trureturingGitHub MATHEMATICAL ATLAS

RESEARCH / Focused target

A closed form for the generalized Stirling numbers of boson ordering

Literature status: not recheckedRepository record: proved

RELEASED FOUNDATIONS / PROPOSED CONNECTION

Research connections

0 released anchors
All source anchors (1)

Problem

The question

Maier uses the Hsu–Shiue generalized Stirling numbers to express orderings of boson creation and annihilation operators. With (y)^{n, α} = y(y − α) ⋯ (y − (n − 1)α), the rescaled numbers Ŝ_{n,k}(α, β; r), 0 ≤ k ≤ n, are defined by (βx + r)^{n, α} = Σ_k Ŝ_{n,k}(α, β; r) C(x, k), equivalently (Theorem 4.1) Ŝ_{n,k} = Σ_{x=0}^{k} (−1)^{k−x} C(k, x) (βx + r)^{n, α}. Conjecture 5.3 of arXiv:2308.10332v4 states, as found heuristically:

For all r ∈ ℤ, Ŝ_{n,k}(−1, 2; r) = Σ_{j=⌊(2−r)/2⌋}^{⌊(n+2−r)/2⌋} C(n − j, n − k) n! C(n + 1, 2j + r − 1).

The source remarks that the upper argument n − j may be negative. Issue #10170 fixes the readings: n, k ∈ ℕ with k ≤ n, r ∈ ℤ; Ŝ is the finite sum of Theorem 4.1 with the rising factorial; C(n − j, n − k) is the generalized binomial coefficient Ring.choose; the sum runs over integers j between the stated floors.

Motivation

Our foothold

The numbers Ŝ_{n,k}(−1, 2; r) enter the ordering identities of §6 of the source. The frozen declaration D5/S3/Quantum/FockSpace/BosonOrderingStirlingClosedForm.result proves the conjectured closed form for every n, every k ≤ n and every integer r.

Gap

Missing bridges

Issue #10170 preregisters the conjecture and its literature check. arXiv lists version 4 (2024-02-09) with the journal reference Adv. in Appl. Math. 156 (2024) 102678. Semantic Scholar reports two citing papers: a paper on squeezing in mesoscopic circuits, and arXiv:2508.13094 by the same author, whose source cites only Theorem 6.5 of the paper.

These readings are not-found-in-searched-scope; they do not establish an exhaustive worldwide literature search, priority, or the absence of an independent proof.

Route

Proposed approach

Extend the sum to all integers j; outside the stated range the factor C(n + 1, 2j + r − 1) vanishes. Write L(n, k, r) for the left side and R(n, k, r) for the sum divided by n!.

  • (K) L(n, k + 1, r) = L(n, k, r + 2) − L(n, k, r), because the forward difference in x commutes with the shift x ↦ x + 1, which replaces r by r + 2. The same holds for R when k < n: shifting j by one and Pascal's rule C(a + 1, b + 1) = C(a, b) + C(a, b + 1) for the generalized binomial coefficient give R(n, k, r + 2) − R(n, k, r) = R(n, k + 1, r).
  • (R) At k = 0, L(n, 0, r) = r(r + 1) ⋯ (r + n − 1), and L(n + 1, 0, r) − L(n + 1, 0, r − 1) = (n + 1) L(n, 0, r). Pascal's rule for C(n + 2, ·), a shift of j and Pascal's rule for the generalized binomial coefficient give R(n + 1, 0, r) − R(n + 1, 0, r − 1) = R(n, 0, r).
  • At r = 1, L(n, 0, 1) = n! and R(n, 0, 1) = 1, because only j = 0 contributes; so L(n, 0, 1) = n! R(n, 0, 1). Induction on n, with an integer induction on r from r = 1, gives the case k = 0; for n = 0, R(0, 0, r) = 1 because exactly one j has 2j + r − 1 ∈ {0, 1}. Induction on k using (K) gives every k ≤ n.

Falsifier

What would falsify this route

The identity fails if the first binomial coefficient is read as zero for a negative upper argument: at (n, k, r) = (1, 1, −1) the left side is 2 while that reading of the right side gives 1. The source's remark that the upper argument may be negative fixes the generalized reading.

Evidence

Evidence to collect

Exact integer computation confirms the identity with the generalized binomial coefficient for n ≤ 10, 0 ≤ k ≤ n, |r| ≤ 25 (3366 cases), and finds 1276 failures in the same range under the zero-for-negative-upper-argument reading. The first rows at r = 0 are 1; 0, 2; 0, 6, 8; 0, 24, 72, 48; 0, 120, 600, 864, 384.

The canonical source is D5/S3/Quantum/FockSpace/BosonOrderingStirlingClosedForm.lean. Its public declarations are stirlingHat, conjectureSum, claim, and result; fB and B are private non-proposition definitions. The binomial coefficient with an integer lower index is the shared D5/S0/Conventions/IntegerIndexBinomial.binom. The frozen module state has statement identity sha256:d4147b129fcc7eee06209fce5e10b131b8f22ed69429327ecf75b7449790e19a. The result declaration has statement identity sha256:297d337df538c5c402555676c2199e22ec0e82a59ad18bbb66bd1b12ac7d72c2. The Freeze event is sha256:00e6e18e6b7a732583f48bdfaa03e7546f7d31655d05d9212a1ae1d711c8af94 and its project-level frozen prerequisite is D5/S0/Conventions/IntegerIndexBinomial. The proof uses only the standard axioms propext, Classical.choice and Quot.sound; no sorry, native_decide, or new axiom.

Triage

Scope assessment

Tier 1 external named conjecture, preregistered in issue #10170 before the probe. theorem; resolution proved under the generalized-binomial reading. The public theorem has proof_shape: content: the recursions (K) and (R) for the conjectured sum and the layered induction are new propositions on the live proof path, not instances of pinned lemmas. Its escape witness is form (2), the public conclusion itself, and its admission basis is open-problem-resolution. Utility none: it is a theorem for every n, k and r.

ASSUMED-UNVERIFIED

Unverified assumptions

Whether the journal version keeps the conjecture and its number was not checked. The bounded literature check does not establish exhaustive worldwide novelty, priority, or the absence of an independent proof. The Lean kernel does not authenticate the external source or its version history.

Dossier & anchor history

Loading release observations...