We introduce the OmCube, a hybrid dynamical system that classifies, stabilizes, and monitors its own convergence through four competing force fields and a self-observing layer. Isotropic Gaussian perturbation is replaced with anisotropic, surprisal-driven exploration derived from a streaming divergence signal. The observer field — center velocity, awareness, collapse flow, and center acceleration — provides second-order self-monitoring, enabling self-terminating computation. Validated across five diverse initial conditions: 100% convergence, 96.6% awareness decay, collapse flow approaching 0.999. Generalizes the Three Laws of Livnium from NLI classification (82.79% on SNLI) to a broader framework for self-stabilizing dynamical computation. Code: https://github.com/chetanxpatil/livnium
Chetan Patil (Fri,) studied this question.