Claude Fable 5.1 Reportedly completed the nine-loop six-particle scattering-amplitude calculation in planar N=4 super-Yang–Mills theory

AT A GLANCE

The report said that Claude completed a nine-loop particle-physics calculation and received independent verification, but the original announcement, paper, and reproduction details were not provided.

Article

Claude completes the task 9 A breakthrough in frontier physics computing; experts initially thought direct calculation was too difficult

According to Beating AI news, Anthropic disclosed that Claude Fable 5.1 completed a nine-loop theoretical particle-physics calculation in Claude Science, calculating the six-particle scattering amplitude in planar N=4 super-Yang–Mills theory. Scattering amplitudes are used to calculate the probability of particle interactions. The higher the number of loops, the more corrections are added and the more difficult the calculation becomes. This specific problem had previously been publicly completed to eight loops.

Humans did not participate hands-on this time. Two Anthropic physicists first told Claude what to calculate, then mainly let it continue working, asking it to report progress every 4 to 6 hours. Claude ultimately calculated the result using two different approaches. The direct approach used Python and SymPy, running on 96 CPU for 1 weeks, at a computing cost of approximately 100 USD; including the cost of calling Claude for extended periods, a complete run of one approach cost approximately 1000 to 2000 USD.

The result was subsequently independently verified by SLAC, a Stanford physicist who had participated in the earlier eight-loop research. Dixon said he had initially thought that a nine-loop amplitude would be very difficult to calculate directly. Claude also did not invent a new physics method; it mainly carried out in full the methods, constraints, and extensive tedious calculations previously established by researchers.

Meanwhile, the team of He Song at the Institute of Theoretical Physics of the Chinese Academy of Sciences also independently advanced to nine loops with the assistance of GPT-6, and has already obtained most of the result.

Original link https://m.theblockbeats.info/flash/369068

Key points

Anthropic disclosed that Claude Fable 5.1 completed the nine-loop calculation of the six-particle scattering amplitude in planar N=4 super-Yang–Mills theory in Claude Science.
The report explained that scattering amplitudes are used to calculate the probability of particle interactions; the higher the number of loops, the more corrections are added and the more difficult the calculation becomes.
This specific problem had previously been publicly completed to eight loops.
Two Anthropic physicists first explained the calculation objective to Claude, after which they mainly let it continue working and required it to report progress every 4 to 6 hours.
Claude obtained the result using two different approaches; the direct approach used Python and SymPy, running on 96 CPU for 1 weeks, at a computing cost of approximately 100 USD。
After including the cost of calling Claude for extended periods, the report said that a complete run of one approach required approximately 1000 to 2000 USD。
The report said that SLAC, a Stanford physicist who participated in the earlier eight-loop research, subsequently independently verified the result.
Lance Dixon said that he had initially thought a nine-loop amplitude would be very difficult to calculate directly.
The report explicitly said that Claude did not invent a new physics method; it mainly fully executed the methods, constraints, and tedious calculations previously established by researchers.
The report said that the He Song team at the Institute of Theoretical Physics of the Chinese Academy of Sciences also independently advanced to nine loops with the assistance of GPT-6, and has already obtained most of the result.

AI-assisted interpretation

The following is analysis, separate from reported facts. Verify important claims independently.

“Nine loops” can be understood as continuously adding further layers of corrections to a base calculation. The higher the loop order, the more cumbersome the formulas and intermediate calculations become. According to news reports, Claude did not propose a new theory; rather, it combined existing methods, constraints, Python tools, and CPU computing power to complete a calculation workflow that had previously been publicly achieved only up to eight loops.

Why it matters to readers

If the results and verification reported here hold, this case shows that AI can execute complex scientific calculations over extended periods and organize established theoretical methods, programming tools, and computing resources into a complete workflow. It demonstrates scientific execution and automation capabilities, rather than the invention of a new physics method.

Beginners can focus on key aspects:AI whether it can continuously handle long tasks, whether the results can be verified through different approaches and by external experts, and how much computing resources and model-invocation costs are required to complete such tasks.

Risks and unknowns

The material is a media retelling and does not provide Anthropic’s original announcement, technical report, or paper.
It does not provide the formulas for the nine-loop result, the scope of its completeness, or the independent verification process, so it cannot be reviewed based on the material alone.

Claude Fable 5.1 Reportedly completed the nine-loop six-particle scattering-amplitude calculation in planar N=4 super-Yang–Mills theory

Article

AI-assisted interpretation

Why it matters to readers

Risks and unknowns

Related Developments

Related concepts

Keep following

Security Center