Erdős·erdos:993
Erdős 993
No independent check recorded yet. A formal artifact, declaration or published object is attached, but no rebuild of it is recorded here.
A Palomar registration names this problem. It is shown below but not counted as a check of the claim.
Fidelity F2: The correspondence is declared through a Comparator challenge, an alignment table and written divergences.
graph theory·Source
AI activity
How grades workNo AI contribution recorded against this statement.
Claimed on erdosproblems.com
2Proof claims posted on erdosproblems.com, which says that listing a claim “is no guarantee of proof correctness”. The register records who claims what, with which systems, and links to each claim there. Nobody has examined them, and none counts in the register's totals.
Fidelity
How fidelity is gradedF2 declared. The correspondence is declared through a Comparator challenge, an alignment table and written divergences.
Checks
1-
verified·Palomar
Registered by Palomar at c020ce11: Comparator confirmed 2 theorems prove the recorded statement within Palomar's axiom policy, replayed through Lean's kernel and the independent nanoda kernel. The project names this problem, which does not establish that it proves the result claimed here, so it is not counted as a check of it
Declared by the projects
1As each project's formalization.yaml states it.
Unimodality of the independence sequence of a finite forest (Erdős Problem 993)
- authors
- Kevin Vallier, Tong Zhang
- method
- agent — Claude Fable 5.1, Claude Opus 5.5, GPT-5.6 Sol, GPT-6 Astra, GPT-6 Astra (Pro mode), GPT-6 Sol, GPT-6.1 Sol
- review
- agent-reviewed (AI review sessions in Claude Code (automated): rebuild from source and axiom audit of each part; comparison of Lean statements and transcribed tables with the manuscript, Kernel checkers (automated): Lean's leanchecker, NanoDa and con-ron on the complete export; the registry's Comparator)
- axioms
- Classical.choice, Quot.sound, propext
- sorry
- 0 unproved goals declared
- results
- 2 main results named, checked with Comparator, with an alignment table
- sources
- Unimodality of Forest Independence Polynomials — adapts, authors participated; Unimodality of Forest Independence Polynomials — formalizes, authors participated; Exact Certificates for Unimodality of Forest Independence Polynomials: A consolidated proof manuscript and reproducible supplement — adapts, authors participated; Unimodality of Forest Independence Polynomials — adapts, authors participated
- related
- google-deepmind/formal-conjectures/issues/5012 — other; google-deepmind/formal-conjectures/pull/4192 — other; selfreferencing/erdos993-lean — builds-on; lixiang90/forest-unimodality — independent
- divergences
- Two comparisons, both made by AI systems on statements (for arXiv v1 also on the authors' change records); Lean proofs were not read for either. PART A. Against arXiv:2610.07943v1, the public version. Same as that paper: the main theorem; the window (Proposition 2.3, Corollary 2.5); the decomposition and mixture (Lemma 2.2, Proposition 3.2); the choice of activity (Lemmas 3.6 to 3.8, Theorem 3.9, Proposition 3.10); the moment criterion (Theorem 3.5, with one Laplace constraint); Lemma 4.30; the statements, constants and starting means of Theorems 4.3, 4.22 and 5.5. Different: (a) Theorem 4.3: other multipliers; exact rational comparisons replace interval checks. (b) Theorem 4.40: only the tilt 0, with an explicit rate. (c) Theorem 4.22: nine activity rows, another local payment. (d) Theorem 5.6 below the starting means: 28 certificates replace 779 boxes; lower edge 4/q. (e) Section 5.3, Appendix B: proved anew, without interval arithmetic. (f) Theorem 6.1: a sharper mode bound; 203 parameter triples. (g) Sections 7 and 8 describe another Lean development (411 compiled evaluations, five axioms). PART B. Against version 2.15, the text the Lean follows (deposited with restricted acces
- checked by
- Palomar
Follow and discuss
All discussionFollow this problem
An email when it has a new claim, check, bounty or discussion. You confirm once and can stop with one click.
Discussion and bounties for this problem load here.
Seen recently
What the monitors picked up in the last thirty days, not yet graded.
Something wrong or missing here? Request a correction or add a claim, with its sources.
Claims and corrections from readers
All of themFormal material
Formal statements · 0
No formal statement located.
Cited proofs · 0
No proof artifact cited by the formal record.
Also known as · 1
- https://www.erdosproblems.com/993
Cite this record
qed.bot, “Erdős 993”, https://qed.bot/s/erdos-993, as of 20 Nov 2026.