The extension of quantum mechanics to a general functional space (“rigged Hilbert space”), which incorporates time-symmetry breaking, is applied to construct extract dynamical models of entropy production and entropy flow. They are illustrated by using a simple conservative Hamiltonian system for multilevel atoms coupled to a time-dependent external force. The external force destroys the monotonicity of the ℋ-function evolution. This leads to a model of the entropy flow that allows a steady nonequilibrium structure of the emitted field around the unstable (...) particles. (shrink)

In recent papers the authors have discussed the dynamical properties of “large Poincaré systems” (LPS), that is, nonintegrable systems with a continuous spectrum (both classical and quantum). An interesting example of LPS is given by the Friedrichs model of field theory. As is well known, perturbation methods analytic in the coupling constant diverge because of resonant denominators. We show that this Poincaré “catastrophe” can be eliminated by a natural time ordering of the dynamical states. We obtain then a dynamical theory (...) which incorporates a privileged direction of time (and therefore the second law of thermodynamics). However, it is only in very simple situations that this time ordering can be performed in an “extended” Hilbert space. In general, we need to go to the Liouville space (superspace) and introduce a time ordering of dynamical states according to the number of particles involved in correlations. This leads then to a generalization of quantum mechanics in which the usual Heisenberg's eigenvalue problem is replaced by a complex eigenvalue problem in the Liouville space. (shrink)

The complex spectral representation of the Liouville operator introduced by Prigogine and others is applied to moderately dense gases interacting through hard-core potentials in arbitrary d-dimensional spaces. Kinetic equations near equilibrium are constructed in each subspace as introduced in the spectral decomposition for collective, renormalized reduced distribution functions. Our renormalization is a nonequilibrium effect, as the renormalization effect disappears at equilibrium. It is remarkable that our renormalized functions strictly obey well-defined Markovian kinetic equations for all d, even though the ordinary (...) distribution functions obey nonMarkovian equations with memory effects. One can now define transport coefficients associated to the collective modes for all dimensional systems including d = 2. Our formulation hence provides a microscopic meaning of the macroscopic transport theory. Moreover, this gives an answer to the long-standing question whether or not transport equations exist in two-dimensional systems. The non-Markovian effects for the ordinary distribution function, such as the long-time tails for arbitrary n-mode coupling, are estimated by superposition of the Markovian evolutions of the dressed distribution functions. (shrink)

The long-time tail effect (i.e., a non-Markovian effect) in a velocity autocorrelation function for moderately dense classical gases in d-dimensional space is estimated for arbitray n-mode coupling by superposition of the Markov equations for the collective modes which has been introduced through the complex spectral representation of the Liouville operator in the previous paper. Taking into account intermediate nonhydrodynamic modes in a transition between hydrodynamic states, we found slower decay processes in the long-time tail. These new processes lead to a (...) critical dimension at d = 4 as in the renormalization group, that is, higher modes processes lead to slower decay process in the autocorrelation function for d = 4, while they lead to quicker decay process for d > 4. This conclusion clashes with the traditional point of view, which leads to the critical dimension d = 2. These slower processes invalidate the traditional kinetic equations for bare distribution functions obtained by a truncation of the BBGKY hierarchy for d < 4, as well as the Green-Kubo formalism, as there appear contributions of order t−1, t−1/2, ... coming from multiple mode-mode couplings even for d = 3. (shrink)