Covering form of the lower bound.
ProvedPrimeFractal.primeFractal_cover_card_geaether-catalognumbertheory
Covering form of the lower bound. For every ε > 0, eventually in m: any
finite family of intervals of length 1/m covering the prime fractal has at least
m ^ (1 - ε) members. The box dimension 1 is therefore not an artefact of the grid.
theorem PrimeFractal.primeFractal_cover_card_ge{ε : ℝ} (hε : 0 < ε) :
∀ᶠ m : ℕ in atTop, ∀ I : Set ℝ, I.Finite →
primeFractal ⊆ (⋃ c ∈ I, Set.Icc c (c + 1 / (m : ℝ))) →
1 - ε ≤ Real.log I.ncard / Real.log m := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source NumberTheory/PrimeFractalCovering.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from NumberTheory/PrimeFractalCovering.lean
import Mathlib
import Definitions.Def_NumberTheory_PrimeFractalHausdorff
import Definitions.Def_NumberTheory_PrimeFractalRefined
/-!
# Robustness of the box dimension: grid boxes versus arbitrary covers
`NumberTheory.PrimeFractalBoxDimension` computes the box dimension of the prime
fractal with *grid* boxes `[k/m, (k+1)/m)`. A critic may object that the value
of a "dimension" must not depend on that choice. It does not: an interval of
length `1/m` meets at most two grid boxes, so any cover of `S` by `K` intervals
of length `1/m` satisfies `boxCountSet S m ≤ 2 K`.
Consequently the dimension-`1` lower bound survives verbatim for the
covering-number definition of the Minkowski dimension
(`primeFractal_cover_card_ge`): however cleverly one covers the primes by
intervals of length `1/m`, one needs `m^{1-o(1)}` of them.
-/
open PrimeFractal
open Filter TopologyFormal statement
theorem PrimeFractal.primeFractal_cover_card_ge{ε : ℝ} (hε : 0 < ε) :
∀ᶠ m : ℕ in atTop, ∀ I : Set ℝ, I.Finite →
primeFractal ⊆ (⋃ c ∈ I, Set.Icc c (c + 1 / (m : ℝ))) →
1 - ε ≤ Real.log I.ncard / Real.log m := by sorrySource