DOCSLIB.ORG

Arxiv:2011.08560V1

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Arxiv:2011.08560V1 Nonflat Histogram Techniques for Spin Glasses Fabio M¨uller,∗ Stefan Schnabel,† and Wolfhard Janke‡ Institut f¨ur Theoretische Physik, Universit¨at Leipzig, IPF 231101, 04081 Leipzig, Germany (Dated: November 18, 2020) We study the bimodal Edwards-Anderson spin glass comparing established methods, namely the multicanonical method, the 1/k-ensemble and parallel tempering, to an approach where the ensemble is modified by simulating power-law-shaped histograms in energy instead of flat histograms as in the standard multicanonical case. We show that by this modification a significant speed-up in terms of mean round-trip times can be achieved for all lattice sizes taken into consideration. I. INTRODUCTION to Simulated Annealing [14] as the system is gradually cooled down according to an annealing schedule. The an- Simulations of systems with rugged free-energy land- nealing is performed on a big population of replicas and scape [1] suffer from massive slowing down of the dy- by introducing intermediate resampling of the population namics in the low-temperature phase. This problem en- of replicas after lowering the temperature the simulation countered in many physical systems, e.g., folding poly- is kept at thermal equilibrium. This permits the evalua- mers or spin glasses, renders the investigation of the tion of thermodynamic observables in contrast to simple thermodynamical properties of such systems in the low- Simulated Annealing. Despite the attempts of optimizing temperature phase a very challenging task. the method for spin glasses [10, 12, 15] it is not able to The Metropolis algorithm [2] is designed to sample con- outperform PT. Its optimization, however, remains more figurations according to their statistical weight in the cumbersome due to the additional complexity. The main canonical ensemble. At low temperatures it fails dra- advantage of this algorithm is its suitability for massively matically for systems with rugged free-energy landscape parallel implementation. For disordered systems, how- because of the effectively broken ergodicity. The simula- ever, this advantage does not come into play, because tions get stuck in local minima (metastable states) and the necessity of simulating many different disorder real- the thermal energy is not sufficient to overcome the huge izations allows the efficient use of parallel computing for free-energy barriers. There have been a wide range of any method. algorithmic developments tackling this problem that can The Multicanonical (MUCA) method [16–18] is an- all be subsumed in the term of broad-energy ensembles. other well established algorithm designed for the simu- One commonly employed method is the Parallel Tem- lation of systems with first-order phase transitions which pering (PT) [3–5] method where Metropolis simulations performs well in the simulation of systems with rugged of copies of the system (replicas) at different tempera- free-energy landscape, too. It has already been applied tures are performed. After certain time intervals ex- to spin glasses in Refs. [19, 20]. In this method the simu- changes of the replicas between the different tempera- lation is set up to visit all possible energies with the same tures are attempted. This procedure enables the replicas probability yielding a flat histogram in energy. However, at low temperature to fully explore deep free-energy val- it has been noted by different researchers that this en- leys and at high temperature to travel freely through the semble is not optimal. One suggested improvement is the phase space and thus to decorrelate. Thereby, the differ- 1/k-ensemble by Hesselbo and Stinchcombe [21], where ent replicas can explore the rugged structure of the free- the sampling distribution is the inverse of the integrated energy landscape much more efficiently than in a simple density of states. As the authors point out, this descrip- Metropolis simulation. Using this method at tempera- tion samples the low-energy region more often than the tures close to the transition, studies of spin-glass sys- high-energy region, resulting in energy histograms which 3 tems of sizes up to 40 spins have been reported [6, 7]. grow towards the low-temperature phase. For ground-state searches systems of about 103 are fea- arXiv:2011.08560v1 [cond-mat.dis-nn] 17 Nov 2020 sible [8]. The great advantage of PT is its simplicity: Another non-parametric optimization of the MUCA al- the algorithm only needs a suitable temperature set and gorithm was proposed in Ref. [22]. The method uses an exhibits good performance which makes it probably the estimator of the local diffusivity in order to maximize the most employed method in the investigation of systems number of performed round trips in energy. The method with rugged free-energy landscape. is among others applied to the ferromagnetic Ising model Another recent development is the Population Anneal- for which it improves the scaling behavior of the round- ing Monte Carlo [9–13] method which proceeds similarly trip times in energy. The improved performance for the models considered in that work and the nature of the algorithm of automatically identifying the bottlenecks of the simulation and concentrating the simulation effort on ∗ [email protected] this region suggest that the round-trip times of the simu- † [email protected] lations should diminish independently of the underlying ‡ [email protected] system. However, in our implementation, in the case of 2 11 the three-dimensional (3D) bimodal Edwards-Anderson 10 5 10 (EA) spin glass [23], the round-trip times did not sys- flat PSH(E) 1/k 4 tematically improve with this method. Instead the simu- PT 10 lation got stuck for some of the considered samples, ren- 9 power law 10 3 dering a comparison to the other methods impossible. 10 ) ) E ( In this work we present a different approach: we pre- E 2 ( 10 scribe parametric profiles for the histograms of the simu- SH H E 7 P lation and adjust the simulation weights accordingly. As 10 ∆ 1 g 10 E for the three previous MUCA variants, it requires the − g 0 knowledge of the underlying density of states, but it is E 10 much more flexible. The profiles are all chosen to be 5 10 shifted power laws having two free parameters. −1000 −800 −600 −400 −200 0 As an example we consider the 3D bimodal EA spin E glass. This is one of the simplest models exhibiting a rugged free-energy landscape and is also interesting from FIG. 1. The recorded histograms H(E) of the different meth- the point of view of an optimization problem where find- ods and the profile function PSH(E) for one disorder realiza- ing ground states of hard disorder realizations is NP- tion of linear lattice size L = 8. The dotted and the dashed hard [24]. Despite the exponential growth of the compu- vertical lines indicate the position of the ground-state energy Eg and the position of the pole of the power law (5), respec- tational resources fundamental questions regarding the tively. nature of the spin-glass phase still remain. For the progress in understanding the open questions the devel- opment of new methods and an improvement of the ex- criterion with an energy dependent weight function isting methods is crucial. The rest of the paper is organized as follows. In Sec. II W (Enew) Pacc = min 1, , (2) the spin-glass model and the simulation methods are ex- W (E ) old plained. The direct comparison of the round-trip times of the individual methods is performed in Sec. III. The where the weight function is proportional to the inverse of the density of states Ω(E), framework of extreme-value statistics is introduced in Sec. IV. In Sec. V benchmarks for the global comparison W (E) ∝ Ω−1(E). (3) are discussed and the different methods are compared in terms of those benchmarks. The results are summarized For the MUCA simulations Ω(E) has to be sufficiently in Sec. VI. well-known a priori for each disorder realization. An es- timator for it can, for instance, be obtained by means of the Wang-Landau algorithm [25] or, as in this work, by II. MODEL AND EMPLOYED METHODS other iterative procedures which are explained, e.g., in Ref. [26]. This ensemble produces histograms which are We take into consideration the 3D bimodal EA model flat in energy and is, therefore, often also referred to as whose Hamiltonian takes the form “flat histogram method”. A straightforward generalization of the flat histogram H = − Jij SiSj, (1) method are the nonflat histogram methods. If the sim- Xhiji ulation weights for the flat MUCA method are multi- plied with the desired energy dependent shape (or profile) where the bonds Jij and the spins Si can take values ±1. function PSH(E) The sum runs over all neighboring spins in the simple- −1 cubic lattice with periodic boundary conditions. W (E) ∝ Ω (E)PSH(E), (4) Due to the disordered nature of spin glasses the study has to take into account a sufficiently large set of disor- the resulting histograms will be shaped according to der realizations on which the averaged quantities can be PSH(E). In this work all the profiles are shifted power computed. In this case one disorder realization consists laws of the form α of a set of 3V couplings Jij which are either positive or E 3 P (E, ∆E, α)= +1 , (5) negative unity with a probability of 50%, where V = L SH ∆E − E is number of spins in a lattice of linear lattice size L. The g disorder realizations are generated prior to the simulation where the exponent α < 0 and ∆E > 0 is the position and then kept fixed for all times (quenched disorder).
Load more

Recommended publications
  • An Efficient Algorithm for Numerical Computations of Continuous Densities of States
    Eur. Phys. J. C (2016) 76:306 DOI 10.1140/epjc/s10052-016-4142-5 Regular Article - Theoretical Physics An efficient algorithm for numerical computations of continuous densities of states K. Langfeld1,a, B. Lucini2,b, R. Pellegrini3,c, A. Rago1,d 1 Centre for Mathematical Sciences, Plymouth University, Plymouth PL4 8AA, UK 2 College of Science, Swansea University, Singleton Park, Swansea SA2 8PP, UK 3 School of Physics and Astronomy, University of Edinburgh, Edinburgh EH9 3FD, UK Received: 15 February 2016 / Accepted: 16 May 2016 © The Author(s) 2016. This article is published with open access at Springerlink.com Abstract In Wang–Landau type algorithms, Monte-Carlo the potential of the method for studies of first order phase updates are performed with respect to the density of states, transitions. Other situations where the method is expected to which is iteratively refined during simulations. The partition be superior to importance sampling techniques are pointed function and thermodynamic observables are then obtained out. by standard integration. In this work, our recently introduced method in this class (the LLR approach) is analysed and fur- ther developed. Our approach is a histogram free method Contents particularly suited for systems with continuous degrees of freedom giving rise to a continuum density of states, as it is 1 Introduction and motivations .............. commonly found in lattice gauge theories and in some sta- 2 Numerical determination of the density of states ... tistical mechanics systems. We show that the method pos- 2.1 The density of states ............... sesses an exponential error suppression that allows us to 2.2 The LLR method ................
    [Show full text]
  • The Measurement of Free Energy by Monte-Carlo Computer Simulation
    The Measurement of Free Energy by MonteCarlo Computer Simulation Graham R Smith A thesis submitted in fullment of the requirements for the degree of Do ctor of Philosophy to the University of Edinburgh Abstract One of the most imp ortant problems in statistical mechanics is the measurement of free energies these b eing the quantities that determine the direction of chemical reactions andthe concern of this thesisthe lo cation of phase transitions While Monte Carlo MC computer simulation is a wellestablished and invaluable aid in statistical mechanical calculations it is well known that in its most commonlypractised form where samples are generated from the Boltzmann distribution it fails if applied directly to the free energy problem This failure o ccurs b ecause the measurement of free energies requires a much more extensive exploration of the systems conguration space than do most statistical mechanical calculations congurations which have a very low Boltzmann probability make a substantial contribution to the free energy and the imp ortant regions of conguration space may b e separated by p otential barriers We b egin the thesis with an introduction and then give a review of the very substantial literature that the problem of the MC measurement of free energy has pro duced explaining and classifying the various dierent approaches that have b een adopted We then pro ceed to present the results of our own investigations First we investigate metho ds in which the congurations of the system are sampled from a distribution other than the
    [Show full text]
  • Markov Chain Monte Carlo Simulations and Their Statistical Analysis – an Overview
    Markov Chain Monte Carlo Simulations and Their Statistical Analysis – An Overview Bernd Berg FSU, August 30, 2005 Content 1. Statistics as needed 2. Markov Chain Monte Carlo (MC) 3. Statistical Analysis of MC Data and Advanced MC. Link to computer code at www.hep.fsu.edu/˜ berg . 1 Probability Distributions and Sampling In N experiments we may find an event A to occur n times. The frequency definition of the probability of the event is n P (A) = lim . N→∞ N Let P (a, b) be the probability that xr ∈ [a, b] where xr is a random variable drawn in the interval (−∞, +∞) with a probability density f(x) > 0. Then, Z b P (y, x) P (a, b) = dx f(x) and f(x) = lim . a y→x x − y The (cumulative) distribution function of the random variable xr is defined as Z x F (x) = P (xr ≤ x) = f(x0) dx0 . −∞ 2 For uniform probability distribution between [0, 1), n 1 for 0 ≤ x < 1; u(x) = 0 elsewhere. The corresponding distribution function is Z x ( 0 for x < 0; U(x) = u(x0) dx0 = x for 0 ≤ x ≤ 1; −∞ 1 for x > 1. It allows for the construction of general probability distributions. Let Z x y = F (x) = f(x0) dx0 . −∞ For yr being a uniformly distributed random variable in [0, 1) xr = F −1(yr) is then distributed according to the probability density f(x) . 3 Example Mapping of the uniform to the Cauchy distribution. 1 0.8 r 0.6 0.4 Uniform y 0.2 0 -10 -5 0 5 10 Cauchy xr 4 Pseudo Random Numbers and Computer Code It is sufficient to generate uniform (pseudo) random numbers.
    [Show full text]
  • Optimized Multicanonical Simulations: a Proposal Based on Classical Fluctuation Theory
    PHYSICAL REVIEW E 74, 046702 ͑2006͒ Optimized multicanonical simulations: A proposal based on classical fluctuation theory J. Viana Lopes Centro de Física do Porto and Departamento de Física, Faculdade de Ciências, Universidade do Porto, 4169-007 Porto, Portugal and Departamento de Física, Instituto Superior de Engenharia, Instituto Politécnico do Porto, Porto, Portugal Miguel D. Costa and J. M. B. Lopes dos Santos Centro de Física do Porto and Departamento de Física, Faculdade de Ciências, Universidade do Porto, 4169-007 Porto, Portugal R. Toral Instituto Mediterráneo de Estudios Avanzados IMEDEA (CSIC-UIB), Edificio Mateu Orfila, Campus UIB, E-07122 Palma de Mallorca, Spain ͑Received 26 May 2006; published 5 October 2006͒ We propose a recursive procedure to estimate the microcanonical density of states in multicanonical Monte Carlo simulations which relies only on measurements of moments of the energy distribution, avoiding entirely the need for energy histograms. This method yields directly a piecewise analytical approximation to the microcanonical inverse temperature ␤͑E͒ and allows improved control over the statistics and efficiency of the simulations. We demonstrate its utility in connection with recently proposed schemes for improving the effi- ciency of multicanonical sampling, either with adjustment of the asymptotic energy distribution or with the replacement of single spin flip dynamics with collective updates. DOI: 10.1103/PhysRevE.74.046702 PACS number͑s͒: 02.70.Rr, 64.60.Cn, 75.10.Hk, 05.10.Ln I. INTRODUCTION studies ͓15,16͔ have shown that very long equilibration times can remain a serious concern. Furthermore, the number of Over the past decade, the use of Monte Carlo methods ͓1͔ independent samples is strongly dependent on energy, mak- has broken the boundaries of statistical physics and has be- ing error estimation rather tricky ͓17͔.
    [Show full text]
  • Multicanonical Jump Walking: a Method for Efficiently Sampling Rough Energy Landscapes Huafeng Xu and B
    JOURNAL OF CHEMICAL PHYSICS VOLUME 110, NUMBER 21 1 JUNE 1999 Multicanonical jump walking: A method for efficiently sampling rough energy landscapes Huafeng Xu and B. J. Berne Department of Chemistry and Center for Biomolecular Simulation, Columbia University, 3000 Broadway, New York, New York 10027 ~Received 5 January 1999; accepted 4 March 1999! The multicanonical sampling and jump walking methods are combined to provide a new, effective means of overcoming quasiergodicity in Monte Carlo simulations. In this new method, configurations generated during a long multicanonical sampling are stored infrequently and a modified jump walking procedure is implemented using this set of configurations to sample phase space at low temperature. Multicanonical jump walking, as this new method is called, is compared with regular jump walking and with straight multicanonical ensemble sampling on two systems: a one-dimensional random potential and an Ar13 cluster. It is shown that for the same number of MC steps, the multicanonical jump walking method more efficiently samples the phase space than either the regular jump walking or the pure multicanonical ensemble sampling method. © 1999 American Institute of Physics. @S0021-9606~99!51920-X# I. INTRODUCTION erated at a high enough temperature to circumvent quasier- godicity. But the configurations at the high temperature tend 1,2 In the study of protein folding, first-order phase tran- to have high energies, and attempts to jump into such con- 3,4 5–7 sition phenomena in clusters and spin glasses, potential figurations from low temperature configurations ~therefore energy minima are often separated by high energy barriers, normally with low energies! are often rejected.
    [Show full text]
  • Arxiv:Hep-Lat/9204001V1 8 Apr 1992 a New Approach to Spin Glass Simulations
    A New Approach to Spin Glass Simulations∗ Bernd A. Berg1,2 and Tarik Celik1,3 1 Supercomputer Computations Research Institute Tallahassee, FL 32306, USA 2 Department of Physics, The Florida State University Tallahassee, FL 32306, USA 3 On leave of absence from Department of Physics Hacettepe University Ankara, Turkey March 19, 2018 arXiv:hep-lat/9204001v1 8 Apr 1992 ∗Submitted to Physical Review Letters. 1 Abstract We present a recursive procedure to calculate the parameters of the recently introduced multicanonical ensemble and explore the approach for spin glasses. Temperature dependence of the energy, the entropy and other physical quantities are easily calculable and we report re- sults for the zero temperature limit. Our data provide evidence that the large L increase of the ergodicity time is greatly improved. The multicanonical ensemble seems to open new horizons for simulations of spin glasses and other systems which have to cope with conflicting constraints. 2 The theoretical understanding of spin glasses, for a review see [1], has re- mained a great challenge. In particular the low temperature limit leaves many open questions about the effects of disorder and frustration. For instance, it has remained controversial whether Parisi’s [2] mean field theory provides the appropriate description for 3D spin glasses. The attractive alternative is the droplet model [3], which in turn is equivalent to a one parameter scaling picture [4]. The simplest spin glass system to study such questions numeri- cally is the Edwards-Anderson model. In its Ising version it is described by the Hamiltonian H = − J s s , (1) X ij i j <ij> where the sum goes over nearest neighbours and the exchange interactions Jij = ±1 between the spins si = ±1 are quenched random variables.
    [Show full text]
  • Biophysics and Physicobiology
    Special Issue Vol. 16, pp. 344–366 (2019) “Progress of Theoretical and Computational Biophysics” doi: 10.2142/biophysico.16.0_344 Biophysics and Physicobiology https://www.jstage.jst.go.jp/browse/biophysico/ Review Article Protein structure predictions by enhanced conformational sampling methods Yuko Okamoto1,2,3,4,5 1Department of Physics, Graduate School of Science, Nagoya University, Nagoya, Aichi 464-8602, Japan 2Structural Biology Research Center, Graduate School of Science, Nagoya University, Nagoya, Aichi 464-8602, Japan 3Center for Computational Science, Graduate School of Engineering, Nagoya University, Nagoya, Aichi 464-8603, Japan 4Information Technology Center, Nagoya University, Nagoya, Aichi 464-8601, Japan 5JST-CREST, Nagoya, Aichi 464-8602, Japan Received July 5, 2019; accepted August 7, 2019 In this Special Festschrift Issue for the celebration of lization of the native conformation of a protein are consis- Professor Nobuhiro Gō’s 80th birthday, we review tent with each other. This principle was deduced from the enhanced conformational sampling methods for protein results of Monte Carlo simulations of a lattice protein model. structure predictions. We present several generalized-en- This principle was reinterpreted as the principle of minimal semble algorithms such as multicanonical algorithm, frustration [2,3]. They showed that the phase space of the replica-exchange method, etc. and parallel Monte Carlo protein system is characterized by two parameters, folding or molecular dynamics method with genetic crossover. temperature T and glass transition temperature T and that a Examples of the results of these methods applied to the F G predictions of protein tertiary structures are also pre- protein can fold into a unique native structure if TG/TF<<1.
    [Show full text]
  • Fast Algorithm to Calculate Density of States
    PHYSICAL REVIEW E 75 , 046701 ͑2007 ͒ Fast algorithm to calculate density of states R. E. Belardinelli and V. D. Pereyra * Departamento de Física, Laboratorio de Ciencias de Superficie, Universidad Nacional de San Luis, CONICET, Chacabuco 917, 5700 San Luis, Argentina ͑Received 13 June 2006; revised manuscript received 6 February 2007; published 5 April 2007 ͒ An algorithm to calculate the density of states, based on the well-known Wang-Landau method, is intro- duced. Independent random walks are performed in different restricted ranges of energy, and the resultant density of states is modified by a function of time, F͑t͒ϰt−1 , for large time. As a consequence, the calculated −1/2 density of state, gm͑E,t͒, approaches asymptotically the exact value gex ͑E͒ as ϰt , avoiding the saturation of the error. It is also shown that the growth of the interface of the energy histogram belongs to the random deposition universality class. DOI: 10.1103/PhysRevE.75.046701 PACS number ͑s͒: 02.70.Rr, 64.60.Cn, 05.50. ϩq I. INTRODUCTION ͓24 –31 ͔ and studies of the efficiency and convergence of this algorithm ͓26 ,30 ͔. However, there are limitations of the The Wang-Landau WL algorithm 1 has been one of ͑ ͒ ͓ ͔ method which remain still unsolved, such as, for example, the most interesting and refreshing improvements in the the behavior of the tunneling time, which is a bound for the Monte Carlo MC simulation scheme in the last decade, ͑ ͒ performance of any flat-histogram algorithm, as is discussed applying to a broad spectrum of interesting problems in sta- in Ref.
    [Show full text]
  • Multicanonical Jump Walking: a Method for Ef®Ciently Sampling Rough
    Multicanonical Jump Walking: A method for ef®ciently sampling rough energy landscapes Huafeng Xu and B. J. Berne Department of Chemistry and Center for Biomolecular Simulation Columbia University 3000 Broadway, New York, New York 10027 (December 29, 1998) Abstract The multicanonical sampling and jump-walking methods are combined to provide a new, effective means of overcoming quasi-ergodicity in Monte Carlo simulations. In this new method, con®gurations generated during a long multicanonical sampling are stored infrequently and a modi®ed jump walking procedure is implemented us- ing this set of con®gurations to sample phase space at low temperature. Multicanon- ical jump-walking, as this new method is called, is compared with regular jump walking and with straight multicanonical ensemble sampling on two systems: a 1- dimensional random potential and an Ar cluster. It is shown that for the same number of MC steps, the multicanonical jump-walking method more ef®ciently sam- ples the phase space than either the regular jump walking or the pure multicanonical ensemble sampling method, and is able to locate the global energy minimum more rapidly than the other two methods. I. INTRODUCTION In the study of protein folding, ®rst order phase transition phenomena in clusters and glasses, ± potential energy minima are often separated by high energy barriers, which are ex- 1 ponentially suppressed because of the Boltzmann factor. Consequently, in a simulation of ®nite length, the high energy barriers will be crossed very rarely. The system can get trapped in some of the local energy basins and the sampling method fails to sample large regions of the thermally accessible phase space.
    [Show full text]
  • An Introduction to Monte Carlo Simulation of Statistical Physics
    AN INTRODUCTION TO MONTE CARLO SIMULATIONS IN STATISTICAL PHYSICS by K. P. N. Murthy Theoretical Studies Section, Materials Science Division. arXiv:cond-mat/0104167v5 [cond-mat.stat-mech] 17 Dec 2003 Indira Gandhi Centre for Atomic Research, Kalpakkam 603 102, Tamilnadu. INDIA 16 December 2003 Preface Science is what we understand well enough to explain to a computer. Donald E. Knuth Computing facilities - like work stations and PCs, with high speed and large memory are becoming increasingly and easily available to researchers and students since recent times. This is particularly true in our country; the growth has been rapid and is most likely to become more rapid in the coming years. As a result, computer simulation methods are gaining in importance and popularity as viable and useful tools for research and education. Monte Carlo is an early and an important computer simulation technique. Named after the city (in the province of Monoco, south of France) famous for its (gambling) casinos, the Monte Carlo method makes extensive use of random numbers. It is em- ployed for simulation of a large variety of phenomena in very many different disciplines. However, in this monograph, I shall confine myself to elucidating the technique for sim- ulation of statistical physics systems. We begin with a quick look at what do we mean by ensembles in general and Gibbs ensemble in particular. We discuss briefly microcanonical ensemble that models isolated system, canonical ensemble that describes closed system and grand canonical ensemble, useful in the study of open system. In the following few sections I present briefly the basic ingredients of the method of Monte Carlo.
    [Show full text]
  • Multicanonical Simulations
    Generalized Ensembles: Multicanonical Simulations 1. Multicanonical Ensemble 2. How to get the Weights? 3. Example Runs and Re-Weighting to the Canonical Ensemble 4. Energy and Specific Heat Calculation 5. Free Energy and Entropy Calculation 6. Complex Systems with Rugged Free Energy Landscapes 7. Summary 1 Multicanonical Ensemble One of the questions which ought to be addressed before performing a large scale computer simulation is “What are suitable weight factors for the problem at hand?” So far we used the Boltzmann weights as this appears natural for simulating the Gibbs ensemble. However, a broader view of the issue is appropriate. Conventional canonical simulations can by re-weighting techniques only be extrapolated to a vicinity of this temperature. For multicanonical simulations this is different. A single simulation allows to obtain equilibrium properties of the Gibbs ensemble over a range of temperatures. Of particular interest are two situations for which canonical simulations do not provide the appropriate implementation of importance sampling: 1. The physically important configurations are rare in the canonical ensemble. 2. A rugged free energy landscape makes the physically important configurations difficult to reach. 2 Multicanonical simulations sample, in an appropriate energy range, with an approximation to the weights (k) (k) (k) 1 w (k) = w (E(k)) = e−b(E ) E +a(E ) = bmu mu n(E(k)) where n(E) is the spectral density. The function b(E) defines the inverse microcanonical temperature and a(E) the dimensionless, microcanonical free energy. The function b(E) has a relatively smooth dependence on its arguments, which makes it a useful quantity when dealing with the weight factors.
    [Show full text]
  • Ensemble Optimization Techniques for the Simulation of Slowly Equilibrating Systems
    Ensemble Optimization Techniques for the Simulation of Slowly Equilibrating Systems S. Trebst1, D.A. Huse2, E. Gull3, H.G. Katzgraber3, U.H.E. Hansmann4,5, and M. Troyer3 1 Microsoft Research and Kavli Institute for Theoretical Physics, University of California, Santa Barbara, CA 93106, USA 2 Department of Physics, Princeton University, Princeton, NJ 08544, USA 3 Theoretische Physik, Eidgen¨ossische Technische Hochschule Z¨urich, CH-8093 Z¨urich, Switzerland 4 Department of Physics, Michigan Technological University, Houghton, MI 49931, USA 5 John-von-Neumann Institute for Computing, Forschungszentrum J¨ulich, D-52425 J¨ulich, Germany Abstract. Competing phases or interactions in complex many-particle systems can result in free energy barriers that strongly suppress thermal equilibration. Here we discuss how extended ensemble Monte Carlo simulations can be used to study the equilibrium behavior of such systems. Special focus will be given to a recently developed adaptive Monte Carlo technique that is capable to explore and overcome the entropic barriers which cause the slow-down. We discuss this technique in the context of broad-histogram Monte Carlo algorithms as well as its application to replica-exchange methods such as parallel tempering. We briefly discuss a number of examples including low-temperature states of magnetic systems with competing interactions and dense liquids. 1 Introduction The free energy landscapes of complex many-body systems with competing phases or interactions are often characterized by many local minima that are separated by entropic barriers. The simulation of such systems with conven- tional Monte Carlo [1] or molecular dynamics [2] methods is slowed down by long relaxation times due to the suppressed tunneling through these barriers.
    [Show full text]