Statistical experiments and their conical measures

Summary In the theory of functional analytic comparison of statistical experiments (started by L. LeCam) one can characterize equivalence classes of experiments by conical measures (in the sense of G. Choquet). We begin with a short proof of the (known) fact that any conical measure with normed resultant belongs to an experiment class. Then we are concerned with the special case of dominated experiments which are characterized by the extendability of their conical measures to finite concrete measures. These results are in close connection with a paper of E.N. Torgersen (Mixtures and products of dominated experiments. Ann. Statist. 5, 44–64 (1977)).After this we study class properties of experiments which are expressible in terms of their conical measures. Simple examples are domination, existence of bounded densities and compactness. It follows the investigation of a more profound class property which we call extremality and which generalizes the concept of an experiment with a sufficient and boundedly complete subalgebra. Finally we prove that the extreme points of the compact convex set of conical measures with normed resultant are just the conical measures of the extremal experiments.     

Pairwise sufficiency

Summary For an arbitrary experiment E we investigate the relation between its pairwise sufficient subalgebras and its sufficient sublattices in the M-space of E (in the sense of L. LeCam). By exhibiting an experiment without minimal pairwise sufficient subalgebra it is shown that this correspondence is in general not bijective. In view of this we introduce the rather large class of majorized experiments. They have a minimal pairwise sufficient subalgebra which can be described explicitely.As a natural subclass of the majorized experiments appear the coherent experiments that are distinguished by the coincidence of sufficiency and pairwise sufficiency. It is shown that the coherent experiments are characterized by the fact that they admit a majorizing measure which is localizable. As a consequence we obtain that the class of coherent experiments coincides with classes of experiments previously introduced by T.S. Pitcher, D. Mußmann, M. Hasegawa and M.D. Perlman.     

Fast recursive algorithm for a class of structured matrices

In this paper, we continue the study of a class of structured matrices which may be treated as a generalization of the class of Hessenberg matrices. For this class fast recursive algorithms for solution of the corresponding linear systems are obtained. The implementation of algorithms is illustrated by numerical experiments.     

Multiple processes in electron-photon cascades

Recent experiments on high energy electron-photon cascades suggest the possibility of the following multiple processes:—

1. (i)
   Electron scattering with pair creation,     
   
   

Parallel Characteristical Algorithms for Solving Problems of Global Optimization

A class of parallel characteristical algorithms for global optimization ofone-dimensional multiextremal functions is introduced. General convergence andefficiency conditions for the algorithms of the class introduced areestablished. A generalization for the multidimensional case is considered.Examples of parallel characteristical algorithms and numerical experiments arepresented.     

Compact representation of the full Broyden class of quasi‐Newton updates

In this paper, we present the compact representation for matrices belonging to the Broyden class of quasi‐Newton updates, where each update may be either rank one or rank two. This work extends previous results solely for the restricted Broyden class of rank‐two updates. In this article, it is not assumed that the same Broyden update is used in each iteration; rather, different members of the Broyden class may be used in each iteration. Numerical experiments suggest that a practical implementation of the compact representation is able to accurately represent matrices belonging to the Broyden class of updates. Furthermore, we demonstrate how to compute the compact representation for the inverse of these matrices and a practical algorithm for solving linear systems with members of the Broyden class of updates. We demonstrate through numerical experiments that the proposed linear solver is able to efficiently solve linear systems with members of the Broyden class of matrices with high accuracy. As an immediate consequence of this work, it is now possible to efficiently compute the eigenvalues of any limited‐memory member of the Broyden class of matrices, allowing for the computation of condition numbers and the ability to perform sensitivity analysis.     

Multisplitting Iteration Schemes for Solving a Class of Nonlinear Complementarity Problems

We consider several synchronous and asynchronous multisplitting iteration schemes for solving aclass of nonlinear complementarity problems with the system matrix being an H-matrix.We establish theconvergence theorems for the schemes.The numerical experiments show that the schemes are efficient forsolving the class of nonlinear complementarity problems.     

Generalization of Backward Differentiation Formulas for Parallel Computers

We analyze an extension of backward differentiation formulas, used as boundary value methods, that generates a class of methods with nice stability and convergence properties. These methods are obtained starting from the boundary value GBDFs class, and are in the class of EBDF-type methods. We discuss different ways of using these linear multistep formulas in order to have efficient parallel implementations. Numerical experiments show their effectiveness.     

An efficient solution method for rank two quasiconcave minimization problems

The technique of dimension reduction earlier developed by the first author is applied to the class of nonconvex minimization problems having the so called rank two property. This class includes in particular the problem of minimizing the product of two affine functions over a polytope. An efficient method for solving this class of problems is presented. Also some results of computational experiments with this method are discussed     

Economical experiments: Bayesian efficient experimental design

We propose and implement a Bayesian optimal design procedure. Our procedure takes as its primitives a class of parametric models of strategic behavior, a class of games (experimental designs), and priors on the behavioral parameters. We select the experimental design that maximizes the information from the experiment. We sequentially sample with the given design and models until only one of the models has viable posterior odds. A model which has low posterior odds in a small collection of models will have an even lower posterior odds when compared to a larger class, and hence we can dismiss it. The procedure can be used sequentially by introducing new models and comparing them to the models that survived earlier rounds of experiments. The emphasis is not on running as many experiments as possible, but rather on choosing experimental designs to distinguish between models in the shortest possible time period. We illustrate this procedure with a simple experimental game with one-sided incomplete information.We acknowledge the financial support from NSF grant #SES-9223701 to the California Institute of Technology. We also acknowledge the research assistance of Eugene Grayver who wrote the software for the experiments.     

Euler method vs. GESS method for dynamical systems

In this paper we introduce GESS method and show that dynamics of the systemy′ =A(s,t,y)y is more faithfully approximated by GESS method than by Euler method. Numerical experiments are given for the comparison of GESS method with Euler method.     

A progressive barrier derivative-free trust-region algorithm for constrained optimization

We study derivative-free constrained optimization problems and propose a trust-region method that builds linear or quadratic models around the best feasible and around the best infeasible solutions found so far. These models are optimized within a trust region, and the progressive barrier methodology handles the constraints by progressively pushing the infeasible solutions toward the feasible domain. Computational experiments on 40 smooth constrained problems indicate that the proposed method is competitive with COBYLA, and experiments on two nonsmooth multidisciplinary optimization problems from mechanical engineering show that it can be competitive with the NOMAD software.     

A new class of test functions for global optimization

In this paper we propose a new class of test functions for unconstrained global optimization problems. The class depends on some parameters through which the difficulty of the test problems can be controlled. As a basis for future comparison, we propose a selected set of these functions, with increasing difficulty, and some computational experiments with two simple global optimization algorithms.     

A combined class of self-scaling and modified quasi-Newton methods

Techniques for obtaining safely positive definite Hessian approximations with self-scaling and modified quasi-Newton updates are combined to obtain ??better?? curvature approximations in line search methods for unconstrained optimization. It is shown that this class of methods, like the BFGS method, has the global and superlinear convergence for convex functions. Numerical experiments with this class, using the well-known quasi-Newton BFGS, DFP and a modified SR1 updates, are presented to illustrate some advantages of the new techniques. These experiments show that the performance of several combined methods are substantially better than that of the standard BFGS method. Similar improvements are also obtained if the simple sufficient function reduction condition on the steplength is used instead of the strong Wolfe conditions.     

A class of exhaustive cone splitting procedures in conical algorithms for concave minmization

A class of exhaustive cone splitting procedures is developed and is shown to perform substantially better than the bisection procedure in conical algorithms for concave minimization. Computational experiments are reported and discussed.     

Speed,Gradient and Workrate in Uphill Running

A series of treadmill experiments is described concerned with a runner's speed, heart-rate and the gradient. Together with the results of similar experiments, some of them carried out over 50 years ago, the results suggest that for a given heart-rate, log(speed) is linearly related to gradient, and that for a given gradient, heart-rate is linearly related to speed. The results suggest:

1. 1)
   that athletes who run p% faster on the level will run p% faster up a slope, if they maintain the same heart-rate;
    
2. 2)
   that athletes will use the same number of heart beats running up a hill of uniform slope no matter how fast or slowly they run;
    
3. 3)
   that athletes should run directly up any slope of less than about 20° and try to zigzag up slopes greater than this.
    

     

High order sliding-mode dynamic control for chaotic intracellular calcium oscillations

In this article the closed-loop stability conditions and the control design of a class of biological system that exhibit chaotic oscillations are addressed. It is proved that the biological system is minimum phase. A class of nonlinear dynamic feedback control is designed using sliding-mode control ideas, which can be used for regulation, tracking and synchronization tasks. Numerical experiments illustrate the satisfactory performance of the proposed control methodology.     

The classI o on abstract structures

We consider the class I_o of elements without indecomposable factor in fdifferent abstract structures and prove in these cases that I_o is strictly included in the class of infinitely divisible elements. The structures investigated are convex cones, q-probability, and statistical experiments. Proceedings of the Seminar on Stability Problems for Stochastic Models, Vologda, Russia, 1998, Part II.     

A posteriori error analysis of multipoint flux mixed finite element methods for interface problems

In this paper, the multipoint flux mixed finite element method is used to approximate the flux of two-dimensional elliptic interface problems. Within the class of modified quasi-monotonically distributed coefficients, we derive uniformly robust residual-type a posteriori error estimators for the flux error. Based on the residual-type estimator, we further develop robust implicit and explicit recovery-type estimators through gradient recovery in H(curl) conforming finite element spaces. Numerical experiments are presented to support the theoretical results.