Hawking–Page transition at weak coupling in N=4 super Yang–Mills is first order, computed by Claude
An Anthropic preprint reports that a Claude research model computed the three-loop quartic coefficient in the Polyakov-loop effective action of super Yang–Mills on , settling a question set up in 2003: the weak-coupling deconfinement transition is first order, like the Hawking–Page transition it is dual to.
- Lab
- Anthropic
- Model
- An internal Claude research model; Anthropic's post presents the results under Claude Fable 5.2
- Field
- Physics
- Date
- 2026-11-07
- Human collaborators
- Hongbin Chen
- Problem posed
- 2003 · open 23 yrs
What was found
In anti-de Sitter space a black hole forms through a first-order phase transition found by Hawking and Page in 1983, which in super Yang–Mills on is deconfinement at strong coupling. The free theory has a transition at its Hagedorn temperature, and Aharony, Marsano, Minwalla, Papadodimas and Van Raamsdonk (hep-th/0310285) suggested in 2003 that the two could be connected either by a single first-order line or through a pair of continuous transitions at weak coupling, with the answer decided by the sign of a quartic coefficient that first appears at three loops. The preprint reports , with the 't Hooft coupling, so the transition is first order and lies slightly below the Hagedorn temperature, with its three saddle points appearing, exchanging dominance and disappearing in the same order as thermal AdS and the small and large black holes. The computation is Hamiltonian, almost every number in it was obtained in two independent ways, and it reproduces the two-loop Hagedorn temperature known from integrability and exact results in closed subsectors.
Novelty check
arXiv, 2026-11-19, abs:"Hawking-Page" AND abs:"weak coupling" returned 7 records and abs:Hagedorn AND abs:"Polyakov loop" AND abs:"three loop" returned 2, all from 2004 to 2019; none reports the three-loop coefficient for . The preprint says it had not been computed in the two decades since 2003, and Anthropic's post says the same group completed the analogous calculation for a simpler theory in 2005. This registry's nearest entry, 2026-09-25-nine-loop-hexagon-amplitude, is a different calculation.
Caveats and known objections
Capped at claimed as a lab announcement. The preprint is on Hexagon, deposited by Anthropic, and labelled primarily AI-generated with human understanding of some parts. Its provenance says the model was prompted by a physicist, who checked it against several known special cases, and that a separate Claude computation by a different method gives the same result. Anthropic's post adds that a physicist outside the project had independently used Claude on the same problem and reached the same answer by a different method; that physicist is not named, and both checks used Claude. The post says the results were formally verified in Lean in September, but no Lean development is linked, so no formal verification is credited here. Autonomy is ai-led: per the post, Hongbin Chen asked Claude to list open problems and set agents running, and the agents worked out the calculation, but the problem itself was framed by physicists in 2003 and a human checked the result.
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). Hawking–Page transition at weak coupling in N=4 super Yang–Mills is first order, computed by Claude. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-11-07-hawking-page-weak-coupling
BibTeX
@misc{whataifound-anthropic-2026-coupling,
title = {Hawking–Page transition at weak coupling in N=4 super Yang–Mills is first order, computed by Claude},
author = {{whataifound.org}},
year = {2026},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by Anthropic. Verification: Claimed. Autonomy: AI-led.},
url = {https://whataifound.org/finding/2026-11-07-hawking-page-weak-coupling}
}