trureturingGitHub MATHEMATICAL ATLAS

RESEARCH / Focused target

First letters of permutations with a fixed inversion number

Literature status: not recheckedRepository record: proved

RELEASED FOUNDATIONS / PROPOSED CONNECTION

Research connections

0 released anchors
All source anchors (1)

Problem

The question

Peter Kagey, “Expected value of letters of permutations with a given number of k-cycles”, arXiv:2112.05281v1, Section 5.2, page 20, Conjecture 5.1 states:

For fixed k and n > k, the expected value of the first letter of a permutation with a given number of inversions satisfies a rational function in n given by E[π(1) | π ∈ S_n, inv(π) = k] = M(n + 1, k)/M(n, k), where M(n, k), as above, is the number of permutations w ∈ S_n such that inv(w) = k.

Statement echo

| Source clause | Formal counterpart | Fidelity | | --- | --- | --- | | Fixed $k$ and $n>k$ | claim universally quantifies both natural parameters and assumes $k<n$. | Exact. | | Permutations in $S_n$ | Equiv.Perm (Fin n) and its finite universal set. | Each permutation occurs once. | | One-based first letter | PSW.word, defined by the existing owner as the list of values plus one; List.headD 0 selects its first entry. | $k<n$ excludes the empty permutation. | | Inversion number $k$ | ListInversions.inv applied to PSW.word. | Strict inverted pairs. | | $M(n,k)$ | M, cardinality of the inversion fiber. | Literal source cardinality. | | Conditional expectation | E, rational first-letter sum divided by rational cardinality. | Literal sum-over-cardinality convention. | | Ratio $M(n+1,k)/M(n,k)$ | The conclusion of claim and result. | Both casts and the strict guard are retained. |

Rational division is total in Lean. The theorem proves the ratio identity directly; its statement does not include a separate positivity assertion for $M(n,k)$.

Motivation

Our foothold

Preregistration #19726 identifies the external tier-1 conjecture and its literal conventions. The settling GID is D5/S3/Combinatorics/Permutation/KageyInversionFirstLetter.result, with admission basis open-problem-resolution (#19726; Proved). Its Scribe carries one OpenProblemResolutionClaim with resolution Proved.

Gap

Missing bridges

The source and the inspected dissertation state this assertion as a conjecture. Finite enumeration alone cannot settle its universal quantifiers. The inspected later sources and repository search supplied no settlement in their searched scope; literature completeness is ASSUMED-UNVERIFIED.

Route

Proposed approach

The order-preserving first-letter decomposition removes exactly the rank of the first letter from the inversion count. Applying this decomposition twice gives two bounded inversion digits. Under $k<n$, a diagonal with total contribution $t$ has $t+1$ admissible digit pairs. Finite sum interchange therefore identifies the natural first-letter weighted numerator with $M(n+1,k)$. Casting this numerator identity to the rationals proves the equality without cancelling a denominator.

Falsifier

What would falsify this route

A pair of natural numbers $k<n$ whose exact rational conditional first-letter mean differs from $M(n+1,k)/M(n,k)$ would refute the source claim. A failed inversion-insertion equation or guarded diagonal count would refute the corresponding proof route.

Evidence

Evidence to collect

The module is D5/S3/Combinatorics/Permutation/KageyInversionFirstLetter.lean. Its sole public theorem D5.S3.Combinatorics.Permutation.KageyInversionFirstLetter.result proves claim for all natural $k<n$. The public definitions are M, E, and claim. The module reuses ListInversions.inv, PSW.word, and Finset.expect_eq_sum_div_card on the proof path.

The four-slot escape audit for result is unfinished: #20067. There is no compiled registration establishing the required source-family reconstruction, readout correspondence, observational dependence and independent interventions; no declared_validated status is asserted.

Experiment

Exact finite enumeration is stored in the experiments repository. Command: python3 docs/reports/kagey-2021-first-letter-expectation-conjectures-5-1-5-2/check.py, run from that repository, exit 0, RESULT PASS. The script SHA-256 is d9a4f92ebf3ce0c7391a2632fd166c4f11b3c8890368268df7665e7708b58beb. Its Conjecture 5.1 checks cover exactly $0\le k<n\le8$, enumerating permutations through size nine to obtain the next Mahonian counts. These finite readings support the universal theorem and do not replace its proof.

Triage

Scope assessment

  • Proved, kernel-checked: the exact binding claim for every natural $k<n$, by KageyInversionFirstLetter.result.
  • Source identification, literature reading: Kagey Conjecture 5.1, arXiv:2112.05281v1, Section 5.2, page 20; external tier 1, preregistration #19726.
  • Mechanism, kernel-checked: the private inv_decompose gives the first-letter inversion contribution; weighted_identity gives the universal natural numerator identity used by result.
  • Computed: the pinned experiment checks the ratio for exactly $0\le k<n\le8$; its command, exit code and script identity are given above.
  • Boundary, proved in prose: for $n=k=3$, the only permutation with three inversions is $321$, so the first-letter mean is $3$. The six permutations of size four with three inversions are $1432$, $2341$, $2413$, $3142$, $3214$, and $4123$. Thus $M(4,3)/M(3,3)=6$, and the ratio identity fails without the strict guard. This finite boundary is not a separate Lean theorem.
  • Sibling, source identification: Conjecture 5.2 concerns the major index and has separate preregistration #19720; this inversion theorem does not settle it.
  • Open: a separately stated denominator-positivity theorem, extensions beyond $k<n$, and exhaustive literature screening. The required integration consists of the frozen settling module, its Scribe and this dossier; the literature boundary remains explicit.

What the settlement shows

The decisive identity is the weighted numerator, established by the private weighted_identity and used in the public result. It covers all $k<n$, including the length-one case, with the original conventions. Two successive insertions count the same objects as the first-letter weight because their inversion digits have an untruncated diagonal under the strict guard. The boundary above shows that removing the guard does not preserve the identity. The source's major-index question is a separate statement; no consequence for its settlement is claimed here.

Declaration classification

The sole public theorem result has proof_shape: content; its surviving intermediate witness is the private weighted_identity, whose derivation uses inv_decompose. The module's admission basis is open-problem-resolution (#19726; Proved); generality: I, utility: none. Its internal length-one decision is an induction base, not a standalone computational certificate.

| Declaration | proof_shape | escape_witness | Direct consumers | | --- | --- | --- | --- | | result | content | weighted_identity | settling conclusion | | E_eq_expect | bind-only | none | result | | M_decompose | content | inv_decompose | double_decompose | | count_ofFn | bind-only | none | inv_decompose, inv_ofFn_congr | | double_decompose | content | inv_decompose | weighted_identity | | inv_decompose | content | inv_ofFn_congr | weighted_decompose, double_decompose, M_decompose | | inv_ofFn_congr | content | inv_ofFn_congr | inv_decompose | | rectangle_count | bind-only | none | weighted_identity | | sum_digit | bind-only | none | rectangle_count | | sum_unit_digit | bind-only | none | rectangle_count | | weighted_decompose | content | inv_decompose | weighted_identity | | weighted_identity | content | inv_decompose | result |

ASSUMED-UNVERIFIED

Unverified assumptions

Literature completeness is ASSUMED-UNVERIFIED. The supplied source screen covers arXiv v1, the author's dissertation, arXiv:2210.17021v2, arXiv:2602.11061v1, Conger's descent paper arXiv:math/0508112, and the OpenishProblems catalog. Unavailable screens include Semantic Scholar (HTTP 429), a thesis source (HTTP 403), and challenged general search engines. Absence of a settlement in this scope establishes neither exhaustive priority nor that no other proof exists.

Dossier & anchor history

Loading release observations...