The Boltzmann Entropy and Randomness Tests.

Area of Science:

• Classical Mechanics
• Information Theory
• Statistical Mechanics

Background:

• Dynamical systems in classical mechanics lack robust entropy measures that bridge algorithmic information theory and thermodynamics.
• Existing entropy definitions, like Boltzmann entropy, can be highly dependent on system partitioning.
• Thermodynamics of computation and unusual spin systems present challenges for current entropy frameworks.

Purpose of the Study:

• Introduce novel "algorithmic fine-grain and coarse-grain entropy" concepts for classical dynamical systems.
• Establish algorithmic entropy as a unifying link between Kolmogorov complexity, Gibbs entropy, and Boltzmann entropy.
• Develop a coarse-grain entropy with reduced partition dependence and improved applicability.

Main Methods:

• Define fine-grain algorithmic entropy as a variant of Martin-Löf randomness tests.
• Develop coarse-grain algorithmic entropy as a correction to Boltzmann's definition.
• Analyze the properties and applications of these new entropy measures in various systems.

Main Results:

• Algorithmic fine-grain entropy links description complexity with classical entropy measures.
• Coarse-grain algorithmic entropy exhibits less partition dependence than traditional Boltzmann entropy.
• The new entropy measures are applicable to a wider range of systems, including thermodynamics of computation and cellular automata spin systems.

Conclusions:

• Algorithmic entropy provides a more unified and robust framework for understanding entropy in dynamical systems.
• The reduced partition dependence of coarse-grain algorithmic entropy enhances its utility across diverse physical systems.
• These findings offer new insights into the thermodynamics of computation and complex spin systems.