Selecting hidden Markov model state number with cross-validated likelihood

The problem of estimating the number of hidden states in a hidden Markov model is considered. Emphasis is placed on cross-validated likelihood criteria. Using cross-validation to assess the number of hidden states allows to circumvent the well-documented technical difficulties of the order identification problem in mixture models. Moreover, in a predictive perspective, it does not require that the sampling distribution belongs to one of the models in competition. However, computing cross-validated likelihood for hidden Markov models for which only one training sample is available, involves difficulties since the data are not independent. Two approaches are proposed to compute cross-validated likelihood for a hidden Markov model. The first one consists of using a deterministic half-sampling procedure, and the second one consists of an adaptation of the EM algorithm for hidden Markov models, to take into account randomly missing values induced by cross-validation. Numerical experiments on both simulated and real data sets compare different versions of cross-validated likelihood criterion and penalised likelihood criteria, including BIC and a penalised marginal likelihood criterion. Those numerical experiments highlight a promising behaviour of the deterministic half-sampling criterion.     

Hydrogenase enzymes: Recent structural studies and active site models

Recent developments in active site structure determination of the three types of hydrogenase enzymes are described. Aspects of recent studies using model complexes relevant to the structure and function of the enzymes are reviewed.     

非合作n人投标报价服从N(μ,σ~2)分布时局中人投标报价策略的研究

推导了矩阵对策模型 ,对非合作 n人投标报价的分布进行 x2 检验。当非合作 n人投标报价服从N (μ,σ2 )分布时 ,根据参数区间公式 ,估计非合作 n人有效报价平均数的范围 ,局中人 最佳报价可根据矩阵对策模型来确定。该方法确定的报价与最佳标底的误差能控制在较小的范围内 ,适用于招标中合成标底的评标办法 ,对确定报价具有一定的实用价值。     

Study of adsorption processes of model drugs at supercritical conditions using partial least squares regression

The characterisation of adsorption or impregnation processes using conventional or supercritical fluid technologies becomes an increasing part of the research on drug formulations. The complexity of the relationships between these adsorption processes and the experimental variables potentially influencing them, however, makes these studies more problematic. In this paper, a chemometric approach based on nonlinear partial least squares (NL-PLS) modelling is applied to characterise the effect of the experimental variables on the supercritical impregnation process. Various adsorbent materials such as silica gel, zeolite and amberlite were investigated using the following model compounds as adsorbates: benzoic, salicylic and acetylsalicylic acids.     

有机化学品的生物降解性及构效关系

介绍了描述生物降解性的主要测定参数，测定过程的影响因素及常用的结构描述符，并结合实例综述了目前应用比较广泛的定量结构－生物降解性关系（QSBR）技术，最后，分析了QSBR工作存在的主要问题及研究前景。     

Facetting and (n × 1) reconstructions of SrTiO3(1 1 0) surfaces

(n × 1) reconstructions and facetting of the (1 1 0) polar surface of SrTiO₃ are studied by means of a combination of shell model and density functional calculations. The polarity compensation can be achieved through the formation of {1 0 0} nano-facets, which play a crucial role in the reconstruction process. The behaviors of various possible terminations (Sr, Ti, and O) are analyzed, as well as their atomic structure and energetics. Their stability in different chemical environments is discussed, with respect to previous formulations and experimental results. The Sr-terminated surface tends to expose large facets, while the TiO and O terminations are marginally stabilized or even destabilized by (n × 1) reconstructions, respectively. Trend to facetting results from a subtle competition between the thermodynamic stability of the ideal non stoichiometric (n × 1) surfaces, and huge atomic relaxations that contribute to the lowering of the surface energy differently for each termination.     

Nucleation of CO2-hydrate in a porous medium

Experiments designed to crystallize gas hydrate from dissolved CO₂ in natural porous media are used to study nucleation under varying thermodynamic conditions. We recover quantitative information from these experiments using a stochastic model for the nucleation process. Estimates of the model parameters are used to determine the average time for nucleation as a function of temperature and composition.     

On regression model selection for the data with correlated errors

A class of regression model selection criteria for the data with correlated errors is proposed. The proposed class of selection criteria is an estimator of weighted prediction risk. In addition, the proposed selection criteria are the generalizations of several commonly used criteria in statistical analysis. The theoretical and asymptotic properties for the class of criteria are established. Further, in the medium-sample case, the results based on a simulation study are quite consistent with the theoretical ones. The proposed criteria perform well in the simulations. Several applications are also given for a variety of statistical models.     

Model checking with Boolean Satisfiability

The evolution of SAT algorithms over the last decade has motivated the application of SAT to model checking, initially through the use of SAT in bounded model checking and, more recently, in unbounded model checking. This paper provides an overview of modern SAT algorithms, SAT-based bounded model checking and some of the most promising approaches for unbounded model checking, namely induction and interpolation. Moreover, the paper details a number of techniques that have proven effective in using SAT solvers in model checking.     

Decomposition approach to model smart suspension struts

Model and simulation study is the starting point for engineering design and development, especially for developing vehicle control systems. This paper presents a methodology to build models for application of smart struts for vehicle suspension control development. The modeling approach is based on decomposition of the testing data. Per the strut functions, the data is dissected according to both control and physical variables. Then the data sets are characterized to represent different aspects of the strut working behaviors. Next different mathematical equations can be built and optimized to best fit the corresponding data sets, respectively. In this way, the model optimization can be facilitated in comparison to a traditional approach to find out a global optimum set of model parameters for a complicated nonlinear model from a series of testing data. Finally, two struts are introduced as examples for this modeling study: magneto-rheological (MR) dampers and compressible fluid (CF) based struts. The model validation shows that this methodology can truly capture macro-behaviors of these struts.