Causal Memory in Two Executable Systems: A DPN Validation Harness and a CMG-Inspired Robotics Prior

This technical addendum to the Causal Memory Gravity (CMG) corpus presents two independently executable computational studies: a validation harness for Dynamic Planck Network (DPN) diagnostics and a simulation-only robotics proof of concept based on a causal relaxation-memory prior. The DPN component provides a reproducible control framework for operator convergence, spectral matching, causal versus acausal kernel selection, and normalization diagnostics. It validates the computational machinery against matched positive and negative controls but does not constitute a derivation of the CMG microscopic model. The robotics component compares an instantaneous risk prior with a two-timescale, first-order relaxation-memory prior under identical downstream CBF/HOCBF control layers. Across seven obstacle-avoidance and seven contact scenarios, with 40 paired random seeds per controller and scenario, all 1,120 archived simulation rows were independently reproduced. In the obstacle sector, the memory prior reduced completion time, path length, command variation, and jerk, while also reducing the minimum obstacle clearance. In the contact sector, its principal effect was a substantial reduction in stiffness-command variation, with only minor changes in contact force, tracking error, and mechanical work. Status: Reproducible simulation-only computational proof of concept and technical addendum. Hardware validation and a concrete microscopic DPN realization remain open.