Hopfield-network proofs

This entrypoint collects the discrete Hopfield proof development:

• Boolean state bookkeeping;
• the classical symmetric-zero-diagonal energy function;
• one-coordinate update energy monotonicity;
• full-sweep progress; and
• convergence to a fixed point within a finite bound.

The proofs follow the standard Lyapunov-energy analysis of Hopfield networks, stated over TorchLean's spec-level model so the theorem layer and model layer share one semantics.

References:

• Hopfield, "Neural networks and physical systems with emergent collective computational abilities", PNAS 1982.
• Hopfield, "Neurons with graded response have collective computational properties like those of two-state neurons", PNAS 1984.