This paper addresses the problem of structural verification for nonequilibrium steady-state systems under declared projection and partial observability, and proposes an EML-NESS framework for generative structural auditing. Existing NESS frameworks have shifted complex-system evaluation from static point-value judgment to structural compatibility assessment, yet they remain insufficient for distinguishing between “the structure still exists” and “the structure is still maintained by a healthy generative path.” To address this gap, this paper takes EML as a prior unified generative language, reformulates observed trajectories as generative-tree reconstruction problems, and uses minimum generative cost to characterize the recursive depth required for trajectory reconstruction. It further introduces the concepts of generative residual capacity, local ellipsoidal shell sequences, flow-structure matching bandwidth, path support rate, generative complexity burden, and generative load index, thereby constructing a continuous analytical chain from symbolic reconstruction and dissipative geometric manifestation to perturbation-based stability verification. On this basis, the paper redefines the basin of attraction, extending it from a geometric domain of return to a four-dimensional feasibility domain simultaneously satisfying geometric recoverability, syntactic compressibility, statistical reproducibility, and dynamical continuity. It also proposes the concepts of generative path rupture and the S/A/X control-interpretation framework. The study argues that the health of a nonequilibrium system depends not only on whether its structure still exists, but also on whether that structure is still maintained by a low-burden, reproducible, bearable, and sustainable generative path.
国家预印本平台
评论