Bernardi–Morales's delta-process conjecture

What Lean checks

Lean proves published Conjectures 27–28 for all admissible parameters: the literal delta-map tree count for every fixed color map, and the complete marked interval-event count and source probability determinant for every fixed surjection. Full subsets, singleton complements and the singular averaging case are included. The 13-module closure passed strict compilation with warnings treated as errors, an all-owned audit of 269 declarations (205 theorem declarations and 23 required endpoints), and independent replay by the pinned stock kernel. Only propext, Classical.choice, and Quot.sound are permitted. Receipt. Combined package gate and hosted validation remain separate.

Audit scope

Search date: 20 November 2026. Targets: the whole delta-process tree-probability identity, published as Conjecture 27 and numbered 21 in the original preprint, and its stronger event-determinant identity, published as Conjecture 28 and numbered 22 in the preprint. These two complete source statements are documented under one named conjecture investigation.

No earlier proof, disproof, or announcement of a resolution was located in the searches below. This records the evidence and its limits; absence from search results cannot certify the absence of unpublished or unindexed work.

Primary statement and author status

Source What was checked Finding
Bernardi–Morales, published paper, EJC 23(2), P2.36 Section 5, pages 22–24; definition of delta; Conjectures 27–28; relation to the earlier theorems; interval-event definitions on page 8 The paper explicitly leaves Conjecture 27 unproved. Conjecture 28 is stronger. Its endpoint cases with no marked rows or every nonroot row marked are already proved there, but the assertion for arbitrary marked sets is left conjectural.
arXiv:1501.01135v1 Section 5, page 23, and submission history The same statements are Conjectures 21 and 22. The arXiv record lists only the 6 January 2015 version. The journal version renumbers them; using only the preprint numbers would be misleading.
Alejandro H. Morales, conjectures The entry for Probabilistic trees with algebraic roots, and surrounding status updates The entry still lists Conjecture 21 without a resolution reference. Other entries explicitly record proofs. This is supporting status evidence, not proof that the list is exhaustive or fully current.
Olivier Bernardi, publications Original article entry and subsequent publications through the displayed 2026 items The original article remains listed; no delta-process resolution is identified on this page.
Morales, papers by topic Original article entry and related enumerative work No later delta-process proof is linked.

The MathDB entry was read in full as a secondary discovery source. It displayed zero solutions and a status summary that had never been refreshed. That status label was not used alone to establish openness.

  1. Lewis–Morales, Factorization problems in complex reflection groups, DOI 10.4153/S0008414X2000022X. The passages citing Bernardi–Morales 2016 concern combinatorial proofs of factorization formulas. The citation discussion near Question 8.5 concerns simpler proofs of those formulas. These passages do not resolve the delta-process conjecture.
  2. Lewis–Morales, GLn(Fq)-analogues of factorization problems in the symmetric group. Indexed text and the bibliography were searched. This is related factorization work. Direct retrieval later returned HTTP 403, so this is recorded as an indexed-text check, not a complete inspection of the paper.
  3. Combinatorial and Algebraic Enumeration: a survey of the work of Ian P. Goulden and David M. Jackson. Search-indexed bibliography cites the 2016 paper. Direct retrieval returned HTTP 403. The retrieved information did not identify a resolution, but this source was not fully inspected.
  4. Bernardi–Morales, Counting trees using symmetries. Its abstract and its placement in the authors' publication lists identify a different multitype-Cayley-tree enumeration problem predating the target conjecture. It is not being recounted as a solution of the delta process.
  5. Dereziński–Liang–Mahoney, Exact expressions for double descent and implicit regularization via surrogate random design. Definition 4, Lemma 4, and Appendix B were inspected. These introduce the all-minor determinant-preserving property and prove closure under independent addition and multiplication. Those general closure facts are prior mathematics and are explicitly credited in the proof. The paper does not identify the Bernardi–Morales cyclic process; a full-text search also found no Schur-complement discussion.
  6. Dereziński–Mahoney, Determinantal Point Processes in Randomized Numerical Linear Algebra, also Notices of the AMS, DOI 10.1090/noti2202. The determinant-preserving framework and its relationship to determinantal point processes were checked. The present work does not claim that framework as new.

The Semantic Scholar citation link from the arXiv record could not be opened through the research tool. The author's old supplementary-code link redirected to sign-in. Neither inaccessible source is counted as inspected.

Search families

Both available search engines were used. Searches included the title, author combinations, both conjecture numbers, spelling variants, and the proof mechanism. Queries returning only the original paper were followed by exclusions of its hosting sites to expose later citations. Representative exact queries were:

  • "Bernardi" "Morales" "delta" conjecture
  • "Bernardi" "Morales" "Conjecture 21"
  • "Bernardi" "Morales" "Conjecture 22"
  • "Bernardi" "Morales" "Conjecture 27"
  • "Bernardi" "Morales" "Conjecture 28"
  • "Bernardi" "Morales" "Conjecture 28" proof
  • "probabilistic trees" "Conjecture 22" proof
  • "probabilistic trees" "Conjecture 28" proof -site:academia.edu -site:combinatorics.org
  • "Bernardi" "Morales" "Pdet"
  • "Morales" "Bernardi" "adjugate"
  • "Morales" "probability determinant" conjecture
  • "Bernardi" "Morales" "determinant" conjecture site:mathdb.com
  • "1501.01135" conjecture
  • "Some probabilistic trees with algebraic roots" proof conjecture
  • "Some probabilistic trees with algebraic roots" "Schur"
  • "Some probabilistic trees with algebraic roots" -site:academia.edu -site:combinatorics.org -site:arxiv.org -site:mathdb.com
  • "delta-process" conjecture proof
  • "delta-process" "proof" "tree" -site:mathdb.com -site:academia.edu
  • "Bernardi-Morales" "tree" conjecture
  • "Bernardi–Morales" "delta"
  • "Bernardi" "Morales" "tree probability"
  • "Bernardi" "Morales" "Schur complement"
  • "cyclic" "Schur complement" "determinant preserving"
  • "determinant-preserving" "Schur complement" expectation

A search restricted to the most recent two years for the exact delta-process name was also used. The similarly named Delta Conjecture in symmetric-function theory is unrelated and was discarded.

Precise novelty claim

The contribution claimed here is a proof of the complete Bernardi–Morales tree-probability identity, together with the stronger statement that the same count holds for every fixed color map, and a separate proof of the paper's full event-determinant conjecture for all marked sets. It is not a claim to invent Schur complements, adjugate identities, Cauchy–Binet, determinant-preserving random matrices, or the matrix-tree theorem. The already proved endpoint cases of the event conjecture are not counted as new results.

The essential identification is specific: a proper subset of a directed cycle produces the delta Laplacian by eliminating that subset, and every proper principal minor of the original cycle Laplacian is 1. The resulting arbitrary-matrix generating identity proves all-minor preservation under precisely the subset weights that encode the original membership constraints. Replacing independent color blocks and extracting those membership coefficients then proves the published conjecture.

The event proof uses a different row-replacement identity. Its full-subset cancellation is essential: the adjugate becomes the all-ones matrix, and the fixed last row makes the determinant vanish. The complementary-column expression vanishes for the same reason. Bernoulli averaging then proves the entire event polynomial, not just its two previously known endpoint cases.

The full written proofs passed separate internal reviews. The parent agent developed the event proof; the analysis agent checked it independently and wrote a direct interval-event verifier. External peer review and author confirmation have not been obtained. The computational checks accompany the proofs and do not substitute for any quantified step.