Motivated by the study of an important data set for understanding the large-scale structure of the universe, this work considers the estimation of the reduced second-moment function, or $K$ function, of a stationary point process on $\mathbb{R}$ observed over a large number of segments of possibly varying lengths. Theory and simulation are used to compare the behavior of isotropic and rigid motion correction estimators and some modifications of these estimators. These results generally support the use of modified versions of the rigid motion correction.When applied to a catalog of astronomical objects known as absorbers, the proposed methods confirm results from earlier analyses of the absorber catalog showing clear evidence of clustering up to $50 h^{-1}$ Mpc and marginal evidence for clustering of matter on spatial scales beyond $100 h^{-1}$ Mpc, which is beyond the distance at which clustering of matter is now generally accepted to exist.

The probability distribution of the cascade generators in a random multiplicative cascade represents a hidden parameter which is reflected in the fine scale limiting behavior of the scaling exponents (sample moments) of a single sample cascade realization as a.s.constants.We identify a large class of cascade generators uniquely determined by these scaling exponents. For this class we provide both asymptotic consistency and confidence intervals for two different estimators of the cumulant generating function (log Laplace transform) of the cascade generator distribution. These results are derived from investigation of the convergence properties of the fine scale sample moments of a single cascade realization.

Consider the model $X = B + S$, where $B$and $S$ are independent Poisson random variables with means $\mu$ and $\nu$, $\nu$ is unknown, but $\mu$ is known. The model arises in particle physics and some recent articles have suggested conditioning on the observed bound on $B$; that is, if $X = n$ is observed, then the suggestion is to base inference on the conditional distribution of $X$ given $B \leq n$. This conditioning is non-standard in that it does not correspond to a partition of the sample space. It is examined here from the view point of decision theory and shown to lead to admissible formal Bayes procedures.

We propose the randomized Generalized Approximate Cross Validation (ranGACV) method for choosing multiple smoothing parameters in penalized likelihood estimates for Bernoulli data. The method is intended for application with penalized likelihood smoothing spline ANOVA models. In addition we propose a class of approximate numerical methods for solving the penalized likelihood variational problem which, in conjunction with the ranGACV method allows the application of smoothing spline ANOVA models with Bernoulli data to much larger data sets than previously possible. These methods are based on choosing an approximating subset of the natural (representer) basis functions for the variational problem. Simulation studies with synthetic data, including synthetic data mimicking demographic risk factor data sets is used to examine the properties of the method and to compare the approach with the GRKPACK code of Wang (1997c). Bayesian “confidence intervals” are obtained for the fits and are shown in the simulation studies to have the “across the function” property usually claimed for these confidence intervals. Finally the method is applied to an observational data set from the Beaver Dam Eye study, with scientifically interesting results.

The Bayesian Information Criterion (BIC) estimates the order of a Markov chain (with finite alphabet $A$) from observation of a sample path $x_1, x_2,\dots, x_n$, as that value $k = \hat{k}$ that minimizes the sum of the negative logarithm of the $k$th order maximum likelihood and the penalty term $\frac{|A|^k(|A|-1)}{2}\log n$. We show that $\hat{k}$ equals the correct order of the chain, eventually almost surely as $n \rightarrow \infty$, thereby strengthening earlier consistency results that assumed an apriori bound on the order. A key tool is a strong ratio-typicality result for Markov sample paths.We also show that the Bayesian estimator or minimum description length estimator, of which the BIC estimator is regarded as an approximation, fails to be consistent for the uniformly distributed i.i.d. process.

This paper addresses the topic of model selection in regression.We emphasize the case of two models, testing which model provides a better prediction based on $n$ observations. Within a family of selection rules, based on maximizing a penalized log-likelihood under a normal model, we search for asymptotically minimax rules over a class $\mathscr{G}$ of possible joint distributions of the explanatory and response variables. For the class $\mathscr{G}$ of multivariate normal joint distributions it is shown that asymptotically minimax selection rules are close to the AIC selection rule when the models’ dimension difference is large. It is further proved that under fairly mild assumptions on $\mathscr{G}$ any asymptotically minimax sequence of procedures satisfies the condition that the difference in their dimension penalties is bounded as the number of observations approaches infinity. The results are then extended to the case of more than two competing models.

Asymptotic approximations for the error probabilities of sequential tests of composite hypotheses in multiparameter exponential families are developed herein for a general class of test statistics, including generalized likelihood ratio statistics and other functions of the sufficient statistics. These results not only generalize previous approximations for Type I error probabilities of sequential generalized likelihood ratio tests, but also pro- vide a unified treatment of both sequential and fixed sample size tests and of Type I and Type II error probabilities. Geometric arguments involving integration over tubes play an important role in this unified theory.

Motivated by an application in computerized adaptive tests, we consider the following sequential design problem.There are $J$ jobs to be processed according to a predetermined order. A single machine is available to process these $J$ jobs. Each job under processing evolves stochastically as a Markov chain and earns rewards as it is processed, not otherwise. The Markov chain has transition probabilities parameterized by an unknown parameter $\theta$. The objective is to determine how long each job should be processed so that the total expected rewards over an extended time interval is maximized. We define the regret associated with a strategy as the shortfall from the maximum expected reward under complete information on $\theta$. Therefore the problem is equivalent to minimizing the regret. The asymptotic lower bound for the regret associated with any uniformly good strategy is characterized by a deterministic constraint minimization problem. In ignorance of the parameter value, we construct a class of efficient strategies, which achieve the lower bound, based on the theory of sequential testing.

We study the continuous time analogue of the Bayes problem of Poor. As usual the results on the optimal solution are more explicit than in the discrete time set-up.

In the common polynomial regression model of degree m we consider the problem of determining the $D$- and $D_1$-optimal designs subject to certain constraints for the $D_1$-efficiencies in the models of degree $m - j, m - j + 1,\dots, m + k(m > j \geq 0, k \geq 0 \text{given})$.We present a complete solution of these problems, which on the one hand allow a fast computation of the constrained optimal designs and, on the other hand, give an answer to the question of the existence of a design satisfying all constraints. Our approach is based on a combination of general equivalence theory with the theory of canonical moments. In the case of equal bounds for the $D_1$-efficiencies the constrained optimal designs can be found explicitly by an application of recent results for associated orthogonal polynomials.

This paper generalizes Kushner’s method for finding optimal repeated measurements designs to find optimal designs under an interference model. The model we assume is for a one-dimensional layout without guard plots and with different left and right neighbor effects. The resulting optimal designs may need many blocks or may not even exist as a finite design. The results give lower bounds for optimality criteria on finite designs and the design structure can be used to suggest efficient small designs.

We consider the situation where one has to maximize a function $\eta(\theta, \mathbf{x})$ with respect to $\mathbf{x} \epsilon \mathbb{R}^q$, when $\theta$ is unknown and estimated by least squares through observations $y_k = \mathbf{f}^{\top(\mathbf{x}_k)\theta + \varepsilon_k$, with $\varepsilon_k$ some random error. Classical applications are regulation and extremum control problems. The approach we adopt corresponds to maximizing the sum of the current estimated objective and a penalization for poor estimation: $\mathbf{x}_{k + 1}$ maximizes $\eta(\hat{\theta}^k, \mathbf{x}) + (\alpha_k/k), d_k(\mathbf{x})$, with $\hat{\theta}^k$ the estimated value of $\theta$ at step $k$ and $d_k$ the penalization function. Sufficient conditions for strong consistency of $\hat{\theta}^k$ and for almost sure convergence of $(1/k) \Sigma_{i=1}^k \eta(\theta, \mathbf{x}_i)$ to the maximum value of $\eta(\theta, \mathbf{x})$ are derived in the case where $d_k(\cdot)$ is the variance function used in the sequential construction of $D$-optimum designs. A classical sequential scheme from adaptive control is shown not to satisfy these conditions, and numerical simulations confirm that it indeed has convergence problems.

A new approximation to the Gaussian likelihood of a multivariate locally stationary process is introduced. It is based on an approximation of the inverse of the covariance matrix of such processes. The new quasi likelihood is a generalization of the classical Whittle likelihood for stationary processes. Several approximation results are proved for the likelihood function. For parametric models, asymptotic normality and efficiency of the resulting estimator are derived for Gaussian locally stationary processes.