Stably numerical solving inverse boundary value problem for data assimilation

In remote sensing data assimilation, inverse methods are often used to retrieve the boundary conditions from some of the observations in the interior scanning region. Needless to say, the inverse boundary value problem (IBVP) is ill-conditioned in general. Ultimately, the data assimilation must include mechanisms that enable them to overcome the numerical instability for solving IBVP. In this paper, we begin by studying the behavior of ill-conditioned IBVP, and find that the condition number varies with the number of the observations and distribution locations of the observations. Next, we define the number of equivalent independent equations as a novel measurement of ill-conditioning of a problem, which can measure the degree of bad conditions. Furthermore, the novel measure can answer how many additional observations are needed to stabilize the retrieving problem, and where additional observations are fixed up. Finally, we illustrate the proposed methodology by applying it to the study of precipitation data assimilation, with a particular emphasis on the analysis of the effect of the number of observations and their distribution locations. The new methodology appears to be particularly efficient in tackling the instability of the retrieving problem in data assimilation.     

Kernel estimation and interpolation for time series containing missing observations

Summary Kernel estimators of conditional expectations are adapted for use in the analysis of stationary time series containing missing observations. Estimators of conditional expectations at fixed points are shown to have an asymptotic distribution with a relatively simple variance-covariance structure. The kernel method is also used to interpolate missing observations, and is shown to converge in probability to the least squares predictor. The results are established under the strong mixing condition and moment conditions, and the methods are applied to a real data set.     

Bayesian inference in Markovian queues

This paper is concerned with the Bayesian analysis of general queues with Poisson input and exponential service times. Joint posterior distribution of the arrival rate and the individual service rate is obtained from a sample consisting inn observations of the interarrival process andm complete service times. Posterior distribution of traffic intensity inM/M/c is also obtained and the statistical analysis of the ergodic condition from a decision point of view is discussed.     

Positive dependence in Markov chains

A general notion of positive dependence among successive observations in a finite-state stationary process is studied, with particular attention to the case of a stationary ergodic Markov chain. This dependence condition can be expressed as a positivity condition on the joint probability matrices of pairs of observations. Some useful conditions equivalent to positive dependence are obtained for reversible chains, but shown not to be equivalent for nonreversible chains.     

Fitting discrete phase-type distribution from censored and truncated observations with pre-specified hazard sequence

Phase-type distribution allows approximation of non-Markovian models, which permits to analyze complex systems under Markovian deterioration. In addition, reliability data is often composed of truncated and censored observations. This paper presents a novel approach that fits a restricted class of discrete phase-type distribution through pre-specified hazard sequence from incomplete observations. Numerical results are shown using Balakrishnan’s mimicked power transformers dataset. Furthermore, it can be used to fit transition probabilities of maintenance optimization’s Markov decision process models from incomplete reliability data.     

Imprecise probability models for learning multinomial distributions from data. Applications to learning credal networks

This paper considers the problem of learning multinomial distributions from a sample of independent observations. The Bayesian approach usually assumes a prior Dirichlet distribution about the probabilities of the different possible values. However, there is no consensus on the parameters of this Dirichlet distribution. Here, it will be shown that this is not a simple problem, providing examples in which different selection criteria are reasonable. To solve it the Imprecise Dirichlet Model (IDM) was introduced. But this model has important drawbacks, as the problems associated to learning from indirect observations. As an alternative approach, the Imprecise Sample Size Dirichlet Model (ISSDM) is introduced and its properties are studied. The prior distribution over the parameters of a multinomial distribution is the basis to learn Bayesian networks using Bayesian scores. Here, we will show that the ISSDM can be used to learn imprecise Bayesian networks, also called credal networks when all the distributions share a common graphical structure. Some experiments are reported on the use of the ISSDM to learn the structure of a graphical model and to build supervised classifiers.     

Learning failure-free PRISM programs

PRISM is a probabilistic logic programming formalism which allows defining a probability distribution over possible worlds. This paper investigates learning a class of generative PRISM programs known as failure-free. The aim is to learn recursive PRISM programs which can be used to model stochastic processes. These programs generalise dynamic Bayesian networks by defining a halting distribution over the generative process. Dynamic Bayesian networks model infinite stochastic processes. Sampling from infinite process can only be done by specifying the length of sequences that the process generates. In this case, only observations of a fixed length of sequences can be obtained. On the other hand, the recursive PRISM programs considered in this paper are self-terminating upon some halting conditions. Thus, they generate observations of different lengths of sequences. The direction taken by this paper is to combine ideas from inductive logic programming and learning Bayesian networks to learn PRISM programs. It builds upon the inductive logic programming approach of learning from entailment.     

Malliavin calculus and asymptotic expansion for martingales

Summary. We present an asymptotic expansion of the distribution of a random variable which admits a stochastic expansion around a continuous martingale. The emphasis is put on the use of the Malliavin calculus; the uniform nondegeneracy of the Malliavin covariance under certain truncation plays an essential role as the Cramér condition did in the case of independent observations. Applications to statistics are presented. Received: 5 September 1995 / In revised form: 20 October 1996     

A goodness-of-fit statistic for Pareto-type behaviour

The fit of a statistical model can be visually assessed by inspection of a quantile–quantile or QQ plot. For the strict Pareto distribution, since log-transformed Pareto random variables are exponentially distributed, it is natural to consider an exponential quantile plot based on the log-transformed data. In case the data originate from a Pareto-type distribution, the Pareto quantile plot will be linear but only in some of the largest observations. In this paper we modify the Jackson statistic, originally proposed as a goodness-of-fit statistic for testing exponentiality, in such a way that it measures the linearity of the k largest observations on the Pareto quantile plot. Further, by taking the second-order tail behaviour of a Pareto-type model into account we construct a bias-corrected Jackson statistic. For both statistics the limiting distribution is derived. Next to these asymptotic results we also evaluate the small sample behaviour on the basis of a simulation study. The method is illustrated on two practical case studies.     

Some data-analytic modifications to Bayes-Stein estimation

Summary The usual Bayes-Stein shrinkages of maximum likelihood estimates towards a common value may be refined by taking fuller account of the locations of the individual observations. Under a Bayesian formulation, the types of shrinkages depend critically upon the nature of the common distribution assumed for the parameters at the second stage of the prior model. In the present paper this distribution is estimated empirically from the data, permitting the data to determine the nature of the shrinkages. For example, when the observations are located in two or more clearly distinct groups, the maximum likelihood estimates are roughly speaking constrained towards common values within each group. The method also detects outliers; an extreme observation will either the regarded as an outlier and not substantially adjusted towards the other observations, or it will be rejected as an outlier, in which case a more radical adjustment takes place. The method is appropriate for a wide range of sampling distributions and may also be viewed as an alternative to standard multiple comparisons, cluster analysis, and nonparametric kernel methods.     

On the existence of minimum contrast estimates in binary response model

When random samples are drawn from a 3-parameter distribution with a shifted origin and the observations corresponding to each sample are binary, criteria for the existence of minimum contrast estimates are given. These criteria can be drived by a method, called the probability contents boundary analysis. The probabilities of the existence of maximum likelihood estimates and least squares estimates are evaluated, by simulation with 1000 replications, in the case where the underlying distribution is a 3-parameter lognormal distribution or a 3-parameter loglogistic distribution.     

Order statistics from mixed exchangeable random variables

Two different exchangeable samples are considered and these two samples are assumed to be independent of each other. From these two samples a new sample is combined and treated as a single set of observations. The distribution of a single order statistic and the joint distribution of two order statistics for a new mixed sample are derived and expressed in terms of joint distribution functions. As a special case the distribution of a single order statistic and the joint distribution of two nonadjacent order statistics from exchangeable random variables are obtained. The results presented in this paper allows widespread applications in modelling of various lifetime data, biomedical sciences, reliability and survival analysis, actuarial sciences etc., where the assumption of independence of data cannot be accepted and the exchangeability is a more realistic assumption.     

Likelihood ratio inference for mean residual life of length-biased random variable

In lifetime data analysis, naturally recorded observations are length-biased data if the probability to select an item is proportional to its length. Based on i.i.d. observations of the true distribution, empirical likelihood (EL) procedure is proposed for the inference on mean residual life (MRL) of naturally recorded item. The limit distribution of the EL based log-likelihood ratio is proved to be the chi-square distribution. Under right censorship, since the EL based log-likelihood ratio leads to a scaled chi-square distribution and estimating the scale parameter leads to lower coverage of confidence interval, we propose an algorithm to calculate the likelihood ratio (LR) directly. The corresponding log-likelihood ratio converges to the standard chi-square distribution and the corresponding confidence interval has a better coverage. Simulation studies are used to support the theoretical results.     

Analysis of variance for high-dimensional time series

Analysis of variance (ANOVA) is tailored for independent observations. Recently, there has been considerable demand for ANOVA of high-dimensional and dependent observations in many fields. For example, it is important to analyze differences among industry averages of financial data. However, ANOVA for these types of observations has been inadequately developed. In this paper, we thus present a study of ANOVA for high-dimensional and dependent observations. Specifically, we present the asymptotics of classical test statistics proposed for independent observations and provide a sufficient condition for them to be asymptotically normal. Numerical examples for simulated and radioactive data are presented as applications of these results.     

����ģ�ͱ������ı�Ҷ˹����

本文主要讨论了变点的先验分布为beta-binomial分布 和Ibrahim等(2003)提出的幂型先验的条件下, 有一个变点的线性模型的贝叶斯统计推断问题, 并且我们假定变点两边的观测值的方差是相等的. 我们得到变点、回归系数、共同方差的后要分布的显示表达式. 本论文不仅把Ferrira(1975)论文从变点先验分布服从离散均匀分布推广到了更好描述变点 的形状的beta-binomial分布, 而且进一步将变点的先验分布推广到包含的历史信息的幂型先验. 当变点的先验分布为beta-binomial分布和幂型先验时, 模拟结果显示了贝叶斯方法具有更高的准确性.     

Steady-state solutions in a nonlinear pool boiling model

We consider a relatively simple model for pool boiling processes. This model involves only the temperature distribution within the heater and describes the heat exchange with the boiling fluid via a nonlinear boundary condition imposed on the fluid–heater interface. This results in a standard heat equation with a nonlinear Neumann boundary condition on part of the boundary. In this paper, we analyse the qualitative structure of steady-state solutions of this heat equation. It turns out that the model allows both multiple homogeneous and multiple heterogeneous solutions in certain regimes of the parameter space. The latter solutions originate from bifurcations on a certain branch of homogeneous solutions. We present a bifurcation analysis that reveals the multiple-solution structure in this mathematical model. In the numerical analysis a continuation algorithm is combined with the method of separation-of-variables and a Fourier collocation technique. For both the continuous and discrete problem a fundamental symmetry property is derived that implies multiplicity of heterogeneous solutions. Numerical simulations of this model problem predict phenomena that are consistent with laboratory observations for pool boiling processes.     

量测带噪声的多个体二阶系统聚集控制

考虑在含有量测噪声情况下的二阶多个体系统聚集控制问题.目的是使得系统中每个个体根据邻居信息构造控制,在只有部分个体能够观测到目标的情况下到达目标.和以前许多多个体同步及聚集问题的研究模型中所考虑的一阶系统不同, 系统的每个个体都只能量测到其邻居个体的部分状态信息,如位置, 并且这些量测还带有噪声.根据这些信息设计了基于局部规则的分散控制律,并且证明当量测噪声和状态量测本身相关时,只要系统在任意给定的时间区域段之内能够保持联合连通,就能够实现系统对目标的跟踪和达到目标.     

Estimating Joint Distributions of Markov Chains

Suppose we observe a stationary Markov chain with unknown transition distribution. The empirical estimator for the expectation of a function of two successive observations is known to be efficient. For reversible Markov chains, an appropriate symmetrization is efficient. For functions of more than two arguments, these estimators cease to be efficient. We determine the influence function of efficient estimators of expectations of functions of several observations, both for completely unknown and for reversible Markov chains. We construct simple efficient estimators in both cases.     

Onm-dependence and Edgeworth expansions

This paper contains two results. The first establishes, under mild assumptions, the validity of an Edgeworth expansion with remaindero(N ^–1/2) for aU-statistic with a kernel of degree two using observations from anm-dependent shift. The second result gives a necessary and sufficient condition for the distribution of a sum ofm-dependent random variables to possess an Edgeworth expansion. This generalizes a result of Bickel and Robinson from the i.i.d. case to them-dependent case.This research was supported in part by National Science Foundation, Grant DMS 89-23071.     

Simulation of vector semibinary homogeneous random fields and modeling of broken clouds

A vector-valued homogeneous random field is said to be semibinary if its single-point marginal distribution is a sum of a singular distribution and a continuous one. In this paper, we present methods of numerical simulation of semibinary fields on the basis of the correlation structure and the marginal distribution. As an example we construct a combined model of cloud top height and optical thickness using satellite observations.