Clark derives full Lyapunov spectrum of random recurrent network at large N

David G. Clark has derived the full Lyapunov spectrum at large N for the Sompolinsky-Crisanti-Sommers random recurrent neural network, reducing the problem to a simple self-consistent single-site calculation.

The result addresses a question in theoretical neuroscience posed about four decades ago.

Andrew Curran reports on the paper and says it was written in collaboration with GPT-6 Astra and Claude Opus 5.5. His post itself is a single sentence, so the description of the paper's content rests on the quoted context from Clark rather than on Curran's own account.