Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-27
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

A summability matrix with only finitely many nonzero entries per row is regular iff each column tends to 0, the row sums tend to 1, and the row absolute sums are uniformly bounded

Statement

Let c be a summability matrix (A summability (Toeplitz) matrix, the transformed sequence yn=∑kcn,kxk, and regularity), so that every row has only finitely many nonzero entries. Then c is regular if and only if all three of the following hold:

  1. (Columns are null.) For every k∈N the k-th column converges with lim⁡ncn,k=0.
  2. (Row sums tend to 1.) The sequence of row sums converges with lim⁡n∑kcn,k=1.
  3. (Row absolute sums are uniformly bounded.) There is M∈R with ∑k∣cn,k∣≤M for every n∈N.

In 1 and 2 the existence of the limit is part of the assertion. The notation is licensed by uniqueness of limits of real sequences (A sequence has at most one limit).

Condition 3 is the one that cannot be seen on any single sequence: 1 and 2 are read off two particular convergent inputs, while the necessity of 3 needs a sequence built against the matrix, by a gliding hump.

Facts & Assumptions

Given: A summability matrix c:N×N→R with finite row support. For a sequence (xk) of reals we write (yn) for its transform, yn=∑kcn,kxk, and rn:=∑k∣cn,k∣ for the row absolute sums.

[L1]

Summability matrices: finite row support, the transform and its independence of the admissible row bound used, the row sum, the row absolute sum, and regularity (A summability (Toeplitz) matrix, the transformed sequence yn=∑kcn,kxk, and regularity, Sequences of reals: bounded, eventually, frequently, tails, subsequences).

[L2]

Finite sums (Finite sums and finite products, by recursion) and their laws: additivity, scaling with ∑k<dλ=dλ, splitting, and monotonicity in the terms (Laws of finite sums and finite products).

[L3]

Triangle inequality for finite sums (Triangle inequality for finite sums); ∣uv∣=∣u∣ ∣v∣, ∣u∣≥0, ∣∣u∣∣=∣u∣ and ∣u∣=u for u≥0 (Basic properties of the absolute value).

[L4]

Convergence: for every real ε>0 there is N beyond which the terms are within ε of the limit, the rational and real formulations agreeing (Limits and Cauchy sequences of reals, Sequences of reals: bounded, eventually, frequently, tails, subsequences, The rationals embed densely in the reals); limits are unique (A sequence has at most one limit); a sequence that is eventually 0 converges to 0.

[L5]

Every convergent sequence of reals is bounded (Every convergent sequence is bounded).

[L6]

Algebra of limits for sums and scalar multiples (Algebra of limits: sums, scalar multiples, products and quotients).

[L7]

Archimedean property of R: for every real z there is a natural n≥1 with z<n⋅1R (Every complete ordered field is Archimedean); equivalently, for every real ε>0 there is a natural J≥1 with 1/J<ε (For every ε>0 in a complete ordered field there is a natural n≥1 with 1/n<ε).

[L8]

Least upper bounds: a nonempty subset of R bounded above has a supremum, which dominates every element of the set (Complete ordered field (least-upper-bound property), Upper bound, least upper bound, and strict upper bound, Lower bound, bounded below, bounded set).

[L9]

Every nonempty finite set of reals has a maximum, which lies in the set and dominates it (Every nonempty finite set of reals has a maximum and a minimum, Maximum and minimum of a set).

[L10]

Recursion theorem (The recursion theorem); well-ordering principle (The well-ordering principle); induction principle (The principle of mathematical induction); totality of the order on N (≤ is a linear order on N); and consecutive comparisons suffice for strict increase, with kj≥j for a strictly increasing index map (A strictly increasing index map satisfies nk≥k).

[L11]

Order arithmetic: a>0 gives a−1>0 and 0<a<b gives 0<b−1<a−1 (Inverses of positives are positive, and reciprocation reverses order); for c>0, a≤b if and only if ac≤bc (Sign rules for products and monotonicity of multiplication); adding a constant preserves the order and inequalities add (Order is preserved by adding a constant and by adding inequalities); canonical naturals are positive and increasing (Canonical naturals are positive and strictly increasing); the order is total and transitive (Complete ordered field (least-upper-bound property), Ordered field). In each clause above, Sign rules for products and monotonicity of multiplication and Order is preserved by adding a constant and by adding inequalities state the STRICT forms and only those; the nonstrict forms used below are those together with the equality cases, which trichotomy settles, the order being total (Ordered field).

Proof

technique · direct
1.1

Sufficiency. Assume conditions 1, 2 and 3, let (xk) converge to L, and let ε>0 be an arbitrary real; fix M≥0 as in condition 3, and fix D≥0 with ∣xk−L∣≤D for every k, which exists because a convergent sequence is bounded.

L1L3L4L5L11choose
1.2

Choose K∈N with ∣xk−L∣<ε (3(M+1))−1 for every k≥K.

L4L11choose
1.3

For every n, choosing an admissible bound R≥K for row n, one has yn−L=∑k=0Rcn,k(xk−L)+(∑kcn,k−1)L, since ∑k=0Rcn,kL=L∑kcn,k.

L1L2
1.4

Necessity. The remaining steps, apart from 2.1, 2.2, 2.3 and 3.1 which finish the sufficiency argument above, assume instead that c is regular.

L1
1.5

Suppose, towards a contradiction, that the row absolute sums (rn) are not bounded above, that is, for every T∈R there is n with rn>T.

L1L11assume-contra
1.6

For each n the set of admissible bounds for row n is a nonempty subset of N, so it has a least element ρ(n); thus cn,k=0 for every k>ρ(n), and every R≥ρ(n) is admissible for row n.

L1L10
2.1

For every n: ∣∑k=0Rcn,k(xk−L)∣≤∑k<K∣cn,k∣ ∣xk−L∣+∑k=KR∣cn,k∣ ∣xk−L∣≤D∑k<K∣cn,k∣+ε (3(M+1))−1rn≤D∑k<K∣cn,k∣+ε/3, the last step because rn≤M<M+1.

step 1.1step 1.2step 1.3L2L3L11
2.2

By condition 1 each of the finitely many columns k<K satisfies cn,k→0, hence ∣cn,k∣→0 since ∣∣cn,k∣−0∣=∣cn,k−0∣; a sum of finitely many null sequences is null, by induction on the number of summands, so ∑k<K∣cn,k∣→0 in n and there is N1 with ∑k<K∣cn,k∣<ε (3(D+1))−1 for every n≥N1.

step 1.1step 1.2L3L4L6L10L11choose
2.3

By condition 2 there is N2 with ∣∑kcn,k−1∣<ε (3(∣L∣+1))−1 for every n≥N2, so that ∣(∑kcn,k−1)L∣<ε/3 for such n.

step 1.1L3L4L11choose
2.4

Condition 1 holds. Fix k and let e be the sequence with ek=1 and ej=0 for j≠k; it is eventually 0, so it converges to 0, and its transform at row n is cn,k because every other term of the row sum vanishes. Regularity gives lim⁡ncn,k=0.

step 1.4L1L2L4
2.5

Condition 2 holds. The constant sequence with value 1 converges to 1 and its transform at row n is the row sum ∑kcn,k, so regularity gives lim⁡n∑kcn,k=1.

step 1.4L1L4
3.1

For every n beyond both N1 and N2: ∣yn−L∣≤D∑k<K∣cn,k∣+ε/3+ε/3<ε/3+ε/3+ε/3=ε; as ε was arbitrary, yn→L, and as (xk) was an arbitrary convergent sequence, c is regular.

step 1.3step 2.1step 2.2step 2.3L1L4L10L11
3.2

Each column converges, hence is bounded, so Mk:=sup⁡{ ∣cn,k∣:n∈N } exists in R for every k; putting Bm:=∑k=0mMk one has ∑k=0m∣cn,k∣≤Bm for every n and every m, and Bm≥0.

step 2.4L2L3L5L8L11
4.1

Define by recursion k0:=0, n0:=0 and, for j≥1, first Tj:= the larger of max⁡{rn:n≤nj−1} and Bkj−1+j (j+Bkj−1), then nj:=min⁡{n:rn>Tj}, which exists by step 1.5 and well-ordering, and then kj:= the larger of kj−1+1 and ρ(nj); then nj>nj−1, because rnj exceeds every rn with n≤nj−1, and kj−1<kj with cnj,k=0 for every k>kj, and rnj>Bkj−1+j (j+Bkj−1).

step 1.5step 1.6step 3.2L9L10L11construct
5.1

Since (kj) is strictly increasing with k0=0, every k≥1 lies in exactly one block kj−1<k≤kj with j≥1; define x0:=0 and, for k in the j-th block, xk:=sgn⁡(cnj,k) j−1, where sgn⁡(t):=1 for t≥0 and sgn⁡(t):=−1 for t<0. Then ∣xk∣≤1 for every k, and xk→0: given ε>0, take J≥1 with 1/J<ε, and every k>kJ lies in a block with index j>J, so ∣xk∣=1/j<1/J<ε.

step 4.1L3L7L10L11construct
6.1

For every j≥1: the terms of ynj with k>kj vanish, so ynj=∑k≤kj−1cnj,kxk+∑kj−1<k≤kjcnj,kxk; the second sum equals j−1∑kj−1<k≤kj∣cnj,k∣=j−1(rnj−∑k≤kj−1∣cnj,k∣)≥j−1(rnj−Bkj−1), while the first has absolute value at most ∑k≤kj−1∣cnj,k∣≤Bkj−1; hence ynj≥j−1(rnj−Bkj−1)−Bkj−1>j−1 j (j+Bkj−1)−Bkj−1=j.

step 3.2step 4.1step 5.1L2L3L11
7.1

But (xk) converges to 0, so regularity makes (yn) converge, hence bounded, so some S∈R has ∣yn∣≤S for every n; taking j with j⋅1R>S, available by the Archimedean property, step 6.1 gives ynj>j>S, a contradiction.

step 1.4step 5.1step 6.1L1L5L7L11
8.1

The assumption of step 1.5 is therefore untenable and condition 3 holds; with steps 2.4 and 2.5, regularity implies all three conditions.

step 2.4step 2.5step 7.1discharge-contradiction
9.1

Sufficiency is step 3.1 and necessity is step 8.1, so c is regular exactly when conditions 1, 2 and 3 all hold.

step 3.1step 8.1∎

Remarks

  • No one of the three conditions follows from the other two, and each is tested by a different input. Condition 1 is what a single nonzero coordinate detects, condition 2 what the constant sequence detects, and condition 3 is invisible to any fixed sequence: for each individual bounded input a matrix with unbounded row absolute sums may behave perfectly well, and the failure only appears against a sequence whose signs are chosen row by row. The substantial case is 3, and it is A summability matrix failing exactly one Silverman-Toeplitz condition and transforming a convergent sequence to a divergent one ↗, which exhibits a matrix satisfying 1 and 2 and failing 3, together with a null sequence whose transform diverges. The other two are settled in a line each and are recorded here rather than given items of their own: the matrix with cn,0=1 and every other entry 0 has row sums and row absolute sums constantly 1, so it satisfies 2 and 3, while its 0-th column is constantly 1 and fails 1; and the zero matrix has null columns and row absolute sums 0, so it satisfies 1 and 3, while its row sums are constantly 0 and fail 2.

  • The gliding hump. The witness of the necessity argument is built in blocks: on the j-th block its terms have modulus 1/j, so the sequence tends to 0, and their signs are chosen to align with the entries of one row nj, so that on that row the transform picks up almost the whole row absolute sum, divided by j. Choosing rnj larger than j times its own head bound makes the transform exceed j there. The bound Bkj−1 on the head is available before nj is chosen, because it depends only on the earlier block boundary, and that is what keeps the construction from circling.

  • No choice is used. Every stage of the recursion takes a least element or a maximum of a finite set; the row bound ρ(n) is the least admissible one; and Mk is a supremum, that is, a definite element of R rather than a selected bound.

Depends on

Used by

Dependency tree · two levels

61 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources