A relation-incidence proof of Tohăneanu's low-exponent supersolvability conjecture
A hyperplane arrangement is a finite collection of flat subspaces of codimension one. Its intersection lattice records how those hyperplanes meet. This note proves that a small set of allowed exponents forces that lattice to have a particularly orderly structure: a maximal chain of modular flats.
The result is a relation-incidence proof of Tohăneanu’s conjecture. Ferudun’s 12 October 2026 preprint proves the same complete theorem by a related dependency-support count and terminal-triple induction. This page makes no claim of first priority or independent development.
Read the manuscript
Open the original PDF · Download PDF · BibTeX · Zenodo record
The PDF is unchanged and dated 13 October 2026. It prints no numeric version; version 1.0 is the Zenodo metadata. The record was created on 13 October 2026 (UTC). Use the direct link if the embedded reader is unavailable.
Original abstract
We prove that every finite central free arrangement over a field of characteristic zero whose nonzero exponents are 1 or 2, with at most one 3, is supersolvable. Factorization and linear-relation formality restrict the incidence graph of the multiple rank-two flats to a tree or a subdivision of K4 with trees attached. A direct modular-coatom argument then gives a maximal modular chain in every rank. This proves the full conjecture stated in Tohăneanu's Oberwolfach contribution. We credit the earlier all-2, tree, rank-three and modular-pruning results and acknowledge Ferudun's later matching full-target preprint. The accompanying package preserves the dated candidate, source audit and exact computational diagnostics. AI tools were used extensively in developing, writing and checking this work. This preprint is unrefereed; its internal AI reviews are not human peer review.
The theorem and its scope
Let A be any finite central free arrangement of distinct linear hyperplanes over any field of characteristic zero. If every nonzero exponent is 1 or 2, apart from at most one exponent equal to 3, then A is supersolvable.
“Central” means that the hyperplanes pass through the origin. “Free” means that the polynomial vector fields tangent to all the hyperplanes form a free module; the coefficient degrees of a homogeneous basis are the exponents. Supersolvability is the stronger lattice conclusion: a maximal chain of flats satisfying the modular rank identity.
The theorem includes zero exponents, arrangements with a common unused direction, and reducible arrangements. Removing unused directions leaves the lattice unchanged; product factors can be handled separately and their modular chains combined. Rank zero has the trivial lattice, and ranks one and two are immediate. The argument therefore reduces to essential irreducible factors of rank at least three.
Why a small cycle bound is enough
Choose one normal vector for each hyperplane. Join each normal to every “rich line” containing it: a two-dimensional span with at least three normals. Freeness supplies two established inputs. Terao’s factorization fixes the first numerical counts, while linear-relation formality, from Yuzvinsky and DiPasquale, says that all relations among the normals come from these rank-two spans. This is linear-relation formality, rather than topological formality.
In an irreducible factor there is exactly one exponent 1. The counts make the incidence graph a tree in the all-2 case and in the one-3 case with a four-element rich line. The remaining case has only triple lines and cycle-space dimension exactly three.
Take a smallest dependent family of triple relations in that remaining case. If it has t triples on s normals, cancellation makes every normal occur at least twice. The global cycle bound then gives
2t − 2 ≤ s ≤ 3t/2.
Thus t ≤ 4. A triple needs three other triples to cover its three normals, since two rich lines share at most one normal, so t ≥ 4. Exactly four triples on six normals remain: the incidence subdivision of K4. This core uses all three cycles, forcing each outside component to be a tree attached at one core normal.
The final step removes terminal pencils while retaining their attachment normal. Formality shows that each removal lowers the rank by one and preserves formality. The retained span is a modular coatom, so a modular chain there lifts back to the full configuration. The trees end at a rank-two pencil; the cyclic case ends at the six-normal rank-three core, where the chosen triple line also meets every ordinary rank-two flat in a represented normal. Reversing these removals gives the required chain in every rank.
Earlier work and the matching theorem
Tohăneanu’s Oberwolfach contribution, printed pages 283–286, states the target. His earlier work supplies the all-1/2 result and precedents for rank-two counts and modular pruning. Denham–Garrousian–Tohăneanu give tree and parallel-connection structure, and Tohăneanu’s quadratic-derivation classification covers earlier rank-three structure. The incidence graph, tree mechanism, pruning and supersolvability of the core are credited ingredients.
DiPasquale’s Corollary 4.10 and Proposition 6.1, numbered in arXiv:1806.05295v1, provide a predecessor for the two tree cases: their local relation spaces are independent and meet the TF2 condition. The cyclic case has a one-dimensional kernel in the local-sum map, so that proposition alone does not cover the complete target.
Tohăneanu’s low-exponent manuscript was withdrawn in 2022 over specified later enumeration and inference steps. The present proof uses neither affected assertion and proves its multiplicity and pruning steps directly. The withdrawal does not erase the earlier contributions credited here.
Ferudun’s matching preprint proves the same full target with a related method. This proof isolates the core using the cycle bound; his argument obtains a removable terminal triple through a leaf count outside a dependency support. The archive also preserves an October 4 candidate record, but that bounded chronology establishes neither exhaustive first priority nor independent development.
Boundaries
Characteristic zero, distinct hyperplanes and the single exponent 3 are retained. The free nonsupersolvable non-Fano arrangement with exponents (1, 3, 3) prevents simply allowing two cubic exponents. No arbitrary-characteristic, affine-arrangement or multiarrangement extension is asserted. The universal conclusion comes from the proof; finite diagnostics in the archive are supporting checks.
Original files and citation
- Manuscript (PDF, 69,859 bytes) · Zenodo original
- Source and verification package (ZIP, 206,394 bytes) · Zenodo original
Kriebel, Alec. A relation-incidence proof of Tohăneanu's low-exponent supersolvability conjecture. 13 October 2026. Zenodo, version 1.0. doi:10.5281/zenodo.23331709. Download BibTeX.
The paper and original authored supplementary work are licensed CC BY 4.0. Third-party works retain their own rights.
References
- S. O. Tohăneanu, Free arrangements with low exponents, Oberwolfach Reports 5/2021, 283–286. DOI.
- S. O. Tohăneanu, Free arrangements with low exponents, 2017 manuscript; 2022 withdrawal notice.
- G. Denham, M. Garrousian and S. Tohăneanu, Modular decomposition of the Orlik–Terao algebra of a hyperplane arrangement, Ann. Comb. 18 (2014), 289–312. arXiv.
- S. Tohăneanu, Projective duality of arrangements with quadratic logarithmic vector fields, Discrete Math. 339 (2016), 54–61. arXiv v2.
- M. DiPasquale, A homological characterization for freeness of multi-arrangements, Math. Ann. 385 (2023), 745–786. DOI; v1 numbering used here.
- A. Ferudun, Free Hyperplane Arrangements with Exponents 1, 2 and at Most One 3 Are Supersolvable: A Proof of a Conjecture of Tohăneanu, 12 October 2026, v1.1. DOI.
- H. Terao, Generalized exponents of a free arrangement of hyperplanes and Shepherd–Todd–Brieskorn formula, Invent. Math. 63 (1981), 159–179. DOI.
- S. Yuzvinsky, The first two obstructions to the freeness of arrangements, Trans. Amer. Math. Soc. 335 (1993), 231–244. DOI.
Status and collaboration
AI tools were used extensively in development, writing and checking. This is an unrefereed preprint; internal AI reviews do not constitute independent human peer review or formal proof certification. The author is responsible for the claims.
Independent verification, substantive extensions and journal preparation would be useful contributions. Collaboration and authorship would be based on actual contributions.