A Typed Measurement Grammar for Cross-Domain Analysis of Interacting Dynamical Systems

Abstract

Scientific fields quantify interactions among dynamical systems with domain-specific tools that are often related but not interchangeable. This preprint introduces the Generative Systems Measurement Grammar (GSMG), a typed cross-domain contract that preserves distinctions among state definition, realized relational structure, predictive versus interventional evidence, estimator fitness, provenance, comparability and causal-unit attribution. Frozen synthetic benchmarks were used adversarially: v1.0 passed 12/15 worlds and was retained as PARTIAL; after prospectively logged amendments, v1.1 passed 14/15 worlds with zero failures on prespecified core-safety tests. Archival applications to mouse V1-LM dynamics, physical oscillator networks and synthetic biological feedback circuits test semantic portability without replacing native domain measures. A candidate universal persistence score was rejected as redundant with established domain-specific stability/resilience measures. GSMG is presented as a methodological grammar, not a universal theory.