Bayesian priors based on a parameter transformation using the distribution function

One of the tasks of the Bayesian consulting statistician is to elicit prior information from his client who may be unfamiliar with parametric statistical models. In some cases it may be more illuminating to base a prior distribution for parameter on the transformed version F(/), where F is the data distribution function and v is a designated reference value, rather than on directly. This approach is outlined and explored in various directions to assess its implications. Some applications are given, including general linear regression and transformed linear models.     

Stochastic comparisons and bounds for aging renewal process shock models and their applications

Sharp comparisons between aging renewal process shock models and the corresponding Esary-Marshall-Proschan (EMP) shock model are considered. The usefulness of such comparisons derive from the simplicity of the latter models. Simple conditions under which such aging renewal process shock models are stochastically ordered relative to a corresponding EMP-model are derived. Applications to renewal functions and single server queues are indicated.     

CAS integration into learning environment

In order to successfully integrate computers into education, it is necessary to organize effectively the prerequisites of human-machine interaction. In the organization of competence-centred education computers could provide valuable assistance for both personal- and group- learning activities. In this paper, we will examine various applications of Computer Algebra Systems (CAS) in classroom settings. The elements of the learning environments are CAS. E-Learning-portal, and Tight VNC remote control system. CAS assisted teaching can become genuinely effective in a complex learning environment if students' instrumental-genesis evolve into instrumental-orchestration. We will demonstrate the evolution of thus process by using one of our developed applications. As an example, we developed, tested, and evaluated our model in the Department of Engineering at the University of Pecs. The study took place during the 2004–2005 academic year with computer science and computer engineering participants.     

Advanced Shell Element Formulations for Coupled Electromechanical Systems

With the significantly increasing applications of smart structures, piezoelectric material is widely used in branches of engineering sciences. Normally, the Finite Element Method is employed in the numerical analysis of these structures [2]. In this contribution, in order to avoid the locking effects and zero energy modes, the Assumed Natural Strain (ANS) Method [4] is implemented into four‐node piezoelectric shallow shell elements, by using the two‐field variational formulation in which displacements and electric potentials serve as independent variables and the three‐field variational formulation in which the dielectric displacement is taken as an independent variable additionally [3]. Moreover, a quadratic variation of the electric potential through the thickness direction is applied in the two‐field formulation. Numerical examples of piezoelectric sensors and actuators are presented, showing the behaviour of the shell elements by using different hybrid finite element formulations. (© 2004 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

On the nested refinement of quadrilateral and hexahedral finite elements and the affine approximation

Summary. It is shown in this paper that the optimal approximation property of bilinear or trilinear finite elements would be retained when the affine mapping is used to replace the Q₁ mapping on each element, if the grids are refined nestedly. The new method truncates the quadratic and cubic terms in reference mappings and produces constant Jacobians and Jacobian matrices. This would avoid a shortcoming of the quadrilateral and hexahedral elements where the integrals of rational functions have to be computed or approximated. Numerical tests verify the analysis.Mathematics Subject Classification (2000): 65N30, 65N50, 65N55Revised version received January 29, 2004     

A Parametric n-Dimensional Markov-Functional Model in the Terminal Measure

Abstract

This article develops and tests an n-dimensional Markov-functional interest rate model in the terminal measure based on parametric functional forms of exponential type. The parametric functional forms enable analytical expressions for forward discount bonds and forward LIBORs at all times and allows for calibration of the model to caplet prices given by a displaced diffusion Black model. The analytical expressions of the model provide a theoretical tool for understanding the structure of standard Markov-functional models (MFMs) as well as comparisons with the LIBOR market model (LMM). In particular, it is shown that for ‘typical’ market data the model is close enough to the LMM to be able to calibrate using the LMM calibration set-up and machinery. This provides further information about the similarities (as well as some of the differences) between MFM and LMM. The parametric n-dimensional MFM may be used for products that require high-dimensional models for appropriate pricing and risk management. When compared with an n-factor LMM, it has the virtue of being (much) faster for certain types of products.     

Sparse Partially Linear Additive Models

The generalized partially linear additive model (GPLAM) is a flexible and interpretable approach to building predictive models. It combines features in an additive manner, allowing each to have either a linear or nonlinear effect on the response. However, the choice of which features to treat as linear or nonlinear is typically assumed known. Thus, to make a GPLAM a viable approach in situations in which little is known a priori about the features, one must overcome two primary model selection challenges: deciding which features to include in the model and determining which of these features to treat nonlinearly. We introduce the sparse partially linear additive model (SPLAM), which combines model fitting and both of these model selection challenges into a single convex optimization problem. SPLAM provides a bridge between the lasso and sparse additive models. Through a statistical oracle inequality and thorough simulation, we demonstrate that SPLAM can outperform other methods across a broad spectrum of statistical regimes, including the high-dimensional (p ? N) setting. We develop efficient algorithms that are applied to real datasets with half a million samples and over 45,000 features with excellent predictive performance. Supplementary materials for this article are available online.     

Learning axes and bridging tools in a technology-based design for statistics

We introduce a design-based research framework, learning axes and bridging tools, and demonstrate its application in the preparation and study of an implementation of a middle-school experimental computer-based unit on probability and statistics, ProbLab (Probability Laboratory, Abrahamson and Wilensky 2002 [Abrahamson, D., & Wilensky, U. (2002). ProbLab. Northwestern University, Evanston, IL: The Center for Connected Learning and Computer-Based Modeling, Northwestern University. ]). ProbLab is a mixed-media unit, which utilizes traditional tools as well as the NetLogo agent-based modeling-and-simulation environment (Wilensky 1999) [Wilensky, U. (1999). NetLogo. Northwestern University, Evanston, IL: The Center for Connected Learning and Computer-Based Modeling ] and HubNet, its technological extension for facilitating participatory simulation activities in networked classrooms (Wilensky and Stroup 1999a) [Wilensky, U., & Stroup, W. (1999a). HubNet. Evanston, IL: The Center for Connected Learning and Computer-Based Modeling, Northwestern University]. We will focus on the statistics module of the unit, Statistics As Multi-Participant Learning-Environment Resource (S.A.M.P.L.E.R.). The framework shapes the design rationale toward creating and developing learning tools, activities, and facilitation guidelines. The framework then constitutes a data-analysis lens on implementation cases of student insight into the mathematical content. Working with this methodology, a designer begins by focusing on mathematical representations associated with a target concept—the designer problematizes and deconstructs each representation into a pair of historical/cognitive antecedents (idea elements), each lying at the poles of a learning axis. Next, the designer creates bridging tools, ambiguous artifacts bearing interaction properties of each of the idea elements, and develops activities with these learning tools that evoke cognitive conflict along the axis. Students reconcile the conflict by means of articulating strategies that embrace both idea elements, thus integrating them into the target concept. The research reported on this paper was funded by NSF ROLE Grant No. REC-0126227. The opinions expressed here are those of the authors and do not necessarily reflect those of NSF. This paper is based on the authors’ AERA 2004 paper titled S.A.M.P.L.E.R.: Statistics As Multi-Participant Learning-Environment Resource.

---

Microlocal Analysis of the Geometric Separation Problem

Image data are often composed of two or more geometrically distinct constituents; in galaxy catalogs, for instance, one sees a mixture of pointlike structures (galaxy superclusters) and curvelike structures (filaments). It would be ideal to process a single image and extract two geometrically “pure” images, each one containing features from only one of the two geometric constituents. This seems to be a seriously underdetermined problem but recent empirical work achieved highly persuasive separations. We present a theoretical analysis showing that accurate geometric separation of point and curve singularities can be achieved by minimizing the ?₁ norm of the representing coefficients in two geometrically complementary frames: wavelets and curvelets. Driving our analysis is a specific property of the ideal (but unachievable) representation where each content type is expanded in the frame best adapted to it. This ideal representation has the property that important coefficients are clustered geometrically in phase space, and that at fine scales, there is very little coherence between a cluster of elements in one frame expansion and individual elements in the complementary frame. We formally introduce notions of cluster coherence and clustered sparsity and use this machinery to show that the underdetermined systems of linear equations can be stably solved by ?₁ minimization; microlocal phase space helps organize the calculations that cluster coherence requires. © 2012 Wiley Periodicals, Inc.     

The physics of categorization

In the context of cognition, categorization is the process through which several elements (i.e., words) are grouped into a single set which by naming becomes an abstraction of its elements. For example, tiger, kitty, and max can be categorized as Cats. In this article, we aim to show how the physical, biological and cognitive dimensions are related in the process of categorization or abstraction through the physics of computation. Drawing on Landauer's principle, we show that the price paid in terms of entropy is higher when grouping elements of low ranking (high probability) than when grouping elements of high ranking (low probability). Therefore, the logic of the cognitive process of abstraction is explained through constraints imposed by memory on the computation of categories. © 2015 Wiley Periodicals, Inc. Complexity 21: 269–274, 2016     

Continuum limits for classical sequential growth models

A random graph order, also known as a transitive percolation process, is defined by taking a random graph on the vertex set {0,…,n ? 1} and putting i below j if there is a path i = i₁…i_k = j in the graph with i₁ < … < i_k. Rideout and Sorkin 14 provide computational evidence that suitably normalized sequences of random graph orders have a “continuum limit.” We confirm that this is the case and show that the continuum limit is always a semiorder. Transitive percolation processes are a special case of a more general class called classical sequential growth models. We give a number of results describing the large‐scale structure of a general classical sequential growth model. We show that for any sufficiently large n, and any classical sequential growth model, there is a semiorder S on {0,…,n ‐ 1} such that the random partial order on {0,…,n ‐ 1} generated according to the model differs from S on an arbitrarily small proportion of pairs. We also show that, if any sequence of classical sequential growth models has a continuum limit, then this limit is (essentially) a semiorder. We give some examples of continuum limits that can occur. Classical sequential growth models were introduced as the only models satisfying certain properties making them suitable as discrete models for spacetime. Our results indicate that this class of models does not contain any that are good approximations to Minkowski space in any dimension ≥ 2. © 2009 Wiley Periodicals, Inc. Random Struct. Alg., 2010     

From Individuals to Populations: A Symbolic Process Algebra Approach to Epidemiology

Is it possible to symbolically express and analyse an individual-based model of disease spread, including realistic population dynamics? This problem is addressed through the use of process algebra and a novel method for transforming process algebra into Mean Field Equations. A number of stochastic models of population growth are presented, exploring different representations based on alternative views of individual behaviour. The overall population dynamics in terms of mean field equations are derived using a formal and rigorous rewriting based method. These equations are easily compared with the traditionally used deterministic Ordinary Differential Equation models and allow evaluation of those ODE models, challenging their assumptions about system dynamics. The utility of our approach for epidemiology is confirmed by constructing a model combining population growth with disease spread and fitting it to data on HIV in the UK population. This work was supported by EPSRC through a Doctoral Training Grant (CM, from 2004–2007), and through System Dynamics from Individual Interactions: A process algebra approach to epidemiology (EP/E006280/1, all authors, 2007–2010).     

The Stone-Čech compactification of a partial frame via ideals and cozero elements

A partial frame is a meet-semilattice in which certain designated subsets are required to have joins, and finite meets distribute over these. The designated subsets are specified by means of a so-called selection function, denoted by S ; these partial frames are called S-frames.

We construct free frames over S-frames using appropriate ideals, called S-ideals. Taking S-ideals gives a functor from S-frames to frames. Coupled with the functor from frames to S-frames that takes S-Lindelöf elements, it provides a category equivalence between S-frames and a non-full subcategory of frames. In the setting of complete regularity, we provide the functor taking S-cozero elements which is right adjoint to the functor taking S-ideals. This adjunction restricts to an equivalence of the category of completely regular S-frames and a full subcategory of completely regular frames. As an application of the latter equivalence, we construct the Stone-? ech compactification of a completely regular S-frame, that is, its compact coreflection in the category of completely regular S-frames.

A distinguishing feature of the study of partial frames is that a small collection of axioms of an elementary nature allows one to do much that is traditional at the level of frames or locales and of uniform or nearness frames. The axioms are sufficiently general to include as examples of partial frames bounded distributive lattices, σ-frames, κ-frames and frames.     

The Intrinsic Enumerability of Linear Orders

We study into the question of which linearly ordered sets are intrinsically enumerable. In particular, it is proved that every countable ordinal lacks this property. To do this, we state a criterion for hereditarily finite admissible sets being existentially equivalent, which is interesting in its own right. Previously, Yu. L. Ershov presented the criterion for elements h ₀ , h ₁ in HF ) to realize a same type as applied to sufficiently saturated models . Incidentally, that criterion fits with every model on the condition that we limit ourselves to 1-types.     

Cut-Free Ordinary Sequent Calculi for Logics Having Generalized Finite-Valued Semantics

The paper presents a method for transforming a given sound and complete n-sequent proof system into an equivalent sound and complete system of ordinary sequents. The method is applicable to a large, central class of (generalized) finite-valued logics with the language satisfying a certain minimal expressiveness condition. The expressiveness condition decrees that the truth-value of any formula φ must be identifiable by determining whether certain formulas uniformly constructed from φ have designated values or not. The transformation preserves the general structure of proofs in the original calculus in a way ensuring preservation of the weak cut elimination theorem under the transformation. The described transformation metod is illustrated on several concrete examples of many-valued logics, including a new application to information sources logics.     

Simulating and visualizing neural networks with Mathematica

Artificial neural networks (ANNs) are both mathematical models of the neural basis of higher-order cognitive functions, such as learning, and adaptive variations of the general linear and nonlinear regression. Students of psychology and cognitive science typically encounter ANNs in both contexts of their studies, especially at the graduate level, however, many of these students do not possess the programming skills to write their own simulations to test their application as cognitive and statistical models. In this paper, simulations using the mathematical programming language Mathematica are used to develop appropriate visualizations of one the foundation topics in ANNs (understanding why linear associative networks cannot learn the nonlinearly separable XOR function). It is argued that Mathematica and similar high-level interpreted packages provide a more accessible environment for nonprogramming students to further their understanding of this key area of psychological science and mathematical modelling.     

Local Rings of Rings of Quotients

The aim of this paper is to characterize those elements in a semiprime ring R for which taking local rings at elements and rings of quotients are commuting operations. If Q denotes the maximal ring of left quotients of R, then this happens precisely for those elements if R which are von Neumann regular in Q. An intrinsic characterization of such elements is given. We derive as a consequence that the maximal left quotient ring of a prime ring with a nonzero PI-element is primitive and has nonzero socle. If we change Q to the Martindale symmetric ring of quotients, or to the maximal symmetric ring of quotients of R, we obtain similar results: an element a in R is von Neumann regular if and only if the ring of quotients of the local ring of R at a is isomorphic to the local ring of Q at a. Partially supported by the Ministerio de Educación y Ciencia and Fondos Feder, jointly, trough projects MTM2004-03845, MTM2007-61978 and MTM2004-06580-C02-02, MTM2007-60333, by the Junta de Andalucía, FQM-264, FQM336 and FQM02467 and by the Plan de Investigación del Principado de Asturias FICYT-IB05-017.     

Optimizing Preventive Maintenance Models

We deal with the problem of scheduling preventive maintenance (PM) for a system so that, over its operating life, we minimize a performance function which reflects repair and replacement costs as well as the costs of the PM itself. It is assumed that a hazard rate model is known which predicts the frequency of system failure as a function of age. It is also assumed that each PM produces a step reduction in the effective age of the system. We consider some variations and extensions of a PM scheduling approach proposed by Lin et al. [6]. In particular we consider numerical algorithms which may be more appropriate for hazard rate models which are less simple than those used in [6] and we introduce some constraints into the problem in order to avoid the possibility of spurious solutions. We also discuss the use of automatic differentiation (AD) as a convenient tool for computing the gradients and Hessians that are needed by numerical optimization methods. The main contribution of the paper is a new problem formulation which allows the optimal number of occurrences of PM to be determined along with their optimal timings. This formulation involves the global minimization of a non-smooth performance function. In our numerical tests this is done via the algorithm DIRECT proposed by Jones et al. [19]. We show results for a number of examples, involving different hazard rate models, to give an indication of how PM schedules can vary in response to changes in relative costs of maintenance, repair and replacement. Part of this work was carried out while the first author was a Visiting Professor in the Department of Mechanical Engineering at the University of Alberta in December 2003.     

Group classification and symmetry reduction of a (2+1) dimensional diffusion-advection equation

Based on classical Lie group method, we consider the continuum problem of the driven diffusive flow of particles past an impenetrable obstacle (rod) of length L. The infinitesimals of the diffusion-advection equation in (2+1) dimensions were found for an arbitrary nonlinear advection. The symmetries corresponding to different forms of the nonlinear advection are obtained. Three models are studied in details. The results show that the presence of an obstacle, whether stationary or moving, in a driven diffusive flow with nonlinear drift will distort the local concentration profile to a state which divided the (x, y)-plane into two regions. The concentration is relatively higher in one side than the other side, apart from the value of where D is the diffusion coefficient and υ is the drift velocity. This problem has relevance for the size segregation of particulate matter which results from the relative motion of different-size particles induced by shaking. Also, the obtained solutions include soliton, periodical, rational and singular solutions. Received: November 10, 2003; revised: February 10, 2004