Hamiltonian and Lagrangian for the Trajectory of the Empirical Distribution and the Empirical Measure of Markov Processes

We compute the Hamiltonian and Lagrangian associated to the large deviations of the trajectory of the empirical distribution for independent Markov processes, and of the empirical measure for translation invariant interacting Markov processes. We treat both the case of jump processes (continuous-time Markov chains and interacting particle systems) as well as diffusion processes. For diffusion processes, the Lagrangian is a quadratic form of the deviation of the trajectory from the solution of the Kolmogorov forward equation. In all cases, the Lagrangian can be interpreted as a relative entropy or relative entropy density per unit time.     

Entropy density in function space and the Onsager-Machlup function

The generalized entropy $\text{S(μ|ν)=∫dν(}\text{dμ}\text{dν}\text{)}\text{ln}\text{(}\text{dμ}\text{dν}\text{)}$ is investigated for measures μ, ν induced by diffusion processes. It is shown that the potential part of the Onsager-Machlup function (lagrangian for the most probable path of a diffusion process) can be interpreted as an entropy production density on the most probable path.     

Functionals of paths of a diffusion process and the Onsager-Machlup function

The Onsager-Machlup function for a one-dimensional diffusion processX _t is obtained by means of the Radon-Nikodym derivative of the measure in function space induced byX _t with respect to the measure corresponding to an auxiliary diffusion processY _t.     

Onsager-Machlup Function for one dimensional nonlinear diffusion processes

Using Ito stochastic differential equations to describe stochastic processes the Onsager-Machlup Function of a nonlinear diffusion process is calculated. It is shown that for two examples the Onsager-Machlup Function calculated directly as limit of finite dimensional probability densities agrees with the formula derived by using the Ito calculus but differs from a formula given by Graham who used the concept of Langevin equations.     

On the path integral for diffusion in curved spaces

The Onsager-Machlup Lagrangian for diffusion processes in curved spaces is determined by evaluating the covariant path integral by means of a spectral analysis of smooth trajectories in Riemannian normal coordinates. The Lagrangian involves a novel curvature scalar potential term $\text{v=?}\text{(}\text{1}\text{8}\text{)}\text{R}$. The present treatment replaces an earlier one.     

Study of a class of non-linear stochastic processes boomerang behaviour of the mean path

Exact results are obtained for a large class of non-linear stochastic processes with constant diffusion. Special attention is given to the intrinsic effects induced by the non-linearities. In particular, a boomerang type behaviour of the mean path and time-controlled transitions from unimodal to bimodal probability densities are observed. Finally, the Onsager-Machlup stationary phase approximation is discussed and is recognised to provide a rather poor information for our particular class of models.     

In one and two dimensions,every stationary measure for a stochastic Ising Model is a Gibbs state

It is shown that one and two dimensional (generalized) stochastic Ising models with finite range potentials have only Gibbs states as their stationary measures. This is true even if the stationary measure or the potential is not translation invariant. This extends previously known results which are restricted to translation invariant stationary measures and potentials. In particular if the potential has only one Gibbs state the stochastic Ising Model must be ergodic.Research supported in part by N.S.F. Grant MPS 74-18926Alfred P. Sloan Fellow     

Projective Invariance and Einstein's Equations

It is shown that a projectively invariant Lagrangian field theory based on linear non-symmetric connections in space-time and arbitrary source fields is equivalent to Einstein's standard theory of gravitation coupled to a source Lagrangian depending solely on the original source fields. A key point is that, as in the case of Lagrangian field theories based on symmetric connections in space-time, the Euler-Lagrange field equations uniquely determine the projective invariant part of the linear connection in terms of the metric, their first-order derivatives, the source fields, and their conjugate momenta.     

On an Algebraic Property of the Disordered Phase of the Ising Model with Competing Interactions on a Cayley Tree

It is known that the disordered phase of the classical Ising model on the Caley tree is extreme in some region of the temperature. If one considers the Ising model with competing interactions on the same tree, then about the extremity of the disordered phase there is no any information. In the present paper, we first aiming to analyze the correspondence between Gibbs measures and QMC’s on trees. Namely, we establish that states associated with translation invariant Gibbs measures of the model can be seen as diagonal quantum Markov chains on some quasi local algebra. Then as an application of the established correspondence, we study some algebraic property of the disordered phase of the Ising model with competing interactions on the Cayley tree of order two. More exactly, we prove that a state corresponding to the disordered phase is not quasi-equivalent to other states associated with translation invariant Gibbs measures. This result shows how the translation invariant states relate to each other, which is even a new phenomena in the classical setting. To establish the main result we basically employ methods of quantum Markov chains.     

Onsager-Machlup Theory and Work Fluctuation Theorem for a Harmonically Driven Brownian Particle

We extend Tooru-Cohen analysis for nonequilibrium steady state (NSS) of a Brownian particle to nonequilibrium oscillatory state (NOS) of Brownian particle by considering time dependent external drive protocol. We consider an unbounded charged Brownian particle in the presence of oscillating electric field and prove work fluctuation theorem, which is valid for any initial distribution and at all times. For harmonically bounded and constantly dragged Brownian particle considered by Tooru and Cohen, work fluctuation theorem is valid for any initial condition (also NSS), but only in large time limit. We use Onsager-Machlup Lagrangian with a constraint to obtain frequency dependent work distribution function, and describe entropy production rate and properties of dissipation functions for the present system using Onsager-Machlup functional.     

Application of the Onsager-Machlup function to nonlinear diffusion processes

The Onsager-Machlup function for a process with nonlinear drift and constant diffusion is considered. The equation of motion for the most probable path is given and its solutions are discussed in the case of the anharmonic oscillator. It is shown that in the case of the laser equation the autocorrelation times are qualitatively reproduced by means of the most probable path.     

First-order invariant Einstein-Cartan variational structures

A first-order invariant Einstein-Cartan structure is a Lagrangian structure on a differential manifold defined by a generally invariant Lagrangian depending on a metric field, a connection field, and the first derivatives of these fields. Moreover, it is assumed that the metric and connection fields satisfy the so-called compatibility condition. In this paper the problem of finding all such invariant Einstein-Cartan structures is discussed. It is shown that each Lagrangian of these structures depends only on certain tensors constructed from the metric and the connection fields, which means that all the Lagrangians can be described within the framework of the classical theory of invariants. The maximal number of functionally independent Lagrangians is determined as a function of the dimension of the underlying manifold.     

Thermal mechanics: A quantum mechanical analogue of nonequilibrium statistical thermodynamics

A formal but not conventional equivalence between stochastic processes in nonequilibrium statistical thermodynamics and Schrödinger dynamics in quantum mechanics is shown. It is found, for each stochastic process described by a stochastic differential equation of Itô type, there exists a Schrödinger-like dynamics in which the absolute square of a wavefunction gives us the same probability distribution as the original stochastic process. In utilizing this equivalence between them, that is, rewriting the stochastic differential equation by an equivalent Schrödinger equation, it is possible to obtain the notion of deterministic limit of the stochastic process as a semi-classical limit of the “Schrödinger” equation. The deterministic limit thus obtained improves the conventional deterministic approximation in the sense of Onsager-Machlup. The present approach is valid for a general class of stochastic equations where local drifts and diffusion coefficients depend on the position. Two concrete examples are given. It should be noticed that the approach in the present form has nothing to do with the conventional one where only a formal similarity between the Fokker-Planck equation and the Schrödinger equation is considered.     

A boundary piecewise constant level set method for boundary control of eigenvalue optimization problems

This paper investigates optimization of the least eigenvalue of ?Δ with the constraint of one-dimension Hausdorff measure of Dirichlet boundary. We propose the boundary piecewise constant level set (BPCLS) method based on the regularity technique to combine two types of boundary conditions into a single Robin boundary condition. We derive the first variation of the least eigenvalue w.r.t. the BPCLS function and propose a penalty BPCLS algorithm and an augmented Lagrangian BPCLS algorithm. Numerical results are reported for experiments on ellipse and L-shape domains.     

Entropy production in open volume-preserving systems

We describe a mechanism leading to positive entropy production in volume-preserving systems under nonequilibrium conditions. We consider volume-preserving systems sustaining a diffusion process like the multibaker map or the Lorentz gas. A continuous flux of particles is imposed across the system resulting in a steady gradient of concentration. In the limit where such flux boundary conditions are imposed at arbitrarily separated boundaries for a fixed gradient, the invariant measure becomes singular. For instance, in the multibaker map, the limit invariant measure has a cumulative function given in terms of the nondifferentiable Takagi function. Because of this singularity of the invariant measure, the entropy must be defined as a coarse-grained entropy instead of the fined-grained Gibbs entropy, which would require the existence of a regular measure with a density. The coarse-grained entropy production is then shown to be asymptotically positive and, moreover, given by the entropy production expected from irreversible thermodynamics.     

Path integrals in Riemannian spaces

The covariant path integral for a free particle in curved space will be evaluated by means of a spectral analysis of smooth paths. No discretization rule will be required to put the action on a lattice. The connection between the resulting quantum hamiltonian and the Onsager-Machlup lagrangian for diffusion processes willbe discussed. The present treatment corrects an earlier version.     

An Analytical Approach to the Turbulence-Relaxed State in Tokamak Plasmas

Starting from the continuity, temperature, and motion equations of the trapped electron fluid in generaltokamak magnetic field with positive or reversed shear and the definition of Lagrangian invariant, dL / dt = ( t u. )L =0, where u is convective velocity, the trapped electron dynamics is considered in the following two assumptions: (i) theturbulence is low frequency electrostatic, and (ii) L is a functional only of the density n, temperature T, and magneticfield B, and the effect of perturbation potential φ is included in the convective velocity u, i.e., u is a functional of n,T, B, and φ. The Lagrangian invariant hidden in the trapped electron dynamics is strictly found: L= ln[(n/B)c1(T/B2/3)c2], where c1 and c2 are dimensionless changeable parameters and c1 ∝ c2. From this Lagrangian invariant thewhich, in the limit of large aspect ratio, reduce to n(r)q(r) = const. and T3/2(r)q(r) = const., respectively. The lattertwo scaling laws are compared with existent experimental results, being in good agreement.     

Onsager-Machlup Theory for Nonequilibrium Steady States and Fluctuation Theorems

A generalization of the Onsager-Machlup theory from equilibrium to nonequilibrium steady states and its connection with recent fluctuation theorems are discussed for a dragged particle restricted by a harmonic potential in a heat reservoir. Using a functional integral approach, the probability functional for a path is expressed in terms of a Lagrangian function from which an entropy production rate and dissipation functions are introduced, and nonequilibrium thermodynamic relations like the energy conservation law and the second law of thermodynamics are derived. Using this Lagrangian function we establish two nonequilibrium detailed balance relations, which not only lead to a fluctuation theorem for work but also to one related to energy loss by friction. In addition, we carried out the functional integral for heat explicitly, leading to the extended fluctuation theorem for heat. We also present a simple argument for this extended fluctuation theorem in the long time limit. PACS numbers: 05.70.Ln, 05.40.-a, 05.10.Gg.     

Large Deviations and¶Variational Principle for Harmonic Crystals

We consider massless Gaussian fields with covariance related to the Green function of a long range random walk on Ê^d. These are viewed as Gibbs measures for a linear-quadratic interaction. We establish thermodynamic identities and prove a version of Gibbs' variational principle, showing that translation invariant Gibbs measures are characterized as minimizers of the relative entropy density. We then study the large deviations of the empirical field of a Gibbs measure. We show that a weak large deviation principle holds at the volume order, with rate given by the relative entropy density.