : pell_unit_squared ≠ 0 := ne_of_gt hP  rw [Real.log_mul hPne hYne]  have hsplit : (Real.log pell_unit_squared + Real.log Y) / Real.log pell_unit_squared      = Real.log Y / Real.log pell_unit_squared + 1 := by    field_simp    ring  rw [hsplit]  exact hF _
See it in the graphDiscuss this resultChallenge it