Exists normalizedTailState collision of value eq rat
ProvedErdosProblems.Erdos269.exists_normalizedTailState_collision_of_value_eq_raterdos-269theorem
Under a rational-value hypothesis for the tail from scale one, two positive-scale normalized states differ by an integer.
Preamble
import Definitions.Def_ErdosProblems_Erdos269_BoundedRadixTailEscape
import Definitions.Def_ErdosProblems_Erdos269_ResidueEscape
import Definitions.Def_ErdosProblems_Erdos269_ThreePrimeRunningLcm
import Definitions.Def_ErdosProblems_Erdos269_RestrictedFloorSum
import Definitions.Def_ErdosProblems_Erdos269_DyadicBlockMassIdentity
import Definitions.Def_ErdosProblems_Erdos269_DyadicBlockThresholdPartition
import Definitions.Def_ErdosProblems_Erdos269_DyadicOrderedTailRecurrence
import Definitions.Def_ErdosProblems_Erdos269_DyadicShellSummability
import Definitions.Def_ErdosProblems_Erdos269_RationalLatticeReduction
import Mathlib.Algebra.BigOperators.Group.Finset.Basic
import Mathlib.Algebra.BigOperators.Group.Finset.Sigma
import Mathlib.Algebra.BigOperators.Group.List.Basic
import Mathlib.Algebra.GCDMonoid.Finset
import Mathlib.Algebra.Ring.Parity
import Mathlib.Data.Finset.Card
import Mathlib.Data.Finset.Prod
import Mathlib.Data.Int.ModEq
import Mathlib.Data.Nat.Log
import Mathlib.Data.Nat.Prime.Basic
import Mathlib.Data.Nat.Prime.Int
import Mathlib.Tactic
import Mathlib.Tactic.Linarith
import Mathlib.Tactic.NormNum
import Mathlib.Tactic.Ring
/-!
# Erdős #269: rationality forces an all-scale lattice, and the collision target
Let `S = ∑_{smooth s ≥ 2} 1/H(s)` with `H` the running `{2,3,5}` LCM height, and
let `X_a = (H(2^a)/2) · T_a` be the normalized dyadic tail state.
Three things are proved here.
1. **Clearing at every prime-power boundary.** For `p ∈ {2,3,5}`, `m ≥ 1` and
every smooth `x < p^m`, one has `p · H(x) ∣ H(p^m)`. The dyadic case
`p = 2` is the clearing used by the tail recurrence; the `p = 3` and
`p = 5` cases are new and give two further families of boundaries at which
the same rational prefix clears.
2. **All-scale rationality lattice.** If `S = p/q` then *every* `X_a` lies on
the `(1/q)`-lattice simultaneously, with an explicit integer witness. This
is the arithmetic reduction, no irrationality claim.
3. **The collision target.** Rationality does not merely produce one
exceptional integral state: by pigeonhole on `(1/q)ℤ / ℤ` it forces two
distinct scales whose tail states differ by an *integer*. Contrapositive:
if the normalized tail states are pairwise incongruent mod `1`, then `S` is
irrational. The `q = 1` "integral state" branch is the special case
`X_a - 0 ∈ ℤ`; the collision statement covers every denominator at once.
Nothing here claims irrationality.
-/
open scoped BigOperators
/-! ### Monomial clearing -/
/-! ### The dyadic half-height normalizer -/
/-! ### Clearing the rational prefix -/
/-! ### The all-scale lattice -/
/-! ### The collision target -/
open ErdosProblems.Erdos269
Formal statement
theorem ErdosProblems.Erdos269.exists_normalizedTailState_collision_of_value_eq_rat
{p q : ℤ} (hq : 0 < q)
(hval : dyadicShellTsumTailR235 1 = (p : ℝ) / (q : ℝ)) :
∃ i j : ℕ, i < j ∧ ∃ z : ℤ,
dyadicNormalizedTailStateR235 dyadicShellTsumTailR235 (1 + j) -
dyadicNormalizedTailStateR235 dyadicShellTsumTailR235 (1 + i) = (z : ℝ) := by sorry
Source
Lean source (Apache-2.0): https://github.com/wcook04/plectis-erdos-lean/blob/cc7e541cf2081c6fef5a5e377d52e365e33b01eb/ErdosProblems/Erdos269/RationalLatticeReduction.lean#L290-L304
Related paper by Will Cook (CC-BY-4.0): https://github.com/wcook04/plectis-erdos/blob/551bae6dc6e732cf85172d66323c8d2bc77ba962/paper/269/erdos-269-three-prime-running-lcm.tex#L1-L260
Paper prior-art bibliography: https://github.com/wcook04/plectis-erdos/blob/551bae6dc6e732cf85172d66323c8d2bc77ba962/paper/269/erdos-269-three-prime-running-lcm.tex#L681-L689