(CWI-CS-R8947; ETN-90-98080) Copyright Avail: NTIS HC/MF A03 The formal description of the Graphical Kernel System (GKS) input model given in Hoare's communicating sequential processes notation is illustrated and extensions in which some of the components in the GKS model are replaced by more interesting ones are explored. Some of the power and flexibility inherent in the component frameworks idea are thus demonstrated. The use of a formal notation led to a deepening of the authors' understanding of the input model and suggested some different ways of looking at the input model. ESA N91-12276# Center for Mathematics and Computer Science, Amsterdam (Netherlands). Dept. of Computer Science. EXTENDING S-INTERPRETATIONS TO LOGIC PROGRAMS WITH NEGATION Daniele Turi Nov. 1989 17 p (CWI-CS-R8948; ETN-90-98081) Copyright Avail: NTIS HC/MF A03 S-interpretations together with the notion of s-truth were previously used to define a least model semantics that resembles very much the least Herbrand model semantics and that provides a strongly complete declarative modeling for positive logic programming (i.e., Horn clause logic + SLD-resolution). Defining cs-interpretations as sets of constrained atoms, s-interpretations are extended to general programs (i.e., logic programs with negation). Chan's constructive negation inspired the definition and provides a procedural support for the declarative framework set up. In this, to define the notion of cs-truth, most general disunifiers between conjunctions of atoms and cs-interpretations are introduced. Stratified programs are general programs much related to positive ones. A fixpoint construction was previously introduced to produce a minimal Herbrand model which extends to stratified programs the least Herbrand model for positive programs, and an alternative characterization of such model was given (perfect Herbrand model). Applying these two constructs to cs-models, the unique perfect cs-model is obtained. A conjecture of a strong completeness for the cs-semantics wrt Przymusinski's interpreter for stratified programs and constructive negation (SLSC-resolution) is given. ESA N91-12277# Center for Mathematics and Computer Science, R. J. vanGlabbeek Jan. 1990 25 p Submitted for publication ST-configurations to represent all the states of concurrent systems are introduced. ST-bisimulation equivalence is proved to be preserved under refinement of actions. This implies that it is possible to abstract from the causal structure of concurrent systems ESA without assuming action atomicity. N91-12278# Technische Univ., Delft (Netherlands). Jacobus Hendricus Weber 1989 180 p Error correcting block codes for reliable transmission or storage of data in a communication system using a binary channel are addressed. Symmetric, unidirectional and asymmetric error types are studied. Conditions for error detection and/or correction capabilities of block codes are derived. Upper bounds on the maximum cardinality of asymmetric or unidirectional error correcting codes are studied. Lower bounds, which can be obtained by constructing codes, are similarly studied. Bounds on the cardinality of codes that are linear are discussed. ESA N91-11767# Royal Dutch Airlines, Amsterdam (Netherlands). N91-12280# Wright Research Development Center, GRAPHICAL INFERENCE IN QUALITATIVE PROBABILISTIC Qualitative probabilistic networks (QPNs) are abstractions of influence diagrams that encode constraints on the probabilistic relation among variables rather than precise numeric distributions. Qualitative relations express monotonicity constraints on direct probabilistic relations between variables, or on interactions among the direct relations. Like their numeric counterpart, QPNs facilitate graphical inference: methods for deriving qualitative relations of interest via graphical transformations of the network model. However, query processing in QPNs exhibits computational properties quite different from basic influence diagrams. In particular, the potential for information loss due to the incomplete specification of probabilities poses the new challenge of minimizing ambiguity. Analysis of the properties of QPN transformations reveals several characteristics of admissible graphical inference GRA procedures. Marc Glenn Slack 1990 202 p Sponsored in part by NASA, Washington and NSWC Avail: Univ. Microfilms Order No. DA9028092 CSCL 09B For mobile robots to autonomously accommodate dynamically changing navigation tasks in a goal-directed fashion, they must employ navigation plans. Any such plan must provide for the robot's immediate and continuous need for guidance while remaining highly flexible in order to avoid costly computation each time the robot's perception of the world changes. Due to the world's uncertainties, creation and maintenance of navigation plans cannot involve arbitrarily complex processes, as the robot's perception of the world will be in constant flux, requiring modifications to be made quickly if they are to be of any use. This work introduces navigation templates (NaT's) which are building blocks for the construction and maintenance of rough navigation plans which capture the relationship that objects in the world have to the current navigation task. By encoding only the critical relationship between the objects in the world and the navigation task, a NaT-based navigation plan is highly flexible; allowing new constraints to be quickly incorporated into the plan and existing constraints to be updated or deleted from the plan. To satisfy the robot's need for immediate local guidance, the NaT's forming the current navigation plan are passed to a transformation function. The transformation function analyzes the plan with respect to the robot's current location to quickly determine (a few times a second) the locally preferred direction of travel. This dissertation presents NaT's and the transformation function as well as the needed support systems to demonstrate the usefulness of the technique for controlling the actions of a mobile robot operating in an uncertain world. Dissert. Abstr. N91-12282 Michigan State Univ., East Lansing. Avail: Univ. Microfilms Order No. DA9028685 Multi-body mechanical systems modeled in non-minimal coordinates have various advantages over systems in minimal sets of generalized coordinates. For example a system in non-minimal coordinates is simpler to model and important system measures (descriptor variables) are directly available. Unfortunately the many advantages of the non-minimal forms are nearly overcome by disadvantages due to the singularities of the system. Especially the standard forms of optimal control are not convenient for the systems represented in non-minimal coordinates (descriptor equations). The goal of this dissertation is to derive and test new and flexible forms of the necessary conditions of optimal control appropriate for multi-body systems in non-minimal coordinate sets. In order to derive the necessary conditions for optimal control, the problem is posed directly on an implicit system represented in non-minimal coordinates sets without reducing the second order form to the first order form. The approach follows the standard method of forming an augmented cost functional wherein the system is adjoined to a nonnegative by introducing adjoint variables. The second order implicit form necessary conditions for optimal control problems on multi-body mechanical systems models of non-minimal or descriptor form are presented. Dissert. Abstr. N91-12284# Oregon Univ., Eugene. Dept. of Psychology. Michael I. Posner and Don M. Tucker 14 Jun. 1990 25 p A new state of the art Evoked Response Potential (ERP) laboratory was developed based on Macintosh computers and labview software that can record up to 64 channels of EEG input. This system is described along with its potential and a manual is included for its operation. In addition, two experiments were completed that show a distinction between words and consonant strings that is maximal for occipital lobe electrodes and that occurs within the first 250 millisec after input. In addition, words also differ from consonant strings in the lateral distribution of electrical activity over posterior temporal leads within the first 200 millisec following input. These two results conform to findings using PET suggesting an area sensitive to orthographic regularity in the left ventral occipital lobe (Snyder et al, 1989). GRA N91-12285# Honeywell, Inc., Minneapolis, MN. Blaise Morton, John Doyle, Andrew Packard, and Allen (Contract N00014-82-C-0157) (AD-A225219) Avail: NTIS HC/MF A03 CSCL 23/2 The topic is the theory of structured singular values in robust control analysis. Classical methods of invariant theory are applied to the problem of computing the structured singular value for the nonrepeated diagonal complex block structure. An algorithm for computing the structured singular value for the four-block problem is described. GRA N91-12286# Michigan Univ., Ann Arbor. ROBUSTNESS OF FEEDBACK SYSTEMS WITH SEVERAL MODELLING ERRORS Final Report, Mar. 1988 - Mar. 1990 James S. Freudenberg Jun. 1990 154 p (Contract F33615-88-C-3601; AF Proj. 2304) (AD-A225365; UM/EECS-025211-T-1; WRDC-TR-90-3018) Avail: NTIS HC/MF A08 CSCL 01/4 In this report we study robustness properties of linear multivariable feedback systems with several sources of modelling uncertainty. We assume that each source of uncertainty is modelled as a stable unstructured perturbation satisfying a frequency dependent norm bound. A stability margin for such an uncertainty description is furnished by the structured singular value. The structured singular value is obtained from a numerical optimization procedure. Consequently, it is difficult to obtain insight into the relation between its size and the plant, compensator, and design specifications upon which it depends. Our approach is to obtain such insight by deriving bounds and approximations to the structured singular value. The results we obtain are appealing in that they are both reasonably tight and in that they do provide the desired insight. We apply this insight first to understand the source of robustness difficulties associated with an ill-conditioned plant, and use the resulting information to develop a procedure for selecting weightings in Linear Quadratic Gaussian with Loop Transfer Recovery (LQG/LTR) and H-infinity synthesis problems so that the structured singular value is minimized. GRA of a supervisor for a nonlinear, adaptive controlled process. Ordinary negative feedback, adaptive and supervisory control systems are addressed. Stability monitoring in supervisors is reported. Software design specifications for a supervisor stability monitor/trend analyzer, the actual design in the form of Jackson diagrams, and a pseudo-code presentation are given. Significant conclusions are that the practical nature of the systems dictate the requirements for an expert system to determine what to do after the supervisor has detected a problem, and the supervisor/expert system combination shows potential for implementation with fast, real time, embedded systems plagued with limited computation time and storage space. 64 NUMERICAL ANALYSIS ESA Includes iteration, difference equations, and numerical approximation. N91-12288 Stanford Univ., CA. ROBUSTNESS OF LINEAR SYSTEMS TO PARAMETER Laurent Marc ElGhaoui 1990 211 p Avail: Univ. Microfilms Order No. DA9024326 The dynamic behavior of many physical systems can be described by linear time-invariant (LTI) models. The stability robustness issue is addressed by defining a parameter margin for a plant with a given controller. It has the smallest I(sub q)-norm of real parameter deviations that produce instability. The I(sub 2) (Euclidian) norm corresponds to the largest stability hypersphere in parameter space. The I(sub infinity) norm corresponds to the largest stability box; it is equal to the inverse of the 'structured singular value with repeated real scalar blocks' found in mu analysis. The formulation is similar to mu analysis, with the new feature that all the input data are kept real. An efficient convex optimization algorithm is introduced to compute an accurate lower bound for the I(sub infinity) parameter margin. The two-parameter case is solved using a quadratically convergent algorithm, for both the I(sub 2) and I(sub infinity) norms; the special case when the parameters enter linearly in the closed-loop characteristic polynomial is solved exactly for all l(sub q)-norms, providing a generalization of the famous Kharitonov theorem. Several robustness analysis problems can be solved using this approach: discrete-time systems, relative stability, and aperiodicity. For robust control synthesis, it is assumed that the designer can select a finite number of 'most sensitive directions' in the parameter space. A heuristic procedure to select them is given. The robustness index J(sub R) is taken as the inverse of the smallest of the one-parameter margins along these directions. A trade off is made against a linear quadratic Gaussian finite-time performance index J(sub P) by minimizing, for a given trade-off weight of O less than or equal to rho and rho less than 1, the index J = (1-rho)J(sub P) + rho(J(sub R)), using the non-linear optimization routine NPSOL. Numerical synthesis examples illustrate the method. Dissert. Abstr. N91-12289# Instituto de Fisica Teorica, Sao Paulo (Brazil). V. C. Aguilera-Navarro, G. A. Estevez, and R. Guardiola (Granada (DE90-637409; IFT-P-10/89) Avail: NTIS (US Sales Only) HC/MF A03 A variational analysis of the spiked harmonic-oscillator Hamiltonian operator -d(exp 2)/dx(exp 2) + x(exp 2) + I(1+1)/x(exp 2) + lambda (absolute value of x)(exp -alpha), where alpha is a real positive parameter, is reported in this work. The formalism makes use of the functional space spanned by the solutions of the Schroedinger equation for the linear harmonic-oscillator Hamiltonian supplemented by a Dirichlet boundary condition, and a standard procedure for diagonalizing symmetric matrices. The eigenvalues obtained by increasing the dimension of the basis set provides accurate approximations for the ground-state energy of the model system, valid for positive and relatively large values of the coupling parameter lambda. Additionally, a large-coupling perturbative expansion is carried out and the contributions up to fourth order to the ground-state energy are explicitly evaluated. Numerical results are compared for the special case alphaequal 5/2. DOE N91-12290# Illinois Univ., Urbana-Champaign. Dept. of Electrical RELIABLE CONTROL OF DECENTRALIZED SYSTEMS: AN A new method is presented of decentralized linear, time invariant control systems synthesis based on the algebraic Riccati equation (ARE). The basic decentralized design guarantees closed-loop stability and a predetermined level of worst case disturbance attenuation. Certain modifications of the basic design guarantee the stability and disturbance attenuation to be robust despite plant uncertainty or reliable despite control component outages. Other modifications guarantee that a subset of the controllers will be open-loop stable. The derived decentralized control law consists of a full order observer of the plant in each control channel. Each observer includes estimates of the controls generated by the other channels and of plant disturbance inputs, based on its own estimate of the state of the plant. All of the observer gains are computed from the solution of a single Riccati-like algebraic equation, while feedback gains are computed from a state feedback design ARE. The existence of appropriate solutions to the design equations is sufficient to guarantee the various properties of the closed-loop system. A convexity property of a certain matrix Riccati function allows parameterization of families of control laws with the same desired properties. Each value of the parameter results in controller of the same order as GRA the plant. N91-12291# Naval Underwater Systems Center, New London, CT. EVALUATION OF INTEGRALS AND SUMS INVOLVING (SIN(MX)/SIN(x))(SUP n) Albert H. Nuttall 18 May 1990 38 p Sponsored by Office of the Chief of Naval Research, Arlington, VA (AD-A224906; NUSC-TR-8689) Avail: NTIS HC/MF A03 CSCL 12/1 The response of equispaced arrays, either linear, planar, or volumetric, to distributed spatial fields, typically encounters, integrals which involve the kernel sin(Mx)/sin(x) or its square. Since this kernel oscillates rather fast with x for large M and does not decay with x, numerical integration of such functions can be very time consuming. By resorting to Parseval's theorem, such integrals can be significantly simplified, requiring only the Fourier transform of the complementary part of the integrand. This procedure is investigated and applied to several typical examples; programs for the examples are also included. GRA N91-12292# Johns Hopkins Univ., Baltimore, MD. Dept. of Electrical and Computer Engineering. SMOOTHING FOR MULTIPOINT BOUNDARY VALUE MODELS Howard L. Weinert 25 Jul. 1990 21 p (Contract N00014-89-J-1642) (AD-A224917; JHU/ECE-90-02) Avail: NTIS HC/MF A03 CSCL 12/2 Researchers show how to solve the linear least-squares smoothing problem for multipoint boundary value models. The complementary model is derived and is used to determine the Hamiltonian equations for the smoothed state estimate and its error covariance. Stable algorithms are obtained using an invariant imbedding/multiple shooting procedure. GRA N91-12293# Rolls-Royce Ltd., Bristol (England). DATA ANALYSIS L. M. Jenkins Derby, England 6 Dec. 1989 38 p Presented at Institute of Metals Seminar on Characterization of High Temperature Materials. 5: Numerical Techniques, 6 Dec. 1989 (PNR-90749; ETN-90-97954) Copyright Avail: NTIS HC/MF A03 The integrity assessment of a component as a fundamental part of the design process is addressed. Materials data forms one of the major inputs into this assessment. The data includes physical and mechanical properties in addition to life estimation data. In most applications failures are at best inconvenient and usually the consequences are far worse. It is therefore necessary to design components to achieve a minimum life for a given level of risk. This requires the factoring of data from typical properties to minimum values associated with the risk level. ESA N91-12294# Center for Mathematics and Computer Science, Amsterdam (Netherlands). Dept. of Numerical Mathematics. ANALYSIS OF THE CONVERGENCE OF ITERATIVE IMPLICIT AND DEFECT-CORRECTION ALGORITHMS FOR HYPERBOLIC PROBLEMS Jean-Antoine Desideri (Institut National de Recherche d'Informatique et d'Automatique, Valbonne, France) and Pieter W. Hemker Mar. 1990 88 p (CWI-NM-R9004; ETN-90-98059) Copyright Avail: NTIS HC/MF A05 The convergence of unfactored implicit schemes for the solution of the steady discrete Euler equations is studied using first and second order accurate discretizations simultaneously. The close resemblance of these schemes with iterative defect correction is shown. Linear model problems are introduced for the one dimensional and the two dimensional cases. These model problems are analyzed in detail both by Fourier by matrix analyses. The convergence behavior appears to be strongly dependent on a parameter that determines the amount of upwinding in the discretization of the second order scheme. The linear convection problem in one and two dimensions is studied in detail. Differences between the various cases are signaled. Experiments for the Euler equations are shown, including comments on how the theory is well or partially verified depending on the problem. ESA N91-12295# Center for Mathematics and Computer Science, P. A. Zegeling and J. G. Blom Mar. 1990 30 p Submitted for publication (Contract CWI-59-0922) (CWI-NM-R9006; ETN-90-98061) Copyright Avail: NTIS HC/MF A03 The Moving Finite Element (MFE) method for solving parabolic and hyperbolic Partial Differential Equations (PDEs) is studied. To investigate to what extent gradient weighted MFE can be called robust, reliable and effective for the automatic solution of time dependent PDEs in one space dimension, the method is tested extensively on a set of five relevant example problems with a different solution behavior. All tests were carried out using the Backward Differentiation Formula (BDF) time integrator SPGEAR of the existing method-of-lines software package SPRINT. ESA N91-12296# Center for Mathematics and Computer Science, Amsterdam (Netherlands). Dept. of Analysis, Algebra and Geometry. interactions. The classical and quantum versions of the integrable systems are addressed. Results on explicit action-angle transformation which lead particularly to duality relations are given. The quantum translations of duality relations are explained in connection with a description of explicit knowledge concerning the action-angle and joint eigenfunction transform. Explicit knowledge concerning the action-angle and joint eigenfunction transforms is emphasized. Connections which have emerged so far are outlined. ESA N91-12297# Center for Mathematics and Computer Science, THE CENTER MANIFOLD FOR DELAY EQUATIONS IN THE Odo Diekmann and Stephan A. vanGils (Universiteit Twente, (CWI-AM-R9007; ETN-90-98063) Copyright Avail: NTIS HC/MF A03 The center manifold theorem for retarded functional differential equations is stated and proved. The method of proof is based on the variation of constants formula in the framework of dual semigroups. As an application, Hopf bifurcation is addressed. ESA N91-12298# Center for Mathematics and Computer Science, Amsterdam (Netherlands). Dept. of Analysis, Algebra and Geometry. FROM BINARY TO GREY-LEVEL MORPHOLOGY H. J. A. M. Heijmans May 1990 37 p The construction of morphological operators on spaces of grey level functions is studied. The approach adopted is to represent a function by a sequence of threshold sets and to transform this sequence into another one. This can be done by applying the same set operator at any level: the resulting operator is called a flat operator. A different set operator could be applied at each level. Or alternatively, the threshold sets can be transformed recursively, using the outcome at a certain grey level for the computation at the next level. ESA N91-12299# Center for Mathematics and Computer Science, ON CUTTING PLANES AND MATRICES Continuing the work of Chvatal and Gomory, Schrijver proved that any rational polyhedron has finite Chvatal rank. This was extended by Cook, Gerards, Schrijver and Tardos, who proved that this Chvatal rank can be bounded from above by a number only depending on A, so independent of b. A study which aims to show that the latter result can be proved quite easily from the result of Chvatal and Schrijver is presented. ESA N91-12300# Center for Mathematics and Computer Science, Evangelos Kranakis and Michel Pocchiola (Centre National de la Enumeration and visibility problems of the d-dimensional integer lattice L (limits d and n) of d-tuples of integers less than or equal to n are studied. Several useful enumeration principles are given and used to study the asymptotic behavior of the number of straight lines traversing a certain fixed number of lattice vertices of L (limits d and n), the line incidence problem and the edge visiblity region. An art gallery problem for point obstacle is studied. More specifically, the camera placement problem for the infinite lattice L (upper limit d) is studied. A lattice point is visible from a camera C (positioned at a vertex of L (upper limit d)) if the line segment joining A and C crosses no other lattice vertex. For any given number s less than or equal to 3 (upper limit d) of cameras, the position they must occupy in the lattice L (upper limit d) in order to maximize their visibility is determined. N91-12301# ESA Center for Mathematics and Computer Science, Amsterdam (Netherlands). Dept. of Computer Science. DERIVING DENOTATIONAL MODELS FOR BISIMULATION FROM STRUCTURED OPERATIONAL SEMANTICS J. J. M. M. Rutten Dec. 1989 20 p (CWI-CS-R8955; ETN-90-98088) Copyright Avail: NTIS HC/MF A03 As a starting point, the notion of Labelled Transition System (LTS) in the style of the structured operational semantics of Plotkin is taken. Every LTS gives rise to a model that maps the states of the LTS (usually terms over some signature) to a representation of their bisimulation equivalence class, namely a so-called process. (Such a model is often called operational). These processes are elements of a metric domain which was first introduced by De Bakker and Zucker. It is shown how the transition system specification (a set of rules for deriving transitions) by which the LTS is defined, induces a denotational model, given that it satisfies certain syntactic restrictions. Both models are proved equal by showing that they are fixed points of the same contraction, which has a unique fixed point by Banach's theorem. ESA N91-12302# Center for Mathematics and Computer Science, Jan Friso Groote and Frits Vaandrager Jan. 1990 16 P An efficient algorithm for the Relational Coarsest Partition with Stuttering problem (RCPS) is presented. The RCPS problem is closely related to the problem of deciding stuttering equivalence on finite state Kripke structures, and to the problem of deciding branching bisimulation equivalence on finite state labelled transition systems. If n is the number of states and m the number of transitions, then the algorithm has time complexity and stuttering equivalence which have the same complexity O(n(n + m)) and space complexity O(n + m). The algorithm induces algorithms for branching bisimulation and stuttering equivalence which have the same complexity. Since for Kripke structures (m is less than or equal to n squared), this confirms a conjecture of Browne, Clarke and Grumberg, that their O(n to the 5th power) time algorithm for ESA stuttering equivalence is not optimal. N91-12303# Universiteit Twente, Enschede (Netherlands). Dept. of Computer Science. ELEMENTARY STRUCTURES OF NON-ATOMIC EVENTS Event structures based on nonatomic events are discussed. Such structures called here organic event structures, possess not one but two ordering relations, representing precedence and influence. Given a hierarchy of events, organic events structures capture precisely the relation over arbitrary sets of events if the hierarchical structure is forgotten. Elementary (atomic) event structures are not powerful enough for this purpose. Some organic event structures can be interpreted by sets of time intervals. Necessary and sufficient conditions for the existence of interpretations in preordered, partially ordered and totally ordered time, are given. The subclass of organic event structures that can be interpreted by non-overlapping intervals is almost, but not quite, equivalent to the class of atomic event structures. ESA N91-12304# Center for Mathematics and Computer Science, (CWI-NM-R9005; ETN-90-98060) Copyright Avail: NTIS A Petrov-Galerkin form of exponential fitting for a mixed finite element formulation of the semiconductor continuity equations on a rectangular domain is discussed. Conditions, that ensure theta (h) bounds for the L squared (omega) norm of the error are given. The constants in the error bound do not contain exponentials of the electrical potential. ESA N91-12305# Center for Mathematics and Computer Science, ON THE TOPOLOGY INDUCED BY THE ADJOINT OF A J. M. A. M. van Neerven May 1990 17 p The adjoint of C sub 0-semigroup on a Banach space X induces a locally convex sigma(X,X dot circled)-topology in X, which is weaker than the weak topology of X. The relation between these two topologies is studied. Among other things a class of subsets of X is given on which they coincide, an Eberlein-Shmulyan type theorem is proved for the sigma(X,X dot circled)-topology and some applications are given. ESA N91-12307# Tennessee Univ., Knoxville. Dept. of Psychology. DIFFERENTIAL WEIGHT PROCEDURE OF THE CONDITIONAL P.D.F. APPROACH FOR ESTIMATING THE OPERATING CHARACTERISTICS OF DISCRETE ITEM RESPONSES Technical Report, 1987 - 1990 Fumiko Samejima 25 Jun. 1990 35 p (Contract N00014-87-K-0320; NR Proj. RR0-4204) (AD-A224697; ONR-RR-90-4) Avail: NTIS HC/MF A03 CSCL 12/3 A new procedure of nonparametric estimation of the operating characteristics of discrete item responses is proposed, and it is called differential weight Procedure of the Conditional Probability Distribution Function (PDF) Approach. Some examples are given, and sensitivities of the resulting estimated operating characteristics to irregularities of the differential weight functions are observed and discussed. Usefulnesses of the method is also discussed. These outcomes suggest the importance of further investigation of the weight function in the future. GRA |