Explicit restricted isometries with k = m^(2/3 − o(1)), first proved in one shot by GPT-6 Astra
Kümmerle and Mixon construct explicit real restricted isometry matrices whose sparsity level reaches , past the square-root barrier that conventional wisdom attached to proofs of this kind, on a problem Tao posed in 2007; their declaration says the main theorem was proved in one shot using GPT-6 Astra.
- Lab
- Independent
- Model
- GPT-6 Astra (main theorem); ChatGPT and Claude Opus 5.5 (exploration, drafting and feedback); an Opus 5.5 orchestrator with GPT-6.1 Sol and GPT-6 Astra subagents (Lean)
- Field
- Mathematics
- Date
- 2026-11-10
- Human collaborators
- Christian Kümmerle, Dustin G. Mixon
- Problem posed
- 2007 · open 19 yrs
What was found
A matrix has the restricted isometry property of order when it nearly preserves the length of every -sparse vector, which is what compressed sensing needs. Random matrices achieve nearly proportional to ; Tao asked in 2007 for comparable explicit constructions, and the abstract notes the conventional wisdom that proofs using only the entrywise absolute value of the Gram matrix stall at . The paper combines Gershgorin's circle theorem with a flat RIP estimate to reduce the problem to choosing a large vertex set in a strongly regular graph in which every small subset spans few edges, and selects that set using an incidence result of Zahl. It also proves that methods using only the entrywise absolute value of the Gram matrix face a barrier at , not as had been believed, and that this construction meets it. The authors write that while digesting the AI proof they distilled the Gershgorin and flat RIP hybrid, identified the 2/3 barrier and explored the general approach over several long conversations with ChatGPT.
Novelty check
arXiv, 2026-11-19, abs:"restricted isometry" AND abs:explicit AND abs:deterministic returned 5 records, including explicit RIP matrices from algebraic geometry (2015) and Gershgorin-type RIP constants (2019); none passes exponent 1/2. Nothing in this registry covers compressed sensing.
Caveats and known objections
Entered at claimed. The paper states that all theorems, lemmas and propositions, including the case of Zahl's theorem used, are formalized in Lean 4 with Mathlib in github.com/ckuemmerle/ExplicitRIPTwoThirds, with the running time of the construction outside the formalization; that repository was not rebuilt for this entry and carries no Palomar registration, so the formalization is not credited here. The authors state that all mathematical claims, citations and computations were checked by them, which could support author verified. It is a partial answer: exponents above 2/3 stay open, and the paper shows this method cannot reach them. Autonomy is ai-led: the central theorem came from the model in one shot, and the humans understood, generalized and rewrote it.
Also recorded at
vibemathedoptimal-explicit-restricted-isometry
Nothing mechanically. It is a parallel listing of the same result, carrying its own verification label rather than an independent check.
No independent check is recorded on this entry yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Claimed and AI-led.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Graded claimed for verification and ai-led for autonomy. What these mean.
Cite this entry
whataifound.org. (2026). Explicit restricted isometries with k = m^(2/3 − o(1)), first proved in one shot by GPT-6 Astra. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-11-10-explicit-rip-two-thirds
BibTeX
@misc{whataifound-independent-2026-thirds,
title = {Explicit restricted isometries with k = m^(2/3 − o(1)), first proved in one shot by GPT-6 Astra},
author = {{whataifound.org}},
year = {2026},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by Independent. Verification: Claimed. Autonomy: AI-led.},
url = {https://whataifound.org/finding/2026-11-10-explicit-rip-two-thirds}
}