Kadison's similarity problem
From OpenAI Aftermath
Kadison's similarity problem asks whether every bounded representation of a C*-algebra on a Hilbert space is similar to a *-representation. A manuscript in OpenAI's public mathematics collection (github.com/openai/math), dated 23 September 2026, claims a positive answer in full generality. This page summarizes the published claim and lists publicly available follow-up work. Statements are the respective authors' claims; the publication of a paper or a formal proof is not presented here as independent verification.
Background
- As recounted in the introduction of the manuscript:
- Kadison asked in 1955 whether every bounded, complex-linear, unital algebra homomorphism from a unital C*-algebra into the bounded operators on a Hilbert space becomes a *-homomorphism after conjugation by a bounded invertible operator.
- Bunce and Christensen settled the nuclear case. Haagerup identified complete boundedness as the criterion for similarity, solved the cyclic case and treated C*-algebras without tracial states; Christensen treated type II1 factors with property Γ. Kirchberg showed that, for each unital C*-algebra, the similarity property is equivalent to innerness of derivations in every faithful *-representation. Pisier organized quantitative bounds through the similarity degree, and Eleftherakis and Paulsen characterized a positive answer by hyperreflexivity of all von Neumann algebras.
The original result
- Claim. Theorem 1.1 of Kadison's similarity theorem through uniform derivation estimates states that, for every unital complex C*-algebra A, every complex Hilbert space H and every bounded, complex-linear, unital algebra homomorphism π from A into the bounded operators on H, there is a bounded invertible operator S such that a ↦ Sπ(a)S−1 is a *-homomorphism. No separability, nuclearity, finite generation, faithfulness or normality is assumed. A corollary states that every unital von Neumann algebra on a complex Hilbert space is hyperreflexive with one universal constant.
- Method, as described: the key input is a uniform commutator estimate: an absolute constant bounds the commutators of every finite matrix amplification of an operator with a von Neumann algebra by its commutators at the scalar level. The hard case, finite factors, goes through a tracial free product with a II1 factor, averaging over finite unitary groups with Schur–Weyl duality, a Hankel matrix estimate, and Popa's relative independence theorem in a tracial ultrapower; central decomposition then gives the general case.
- Source (openai/math repository, commit
7ef5c0c): Kadison's similarity theorem through uniform derivation estimates, dated 23 September 2026. - Formalization, as described by the repository. The scope note lean/docs/288.md states that the formalization proves the similarity theorem for arbitrary Hilbert spaces and the universal hyperreflexivity constant, and that the linked commutator estimate is uniform over all finite matrix amplifications. It links the comparator statements KadisonSimilarity.lean and UniformCommutator.lean.
Public follow-up work
| Date | Work | Relation | Public claim | Public status (checked 28 October 2026) |
|---|---|---|---|---|
| 2026-10-07 | Kadison similarity: uniform commutators and free-product corners, an article of the Latent Seminar blog at lowrank.github.io | Reading (exposition) | Explains the manuscript's affirmative answer to Kadison's similarity problem without separability assumptions: a central estimate controls every matrix amplification of a commutator by its scalar norm, and the proof passes through cyclic modules, free-product words and a Fourier compression whose trace factors cancel. Sections treat complete boundedness, rectangular derivations, the free-product corner estimate, the transfer to arbitrary algebras and universal hyperreflexivity. | Blog article; it states that its review covers statements, introductions and selected proof mechanisms and is not an independent verification of the complete proofs. |
| 2026-10-09 | Kadison's similarity problem, a course on Open Mathematics Courses (kokunoyumeto/open-math-courses-public) | Reading (exposition) | Ten lessons: completely bounded homomorphisms and Paulsen's similarity theorem, implementation of derivations and Arveson's distance formula, the noncommutative little Grothendieck inequality with row and column estimates, Kirchberg's passage from inner derivations to similarity, words in free subalgebras, Schur–Weyl invariants, propagation from small corners and a Hankel test, transfer to II1 factors through Popa's free independence in ultrapowers, the uniform commutator estimate, and the similarity theorem with universal hyperreflexivity of von Neumann algebras. | The lessons state that they were written by Claude Opus 5.5 (Anthropic) in October 2026 and self-checked by the writing AI; the course lists its status as author self-checked. |
| 2026-10-18 | AI Proved What?, episode 288 Kadison’s question answered, proved at last?, remade version of 23 October 2026 | Reading (exposition) | Video explainer of the claim that every bounded unital homomorphism from a unital complex C*-algebra into operators on a Hilbert space becomes a *-homomorphism after conjugation by a bounded invertible operator, through one absolute constant that bounds matrix-level commutators with any von Neumann algebra. The episode states that a value for Pisier's similarity degree, a bound on how much the conjugation distorts, and Dixmier's separate problem for groups are not claimed. | Web page and video; the site states that its scripts and animations are produced with AI assistance and that the narration is an AI-generated voice. Status record last checked 27 October 2026: posted as a preprint: OpenAI, 31 pages, dated September 23, 2026; under expert review: not yet; peer reviewed: not yet; checked in Lean: twice, the full theorem and the key estimate. The page also states that the site scanned the two Lean formalizations, of the full theorem and of the commutator estimate, without rebuilding them and found no unproved steps, and that it found no comment from specialists. |
Notes
- The background summarizes the manuscript's own account of earlier work.
- The list covers follow-ups found in public sources up to the date above; it is not a priority or completeness claim.
