Exponential Relaxation out of Nonequilibrium

Simulation results are presented for a quench from a disordered state to a state below the coexistence curve. The model which we consider is the Ising model but with the dynamics governed by the Swendsen-Wang transition probabilities. We show that the resulting domain growth has an exponential instead of a power law behaviour and that the system is non-self-averaging while in nonequilibrium. The simulations were carried out on a parallel computer with up to 128 processors.

Rejection-Free Monte Carlo

So far, we have been using the rejection Monte Carlo algorithms. To remind us, the algorithms proceed from state x to possible state \(x'\) as outlined in Algorithm 1.

Critical wetting in the square Ising model with a boundary field

The Ising square lattice with nearest-neighbor exchangeJ>0 and a free surface at which a boundary magnetic fieldH1 acts has a second-order wetting transition. We study the surface excess magnetization and the susceptibility ofL×M lattices by Monte Carlo simulation and probe the critical behavior of this wetting transition, applying finite-size scaling methods. For the cases studied, the results are not consistent with the presumably exactly known values of the critical exponents, because the asymptotic critical region has not yet been reached. Implication of our results for critical wetting in three dimensions and for the application of the present model to adsorbed wetting layers at surfac…

Growth of Domains and Scaling in the Late Stages of Phase Separation and Diffusion-Controlled Ordering Phenomena

These lectures consider the kinetics of phase changes, induced by a sudden change of external thermodynamic parameters. E.g., we treat a system with a second-order transition at a critical temperature Tc (Fig. 1, left part). For T0 > Tc the system is disordered, while for T < Tc there is an order parameter ± ψ (implying one-component orderings, e.g., an Ising model; later we discuss generalizations). We consider a “quenching experiment”: The system is brought from an initially disordered state at T0 to a state at T where in equilibrium the system should be orderedl. Since no sign of ψ is preferred, the system starts forming locally ordered regions of either sign, separated by domain walls. …

Theoretical Foundations of the Monte Carlo Method and Its Applications in Statistical Physics

In this chapter we first introduce the basic concepts of Monte Carlo sampling, give some details on how Monte Carlo programs need to be organized, and then proceed to the interpretation and analysis of Monte Carlo results.

Crossover scaling in semidilute polymer solutions: a Monte Carlo test

Finite-size scaling in a microcanonical ensemble

The finite-size scaling technique is extended to a microcanonical ensemble. As an application, equilibrium magnetic properties of anL×L square lattice Ising model are computed using the microcanonical ensemble simulation technique of Creutz, and the results are analyzed using the microcanonical ensemble finite-size scaling. The computations were done on the multitransputer system of the Condensed Matter Theory Group at the University of Mainz.

Finite-size scaling analysis of the ?4 field theory on the square lattice

Monte-Carlo calculations are performed for the model Hamiltonian ℋ = ∑i[(r/2)Φ 2(i)+(u/4)/gF4(i)]+∑ (C/2)[Φ (i)−Φ(j)]2 for various values of the parametersr, u, C in the crossover region from the Ising limit (r→-∞,u+∞) to the displacive limit (r=0). The variableφ(i) is a scalar continuous spin variable which can lie in the range-∞<φ(i)<+∞, for each lattice site (i).φ(i) is a priori selected proportional to the single-site probability in our Monte Carlo algorithm. The critical line is obtained in very good agreement with other previous approaches. A decrease of apparent critical exponents, deduced from a finite-size scaling analysis, is attributed to a crossover toward mean-field values at t…

Structure of Polymer Brushes in Cylindrical Tubes: A Molecular Dynamics Simulation

Molecular Dynamics simulations of a coarse-grained bead-spring model of flexible macromolecules tethered with one end to the surface of a cylindrical pore are presented. Chain length $N$ and grafting density $\sigma$ are varied over a wide range and the crossover from ``mushroom'' to ``brush'' behavior is studied for three pore diameters. The monomer density profile and the distribution of the free chain ends are computed and compared to the corresponding model of polymer brushes at flat substrates. It is found that there exists a regime of $N$ and $\sigma$ for large enough pore diameter where the brush height in the pore exceeds the brush height on the flat substrate, while for large enoug…

Kinetics of domain growth in finite Ising strips

Abstract Monte Carlo simulations are presented for the kinetics of ordering of the two-dimensional nearest-neighbor Ising models in an L x M geometry with two free boundaries of length M ⪢ L . This geometry models a “terrace” of width L on regularly stepped surfaces, adatoms adsorbed on neighboring terraces being assumed to be noninteracting. Starting out with an initially random configuration of the atoms in the lattice gas at coverage θ = 1 2 in the square lattice, quenching experiments to temperatures in the range 0.85⩽ T / T c ⩽1 are considered, assuming a dynamics of the Glauber model type (no conservation laws being operative). At T c the ordering behavior can be described in terms of…

Dynamical block analysis in a non-equilibrium system

Abstract We present molecular dynamics simulation results of quenches into the unstable region of a two-dimensional Lennard-Jones system. The evolution of the system from the non-equilibrium state into equilibrium was analyzed with a dynamical block analysis. This can lead to a new approach in the study of non-equilibrium phenomena. We show that with such an analysis one can obtain results on the dynamic evolution as the system evolves, consistent with those obtained from and analysis of the pair-distribution function, structure factor and excess energy. The simulations were carried out on the parallel computer of the condensed matter theory group at the University of Mainz.

Guide to Practical Work with the Monte Carlo Method

The guide is structured such that we proceed from the “easy” simulation methods and algorithms to the more sophisticated. For each method the algorithms are presented by the technique of stepwise refinement. We first present the idea and the basic outline. From then on we proceed by breaking up the larger logical and algorithmic structures into smaller ones, until we have reached the level of single basic statements. Sometimes we may elect not to go to such a depth and the reader is asked to fill in the gaps.

Monte Carlo Methods for the Sampling of Free Energy Landscapes

In this chapter, we return to classical statistical mechanics, wherein the canonical ensemble averages of an observable \(A(\overrightarrow{x})\), where \(\overrightarrow{x} \) stands symbolically for the “microstate” coordinate in the configurational part of the phase space of the system, are given by (cf. Sect. 2.1.1)

The Ising square lattice in aL�M geometry: A model for the effect of surface steps on phase transitions in adsorbed monolayers

Critical phenomena in adsorbed monolayers on surfaces are influenced by limited substrate homogeneity, such as surface steps. We consider the resulting finite-size and boundary effects in the framework of a lattice gas system with nearest neighbor attraction in aL×M geometry, with two free boundaries of lengthM≫L, and periodic boundary conditions in the other direction (along the direction of the steps). This geometry thus models a “terrace” of the stepped surface, and adatoms adsorbed on neighboring terraces are assumed to be non-interacting. Also the effect of boundary “fields” is considered (describing the effects of missing neighbors and changed binding energy to the substrate near the …

Influence of a continuous quenching procedure on the initial stages of spinodal decomposition

Instead of the standard assumption in the theory of phase separation where an instantaneous quench from an initial equilibrium state to the final state in the two-phase region is assumed, we consider the more realistic situation that the change of the external control parameter (e.g. temperature) can only be performed with finite rates. During the initial stages of spinodal decomposition the system then has some “memory” of the states intermediate between the initial and the final one. This influence of the finite quench rate in continuous quenching procedures is studied within the linearized theory of spinodal decomposition, with the Langer-Baron-Miller decoupling, and with Monte Carlo sim…

Scaling and mean-field-like behaviour in phase separation processes

Extensive Monte Carlo calculations are used to study the still not well understood process of phase separation in binary systems. The results show that for a separation dominated by long-wavelength fluctuations a mean-field description holds in certain concentration regions. This, however, is only true for short times after the system has been brought into a non-equilibrium state. A crucial parameter is the interaction range. It determines the region and the time where the mean field description is valid. At later times the structure factor exhibits dynamical scaling. Scaling is also investigated for the metastable states. The results are applicable to polymer blends with long chains or bin…

Formation of Ordered Structures in Quenching Experiments: Scaling Theory and Simulations

In this note we want to address the particular problem of the formation of ordered structures resulting from “quenching experiments”. The generic experimental situation is depicted in Figure 1. Initially the system is in an unordered random state in the one-phase region. Then the temperature is lowered (for some systems like polymers the coexistence curve is inverted so that the temperature must be raised) until the system is in the two phase region. The system is now in a non-equilibrium situation and evolves toward equilibrium. It is during the evolution toward equilibrium that the system develops ordered structures /1,2/.

Self-diffusion in polymer solutions using the bond-fluctuation MC-algorithm

Abstract A lattice Monte Carlo study of the self-diffusion of polymer chains in an athermal solution of equal chains is presented. The examined chain lengths, N (= 20–200), and volume fractions, φ (= 0.025-0.5), cover the range from dilute solution to concentrated solution, respectively. The dynamics show a gradual crossover from Rouse to reptation-like behaviour. Analysing the data according to a scaling theory and taking into account the density dependence of the microscopic length and time-scales, an almost perfect scaling of the self-diffusion coefficient is achieved. The high statistical accuracy of the data (103–104 chains per parameter combination) was obtainable by using a transpute…

Monte Carlo study of the ising model phase transition in terms of the percolation transition of “physical clusters”

Finite squareL×L Ising lattices with ferromagnetic nearest neighbor interaction are simulated using the Swendsen-Wang cluster algorithm. Both thermal properties (internal energyU, specific heatC, magnetization 〈|M|〉, susceptibilityχ) and percolation cluster properties relating to the “physical clusters,” namely the Fortuin-Kasteleyn clusters (percolation probability 〈P∞〉, percolation susceptibilityχp, cluster size distributionnl) are evaluated, paying particular attention to finite-size effects. It is shown that thermal properties can be expressed entirely in terms of cluster properties, 〈P∞〉 being identical to 〈|M|〉 in the thermodynamic limit, while finite-size corrections differ. In contr…

Critical phenomena in polymer mixtures: Monte Carlo simulation of a lattice model

A lattice model of a symmetrical binary (AB) polymer mixture is studied, modelling the polymer chains by self-avoiding walks withN A =N B =N steps on a simple cubic lattice. If a pair of nearest neighbour sites is taken by different monomersAB orBA, an energye ab is won; if the pair of sites is taken by anAA or aBB pair, an energye is won, while the energy is reduced to zero if at least one of the sites of the pair is vacant. To allow enough chain mobility, 20% of the lattice sites are vacancies. In addition to local motions of the chain segments we use a novel “grand-canonical” simulation technique:A chains are transformed intoB chains and vice versa, keeping the chemical potential differe…

Transient Reversible Growth and Percolation During Phase Separation

Binary mixtures when quenched into the two-phase region exhibit transient percolation phenomena. These transient percolation phenomena and the underlying mechanism of transient reversible growth are investigated. In particular, one of the possible dynamical percolation lines between the dynamical spinodal and the line of macroscopic percolation is traced out. Analyzing the finite size effects with the usual scaling theory one finds exponents which seem to be inconsistent with the universality class of percolation. However, at zero temperature, where the growth is non-reversible and the transition of a sol-gel type, the exponents are consistent with those of random percolation.

Quantum Monte Carlo Simulations: An Introduction

To be specific, let us consider for the moment the problem of N atoms in a volume V at temperature T, and we wish to calculate the average of some observable A which in quantum mechanics is described by an operator Â.

Fluctuations and lack of self-averaging in the kinetics of domain growth

The fluctuations occurring when an initially disordered system is quenched at timet=0 to a state, where in equilibrium it is ordered, are studied with a scaling theory. Both the mean-sizel(t)d of thed-dimensional ordered domains and their fluctuations in size are found to increase with the same power of the time; their relative size fluctuations are independent of the total volumeLd of the system. This lack of self-averaging is tested for both the Ising model and the φ4 model on the square lattice. Both models exhibit the same lawl(t)=(Rt)x withx=1/2, although the φ4 model has “soft walls”. However, spurious results withx≷1/2 are obtained if “bad” pseudorandom numbers are used, and if the n…

Some Important Recent Developments of the Monte Carlo Methodology

Roughly at the time (1987) when the manuscript for the first three chapters of the present book was completed, several breakthroughs occurred. They had a profound influence on the scope of Monte Carlo simulations in statistical physics, particularly for the study of phase transitions in lattice models.

System size dependence of the autocorrelation time for the Swendsen-Wang Ising model

Abstract We present Monte Carlo simulation results of the autocorrelation time for the Swendsen-Wang method for the simulation of the Ising model. We have calculated the exponential and the integrated autocorrelation time at the critical point T c of the two-dimensional Ising model. Our results indicate that both autocorrelation times depend logarithmically on the linear system size L instead of a power law. The simulations were carried out on the parallel computer of the condensed matter theory group at the University of Mainz.

Computer simulation of the glass transition of polymer melts

Bond fluctuation models on square and simple cubic lattices at melt densities are simulated, using potentials depending on the length of the (effective) bond (and also on the bond angle, in d=3 dimensions). Various relaxation functions have the Kohlrausch-Williams-Watts (KWW) form; the associated relaxation time diverges as exp (const/T 2) in d=2 and as exp [const/T−T 0)] in d=3. For d=3 the self-diffusion constant also follows the Vogel-Fulcher law, with T 0=250 K for chain lengths N=20 and potentials adapted to bisphenol-A-polycarbonate [BPA-PC].

Introduction: Purpose and Scope of this Volume, and Some General Comments

In recent years the method of “computer simulation” has started something like a revolution of science: the old division of physics (as well as chemistry, biology, etc.) into “experimental” and “theoretical” branches is no longer really complete. Rather, “computer simulation” has become a third branch complementary to the first two traditional approaches.

Dynamic percolation transition induced by phase separation: A Monte Carlo analysis

The percolation transition of geometric clusters in the three-dimensional, simple cubic, nearest neighbor Ising lattice gas model is investigated in the temperature and concentration region inside the coexistence curve. We consider “quenching experiments,” where the system starts from an initially completely random configuration (corresponding to equilibrium at infinite temperature), letting the system evolve at the considered temperature according to the Kawasaki “spinexchange” dynamics. Analyzing the distributionnl(t) of clusters of sizel at timet, we find that after a time of the order of about 100 Monte Carlo steps per site a percolation transition occurs at a concentration distinctly l…

Finite Size Scaling Tools for the Study of Interfacial Phenomena and Wetting

In this chapter, we use the word “interface” in the sense of a boundary between coexisting bulk phases (in thermal equilibrium). An example is the interface between liquid (e.g. water) and gas phases (water vapor) but also interfaces between fluid and solid phases (e.g. water and ice) can be considered, as well as interfaces between coexisting solid phases. The generic example are “domain walls” in magnets, separating domains with opposite orientation of the magnetization, a case that can already be studied in the framework of the simple Ising model (Chaps. 2 and 3) where one has spins on the sites of a rigid perfect lattice pointing up or down.

Cluster Algorithms and Reweighting Methods

Roughly at the time (1987) when the manuscript for the first three chapters of the present book was completed, several breakthroughs occurred. They had a profound influence on the scope of Monte Carlo simulations in statistical physics, particularly for the study of phase transitions in lattice models.

Cover Picture: Macromol. Theory Simul. 7/2006

General Introduction to Computer Simulation Methods

Computer simulation methods are now an established tool in many branches of science. The motivation for computer simulations of physical systems are manifold. One of the main motivations is that one eliminates approximations with computer simulations. Usually to treat a problem analytically (if it can be done at all) one needs to resort to some kind of approximation; for exam- ple a mean-field-type approximation. With a computer simulation we have the ability to study systems not yet tractable with analytical methods. The computer simulation approach allows one to study complex systems and gain insight into their behaviour. Indeed, the complexity can go far beyond the reach of present analy…