
We model and compute the average response of truncated BurgersHopf dynamics
to finite perturbations away from the Gibbs equipartition energy spectrum using
a dynamical optimization framework recently conceptualized in a series of
papers. Nonequilibrium averages are there approximated in terms of geodesic
paths in probability space that bestfit the Liouvillean dynamics over a family
of quasiequilibrium trial densities. By recasting the geodesic principle as an
optimal control problem, we solve numerically for the nonequilibrium responses
using an augmented Lagrangian, nonlinear conjugate gradient descent method.
For moderate perturbations, we find an excellent agreement between the optimal
predictions and the direct numerical simulations of the truncated BurgersHopf
dynamics. In this nearequilibrium regime, we argue that the optimal response
theory provides an approximate yet predictive counterpart to
fluctuationdissipation identities.

A model reduction technique based on an optimization principle is employed to
coarsegrain inviscid, incompressible fluid dynamics in two dimensions. In this
reduction the spectrallytruncated vorticity equation defines the
microdynamics, while the macroscopic state space consists of quasiequilibrium
trial probability densities on the microscopic phase space, which are
parameterized by the means and variances of the low modes of the vorticity. A
macroscopic path therefore represents a coarsegrained approximation to the
evolution of a nonequilibrium ensemble of microscopic solutions. Closure in
terms of the vector of resolved variables, namely, the means and variances of
the low modes, is achieved by minimizing over all feasible paths the time
integral of their meansquared residual with respect to the Liouville equation.
The equations governing the optimal path are deduced from HamiltonJacobi
theory. The coarsegrained dynamics derived by this optimization technique
contains a scaledependent eddy viscosity, modified nonlinear interactions
between the low mode means, and a nonlinear coupling between the mean and
variance of each low mode. The predictive skill of this optimal closure is
validated quantitatively by comparing it against direct numerical simulations.
These tests show that good agreement is achieved without adjusting any closure
parameters.

A general method for deriving closed reduced models of Hamiltonian dynamical
systems is developed using techniques from optimization and statistical
estimation. As in standard projection operator methods, a set of resolved
variables is selected to capture the slow, macroscopic behavior of the system,
and the family of quasiequilibrium probability densities on phase space
corresponding to these resolved variables is employed as a statistical model.
The macroscopic dynamics of the mean resolved variables is determined by
optimizing over paths of these probability densities. Specifically, a cost
function is introduced that quantifies the lackoffit of such paths to the
underlying microscopic dynamics; it is an ensembleaveraged, squarednorm of
the residual that results from submitting a path of trial densities to the
Liouville equation. The evolution of the macrostate is estimated by minimizing
the time integral of the cost function. The value function for this
optimization satisfies the associated HamiltonJacobi equation, and it
determines the optimal relation between the statistical parameters and the
irreversible fluxes of the resolved variables, thereby closing the reduced
dynamics. The resulting equations for the macroscopic variables have the
generic form of governing equations for nonequilibrium thermodynamics, and they
furnish a rational extension of the classical equations of linear irreversible
thermodynamics beyond the nearequilibrium regime. In particular, the value
function is a thermodynamic potential that extends the classical dissipation
function and supplies the nonlinear relation between thermodynamics forces and
fluxes.

A new method of deriving reduced models of Hamiltonian dynamical systems is
developed using techniques from optimization and statistical estimation. Given
a set of resolved variables that define a model reduction, the
quasiequilibrium ensembles associated with the resolved variables are employed
as a family of trial probability densities on phase space. The residual that
results from submitting these trial densities to the Liouville equation is
quantified by an ensembleaveraged cost function related to the information
loss rate of the reduction. From an initial nonequilibrium state, the
statistical state of the system at any later time is estimated by minimizing
the time integral of the cost function over paths of trial densities.
Statistical closure of the underresolved dynamics is obtained at the level of
the value function, which equals the optimal cost of reduction with respect to
the resolved variables, and the evolution of the estimated statistical state is
deduced from the HamiltonJacobi equation satisfied by the value function. In
the nearequilibrium regime, or under a local quadratic approximation in the
farfromequilibrium regime, this bestfit closure is governed by a
differential equation for the estimated state vector coupled to a Riccati
differential equation for the Hessian matrix of the value function. Since
memory effects are not explicitly included in the trial densities, a single
adjustable parameter is introduced into the cost function to capture a
timescale ratio between resolved and unresolved motions. Apart from this
parameter, the closed equations for the resolved variables are completely
determined by the underlying deterministic dynamics.

We illustrate a novel characterization of nonequivalent statistical
mechanical ensembles using the meanfield BlumeEmeryGriffiths (BEG) model as
a test model. The novel characterization takes effect at the level of the
microcanonical and canonical equilibrium distributions of states. For this
reason it may be viewed as a statistical characterization of nonequivalent
ensembles which extends and complements the common thermodynamic
characterization of nonequivalent ensembles based on nonconcave anomalies of
the microcanonical entropy. By computing numerically both the microcanonical
and canonical sets of equilibrium distributions of states of the BEG model, we
show that for values of the mean energy where the microcanonical entropy is
nonconcave, the microcanonical distributions of states are nowhere realized in
the canonical ensemble. Moreover, we show that for values of the mean energy
where the microcanonical entropy is strictly concave, the equilibrium
microcanonical distributions of states can be put in onetoone correspondence
with equivalent canonical equilibrium distributions of states. Our numerical
computations illustrate general results relating thermodynamic and statistical
equivalence and nonequivalence of ensembles proved by Ellis, Haven, and
Turkington [J. Stat. Phys. 101, 999 (2000)].

A statistical model of selforganization in a generic class of
onedimensional nonlinear Schrodinger (NLS) equations on a bounded interval is
developed. The main prediction of this model is that the statistically
preferred state for such equations consists of a deterministic coherent
structure coupled with finescale, random fluctuations, or radiation. The model
is derived from equilibrium statistical mechanics by using a meanfield
approximation of the conserved Hamiltonian and particle number for
finitedimensional spectral truncations of the NLS dynamics. The continuum
limits of these approximated statistical equilibrium ensembles on
finitedimensional phase spaces are analyzed, holding the energy and particle
number at fixed, finite values. The analysis shows that the coherent structure
minimizes total energy for a given value of particle number and hence is a
solution to the NLS ground state equation, and that the remaining energy
resides in Gaussian fluctuations equipartitioned over wavenumbers. Some results
of direct numerical integration of the NLS equation are included to validate
empirically these properties of the most probable states for the statistical
model. Moreover, a theoretical justification of the meanfield approximation is
given, in which the approximate ensembles are shown to concentrate on the
associated microcanonical ensemble in the continuum limit.