Portmanteau model diagnostics and tests for nonlinearity: a comparative Monte Carlo study of two alternative methods

This paper employs an extensive Monte Carlo study to test the size and power of the BDS and close return methods of testing for departures from independent and identical distribution. It is found that the finite sample properties of the BDS test are far superior and that the close return method cannot be recommended as a model diagnostic. Neither test can be reliably used for very small samples, while the close return test has low power even at large sample sizes

Reducing noise associated with the Monte Carlo Independent Column Approximation for weather forecasting models

The Monte Carlo Independent Column Approximation (McICA) is a flexible method for representing subgrid-scale cloud inhomogeneity in radiative transfer schemes. It does, however, introduce conditional random errors but these have been shown to have little effect on climate simulations, where spatial and temporal scales of interest are large enough for effects of noise to be averaged out. This article considers the effect of McICA noise on a numerical weather prediction (NWP) model, where the time and spatial scales of interest are much closer to those at which the errors manifest themselves; this, as we show, means that noise is more significant. We suggest methods for efficiently reducing the magnitude of McICA noise and test these methods in a global NWP version of the UK Met Office Unified Model (MetUM). The resultant errors are put into context by comparison with errors due to the widely used assumption of maximum-random-overlap of plane-parallel homogeneous cloud. For a simple implementation of the McICA scheme, forecasts of near-surface temperature are found to be worse than those obtained using the plane-parallel, maximum-random-overlap representation of clouds. However, by applying the methods suggested in this article, we can reduce noise enough to give forecasts of near-surface temperature that are an improvement on the plane-parallel maximum-random-overlap forecasts. We conclude that the McICA scheme can be used to improve the representation of clouds in NWP models, with the provision that the associated noise is sufficiently small.

Estimation of DSGE Models: A Monte Carlo Analysis

Neste trabalho investigamos as propriedades em pequena amostra e a robustez das estimativas dos parâmetros de modelos DSGE. Tomamos o modelo de Smets and Wouters (2007) como base e avaliamos a performance de dois procedimentos de estimação: Método dos Momentos Simulados (MMS) e Máxima Verossimilhança (MV). Examinamos a distribuição empírica das estimativas dos parâmetros e sua implicação para as análises de impulso-resposta e decomposição de variância nos casos de especificação correta e má especificação. Nossos resultados apontam para um desempenho ruim de MMS e alguns padrões de viés nas análises de impulso-resposta e decomposição de variância com estimativas de MV nos casos de má especificação considerados.

Simulações de Monte Carlo para os modelos Ising e Blume-Capel em redes complexa

We studied the Ising model ferromagnetic as spin-1/2 and the Blume-Capel model as spin-1, > 0 on small world network, using computer simulation through the Metropolis algorithm. We calculated macroscopic quantities of the system, such as internal energy, magnetization, specific heat, magnetic susceptibility and Binder cumulant. We found for the Ising model the same result obtained by Koreans H. Hong, Beom Jun Kim and M. Y. Choi [6] and critical behavior similar Blume-Capel model

The STATFLUX code: a statistical method for calculation of flow and set of parameters, based on the Multiple-Compartment Biokinetical Model

The code STATFLUX, implementing a new and simple statistical procedure for the calculation of transfer coefficients in radionuclide transport to animals and plants, is proposed. The method is based on the general multiple-compartment model, which uses a system of linear equations involving geometrical volume considerations. Flow parameters were estimated by employing two different least-squares procedures: Derivative and Gauss-Marquardt methods, with the available experimental data of radionuclide concentrations as the input functions of time. The solution of the inverse problem, which relates a given set of flow parameter with the time evolution of concentration functions, is achieved via a Monte Carlo Simulation procedure.Program summaryTitle of program: STATFLUXCatalogue identifier: ADYS_v1_0Program summary URL: http://cpc.cs.qub.ac.uk/summaries/ADYS_v1_0Program obtainable from: CPC Program Library, Queen's University of Belfast, N. IrelandLicensing provisions: noneComputer for which the program is designed and others on which it has been tested: Micro-computer with Intel Pentium III, 3.0 GHzInstallation: Laboratory of Linear Accelerator, Department of Experimental Physics, University of São Paulo, BrazilOperating system: Windows 2000 and Windows XPProgramming language used: Fortran-77 as implemented in Microsoft Fortran 4.0. NOTE: Microsoft Fortran includes non-standard features which are used in this program. Standard Fortran compilers such as, g77, f77, ifort and NAG95, are not able to compile the code and therefore it has not been possible for the CPC Program Library to test the program.Memory, required to execute with typical data: 8 Mbytes of RAM memory and 100 MB of Hard disk memoryNo. of bits in a word: 16No. of lines in distributed program, including test data, etc.: 6912No. of bytes in distributed Program, including test data, etc.: 229 541Distribution format: tar.gzNature of the physical problem: the investigation of transport mechanisms for radioactive substances, through environmental pathways, is very important for radiological protection of populations. One such pathway, associated with the food chain, is the grass-animal-man sequence. The distribution of trace elements in humans and laboratory animals has been intensively studied over the past 60 years [R.C. Pendlenton, C.W. Mays, R.D. Lloyd, A.L. Brooks, Differential accumulation of iodine-131 from local fallout in people and milk, Health Phys. 9 (1963) 1253-1262]. In addition, investigations on the incidence of cancer in humans, and a possible causal relationship to radioactive fallout, have been undertaken [E.S. Weiss, M.L. Rallison, W.T. London, W.T. Carlyle Thompson, Thyroid nodularity in southwestern Utah school children exposed to fallout radiation, Amer. J. Public Health 61 (1971) 241-249; M.L. Rallison, B.M. Dobyns, F.R. Keating, J.E. Rall, F.H. Tyler, Thyroid diseases in children, Amer. J. Med. 56 (1974) 457-463; J.L. Lyon, M.R. Klauber, J.W. Gardner, K.S. Udall, Childhood leukemia associated with fallout from nuclear testing, N. Engl. J. Med. 300 (1979) 397-402]. From the pathways of entry of radionuclides in the human (or animal) body, ingestion is the most important because it is closely related to life-long alimentary (or dietary) habits. Those radionuclides which are able to enter the living cells by either metabolic or other processes give rise to localized doses which can be very high. The evaluation of these internally localized doses is of paramount importance for the assessment of radiobiological risks and radiological protection. The time behavior of trace concentration in organs is the principal input for prediction of internal doses after acute or chronic exposure. The General Multiple-Compartment Model (GMCM) is the powerful and more accepted method for biokinetical studies, which allows the calculation of concentration of trace elements in organs as a function of time, when the flow parameters of the model are known. However, few biokinetics data exist in the literature, and the determination of flow and transfer parameters by statistical fitting for each system is an open problem.Restriction on the complexity of the problem: This version of the code works with the constant volume approximation, which is valid for many situations where the biological half-live of a trace is lower than the volume rise time. Another restriction is related to the central flux model. The model considered in the code assumes that exist one central compartment (e.g., blood), that connect the flow with all compartments, and the flow between other compartments is not included.Typical running time: Depends on the choice for calculations. Using the Derivative Method the time is very short (a few minutes) for any number of compartments considered. When the Gauss-Marquardt iterative method is used the calculation time can be approximately 5-6 hours when similar to 15 compartments are considered. (C) 2006 Elsevier B.V. All rights reserved.

Estudo de solvatocromismo em líquidos moleculares orgânicos via método seqüencial Monte Carlo/Mecânica Quântica

Utilizou-se o método seqüencial Monte Carlo / Mecânica Quântica para obterem-se os desvios de solvatocromismo e os momentos de dipolo dos sistemas de moléculas orgânicas: Uracil em meio aquoso, -Caroteno em Ácido Oléico, Ácido Ricinoléico em metanol e em Etanol e Ácido Oléico em metanol e em Etanol. As otimizações das geometrias e as distribuições de cargas foram obtidas através da Teoria do Funcional Densidade com o funcional B3LYP e os conjuntos de funções de base 6-31G(d) para todas as moléculas exceto para a água e Uracil, as quais, foram utilizadas o conjunto de funções de base 6-311++G(d,p). No tratamento clássico, Monte Carlo, aplicou-se o algoritmo Metropólis através do programa DICE. A separação de configurações estatisticamente relevantes para os cálculos das propriedades médias foi implementada com a utilização da função de auto-correlação calculada para cada sistema. A função de distribuição radial dos líquidos moleculares foi utilizada para a separação da primeira camada de solvatação, a qual, estabelece a principal interação entre soluto-solvente. As configurações relevantes da primeira camada de solvatação de cada sistema foram submetidas a cálculos quânticos a nível semi-empírico com o método ZINDO/S-CI. Os espectros de absorção foram obtidos para os solutos em fase gasosa e para os sistemas de líquidos moleculares comentados. Os momentos de dipolo elétrico dos mesmos também foram obtidos. Todas as bandas dos espectros de absorção dos sistemas tiveram um desvio para o azul, exceto a segunda banda do sistema de Beta-Caroteno em Ácido Oléico que apresentou um desvio para o vermelho. Os resultados encontrados apresentam-se em excelente concordância com os valores experimentais encontrados na literatura. Todos os sistemas tiveram aumento no momento de dipolo elétrico devido às moléculas dos solventes serem moléculas polares. Os sistemas de ácidos graxos em álcoois apresentaram resultados muito semelhantes, ou seja, os ácidos graxos mencionados possuem comportamentos espectroscópicos semelhantes submetidos aos mesmos solventes. As simulações através do método seqüencial Monte Carlo / Mecânica Quântica estudadas demonstraram que a metodologia é eficaz para a obtenção das propriedades espectroscópicas dos líquidos moleculares analisados.

Investigação da dinâmica sequêncial Monte Carlo/Mecânica Quântica para sitemas moleculares orgânicos

O principal objetivo deste trabalho é investigar teóricamente as propriedades eletrônicas e óticas de compostos orgânicos em ambiente líquido. A compreensão das interações em sistemas líquidos é muito importante para a descrição de fenômenos em muitas partes da ciência, como Física, Química, Biologia e Materiais, bem como no desenvolvimento de materiais óticos. As flutuações térmicas fazem que inumaram das configurações para os líquidos possam ser gerados. Esta é a razão do comportamento estatístico observado em sistemas líquidos. Considerando estas dificuldades, o tratamento sequencial Monte Carlo / Mecânica Quântica (SMC/QM) é usado neste trabalho. Neste procedimento, a estrutura líquida é gerada primeiramente por simulações clássicas de MC e mais tarde, somente a parte mais importante do sistema é tratada com mecânica quântica. Usando o procedimento acima, os propriedades do quantum o pirazine dimethyl do thiene Methyl da laranja (MO) e do O 2,3-dimetil tieno[3,4-b] pirazina (DTP) foram investigados. O MO é um conhecido indicador de pH e pode ser encontrado sob circunstâncias básicas e acidas. Suas geometrias de mínima energia foram obtidas mediante a Teoria do Funcional da Densidade pelo funcional B3LYP, sendo o sistema descrito pelas bases de Pople com uma função de polarização (6-31G*). Para obter as propriedades médias dos observaveis, cálculos de química quântica foram executados dentro da aproximação semi-empírica de INDO/S-CI. Com respeito ao espectro de absorção, os dados experimentais existentes na literatura científica reportam a existência de uma larga banda localizada na região de baixas energias, mais precisamente entre 400 e 600 nm. Nossos resultados teóricos para a forma alcalina mostram uma transição intensa transição π → π* aproximadamente à 432.4 ± 0.03 nm e, sob condições ácidas, esta transição aparece aproximadamente à 507.4 ± 0.12, 496.4 ± 0.28 ou 545.3 ± 0.10 nm, dependendo da estrutura, mostrando bom acordo com resultados experimentais. O DTP é um sistema particular usado na produção de polímeros de baixo gap. Suas propriedades elétricas e óticas foram obtidas através de um novo procedimento conhecido por Configuração Eletrostática Média do Solvent (ASEC). O procedimento ASEC inclui moléculas do solvente como cargas pontuais e permite o obtenção das quantidades quânticas executando somente poucos cálculos de mecânica quântica. Para o DTP, usando a teoria das perturbações de segunda ordem Mφller-Plesset (MP2) e o conjunto de bases aug-cc-pVDZ, a convergência do momento de dipolo foi alcançada com apenas quatro cálculos de mecânica quântica à 1.16 D, apresentando um aumento de 42% quando comparado ao dipolo isolado. O polarizabilidade corresponde à outra característica elétrica que pôde ser medida. Considerando o mesmo nível empregado ao cálculo do dipolo, o valor médio 132.7 a30 foi observado. A região de mais baixas da energias do espectro de absorption foi investigada também atravé de procedimento de ASEC usando ambos as aproximações, semi-empírico e DFT. Esta região de absorção é motivo de conclusões contraditórias com relação à natureza das transições n → π* e π → π*. Nossos resultados mostram que realmente que essas excitações são realmente observadas simultanemente podendo sobrepôr-se. Como exemplo, nossos resultados para DFT, encontrados usando o funcional B3LYP nos mostra que estas transições aparecem aproximadamente à 360.6 e 351.1 nm.

Uso do método de Monte Carlo para simulação e detecção do espectro de raios-X gerado com feixes de elétrons em alvo de Molibdênio, utilizando o Geant4

The huge demand for procedures involving ionizing radiation promotes the need for safe methods of experimentation considering the danger of their biological e ects with consequent risk to humans. Brazilian's legislation prohibits experiments involving this type of radiation in humans through Decree 453 of Ministry of Health with determines that such procedures comply with the principles of justi cation, optimization and dose limitation. In this line, concurrently with the advancement of available computer processing power, computing simulations have become relevant in those situations where experimental procedures are too cost or impractical. The Monte Carlo method, created along the Manhattan Project duringWorldWar II, is a powerful strategy to simulations in computational physics. In medical physics, this technique has been extensively used with applications in diagnostics and cancer treatment. The objective of this work is to simulate the production and detection of X-rays for the energy range of diagnostic radiology, for molybdenum target, using the Geant4 toolkit. X-ray tubes with this kind of target material are used in diagnostic radiology, speci cally in mammography, one of the most used techniques for screening of breast cancer in women. During the simulations, we used di erent models for bremsstrahlung available in physical models for low energy, in situations already covered by the literature in earlier versions of Geant4. Our results show that although the physical situations seems qualitatively adequate, quantitative comparisons to available analytical data shows aws in the code of Geant4 Low Energy source

Simulação de Monte Carlo: de modelos de spin à teoria de campos na rede

Pós-graduação em Física - IFT

Quantum Monte Carlo study of small aluminum clusters Al-n (n=2-13)

Using fixed node diffusion quantum Monte Carlo (FN-DMC) simulations and density functional theory (DFT) within the generalized gradient approximations, we calculate the total energies of the relaxed and unrelaxed neutral, cationic, and anionic aluminum clusters, Al-n (n = 1-13). From the obtained total energies, we extract the ionization potential and electron detachment energy and compare with previous theoretical and experimental results. Our results for the electronic properties from both the FN-DMC and DFT calculations are in reasonably good agreement with the available experimental data. A comparison between the FN-DMC and DFT results reveals that their differences are a few tenths of electron volt for both the ionization potential and the electron detachment energy. We also observe two distinct behaviors in the electron correlation contribution to the total energies from smaller to larger clusters, which could be assigned to the structural transition of the clusters from planar to three-dimensional occurring at n = 4 to 5.

Monte Carlo Semi-Empirical Model for Si(Li) X-Ray Detector: Differences between Nominal and Fitted Parameters.

A detailed characterization of a X-ray Si(Li) detector was performed to obtain the energy dependence of efficiency in the photon energy range of 6.4 - 59.5 keV. which was measured and reproduced by Monte Carlo (MC) simulations. Significant discrepancies between MC and experimental values were found when lhe manufacturer parameters of lhe detector were used in lhe simulation. A complete Computerized Tomagraphy (CT) detector scan allowed to find the correct crystal dimensions and position inside the capsule. The computed efficiencies with the resulting detector model differed with the measured values no more than 10% in most of the energy range.

Monte Carlo-Simulationen zum kritischen Verhalten dünner Ising-Filme

Monte Carlo-Simulationen zum kritischen Verhalten dünnerIsing-Filme Dünne Ising-Filme können als vereinfachtes Modell zurBeschreibung von binären Mischungen oder von Flüssigkeitenin schlitzartigen Kapillaren dienen. Infolge dereingeschränkten Geometrie unterscheidet sich das kritischeVerhalten dieser Systeme signifikant von dem einesBulk-Systems, es kommt zu einem Crossover von zwei- zudreidimensionalem kritischen Verhalten. Zusätzlichverschiebt sich der Phasenübergang in den ungesättigtenBereich, ein Effekt, der als 'capillary condensation'bezeichnet wird. In der vorliegenden Arbeit wurden die kritischenEigenschaften von Ising-Filmen im Rahmen einer MonteCarlo-Simulation untersucht. Zur Verbesserung der Effizienzwurde ein Cluster-Algorithmus verwendet, der um einenGhost-Spin-Term zur Behandlung der Magnetfelder erweitertwar. Bei der Datenanalyse kamen moderneMulti-Histogramm-Techniken zur Anwendung. Für alle untersuchten Schichtdicken konnten kritischeTemperatur und Magnetfeld sehr präzise bestimmt werden. DieSkalenhypothese von Fisher und Nakanishi, die dieVerschiebung des kritischen Punktes gegenüber seinesBulk-Wertes beschreibt, wurde sowohl für Systeme mit freienOberflächen als auch für Systeme mit schwachemOberflächenfeld bestätigt. Der Wert des Gap-Exponenten derOberfläche wurde mit $Delta_1$=0.459(13) in Übereinstimmungmit den Literaturwerten abgeschätzt. Die Observablen Magnetisierung und magnetischeSuszeptibilität sowie deren auf die Oberfläche bezogenenEntsprechungen zeigen kein reines zweidimensionaleskritisches Verhalten. Zu ihrer Beschreibung in der Nähe deskritischen Punktes wurden effektive Exponenten für dieeinzelnen Schichtdicken bestimmt.

Oberflächengebundene flüssigkristalline Polymere in nematischer Lösung - eine Monte Carlo-Untersuchung

Oberflächengebundene flüssigkristalline Polymere innematischer Lösung - eine Monte Carlo-Untersuchung In der vorliegenden Arbeit wurde ein System vonoberflächengebundenen flüssigkristallinen Polymeren innematischer Lösung mittels Monte Carlo-Simulationen imNpT-Ensemble untersucht. Dabei wurden die Lösungsmittelteilchen und Polymerteilchen durchweiche abstoßende Ellipsoide des Durchmessers sigma_s undder Länge sigma_l modelliert. Darüber hinaus wirktenzwischen Teilchen innerhalb einer Kette zusätzliche Bondlängen- und Bondwinkelpotentiale. Die Kopfenden der Ketten waren mit ihren Mittelpunkten auf einen kubischen Gitter in der Oberfläche fixiert und wechselwirkten nicht mit den übrigen Teilchen. Für die Simulation kam ein neuartiger 'Configurationalbiased Monte Carlo'-Algorithmus zum Einsatz, mit dem esmöglich war, die Simulation stark zu beschleunigen. Dieswurde erreicht, indem die Bindungen einer Polymerkette entfernt und die ehemaligen Kettenteilchen somit zu Lösungsmittelteilchen werden. Aus den Lösungsmittelteilchen wurde dann durch das Ziehen neuer Bindungen eine neue Kette generiert, über deren Annahme dann eine Metropolisabfrage entschied. In den Simulationen sollte untersucht werden, inwieweit esmöglich ist durch Variation der Pfropfdichte einer auf eineglatte Wand aufgebrachten Polymerbürste die Richtung derOrientierung des sie quellenden Lösungsmittels zu einflussen. Neben Systemen ohne Ketten wurden Systeme mit Pfropfdichtenbis zu sigma = 0.84/sigma_s^2 über mehrere Millionen MonteCarlo-Schritte simuliert. Mit den durchgeführten Simulationen konnte gezeigt werden,daß ein Phasenübergang zwischen einer Phase mit flachentlang den Wänden ausgerichteten Teilchen und einer Phasemit Teilchen, deren Orientierung nicht parallel zur Wand ist, existiert.Dieser Phasenübergang tritt im hier betrachteten System beieiner Pfropfdichte von sigma = 0.13/sigma_s^2 auf. Weiterhin konnte bei hohen Pfropfdichten (sigma >=0.8/sigma_s^2) eine Phase mit einer Ausrichtung derLösungsmittelteilchen senkrecht zur Grundfläche gefundenwerden, obwohl die Ketten auch bei diesen Pfropfdichten nicht senkrecht stehen und die Bürste und das Lösungsmittel entkoppeln.

Monte Carlo-Simulationen zum Adsorptionsverhalten von Homo- und Copolymerlösungen

Die vorliegende Arbeit beschäftigt sich mit demAdsorptionsverhalten von Homo- und Blockopolymeren anglatten Wänden im guten Lösungsmittel. Zur Modellierung wirddabei ein vergröbertes Kugel-Feder-Modell benutzt, das diePolymere als Kette von effektiven Monomeren auffasst.Benachbarte Monomere einer Kette werden durch einBindungspotential aneinander gekoppelt; zusätzlich wird eineVolumenausschlusswechselwirkung zwischen allen Monomeren derLösung berücksichtigt. Das Lösungsmittel wird dabei implizitdurch diese effektiven Potentiale berücksichtigt, wobei dieLösungsmittelqualität über die Temperatur geregelt wird.Während das System in z-Richtung durch zwei ebene Wändebegrenzt wird, von denen eine attraktiv mit den Monomerenwechselwirkt, werden in x- und y-Richtung periodischeRandbedingungen vorgegeben. Die Bewegung der Polymereerfolgt durch Monte Carlo-Simulationen im räumlichenKontinuum, welche auf (großkanonischen) Configurational BiasMonte Carlo-Algorithmen sowie lokalen Monomerverschiebungenund Reptationsbewegungen basieren.Über verschiedene Auswertemethoden werden anhand vonEinzelkettensimulationen die (modellabhängige)Adsorptionsschwelle - welche durch Simualtionen derAdsorption aus der Lösung bestätigt wird - sowie der(universelle) Crossoverexponent - dessen gefundener Wertsehr gut mit den neusten analytischen Berechnungenübereinstimmt - bestimmt. Bei der Simulation derBlockcopolymere bilden sich bei niedrigerOberflächenbelegungsdichte pilzartige Konformationen('mushroom'-Bereich), bei hoher OberflächenbelegungsdichtePolymerbürsten ('brush'-Bereich). Die theoretischenVorhersagen für den jeweiligen Bereich werden von denSimulationen im Wesentlichen gut wiedergegeben.

Monte Carlo simulations of Potts glasses

A complete understanding of the glass transition isstill a challenging problem. Some researchers attributeit to the (hypothetical) occurrence of a static phasetransition, others emphasize the dynamical transitionof mode coupling-theory from an ergodic to a non ergodicstate. A class of disordered spin models has been foundwhich unifies both scenarios. One of these models isthe p-state infinite range Potts glass with p>4, whichexhibits in the thermodynamic limit both a dynamicalphase transition at a temperature T_D, and a static oneat T_0 < T_D. In this model every spins interacts withall the others, irrespective of distance. Interactionsare taken from a Gaussian distribution.In order to understand better its behavior forfinite number N of spins and the approach to thethermodynamic limit, we have performed extensive MonteCarlo simulations of the p=10 Potts glass up to N=2560.The time-dependent spin-autocorrelation function C(t)shows strong finite size effects and it does not showa plateau even for temperatures around the dynamicalcritical temperature T_D. We show that the N-andT-dependence of the relaxation time for T > T_D can beunderstood by means of a dynamical finite size scalingAnsatz.The behavior in the spin glass phase down to atemperature T=0.7 (about 60% of the transitiontemperature) is studied. Well equilibratedconfigurations are obtained with the paralleltempering method, which is also useful for properlyestablishing static properties, such as the orderparameter distribution function P(q). Evidence is givenfor the compatibility with a one step replica symmetrybreaking scenario. The study of the cumulants of theorder parameter does not permit a reliable estimation ofthe static transition temperature. The autocorrelationfunction at low T exhibits a two-step decay, and ascaling behavior typical of supercooled liquids, thetime-temperature superposition principle, is observed. Inthis region the dynamics is governed by Arrheniusrelaxations, with barriers growing like N^{1/2}.We analyzed the single spin dynamics down to temperaturesmuch lower than the dynamical transition temperature. We found strong dynamical heterogeneities, which explainthe non-exponential character of the spin autocorrelationfunction. The spins seem to relax according to dynamicalclusters. The model in three dimensions tends to acquireferromagnetic order for equal concentration of ferro-and antiferromagnetic bonds. The ordering has differentcharacteristics from the pure ferromagnet. The spinglass susceptibility behaves like chi_{SG} proportionalto 1/T in the region where a spin glass is predicted toexist in mean-field. Also the analysis of the cumulantsis consistent with the absence of spin glass orderingat finite temperature. The dynamics shows multi-scalerelaxations if a bimodal distribution of bonds isused. We propose to understand it with a model based onthe local spin configuration. This is consistent with theabsence of plateaus if Gaussian interactions are used.