← AMRAL · PROGRAM-P-VS-NP
Dynamic Four-Layer Closure Framework · Led by Neo.K, collaboratively organized with Aletheia
The next reformulation the researcher proposed after the 24-round dual rehearsal:
splitting the traditional P vs. NP problem into four interrelated, functionally distinct observation planes —
the Global Computational Complexity state GCC, the Universal State-Rate Transformation state
USRT, the Universal Sufficient-Sequence Generation state USEG, and the Global Lossless
Closure state GLC, which serves as the capping acceptance condition. The core engineering-style
statement is: "the process is free, the final ledger is not free." An algorithm may switch
paths, change representations, roll back, and recompute midway, but must ultimately deliver a result that is
correct, complete, has zero semantic loss, and passes resource-ledger accounting. The document explicitly
states its own position — a heuristic re-description plus a formalization research agenda, not a
completed proof of P vs. NP — and lists seven or eight research red lines it explicitly does not
license.
This page presents both "research handoff" documents side by side rather than picking a side for the researcher — the two are different execution-order proposals for the same four-layer framework, differing on whether GLC should serve as the final acceptance layer or be moved to the front as the semantic foundation.
Same framework, two different research execution orders — both written for "the AI picking this up next," so the handoff lets the work continue without re-guessing the project's intent.
GCC→USRT→USEG→GLC. Eight parallel research lines (axiomatization, equivalence arrows, non-circularity, formal proof, algorithmic implementation, counterexample testing, model invariance, obstruction mapping), a four-phase research plan, seven research red lines.
Plan B · GLC-FirstGLC→{GCC,USRT,USEG}. Defines "what genuine completion means" first (five GLC axioms, standard/robust versions, the non-circularity principle) — the other three layers are then built on top of GLC.
A separate research line independent of the two documents above: adversarial verification of a related engineering candidate by 7 AI roles, ongoing since 2026-08-09.