What was already known
The N=d=4 case was already known from Kevin M.’s 2023 work. Earlier restricted results did not settle the full N=6 and N=10 diagonal cases.
AI involvement
The DeepMind researchers and Mario Krenn supplied the questions and formal targets; the report credits the agent with resolving the diagonal cases. Krenn later summarized their relation to earlier work.
The paper credits the agent with resolving the central diagonal cases, while the research team supplied the questions and formal statements.
Importance
The reported N=6 and N=10 exclusions extend the known special cases for quantum-optical GHZ graph constructions.
Limitations
The report does not identify the exact coefficient domains or theorem-matched proof artifacts for these cases. A later sharper Lean result is separate and does not check the agent’s original proof.
Technical verification
The agent runs in a Lean-checked framework, and Krenn calls this a proof, but the reviewed sources do not identify a completed theorem-matching check or accessible proof for these specific diagonal cases. Reported is the evidence default, not a failed check.
Human understanding
Krenn explains the graph-to-quantum-state interpretation and baseline, but a human explanation of the new diagonal obstruction is not documented; Few is the evidence default.
Correction and publication history
The reported N=6 and N=10 exclusions extend the known special cases for quantum-optical GHZ graph constructions.
Read this published revision