Searcharxiv⌕ Search

SEARCH · Searcharxiv

Results for “adap-org”

Search indexed arXiv papers on artificial intelligence, large language models, computer vision and robotics. Read source abstracts and follow links to arXiv.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 415 records · Page 23Linked to original sources

Renormalization Group Analysis of a Noisy Kuramoto-Sivashinsky Equation

We have analyzed the Kuramoto-Sivashinsky equation with a stochastic noise term through a dynamic renormalization group calculation. For a system in which the lattice spacing is smaller than the typical wavelength of the linear instability occurring in the system, the large-distance and long-time behavior of this equation is the same as for the Kardar-Parisi-Zhang equation in one and two spatial dimensions. For the $d=2$ case the agreement is only qualitative. On the other hand, when coarse-graining on larger scales the asymptotic flow depends on the initial values of the parameters.

cond-mat↗

Correlations and Renormalization in Lattice Gases

A complete formulation is given of an exact kinetic theory for lattice gases. This kinetic theory makes possible the calculation of corrections to the usual Boltzmann / Chapman-Enskog analysis of lattice gases due to the buildup of correlations. It is shown that renormalized transport coefficients can be calculated perturbatively by summing terms in an infinite series. A diagrammatic notation for the terms in this series is given, in analogy with the diagrammatic expansions of continuum kinetic theory and quantum field theory. A closed-form expression for the coefficients associated with the vertices of these diagrams is given. This method is applied to several standard lattice gases, and the results are shown to correctly predict experimentally observed deviations from the Boltzmann analysis.

comp-gas↗

Kolmogorov turbulence in a random-force-driven Burgers equation: anomalous scaling and probability density functions

High-resolution numerical experiments, described in this work, show that velocity fluctuations governed by the one-dimensional Burgers equation driven by a white-in-time random noise with the spectrum $\overline{|f(k)|^2}\propto k^{-1}$ exhibit a biscaling behavior: All moments of velocity differences $S_{n\le 3}(r)=\overline{|u(x+r)-u(x)|^n}\equiv\overline{|Δu|^n}\propto r^{n/3}$, while $S_{n>3}\propto r^{ζ_n}$ with $ζ_n\approx 1$ for real $n>0$ (Chekhlov and Yakhot, Phys. Rev. E {\bf 51}, R2739, 1995). The probability density function, which is dominated by coherent shocks in the interval $Δu<0$, is ${\cal P}(Δu,r)\propto (Δu)^{-q}$ with $q\approx 4$.

adap-org↗

Chromosome mapping: radiation hybrid data and stochastic spin models

This work approaches human chromosome mapping by developing algorithms for ordering markers associated with radiation hybrid data. Motivated by recent work of Boehnke et al. [1], we formulate the ordering problem by developing stochastic spin models to search for minimum-break marker configurations. As a particular application, the methods developed are applied to 14 human chromosome-21 markers tested by Cox et al. [2]. The methods generate configurations consistent with the best found by others. Additionally, we find that the set of low-lying configurations is described by a Markov-like ordering probability distribution. The distribution displays cluster correlations reflecting closely linked loci.

adap-org↗

A Pinned Polymer Model of Posture Control

A phenomenological model of human posture control is posited. The dynamics are modelled as an elastically pinned polymer under the influence of noise. The model accurately reproduces the two-point correlation functions of experimental posture data and makes predictions for the response function of the postural control system. The physiological and clinical significance of the model is discussed.

adap-org↗

Long Range Anticorrelations and Non-Gaussian Behavior of a Leaky Faucet

We find that intervals between successive drops from a leaky faucet display scale-invariant, long-range anticorrelations characterized by the same exponents of heart beat-to-beat intervals of healthy subjects. This behavior is also confirmed by numerical simulations on lattice and it is faucet-width- and flow-rate-independent. The histogram for the drop intervals is also well described by a Lévy distribution with the same index for both histograms of healthy and diseased subjects. This additional result corroborates the evidence for similarities between leaky faucets and healthy hearts underlying dynamics.

adap-org↗

Mutation Accumulation and the Catastrophic Senescence of Pacific Salmon

The bit-string model of biological aging is used to simulate the catastrophic senescence of Pacific Salmon. We have shown that reproduction occuring only once and at a fixed age is the only ingredient needed to explain the catastrophic senescence according the mutation accumulation theory. Several results are presented, some of them with up to $10^8$ fishes, showing how the survival rates in catastrophic senescence are affected by changes in the parameters of the model.

adap-org↗

Universal Short-Time Dynamics in the Kosterlitz-Thouless Phase

We study the short-time dynamics of systems that develop ``quasi long-range order'' after a quench to the Kosterlitz-Thouless phase. With the working hypothesis that the ``universal short-time behavior'', previously found in Ising-like systems, also occurs in the Kosterlitz-Thouless phase, we explore the scaling behavior of thermodynamic variables during the relaxational process following the quench. As a concrete example, we investigate the two-dimensional $6$-state clock model by Monte Carlo simulation. The exponents governing the magnetization, the second moment, and the autocorrelation function are calculated. From them, by means of scaling relations, estimates for the equilibrium exponents $z$ and $η$ are derived. In particular, our estimates for the temperature-dependent anomalous dimension $η$ that governs the static correlation function are consistent with existing analytical and numerical results and, thus, confirm our working hypothesis.

cond-mat↗

Apparent Fractality Emerging from Models of Random Distributions

The fractal properties of models of randomly placed $n$-dimensional spheres ($n$=1,2,3) are studied using standard techniques for calculating fractal dimensions in empirical data (the box counting and Minkowski-sausage techniques). Using analytical and numerical calculations it is shown that in the regime of low volume fraction occupied by the spheres, apparent fractal behavior is observed for a range of scales between physically relevant cut-offs. The width of this range, typically spanning between one and two orders of magnitude, is in very good agreement with the typical range observed in experimental measurements of fractals. The dimensions are not universal and depend on density. These observations are applicable to spatial, temporal and spectral random structures. Polydispersivity in sphere radii and impenetrability of the spheres (resulting in short range correlations) are also introduced and are found to have little effect on the scaling properties. We thus propose that apparent fractal behavior observed experimentally over a limited range may often have its origin in underlying randomness.

cond-mat↗

Reconstruction of a first-order phase transition from computer simulations of individual phases and subphases

We present a new method for investigating first-order phase transitions using Monte Carlo simulations. It relies on the multiple-histogram method and uses solely histograms of individual phases. In addition, we extend the method to include histograms of subphases. The free energy difference between phases, necessary for attributing the correct statistical weights to the histograms, is determined by a detour in control parameter space via auxiliary systems with short relaxation times. We apply this method to a recently introduced model for structure formation in polypeptides for which other methods fail.

adap-org↗

Avalanche Dynamics in Evolution, Growth, and Depinning Models

The dynamics of complex systems in nature often occurs in terms of punctuations, or avalanches, rather than following a smooth, gradual path. A comprehensive theory of avalanche dynamics in models of growth, interface depinning, and evolution is presented. Specifically, we include the Bak-Sneppen evolution model, the Sneppen interface depinning model, the Zaitsev flux creep model, invasion percolation, and several other depinning models into a unified treatment encompassing a large class of far from equilibrium processes. The formation of fractal structures, the appearance of $1/f$ noise, diffusion with anomalous Hurst exponents, Levy flights, and punctuated equilibria can all be related to the same underlying avalanche dynamics. This dynamics can be represented as a fractal in $d$ spatial plus one temporal dimension. We develop a scaling theory that relates many of the critical exponents in this broad category of extremal models, representing different universality classes, to two basic exponents characterizing the fractal attractor. The exact equations and the derived set of scaling relations are consistent with numerical simulations of the above mentioned models.

adap-org↗

Effect of time-correlation of input patterns on the convergence of on-line learning

We studied the effects of time correlation of subsequent patterns on the convergence of on-line learning by a feedforward neural network with backpropagation algorithm. By using chaotic time series as sequences of correlated patterns, we found that the unexpected scaling of converging time with learning parameter emerges when time-correlated patterns accelerate learning process.

adap-org↗

Particle hopping models and traffic flow theory

This paper shows how particle hopping models fit into the context of traffic flow theory. Connections between fluid-dynamical traffic flow models, which derive from the Navier-Stokes-equations, and particle hopping models are shown. In some cases, these connections are exact and have long been established, but have never been viewed in the context of traffic theory. In other cases, critical behavior of traffic jam clusters can be compared to instabilities in the partial differential equations. Finally, it is shown how this leads to a consistent picture of traffic jam dynamics.

cond-mat↗

Thermodynamic Description of the Relaxation of Two-Dimensional Euler Turbulence Using Tsallis Statistics

Euler turbulence has been experimentally observed to relax to a metaequilibrium state that does not maximize the Boltzmann entropy, but rather seems to minimize enstrophy. We show that a recent generalization of thermodynamics and statistics due to Tsallis is capable of explaining this phenomenon in a natural way. The maximization of the generalized entropy $S_{1/2}$ for this system leads to precisely the same profiles predicted by the Restricted Minimum Enstrophy theory of Huang and Driscoll. This makes possible the construction of a comprehensive thermodynamic description of Euler turbulence.

chao-dyn↗

Learning from Minimum Entropy Queries in a Large Committee Machine

In supervised learning, the redundancy contained in random examples can be avoided by learning from queries. Using statistical mechanics, we study learning from minimum entropy queries in a large tree-committee machine. The generalization error decreases exponentially with the number of training examples, providing a significant improvement over the algebraic decay for random examples. The connection between entropy and generalization error in multi-layer networks is discussed, and a computationally cheap algorithm for constructing queries is suggested and analysed.

adap-org↗

Ultrametricity and Memory in a Solvable Model of Self-Organized Criticality

Slowly driven dissipative systems may evolve to a critical state where long periods of apparent equilibrium are punctuated by intermittent avalanches of activity. We present a self-organized critical model of punctuated equilibrium behavior in the context of biological evolution, and solve it in the limit that the number of independent traits for each species diverges. We derive an exact equation of motion for the avalanche dynamics from the microscopic rules. In the continuum limit, avalanches propagate via a diffusion equation with a nonlocal, history-dependent potential representing memory. This nonlocal potential gives rise to a non-Gaussian (fat) tail for the subdiffusive spreading of activity. The probability for the activity to spread beyond a distance $r$ in time $s$ decays as $\sqrt{24\overπ}s^{-3/2}x^{1/3} \exp{[-{3\over 4}x^{1/3}]}$ for $x={r^4\over s} \gg 1$. The potential represents a hierarchy of time scales that is dynamically generated by the ultrametric structure of avalanches, which can be quantified in terms of ``backward'' avalanches. In addition, a number of other correlation functions characterizing the punctuated equilibrium dynamics are determined exactly.

cond-mat↗

On War: The Dynamics of Vicious Civilizations

The dynamics of ``vicious'', continuously growing civilizations (domains), which engage in ``war'' whenever two domains meet, is investigated. In the war event, the smaller domain is annihilated, while the larger domain is reduced in size by a fraction $\e$ of the casualties of the loser. Here $\e$ quantifies the fairness of the war, with $\e=1$ corresponding to a fair war with equal casualties on both side, and $\e=0$ corresponding to a completely unfair war where the winner suffers no casualties. In the heterogeneous version of the model, evolution begins from a specified initial distribution of domains, while in the homogeneous system, there is a continuous and spatially uniform input of point domains, in addition to the growth and warfare. For the heterogeneous case, the rate equations are derived and solved, and comparisons with numerical simulations are made. An exact solution is also derived for the case of equal size domains in one dimension. The heterogeneous system is found to coarsen, with the typical cluster size growing linearly in time $t$ and the number density of domains decreases as $1/t$. For the homogeneous system, two different long-time behaviors arise as a function of $\e$. When $1/2<\e\leq 1$ (relatively fair wars), a steady state arises which is characterized by egalitarian competition between domains of comparable size. In the limiting case of $\e=1$, rate equations which simultaneously account for the distribution of domains and that of the intervening gaps are derived and solved. The steady state is characterized by domains whose age is typically much larger than their size. When $0\leq\e<1/2$ (unfair wars), a few ``superpowers'' ultimately dominate. Simulations indicate that this coarsening process is characterized by power-law temporal behavior, with non-universal

cond-mat↗