Permanent summability methods
DefinitionIndex sequences of density one#✓fully proved
A sequence \(s \colon \mathbb{N} \to \mathbb{N}\) has density one when it is strictly increasing and \(s(n+1)/(n+1) \to 1\) as \(n \to \infty\).
For a strictly increasing sequence of natural numbers this ratio condition is equivalent to the usual counting definition of density \(1\) for its set of values.
def HasDensityOne (s : ℕ → ℕ) : Prop :=
StrictMono s ∧
Filter.Tendsto (fun n => ((s (n + 1) : ℝ)) / ((n : ℝ) + 1)) Filter.atTop (nhds 1)
Used by 1
DefinitionAsymptotically convergent sequences#✓fully proved
A real sequence \((x_n)_{n \in \mathbb{N}}\) is asymptotically convergent to a real number \(S\) when there is an index sequence \(s\) of density one such that \(x_{s(n)} \to S\) as \(n \to \infty\).
def AsymptoticallyConvergent (x : ℕ → ℝ) (S : ℝ) : Prop :=
∃ s : ℕ → ℕ, HasDensityOne s ∧ Filter.Tendsto (x ∘ s) Filter.atTop (nhds S)
def HasDensityOne (s : ℕ → ℕ) : Prop :=
StrictMono s ∧
Filter.Tendsto (fun n => ((s (n + 1) : ℝ)) / ((n : ℝ) + 1)) Filter.atTop (nhds 1)
WantedExt/Analysis/Series/ScottishBook5BoundedWanted.lean:26
Sources
- this is the typescript's definition, verbatim in content.
Uses 1 · Used by 1
DefinitionSummation by a matrix method#✓fully proved
A real matrix \((a_{ik})_{i,k \in \mathbb{N}}\) sums a real sequence \((x_k)_{k \in \mathbb{N}}\) to a real number \(S\) when for every row \(i\) the series \(\sum_k a_{ik} x_k\) converges unconditionally (equivalently, absolutely) and the row sums \(\sum_k a_{ik} x_k\) tend to \(S\) as \(i \to \infty\).
def MatrixSums (a : ℕ → ℕ → ℝ) (x : ℕ → ℝ) (S : ℝ) : Prop :=
(∀ i : ℕ, Summable (fun k => a i k * x k)) ∧
Filter.Tendsto (fun i => ∑' k, a i k * x k) Filter.atTop (nhds S)
Used by 2
DefinitionPermanent matrix methods#✓fully proved
A real matrix \((a_{ik})_{i,k \in \mathbb{N}}\) is a permanent (regular, or Toeplitz) method when it sums every convergent real sequence to its limit: for every real sequence \(x\) and real number \(S\) with \(x_n \to S\), the matrix sums \(x\) to \(S\).
The Silverman–Toeplitz conditions characterise exactly these methods.
def IsPermanent (a : ℕ → ℕ → ℝ) : Prop :=
∀ x : ℕ → ℝ, ∀ S : ℝ, Filter.Tendsto x Filter.atTop (nhds S) → MatrixSums a x S
def MatrixSums (a : ℕ → ℕ → ℝ) (x : ℕ → ℝ) (S : ℝ) : Prop :=
(∀ i : ℕ, Summable (fun k => a i k * x k)) ∧
Filter.Tendsto (fun i => ∑' k, a i k * x k) Filter.atTop (nhds S)
WantedExt/Analysis/Series/ScottishBook5BoundedWanted.lean:38
Sources
- "Toeplitz method" is pinned to this standard reading; the typescript does not define it.
Uses 1 · Used by 1
DefinitionBounded sequences#✓fully proved
A real sequence \((x_n)_{n \in \mathbb{N}}\) is bounded when there is a real number \(B\) with \(|x_n| \le B\) for all \(n\).
def IsBounded (x : ℕ → ℝ) : Prop :=
∃ B : ℝ, ∀ n : ℕ, |x n| ≤ B
Used by 1
DefinitionThe bounded case of Scottish Book Problem 5#✓fully proved
There is a real matrix \((a_{ik})_{i,k \in \mathbb{N}}\) that is a permanent matrix method such that for every bounded real sequence \(x\) and every real number \(S\), the matrix sums \(x\) to \(S\) if and only if \(x\) is asymptotically convergent to \(S\).
def ScottishBookProblem5Claim : Prop :=
∃ a : ℕ → ℕ → ℝ, IsPermanent a ∧
∀ x : ℕ → ℝ, IsBounded x →
∀ S : ℝ, MatrixSums a x S ↔ AsymptoticallyConvergent x S
def IsPermanent (a : ℕ → ℕ → ℝ) : Prop :=
∀ x : ℕ → ℝ, ∀ S : ℝ, Filter.Tendsto x Filter.atTop (nhds S) → MatrixSums a x S
WantedExt/Analysis/Series/ScottishBook5BoundedWanted.lean:47
def IsBounded (x : ℕ → ℝ) : Prop :=
∃ B : ℝ, ∀ n : ℕ, |x n| ≤ B
WantedExt/Analysis/Series/ScottishBook5BoundedWanted.lean:52
def MatrixSums (a : ℕ → ℕ → ℝ) (x : ℕ → ℝ) (S : ℝ) : Prop :=
(∀ i : ℕ, Summable (fun k => a i k * x k)) ∧
Filter.Tendsto (fun i => ∑' k, a i k * x k) Filter.atTop (nhds S)
WantedExt/Analysis/Series/ScottishBook5BoundedWanted.lean:38
def AsymptoticallyConvergent (x : ℕ → ℝ) (S : ℝ) : Prop :=
∃ s : ℕ → ℕ, HasDensityOne s ∧ Filter.Tendsto (x ∘ s) Filter.atTop (nhds S)
WantedExt/Analysis/Series/ScottishBook5BoundedWanted.lean:33
Uses 4 · Used by 1
TheoremNegative answer to Scottish Book Problem 5 for bounded sequences#●ready to prove
There is no real matrix \((a_{ik})_{i,k \in \mathbb{N}}\) that is a permanent matrix method such that for every bounded real sequence \(x\) and every real number \(S\), the matrix sums \(x\) to \(S\) if and only if \(x\) is asymptotically convergent to \(S\).
theorem_wanted mazur5BoundedRefuted : ¬ ScottishBookProblem5Claim
def ScottishBookProblem5Claim : Prop :=
∃ a : ℕ → ℕ → ℝ, IsPermanent a ∧
∀ x : ℕ → ℝ, IsBounded x →
∀ S : ℝ, MatrixSums a x S ↔ AsymptoticallyConvergent x S
WantedExt/Analysis/Series/ScottishBook5BoundedWanted.lean:56
Sources
- the typescript itself, Ad 5 (Mazur), PDF p.5 / printed p.2, verified against the typescript page. Ad 5 (2) states verbatim: "If a method (a_{ik}) sums all the convergent sequences and every bounded sequence summable by the sequence [sic — the method] is asymptotically convergent, then there exists a sequence of increasing integers {kₙ} with density 1, such that for every bounded sequence {ξₙ} summable by this method, the sequence {ξ_{kₙ}} is convergent." The typescript then draws the corollary verbatim: "from (2) it follows that a permanent method summing all bounded asymptotically convergent sequences must also sum some other bounded sequences" — i.e. no permanent method sums exactly the bounded asymptotically convergent sequences, which refutes the equivalence formalized in
ScottishBookProblem5Claim. Proof sketch of the corollary from (2): suppose a permanent methodasummed exactly the bounded asymptotically convergent sequences (to the right limits). Thenasatisfies both hypotheses of (2) (permanent, and every boundeda-summable sequence is asymptotically convergent, by the assumed equivalence), so (2) gives a fixed density-1 sequence {kₙ} along which every boundeda-summable sequence converges. Let J = ℕ ∖ {k_{m²} : m ∈ ℕ}; J has density 1 (since k_{m²} ≥ m², the excluded set has density 0), and let ξ be 0 on J and 1 elsewhere. Then ξ is bounded and asymptotically convergent to 0 (via J), soasums ξ to 0; but ξ_{kₙ} equals 1 when n is a square and 0 otherwise, so it diverges — contradicting (2). (Cesàro does not satisfy (2)'s hypothesis — e.g. (-1)^n is Cesàro-summable but not asymptotically convergent — so it is not a counterexample to (2).) Note: the earlier Phase 3c entry recorded only that Ad 5 (1) does not settle the bounded question; Ad 5 (2) and its corollary do.
Uses 1
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.