Gradient and GENERIC evolution towards reduced dynamics

Abstract

Let (M,J) be a dynamical model of macroscopic systems and (N,K) a less microscopic model (i.e. a model involving less details) of the same macroscopic systems; M and N are manifolds, J are vector fields on M, and K are vector fields on N. Let P be the phase portrait corresponding to (M,J) (i.e. P is the set of all trajectories in M generated by a family of vector fields in J), and R the phase portrait corresponding to (N,K). Thermodynamics in its general sense is a pattern recognition process in which R is recognized as a pattern in P. In particular, the classical (both equilibrium and nonequilibrium) thermodynamics arises in the investigation of relations between models (M,J) and models without time evolution, i.e. models with K= 0. In such case R is a submanifold of M composed of fixed points. Let Su mapping M to R be a potential, called an upper entropy, generating the vector field J. The equilibrium thermodynamic relation in N is the lower entropy Sd(y) defined by Sd(y)=Su(x=y), where x is in M,y in N, and y is given by a final destination (i.e. when the time goes to infinity) of x in the time evolution generated by the vector field J. In this paper we show that if K is not zero (e.g. in externally forced or, in other words, open systems), then the reduction also provides thermodynamics (we call it flux-thermodynamics). If certain conditions are satisfied, then the lower entropy Ss, that arises in the investigation of the approach of J to K, is the time derivative of the lower entropy Sd arising in the investigation of the approach of M to N as t goes to infinity.

submitted

References in corpus (17)

• The large deviation approach to statistical mechanics
• On the relation between gradient flows and the large-deviation principle, with applications to Markov chains and diffusion
• Maximum Entropy Principle in statistical inference: case for non-Shannonian entropies
• Continuum Mechanics and Thermodynamics in the Hamilton and the Godunov-type Formulations
• Hilbert's 6th Problem: Exact and Approximate Hydrodynamic Manifolds for Kinetic Equations
• A generalization of Onsager's reciprocity relations to gradient flows with nonlinear mobility
• Inconsistency of a dissipative contribution to the mass flux in hydrodynamics
• A hierarchy of Poisson brackets in non-equilibrium thermodynamics
• An optimization principle for deriving nonequilibrium statistical models of Hamiltonian dynamics
• Thermodynamics and stability of non-equilibrium steady states in open systems
• Hamiltonian and Godunov Structures of the Grad Hierarchy
• Irreversibility in the short memory approximation
• Non-convex dissipation potentials in multiscale non-equilibrium thermodynamics
• Extra mass flux in fluid mechanics
• Ehrenfest regularization of Hamiltonian systems
• Entropy and Entropy Production in Multiscale Dynamics
• Braun-Le Chatelier principle in dissipative thermodynamics

Cited by in corpus (5)

• Structure-preserving neural networks
• Variational principles and thermodynamics
• Nonequilibrium thermodynamics: emergent and fundamental
• Multiscale Thermodynamics
• Port-metriplectic neural networks: thermodynamics-informed machine learning of complex physical systems