arXiv subjects
Publications and source records attributed to Reuven Zeitak.
We introduce a new formalism for dealing with networks of queues. The formalism is based on the Doi-Peliti second quantization method for reaction diffusion systems. As a demonstration of the method's utility we compute perturbatively the different time busy-busy correlations between two servers in a Jackson network.
We calculate, using numerical methods, the Lyapounov exponent gamma(E) and the density of states rho(E) at energy E of a one-dimensional non-Hermitian Schroedinger equation with off-diagonal disorder. For the particular case we consider, both gamma(E) and rho(E) depend only on the modulus of E. We find a pronounced maximum of rho(|E|) at energy E=2/sqrt(3), which seems to be linked to the fixed point structure of an associated random map. We show how the density of states rho(E) can be expanded in powers of E. We find rho(|E|) = 1/pi^2 + 4/(3 pi^3) |E|^2 + ... This expansion, which seems to be asymptotic, can be carried out to an arbitrarily high order.
We present a numerical method which is used to calculate anomalous scaling exponents of structure functions in the Kraichnan passive scalar advection model (R. H. Kraichnan, Phys. Fluids {\bf11}, 945 (1968)). This Monte-Carlo method, which is applicable in any space dimension, is based on the Lagrangian path interpretation of passive scalar dynamics, and uses the recently discovered equivalence between scaling exponents of structure functions and relaxation rates in the stochastic shape dynamics of groups of Lagrangian particles. We calculate third and fourth order anomalous exponents for several dimensions, comparing with the predictions of perturbative calculations in large dimensions. We find that Kraichnan's closure appears to give results in close agreement with the numerics. The third order exponents are compatible with our own previous nonperturbative calculations.
The Kraichnan rapid advection model is recast as the stochastic dynamics of tracer trajectories. This framework replaces the random fields with a small set of stochastic ordinary differential equations. Multiscaling of correlation functions arises naturally as a consequence of the geometry described by the evolution of N trajectories. Scaling exponents and scaling structures are interpreted as excited states of the evolution operator. The trajectories become nearly deterministic in high dimensions allowing for perturbation theory in this limit. We calculate perturbatively the anomalous exponent of the third and fourth order correlation functions. The fourth order result agrees with previous calculations.
For the zero temperature Glauber dynamics of the $q$-state Potts model, we calculate the exact distribution of domain sizes by mapping the problem on an exactly soluble one-species coagulation model ($A+A\rightarrow A$). In the long time limit, this distribution is universal and from its (complicated) exact expression, we extract its behavior in various regimes. Our results are tested in a simulation and compared to the predictions of a simple approximation proposed recently. Considering the dynamics of domain walls as a reaction diffusion model $ A+A \to \ A $ with probability $ (q-2)/(q-1)$ and $A+A \to \emptyset $ with probability $1/(q-1)$, we calculate the pair correlation function in the long time regime.
The fraction r(t) of spins which have never flipped up to time t is studied within a linear diffusion approximation to phase ordering. Numerical simulations show that, even in this simple context, r(t) decays with time like a power-law with a non-trival exponent $θ$ which depends on the space dimension. The local dynamics at a given point is a special case of a stationary gaussian process of known correlation function and the exponent $θ$ is shown to be determined by the asymptotic behavior of the probability distribution of intervals between consecutive zero-crossings of this process. An approximate way of computing this distribution is proposed, by taking the lengths of the intervals between successive zero-crossings as independent random variables. The approximation gives values of the exponent $θ$ in close agreement with the results of simulations.
We present an analytic and numerical analysis of the Gledzer-Ohkitani-Yamada (GOY) cascade model for turbulence. We concentrate on the dynamic correlations, and demonstrate both numerically and analytically, using resummed perturbation theory, that the correlations do not follow a dynamic scaling ansatz. The basic reason for this is the existence of a second quadratic invariant, in addition to energy. This implies the breakdown of the Kolmogorov type scaling law, in a manner different from the conventional mechanisms proposed for Navier-Stokes intermittency. By modifying the model equation so as to eliminate the spurious invariant, we recover to good accuracy, both dynamic scaling and the Kolmogorov exponents. We conclude that intermittency in the GOY model may be attributed to the effects of the spurious invariant which does not exist in the 3-dimensional Navier-Stokes flow.