Randomized trial analyzes the relationship between coarse-graining and entropy in thermodynamic systems, suggesting new criteria for admissibility.
Key Points
The aim is to establish a precise criterion for coarse-graining in statistical mechanics, defining permissible macrostate descriptions and their relation to entropy.
Utilized finite-Markov machinery, including concepts like strong lumpability and partition refinement.
Derived criteria for admissible transitions and the essential distinctions within physical systems.
Proved necessary and sufficient conditions for the existence of unique coarse partitions.
Established that a macrostate preserves constitutive labels and is strongly lumpable for dynamic closure.
Demonstrated that relative entropy toward stationary distributions is monotonic, supporting a new distributional H-theorem.
Showed the second law is precise under specific coarse-graining conditions, not applicable when essential information is discarded.