[Paper Review] Thermodynamics and the moment map

Mikhail Kapranov|arXiv (Cornell University)|Aug 17, 2011

Advanced Thermodynamics and Statistical Mechanics2 references17 citations

TL;DR

This paper establishes a deep connection between thermodynamics and algebraic geometry by interpreting the moment map in toric varieties as the mean energy map in a generalized thermodynamic framework with multiple Hamiltonians. The key result is that the convexity properties of the moment map correspond to fundamental thermodynamic principles, such as the concavity of the entropy functional, and the inverse temperature parameter generalizes to a vector, linking statistical mechanics to tropical geometry in the low-temperature limit.

ABSTRACT

We give a thermodynamical interpretation of the moment map for toric varieties.

Motivation & Objective

• To generalize standard statistical thermodynamics to systems with multiple commuting Hamiltonians, leading to a vector-valued energy function.
• To interpret the moment map in toric varieties as the mean energy map in this generalized thermodynamic framework.
• To show that the convexity properties of the moment map correspond to fundamental thermodynamic principles, such as concavity of the entropy functional.
• To connect the low-temperature limit of statistical mechanics to tropical geometry, clarifying that this limit corresponds to the tropical (ultraviolet) regime.
• To establish a Legendre duality between the free energy and entropy functions in this higher-dimensional thermodynamic setting.

Proposed method

• The paper uses the Gibbs distribution with a vector-valued inverse temperature parameter to model systems with multiple Hamiltonians.
• It defines the mean energy map as the expectation of the energy vector under the Gibbs measure, which corresponds to the moment map in toric geometry.
• The entropy function on the probability simplex is restricted to the fiber over a given mean energy, and its maximum is shown to be unique and critical.
• The direct image of the entropy function under the mean energy map yields a strictly concave function on the image polytope, which is identified with the entropy in the generalized thermodynamic setting.
• The method applies the Lagrange multiplier technique to derive the inverse relation between the inverse temperature and mean energy, proving the mean energy map is a diffeomorphism.
• The Legendre transform duality between the free energy and entropy is established via the gradient maps of these functions.

Experimental results

Research questions

• RQ1 How does the standard thermodynamic formalism extend to systems with multiple commuting Hamiltonians?
• RQ2 What is the geometric interpretation of the mean energy map in such systems, and how does it relate to known structures in algebraic geometry?
• RQ3 How do convexity properties of the moment map reflect thermodynamic principles such as entropy concavity?
• RQ4 What is the role of the low-temperature limit in this generalized framework, and how does it connect to tropical geometry?
• RQ5 Can the Legendre duality between free energy and entropy be recovered in this higher-dimensional setting?

Key findings

• The mean energy map in the generalized thermodynamics with multiple Hamiltonians is equivalent to the moment map in toric varieties.
• The entropy functional restricted to fibers over a given mean energy achieves a unique maximum, corresponding to the Gibbs distribution.
• The direct image of the entropy function under the mean energy map is strictly concave, confirming thermodynamic stability.
• The inverse of the mean energy map is given by the gradient of the entropy function, establishing a Legendre duality with the free energy.
• The low-temperature limit of the system corresponds to the tropical limit near vertices of the moment polytope, linking thermodynamics to tropical geometry.
• The vector-valued inverse temperature parameter generalizes the standard scalar inverse temperature, and its components correspond to the Lagrange multipliers in the constrained optimization of entropy.