Bayesian graphical model determination using decision theory

Bayesian model determination in the complete class of graphical models is considered using a decision theoretic framework within the regular exponential family. The complete class contains both decomposable and non-decomposable graphical models. A utility measure based on a logarithmic score function is introduced under reference priors for the model parameters. The logarithmic utility of a model is decomposed into predictive performance and relative complexity. Axioms of decision theory lead to the judgement of the plausibility of a model in terms of the posterior expected utility. This quantity has an analytic expression for decomposable models when certain reference priors are used and the exponential family is closed under marginalization. For non-decomposable models, a simulation consistent estimate of the expectation can be obtained. Both real and simulated data sets are used to illustrate the introduced methodology.     

Sufficiently general framework for simple models of the net immune response

A general framework for modelling the immune response is proposed and analyzed. The models have simple formulation in terms of only few measurable quantities and the complexity of the immune response is accounted for by using nonlinear functions with time-delayed arguments. It is illustrated that such models are capable of describing various typical cases of the immune response. Standard analytical tools are used to show how such models can assist in the problem of parameter estimation.     

Bayesian Multivariate Distributional Regression With Skewed Responses and Skewed Random Effects

The normal and the t distribution are classical tools for building random effects regression models where both can be used for the specification of either the conditional response distribution or the random effects distribution. However, the underlying assumption of symmetry can be questionable in many applications. We, therefore, propose regression models where the skew-normal and skew-t distribution are considered for both the response and the random effects specification and embed these models in the framework of distributional regression such that regression predictors can be specified for all distributional parameters. The distributional regression framework also allows us to consider multivariate versions of the skew-normal and the skew-t distribution. For Bayesian inference, we adapt iteratively weighted least-square proposals within Markov chain Monte Carlo simulations such that they can also facilitate the inclusion of nonnormal random effects specifications. Model choice is based on the Watanabe–Akaike information criterion, in particular, to differentiate between skew and nonskew distributional specifications in a number of simulation studies. Finally, to illustrate their practical applicability, the developed models are applied to a study on cholesterol levels originating from the Framingham Heart Study and a dataset from the Demographic and Health Surveys on undernutrition among children in Nigeria. Supplementary material for this article is available online.     

半参数随机效应模型的异方差检验

在许多实际问题中,检验观察数据是否出现异方差性是一个相当感兴趣的问题.该文研究了半参数随机效应模型的异方差检验问题.基于Lin(1997)的方法,得到了检验方差成分都为零的Score检验统计量.通过随机模拟和实际数值例子,论证了方法的有效性.利用现有的统计软件,容易实现该文所提出的检验方法.     

Modeling,simulation and adaptive process control of biotechnological processes in decentralized anaerobic treatment of domestic wastewater

Control in anaerobic wastewater treatment plants is difficult to achieve but necessary due to a high sensitivity to disturbances and process complexity. With the help of different mathematical tools, control strategies can be developed. Particularly, a well-defined mathematical model can be highly effective for design, assessment and optimization of treatment plants. However, applications directly in the control system of a treatment plant are hard to achieve due to model complexity and usually require specialized software and the engagement of experts in the subject. The objective of the present study was the development of less empirical methods for assessment and control of a decentralized anaerobic plant for the treatment of domestic wastewater. A lab-scale plant, which consisted of a two-stage anaerobic digestion process followed by an anaerobic ammonium oxidation (ANAMMOX) reactor for nitrogen removal, was used as object of study. Ordinary differential equation models were implemented to simulate the processes that took place in the treatment plant. With the help of the implemented models, control tools were developed. These tools include a standalone application for monitoring of the two-stage anaerobic digestion process and an ammonium estimator for the ANAMMOX reactor by means of artificial neural networks (ANNs). The procedures followed aimed to reduce the amount of experimental work required so they can be easily transferred from laboratory to full-scale conditions. (© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Pricing range notes within Wishart affine models

We provide analytic pricing formulas for Fixed and Floating Range Accrual Notes within the multifactor Wishart affine framework which extends significantly the standard affine model. Using estimates for three short rate models, two of which are based on the Wishart process whilst the third one belongs to the standard affine framework, we price these structured products using the FFT methodology. Thanks to the Wishart tractability the hedge ratios are also easily computed. As the models are estimated on the same dataset, our results illustrate how the fit discrepancies (meaning differences in the likelihood functions) between models translate in terms of derivatives pricing errors, and we show that the models can produce different price evolutions for the Range Accrual Notes. The differences can be substantial and underline the importance of model risk both from a static and a dynamic perspective. These results are confirmed by an analysis performed at the hedge ratios level.     

Complex fluid simulations with the parallel tree-based Lattice Boltzmann solver Musubi

We present the open source Lattice Boltzmann solver Musubi. It is part of the parallel simulation framework APES, which utilizes octrees to represent sparse meshes and provides tools from automatic mesh generation to post-processing. The octree mesh representation enables the handling of arbitrarily complex simulation domains, even on massively parallel systems. Local grid refinement is implemented by several interpolation schemes in Musubi. Various kernels provide different physical models based on stream-collide algorithms. These models can be computed concurrently and can be coupled with each other. This paper explains our approach to provide a flexible yet scalable simulation environment and elaborates its design principles and implementation details. The efficiency of our approach is demonstrated with a performance evaluation on two supercomputers and a comparison to the widely used Lattice Boltzmann solver Palabos.     

Architecture for development of adaptive on-line prediction models

This work presents an architecture for the development of on-line prediction models. The architecture defines unified modular environment based on three concepts from machine learning, these are: (i) ensemble methods, (ii) local learning, and (iii) meta learning. The three concepts are organised in a three layer hierarchy within the architecture. For the actual prediction making any data-driven predictive method such as artificial neural network, support vector machines, etc. can be implemented and plugged in. In addition to the predictive methods, data pre-processing methods can also be implemented as plug-ins. Models developed according to the architecture can be trained and operated in different modes. With regard to the training, the architecture supports the building of initial models based on a batch of training data, but if this data is not available the models can also be trained in incremental mode. In a scenario where correct target values are (occasionally) available during the run-time, the architecture supports life-long learning by providing several adaptation mechanisms across the three hierarchical levels. In order to demonstrate its practicality, we show how the issues of current soft sensor development and maintenance can be effectively dealt with by using the architecture as a construction plan for the development of adaptive soft sensing algorithms.     

Parallel simulation of individual-based, physiologically structured population models

A general scheme for parallel simulation of individual-based, structured population models is proposed. Algorithms are developed to simulate such models in a parallel computing environment. The simulation model consists of an individual model and a population model that incorporates the individual dynamics. The individual model is a continuous time representation of organism life history for growth with discrete allocations for reproductive processes. The population model is a continuous time simulation of a nonlinear partial differential equation of extended McKendrick-von Foerster-type.

As a prototypical example, we show that a specific individual-based, physiologically structured model for Daphnia populations is well suited for parallelization, and significant speed-ups can be obtained by using efficient algorithms developed along our general scheme. Because the parallel algorithms are applicable to generic structured populations which are the foundation for populations in a more complex community or food-web model, parallel computation appears to be a valuable tool for ecological modeling and simulation.     

Bayesian modeling of forestry data with R using optimization and simulation tools

Data generated in forestry biometrics are not normal in statistical sense as they rarely follow the normal regression model. Hence, there is a need to develop models and methods in forest biometric applications for non-normal models. Due to generality of Bayesian methods it can be implemented in the situations when Gaussian regression models do not fit the data. Data on diameter at breast height (dbh), which is a very important characteristic in forestry has been fitted to Weibull and gamma models in Bayesian paradigm and comparisons have also been made with its classical counterpart. It may be noted that MCMC simulation tools are used in this study. An attempt has been made to apply Bayesian simulation tools using \textbf{R} software.     

Air path identification of diesel engines by LPV techniques for gain scheduled control

Modelling is a key element to improve the performance of engine control systems, but many factors like non-linearity and complexity complicate the derivation of sufficiently precise physical models. This motivates an increasing interest in data based models. Linear models can successfully represent the engine operation in some reduced regions, but tend to fail when large operating regions must be considered. This motivates the interest in deriving and using gain scheduling models or their natural extension, the linear parameter varying (LPV) models. In this article we propose to model the air path of diesel engines using input–output LPV models with a physically motivated structure and parameters estimated from data. These models are shown to combine good precision with simplicity and allow the systematic design of optimal and robust control systems, and can be determined in a very short time if sufficient data are available.     

A participative and facilitative conceptual modelling framework for discrete event simulation studies in healthcare

Existing approaches to conceptual modelling (CM) in discrete-event simulation do not formally support the participation of a group of stakeholders. Simulation in healthcare can benefit from stakeholder participation as it makes possible to share multiple views and tacit knowledge from different parts of the system. We put forward a framework tailored to healthcare that supports the interaction of simulation modellers with a group of stakeholders to arrive at a common conceptual model. The framework incorporates two facilitated workshops. It consists of a package including: three key stages and sub-stages; activities and guidance; tools and prescribed outputs. The CM framework is tested in a real case study of an obesity system. The benefits of using this framework in healthcare studies and more widely in simulation are discussed. The paper also considers how the framework meets the CM requirements.     

Symbolic models for infinite networks of control systems: A compositional approach

This paper presents a compositional framework for the construction of symbolic models for a network composed of a countably infinite number of finite-dimensional discrete-time control subsystems. We refer to such a network as infinite network. The proposed approach is based on the notion of alternating simulation functions. This notion relates a concrete network to its symbolic model with guaranteed mismatch bounds between their output behaviors. We propose a compositional approach to construct a symbolic model for an infinite network, together with an alternating simulation function, by composing symbolic models and alternating simulation functions constructed for subsystems. Assuming that each subsystem is incrementally input-to-state stable and under some small-gain type conditions, we present an algorithm for orderly constructing local symbolic models with properly designed quantization parameters. In this way, the proposed compositional approach can provide us a guideline for constructing an overall symbolic model with any desired approximation accuracy. A compositional controller synthesis scheme is also provided to enforce safety properties on the infinite network in a decentralized fashion. The effectiveness of our result is illustrated through a road traffic network consisting of infinitely many road cells.     

Return and Value at Risk using the Dirichlet Process

There exists a wide variety of models for return, and the chosen model determines the tool required to calculate the value at risk (VaR). This paper introduces an alternative methodology to model‐based simulation by using a Monte Carlo simulation of the Dirichlet process. The model is constructed in a Bayesian framework, using properties initially described by Ferguson. A notable advantage of this model is that, on average, the random draws are sampled from a mixed distribution that consists of a prior guess by an expert and the empirical process based on a random sample of historical asset returns. The method is relatively automatic and similar to machine learning tools, e.g. the estimate is updated as new data arrive.     

非线性贝叶斯动态模型的重要性再抽样(英文)

本文研究了非线性贝叶斯动态模型的随机模拟.在更宽泛的先验分布假设下.利用重要性再抽样的方法,以"样本"代替"分布",实现了对模型的后验推断、预测和模型选择,扩张了贝叶斯动态模型的应用领域.     

System dynamic modelling and simulation of hydrodynamic machines

A graph-theoretic framework for the dynamic simulation of hydrodynamic (both axial and radial flow) machines is presented in this article. The physics based analytical models are developed by considering the dynamics of the hydraulic fluid flow and its interaction with the mechanical components. A linear graph is used to capture the topology of the system and the interconnection of the constituent components. Using the graph-theoretic framework, a dynamic model of an automotive hydrodynamic torque converter is developed to simulate its behaviour under different flow conditions. The ability of the model to capture different features of the torque converter will also be demonstrated by simulation. The simulation results are compared with and validated by experimental results in the literature.     

Semiqualitative modelling of bioprocesses

This paper deals with semiqualitative modelling of bioprocesses with a view to their supervision. An analysis of several approaches for modelling shows the difficulties involved in taking into account in a same framework, quantitative and qualitative knowledge, generally available about a process that we want to control. We propose an original approach, placed in the context of semiqualitative modelling, that is supported by a knowledge model the variables and parameters of which are defined by intervals. For these semiqualitative models, we study their properties in simulation and prediction, and more precisely, their fitting based on experimental data. We show that pertinent predictions in a short time can be obtained, making of these semiqualitative models interesting tools for the development of systems for bioprocess supervision     

Stochastic actor‐oriented models for network change

A class of models is proposed for longitudinal network data. These models are along the lines of methodological individualism: actors use heuristics to try to achieve their individual goals, subject to constraints. The current network structure is among these constraints. The models are continuous time Markov chain models that can be implemented as simulation models. They incorporate random change in addition to the purposeful change that follows from the actors’ pursuit of their goals, and include parameters that must be estimated from observed data. Statistical methods are proposed for estimating and testing these models. These methods can also be used for parameter estimation for other simulation models. The statistical procedures are based on the method of moments, and use computer simulation to estimate the theoretical moments. The Robbins‐Monro process is used to deal with the stochastic nature of the estimated theoretical moments. An example is given for Newcomb's fraternity data, using a model that expresses reciprocity and balance.     

Multivariate analysis using ‘and’ and ‘or’

The connectives ‘and’ and ‘or’ are potentially useful in multivariate analysis and theory construction. They are simple, logical ways to connect two or more variables together. However, until recently there has been no framework for operationalizing these connectives for continuous variables, and this lack has severely limited their use. Using fuzzy set theory as a basis for such a framework, this paper lays out the necessary tools and models to permit the use of ‘and’ and ‘or’ in multivariate analysis.Section 1 introduces conventional operators for ‘and’ and ‘or’, and Section 2 provides suitable extensions and generalizations of them. Section 3 sets out the required least-squares techniques for fitting these generalized operators to data, first in the context of ANOVA problems and then in regression contexts, for single-connective (three-variable) models. The theoretical developments and examples from real data-sets demonstrate the utility of ‘and’ and ‘or’ as a means to cell-specific interpretations of interaction effects which can also readily be translated into English. Section 4 extends these developments to multivariate, multiple-connective models and discusses issues of generalizability. The paper concludes (Section 5) with a brief discussion of remaining unsolved problems, future prospects for more sophisticated models, and computer programs.