Real reciprocal tails and analytic bridges
DefinitionErdosProblems_Erdos243_PaperCompleteR7_RealTail_v243UnionFbf41fb8erdos-243formalizationreciprocal-tail
For a real sequence t, defines its tail at n as the infinite sum of t_(k+n) over k≥0. Included source proofs establish tail decomposition under summability, positivity under positive terms, geometric comparison, and the leading-term asymptotic when successive ratios tend to zero. The reciprocal-growth bridge carries its stated analytic hypotheses.
Definition code
import Definitions.Def_ErdosProblems_Erdos243_ReciprocalTailRigidity
import Definitions.Def_ErdosProblems_Erdos243_PaperCompleteR7_Limits
import Mathlib.Algebra.Order.BigOperators.GroupWithZero.Finset
import Mathlib.Analysis.SpecificLimits.Basic
import Mathlib.Analysis.SpecificLimits.Normed
import Mathlib.Data.Fin.Pigeonhole
import Mathlib.Data.Nat.ChineseRemainder
import Mathlib.Data.Nat.Find
import Mathlib.Data.Nat.Prime.Basic
import Mathlib.Data.ZMod.Basic
import Mathlib.Order.Filter.AtTopBot.Basic
import Mathlib.Tactic
import Mathlib.Tactic.Ring
import Mathlib.Topology.Algebra.InfiniteSum.NatInt
/-!
# Analytic part of the canonical tail bridge
Compiled candidates. The principal analytic assertion is proved, not
postulated: if positive summable terms have successive ratio tending to
zero, their tail divided by the leading term tends to one.
The final theorem derives normalised vanishing from quadratic growth and
an exact tail representation. It does NOT claim to have constructed the
canonical integer state, proved its integrality, or obtained the sharper
O(1/a_n) expansion required elsewhere in the papers.
-/
namespace ErdosProblems.Erdos243.PaperCompleteR7
open Filter
open scoped BigOperators
noncomputable def realTail (t : ℕ → ℝ) (n : ℕ) : ℝ :=
∑' k : ℕ, t (k + n)
theorem realTail_step (t : ℕ → ℝ) (ht : Summable t) (n : ℕ) :
realTail t n = t n + realTail t (n + 1) := by
have hs : Summable (fun k : ℕ ↦ t (k + n)) := (summable_nat_add_iff n).mpr ht
simpa only [realTail, Nat.zero_add, Nat.add_assoc, Nat.add_left_comm,
Nat.add_comm] using hs.tsum_eq_zero_add
theorem realTail_nonneg (t : ℕ → ℝ) (hpos : ∀ n, 0 ≤ t n) (n : ℕ) :
0 ≤ realTail t n := by
exact tsum_nonneg (fun k ↦ hpos (k + n))
theorem realTail_leading_term (t : ℕ → ℝ) (ht : Summable t)
(hpos : ∀ n, 0 ≤ t n) (n : ℕ) : t n ≤ realTail t n := by
rw [realTail_step t ht n]
have hp := realTail_nonneg t hpos (n + 1)
linarith
theorem realTail_pos (t : ℕ → ℝ) (ht : Summable t)
(hpos : ∀ n, 0 < t n) (n : ℕ) : 0 < realTail t n :=
(hpos n).trans_le (realTail_leading_term t ht (fun k ↦ (hpos k).le) n)
/-- A uniform one-step ratio bound on a tail gives the full geometric
majorant. Both infinite sums in the comparison have summability proofs. -/
theorem realTail_geometric_bound
(t : ℕ → ℝ) (ht : Summable t) (hpos : ∀ n, 0 ≤ t n)
(r : ℝ) (hr : 0 ≤ r) (hr1 : r < 1) (N n : ℕ) (hn : N ≤ n)
(hstep : ∀ j, N ≤ j → t (j + 1) ≤ r * t j) :
realTail t n ≤ (1 - r)⁻¹ * t n := by
have hmajor : ∀ k : ℕ, t (k + n) ≤ r ^ k * t n := by
intro k
induction k with
| zero => simp
| succ k ih =>
calc
t (k + 1 + n) = t ((k + n) + 1) := by congr 1 <;> omega
_ ≤ r * t (k + n) := hstep (k + n) (by omega)
_ ≤ r * (r ^ k * t n) := mul_le_mul_of_nonneg_left ih hr
_ = r ^ (k + 1) * t n := by rw [pow_succ]; ring
have hg : HasSum (fun k : ℕ ↦ r ^ k) ((1 - r)⁻¹) :=
hasSum_geometric_of_abs_lt_one (by rwa [abs_of_nonneg hr])
have hgm : HasSum (fun k : ℕ ↦ r ^ k * t n) ((1 - r)⁻¹ * t n) :=
hg.mul_right (t n)
have hs : Summable (fun k : ℕ ↦ t (k + n)) := (summable_nat_add_iff n).mpr ht
calc
realTail t n ≤ ∑' k : ℕ, r ^ k * t n :=
hs.tsum_le_tsum hmajor hgm.summable
_ = (1 - r)⁻¹ * t n := hgm.tsum_eq
/-- Positive terms with successive ratio tending to zero have tails
asymptotic to their leading term. This is a complete analytic lemma,
not a bound supplied as a hypothesis to a formal endpoint. -/
theorem realTail_div_term_tendsto_one
(t : ℕ → ℝ) (ht : Summable t) (hpos : ∀ n, 0 < t n)
(hratio : Tendsto (fun n ↦ t (n + 1) / t n) atTop (nhds 0)) :
Tendsto (fun n ↦ realTail t n / t n) atTop (nhds 1) := by
apply Metric.tendsto_atTop.mpr
intro ε hε
let r : ℝ := min (ε / 4) (1 / 4)
have hrpos : 0 < r := lt_min (by positivity) (by norm_num)
have hrquarter : r ≤ 1 / 4 := min_le_right _ _
have hreps : r ≤ ε / 4 := min_le_left _ _
have hr1 : r < 1 := by linarith
obtain ⟨N, hN⟩ := Metric.tendsto_atTop.mp hratio r hrpos
have hstep : ∀ j, N ≤ j → t (j + 1) ≤ r * t j := by
intro j hj
have hnonneg : 0 ≤ t (j + 1) / t j := div_nonneg (hpos _).le (hpos _).le
have hh : t (j + 1) / t j < r := by
simpa only [Real.dist_eq, sub_zero, abs_of_nonneg hnonneg] using hN j hj
exact (div_le_iff₀ (hpos j)).mp hh.le
refine ⟨N, fun n hn ↦ ?_⟩
have hlow : 1 ≤ realTail t n / t n := by
apply (le_div_iff₀ (hpos n)).mpr
simpa only [one_mul] using
realTail_leading_term t ht (fun k ↦ (hpos k).le) n
have hhigh : realTail t n / t n ≤ (1 - r)⁻¹ := by
apply (div_le_iff₀ (hpos n)).mpr
exact realTail_geometric_bound t ht (fun k ↦ (hpos k).le)
r hrpos.le hr1 N n hn hstep
have hden : 0 < 1 - r := by linarith
have hinv : (1 - r)⁻¹ ≤ 1 + 2 * r := by
rw [← one_div]
apply (div_le_iff₀ hden).mpr
have hp : 0 ≤ r * (1 - 2 * r) := mul_nonneg hrpos.le (by linarith)
nlinarith
have habs : |realTail t n / t n - 1| < ε := by
rw [abs_of_nonneg (by linarith : 0 ≤ realTail t n / t n - 1)]
linarith
simpa only [Real.dist_eq] using habs
theorem shift_tendsto_atTop (k : ℕ) :
Tendsto (fun n : ℕ ↦ n + k) atTop atTop := by
rw [Filter.tendsto_atTop_atTop]
intro B
exact ⟨B, fun n hn ↦ by omega⟩
/-- Strict increase and positivity give the real divergence needed for
reciprocals; it is not inserted as an additional growth assumption. -/
theorem strictMono_nat_cast_tendsto_atTop
(a : ℕ → ℕ) (ha : StrictMono a) (hpos : ∀ n, 0 < a n) :
Tendsto (fun n ↦ (a n : ℝ)) atTop atTop := by
have hlarge : ∀ n, n < a n := by
intro n
induction n with
| zero => exact hpos 0
| succ n ih =>
have hs : a n < a (n + 1) := ha (Nat.lt_succ_self n)
omega
have hnat : Tendsto a atTop atTop := by
rw [Filter.tendsto_atTop_atTop]
intro B
exact ⟨B, fun n hn ↦ hn.trans (hlarge n).le⟩
exact tendsto_natCast_atTop_atTop.comp hnat
/-- Quadratic denominator growth implies successive reciprocal terms
have ratio zero. The hypothesis uses the paper's a_(n+1)/a_n^2 ratio. -/
theorem reciprocal_successive_ratio_tendsto_zero
(a : ℕ → ℕ) (ha : StrictMono a) (hpos : ∀ n, 0 < a n)
(hgrowth : Tendsto (fun n ↦ (a (n + 1) : ℝ) / (a n : ℝ) ^ 2)
atTop (nhds 1)) :
Tendsto (fun n ↦ (1 / (a (n + 1) : ℝ)) / (1 / (a n : ℝ)))
atTop (nhds 0) := by
have hinv : Tendsto (fun n ↦ (a n : ℝ)⁻¹) atTop (nhds 0) :=
tendsto_inv_atTop_zero.comp (strictMono_nat_cast_tendsto_atTop a ha hpos)
have hq : Tendsto (fun n ↦ (a n : ℝ) ^ 2 / (a (n + 1) : ℝ))
atTop (nhds 1) := by
simpa only [inv_div, inv_one] using hgrowth.inv₀ (by norm_num : (1 : ℝ) ≠ 0)
have hm := hq.mul hinv
have hlim : Tendsto
(fun n ↦ (a n : ℝ) ^ 2 / (a (n + 1) : ℝ) * (a n : ℝ)⁻¹)
atTop (nhds 0) := by simpa only [one_mul] using hm
apply hlim.congr'
exact Filter.Eventually.of_forall fun n ↦ by
have h0 : (a n : ℝ) ≠ 0 := by exact_mod_cast (hpos n).ne'
have h1 : (a (n + 1) : ℝ) ≠ 0 := by exact_mod_cast (hpos (n + 1)).ne'
field_simp [h0, h1]
<;> ring
/-- The scaled consecutive real-tail ratio tends to one under the
actual growth condition, using only summability of the reciprocal series. -/
theorem scaled_reciprocal_tail_ratio_tendsto_one
(a : ℕ → ℕ) (ha : StrictMono a) (hpos : ∀ n, 0 < a n)
(hsum : Summable (fun n ↦ 1 / (a n : ℝ)))
(hgrowth : Tendsto (fun n ↦ (a (n + 1) : ℝ) / (a n : ℝ) ^ 2)
atTop (nhds 1)) :
Tendsto (fun n ↦ (a n : ℝ) * realTail (fun k ↦ 1 / (a k : ℝ)) (n + 1) /
realTail (fun k ↦ 1 / (a k : ℝ)) n) atTop (nhds 1) := by
let t : ℕ → ℝ := fun n ↦ 1 / (a n : ℝ)
have htpos : ∀ n, 0 < t n := fun n ↦ by
dsimp [t]
exact one_div_pos.mpr (by exact_mod_cast hpos n)
have hratio := reciprocal_successive_ratio_tendsto_zero a ha hpos hgrowth
have hrel : Tendsto (fun n ↦ realTail t n / t n) atTop (nhds 1) :=
realTail_div_term_tendsto_one t hsum htpos hratio
have hq : Tendsto (fun n ↦ (a n : ℝ) ^ 2 / (a (n + 1) : ℝ))
atTop (nhds 1) := by
simpa only [inv_div, inv_one] using hgrowth.inv₀ (by norm_num : (1 : ℝ) ≠ 0)
have hrnext := hrel.comp (shift_tendsto_atTop 1)
have hlim := (hq.mul hrnext).div hrel (by norm_num : (1 : ℝ) ≠ 0)
have hlim' : Tendsto (fun n ↦
((a n : ℝ) ^ 2 / (a (n + 1) : ℝ)) * (realTail t (n + 1) / t (n + 1)) /
(realTail t n / t n)) atTop (nhds 1) := by
simpa only [one_mul, div_one, Function.comp_apply] using hlim
apply hlim'.congr'
exact Filter.Eventually.of_forall fun n ↦ by
have h0 : (a n : ℝ) ≠ 0 := by exact_mod_cast (hpos n).ne'
have h1 : (a (n + 1) : ℝ) ≠ 0 := by exact_mod_cast (hpos (n + 1)).ne'
have hx : realTail t n ≠ 0 := (realTail_pos t hsum htpos n).ne'
dsimp only [t]
field_simp [h0, h1, hx]
<;> ring
/-- Analytic-to-arithmetic bridge with an explicit real-tail
representation. This finishes the normalised-vanishing implication once
the canonical integer state has been constructed; that construction is
not smuggled into the conclusion. -/
theorem normalized_vanishing_of_tail_representation
(a C D : ℕ → ℕ) (E : ℕ → ℤ)
(ha : StrictMono a) (hapos : ∀ n, 0 < a n)
(hCpos : ∀ n, 0 < C n) (hDpos : ∀ n, 0 < D n)
(hC : ∀ n, C (n + 1) + D n = a n * C n)
(hD : ∀ n, D (n + 1) = a n * D n)
(hE : ∀ n, E n = centeredState (a n : ℤ) (D n : ℤ) (C n : ℤ))
(hsum : Summable (fun n ↦ 1 / (a n : ℝ)))
(hrep : ∀ n, (C n : ℝ) = (D n : ℝ) *
realTail (fun k ↦ 1 / (a k : ℝ)) n)
(hgrowth : Tendsto (fun n ↦ (a (n + 1) : ℝ) / (a n : ℝ) ^ 2)
atTop (nhds 1)) :
(∀ K : ℕ, ∃ N, ∀ n, N ≤ n → K * Int.natAbs (E n) < C n) ∧
(∃ N, ∀ n, N ≤ n → Int.natAbs (E n) < C n) := by
have hr := scaled_reciprocal_tail_ratio_tendsto_one a ha hapos hsum hgrowth
have hquot : Tendsto (fun n ↦ (C (n + 1) : ℝ) / (C n : ℝ))
atTop (nhds 1) := by
apply hr.congr'
exact Filter.Eventually.of_forall fun n ↦ by
dsimp only
rw [hrep (n + 1), hrep n, hD n, Nat.cast_mul]
have hd0 : (D n : ℝ) ≠ 0 := by exact_mod_cast (hDpos n).ne'
have hx : realTail (fun k ↦ 1 / (a k : ℝ)) n ≠ 0 := by
exact (realTail_pos _ hsum (fun k ↦ one_div_pos.mpr
(by exact_mod_cast hapos k)) n).ne'
field_simp [hd0, hx]
<;> ring
have herr : Tendsto (fun n ↦ (E n : ℝ) / (C n : ℝ)) atTop (nhds 0) := by
have hz := (tendsto_const_nhds (x := (1 : ℝ))).sub hquot
have hz' : Tendsto (fun n ↦ 1 - (C (n + 1) : ℝ) / (C n : ℝ))
atTop (nhds 0) := by simpa only [sub_self] using hz
apply hz'.congr'
exact Filter.Eventually.of_forall fun n ↦ by
dsimp only
have hs : (C (n + 1) : ℝ) = (C n : ℝ) - (E n : ℝ) := by
exact_mod_cast natTail_eq_sub_centeredState a C D E hC hE n
have hc0 : (C n : ℝ) ≠ 0 := by exact_mod_cast (hCpos n).ne'
rw [hs]
field_simp [hc0]
<;> ring
have habs : Tendsto (fun n ↦ |(E n : ℝ)| / (C n : ℝ)) atTop (nhds 0) := by
have hx : Tendsto (fun n ↦ |(E n : ℝ) / (C n : ℝ)|) atTop (nhds 0) := by
simpa only [abs_zero] using herr.abs
refine hx.congr fun n ↦ ?_
rw [abs_div, abs_of_nonneg (by positivity : (0 : ℝ) ≤ (C n : ℝ))]
have hv := (normalized_vanishing_iff C E hCpos).mp habs
refine ⟨hv, ?_⟩
obtain ⟨N, hN⟩ := hv 1
exact ⟨N, fun n hn ↦ by simpa only [one_mul] using hN n hn⟩
end ErdosProblems.Erdos243.PaperCompleteR7
Source
Pinned original Lean source: https://github.com/wcook04/plectis-erdos-lean/blob/fbf41fb8b571d217b1648070dcc39d5eda5394ad/ErdosProblems/Erdos243/PaperCompleteR7/RealTail.lean#L1-L250
Paper context and prior-art bibliography: https://github.com/wcook04/plectis-erdos/blob/551bae6dc6e732cf85172d66323c8d2bc77ba962/paper/243/erdos-243-reciprocal-tail-rigidity.tex#L1151-L1235
AI-assisted formalization and editorial review; no human review attested. Hosted import name ErdosProblems_Erdos243_PaperCompleteR7_RealTail_v243UnionFbf41fb8 is the versioned native alias of original module ErdosProblems.Erdos243.PaperCompleteR7.RealTail.