arXiv subjects

Werner Krauth

Publications and source records attributed to Werner Krauth.

At least 73 records · Page 4Linked to original sources

arXiv2008

Renormalization group approach to exact sampling

In this Letter, we use a general renormalization-group algorithm to implement Propp and Wilson's "coupling from the past" approach to complex physical systems. Our algorithm follows the evolution of the entire configuration space under the Markov chain Monte Carlo dynamics from parts of the configurations (patches) on increasing length scales, and it allows us to generate "exact samples" of the Boltzmann distribution, which are rigorously proven to be uncorrelated with the initial condition. We validate our approach in the two-dimensional Ising spin glass on lattices of size 64 x 64.

cond-mat.stat-mech↗

arXiv2008

Semiclassical theory of the quasi two-dimensional trapped Bose gas

We discuss the quasi two-dimensional trapped Bose gas where the thermal occupation of excited states in the tightly confined direction is small but remains finite in the thermodynamic limit. We show that the semiclassical theory describes very accurately the density profile obtained by Quantum Monte Carlo calculations in the normal phase above the Kosterlitz--Thouless temperature T_KT, but differs strongly from the predictions of strictly two-dimensional mean-field theory, even at relatively high temperature. We discuss the relevance of our findings for analyzing ultra-cold-atom experiments in quasi two-dimensional traps.

cond-mat.stat-mech↗

arXiv2007

Kosterlitz-Thouless transition of the quasi two-dimensional trapped Bose gas

We present Quantum Monte Carlo calculations with up to N=576000 interacting bosons in a quasi two-dimensional trap geometry closely related to recent experiments with atomic gases. The density profile of the gas and the non-classical moment of inertia yield intrinsic signatures for the Kosterlitz--Thouless transition temperature T_KT. From the reduced one-body density matrix, we compute the condensate fraction, which is quite large for small systems. It decreases slowly with increasing system sizes, vanishing in the thermodynamic limit. We interpret our data in the framework of the local-density approximation, and point out the relevance of our results for the analysis of experiments.

cond-mat.stat-mech↗

arXiv2007

Off-diagonal long-range order, cycle probabilities, and condensate fraction in the ideal Bose gas

We discuss the relationship between the cycle probabilities in the path-integral representation of the ideal Bose gas, off-diagonal long-range order, and Bose--Einstein condensation. Starting from the Landsberg recursion relation for the canonic partition function, we use elementary considerations to show that in a box of size L^3 the sum of the cycle probabilities of length k >> L^2 equals the off-diagonal long-range order parameter in the thermodynamic limit. For arbitrary systems of ideal bosons, the integer derivative of the cycle probabilities is related to the probability of condensing k bosons. We use this relation to derive the precise form of the π_k in the thermodynamic limit. We also determine the function π_k for arbitrary systems. Furthermore we use the cycle probabilities to compute the probability distribution of the maximum-length cycles both at T=0, where the ideal Bose gas reduces to the study of random permutations, and at finite temperature. We close with comments on the cycle probabilities in interacting Bose gases.

cond-mat.stat-mech↗

arXiv2007

Selective-pivot sampling of radial distribution functions in asymmetric liquid mixtures

We present a Monte Carlo algorithm for selectively sampling radial distribution functions and effective interaction potentials in asymmetric liquid mixtures. We demonstrate its efficiency for hard-sphere mixtures, and for model systems with more general interactions, and compare our simulations with several analytical approximations. For interaction potentials containing a hard-sphere contribution, the algorithm yields the contact value of the radial distribution function.

cond-mat.soft↗

arXiv2007

Depinning exponents of the driven long-range elastic string

We perform a high-precision calculation of the critical exponents for the long-range elastic string driven through quenched disorder at the depinning transition, at zero temperature. Large-scale simulations are used to avoid finite-size effects and to enable high precision. The roughness, growth, and velocity exponents are calculated independently, and the dynamic and correlation length exponents are derived. The critical exponents satisfy known scaling relations and agree well with analytical predictions.

cond-mat.dis-nn↗

arXiv2006

Introduction To Monte Carlo Algorithms

In these lectures, given in '96 summer schools in Beg-Rohu (France) and Budapest, I discuss the fundamental principles of thermodynamic and dynamic Monte Carlo methods in a simple light-weight fashion. The keywords are MARKOV CHAINS, SAMPLING, DETAILED BALANCE, A PRIORI PROBABILITIES, REJECTIONS, ERGODICITY, "FASTER THAN THE CLOCK ALGORITHMS". The emphasis is on ORIENTATION, which is difficult to obtain (all the mathematics being simple). A firm sense of orientation helps to avoid getting lost, especially if you want to leave safe trodden-out paths established by common usage. Even though I remain quite basic (and, I hope, readable), I make every effort to drive home the essential messages, which are easily explained: the crystal-clearness of detail balance, the main problem with Markov chains, the great algorithmic freedom, both in thermodynamic and dynamic Monte Carlo, and the fundamental differences between the two problems.

cond-mat.stat-mech↗

arXiv2006

Universal width distributions in non-Markovian Gaussian processes

We study the influence of boundary conditions on self-affine random functions u(t) in the interval t/L \in [0,1], with independent Gaussian Fourier modes of variance ~ 1/q^α. We consider the probability distribution of the mean square width of u(t) taken over the whole interval or in a window t/L \in [x, x+δ]. Its characteristic function can be expressed in terms of the spectrum of an infinite matrix. This distribution strongly depends on the boundary conditions of u(t) for finite δ, but we show that it is universal (independent of boundary conditions) in the small-window limit. We compute it directly for all values of α, using, for α<3, an asymptotic expansion formula that we derive. For α> 3, the limiting width distribution is independent of α. It corresponds to an infinite matrix with a single non-zero eigenvalue. We give the exact expression for the width distribution in this case. Our analysis facilitates the estimation of the roughness exponent from experimental data, in cases where the standard extrapolation method cannot be used

cond-mat.stat-mech↗

arXiv2006

Dynamics below the depinning threshold

We study the steady-state low-temperature dynamics of an elastic line in a disordered medium below the depinning threshold. Analogously to the equilibrium dynamics, in the limit T->0, the steady state is dominated by a single configuration which is occupied with probability one. We develop an exact algorithm to target this dominant configuration and to analyze its geometrical properties as a function of the driving force. The roughness exponent of the line at large scales is identical to the one at depinning. No length scale diverges in the steady state regime as the depinning threshold is approached from below. We do find, a divergent length, but it is associated only with the transient relaxation between metastable states.

cond-mat.dis-nn↗

arXiv2005

Variant Monte Carlo algorithm for driven elastic strings in random media

We discuss the non-local Variant Monte Carlo algorithm which has been successfully employed in the study of driven elastic strings in disordered media at the depinning threshold. Here we prove two theorems, which establish that the algorithm satisfies the crucial no-passing rule and that, after some initial time, the string exclusively moves forward. The Variant Monte Carlo algorithm overcomes the shortcomings of local methods, as we show by analyzing the depinning threshold of a single-pin problem.

cond-mat.dis-nn↗

arXiv2005

Geometry of Gaussian signals

We consider Gaussian signals, i.e. random functions $u(t)$ ($t/L \in [0,1]$) with independent Gaussian Fourier modes of variance $\sim 1/q^α$, and compute their statistical properties in small windows $[x, x+δ]$. We determine moments of the probability distribution of the mean square width of $u(t)$ in powers of the window size $δ$. We show that the moments, in the small-window limit $δ\ll 1$, become universal, whereas they strongly depend on the boundary conditions of $u(t)$ for larger $δ$. For $α> 3$, the probability distribution is computed in the small-window limit and shown to be independent of $α$.

cond-mat.dis-nn↗

arXiv2005

Critical exponents of the driven elastic string in a disordered medium

We analyze the harmonic elastic string driven through a continuous random potential above the depinning threshold. The velocity exponent beta = 0.33(2) is calculated. We observe a crossover in the roughness exponent zeta from the critical value 1.26 to the asymptotic (large force) value of 0.5. We calculate directly the velocity correlation function and the corresponding correlation length exponent nu = 1.29(5), which obeys the scaling relation nu = 1/(2-zeta), and agrees with the finite-size-scaling exponent of fluctuations in the critical force. The velocity correlation function is non-universal at short distances.

cond-mat.stat-mech↗

arXiv2004

Jamming and geometric representations of graphs

We expose a relationship between jamming and a generalization of Tutte's barycentric embedding. This provides a basis for the systematic treatment of jamming and maximal packing problems on two-dimensional surfaces.

math.CO↗

arXiv2003

Cluster Monte Carlo algorithms

In recent years, a better understanding of the Monte Carlo method has provided us with many new techniques in different areas of statistical physics. Of particular interest are so called cluster methods, which exploit the considerable algorithmic freedom given by the detailed balance condition. Cluster algorithms appear, among other systems, in classical spin models, such as the Ising model, in lattice quantum models (bosons, quantum spins and related systems) and in hard spheres and other `entropic' systems for which the configurational energy is either zero or infinite. In this chapter, we discuss the basic idea of cluster algorithms with special emphasis on the pivot cluster method for hard spheres and related systems, for which several recent applications are presented.We provide less technical detail but more context than in the original papers.

cond-mat.stat-mech↗

arXiv2003

Coulomb and Liquid Dimer Models in Three Dimensions

We study classical hard-core dimer models on three-dimensional lattices using analytical approaches and Monte Carlo simulations. On the bipartite cubic lattice, a local gauge field generalization of the height representation used on the square lattice predicts that the dimers are in a critical Coulomb phase with algebraic, dipolar, correlations, in excellent agreement with our large-scale Monte Carlo simulations. The non-bipartite FCC and Fisher lattices lack such a representation, and we find that these models have both confined and exponentially deconfined but no critical phases. We conjecture that extended critical phases are realized only on bipartite lattices, even in higher dimensions.

cond-mat.stat-mech↗

arXiv2003

Width distribution of contact lines on a disordered substrate

We have studied the roughness of a contact line of a liquid meniscus on a disordered substrate by measuring its width distribution. The comparison between the measured width distribution and the width distribution calculated in previous works, extended here to the case of open boundary conditions, confirms that the Joanny-de Gennes model is not sufficient to describe the dynamics of contact lines at the depinning threshold. This conclusion is in agreement with recent measurements which determine the roughness exponent by extrapolation to large system sizes.

cond-mat.dis-nn↗

arXiv2003

Universal interface width distributions at the depinning threshold

We compute the probability distribution of the interface width at the depinning threshold, using recent powerful algorithms. It confirms the universality classes found previously. In all cases, the distribution is surprisingly well approximated by a generalized Gaussian theory of independant modes which decay with a characteristic propagator G(q)=1/q^(d+2 zeta); zeta, the roughness exponent, is computed independently. A functional renormalization analysis explains this result and allows to compute the small deviations, i.e. a universal kurtosis ratio, in agreement with numerics. We stress the importance of the Gaussian theory to interpret numerical data and experiments.

cond-mat↗

arXiv2002

Disks on a Sphere and two-dimensional Glasses

I describe the classic circle-packing problem on a sphere, and the analytic and numerical approaches that have been used to study it. I then present a very simple Markov-chain Monte Carlo algorithm, which succeeds in finding the best solutions known today. The behavior of the algorithm is put into the context of the statistical physics of glasses.

cond-mat.dis-nn↗