|
|
|
|
|
|
|
|
F. V. Atkinson, ``Self-Intersecting Solutions of the
Prescribed Mean Curvature Equation,'' Proc. of the Focused Research Program on
Spectral Theory and Boundary Value Problems: Nonlinear Differential Equations, Vol.
4, ANL-87-26 (August 1989), pp. 1-20. |
Inequalities and asymptotic estimates are obtained for
solutions of the prescribed mean curvature equation in a group of cases in which f(u) is
positive, increasing, and unbounded. This system is the intrinsic version of the radially
symmetric prescribed mean curvature equation and permits the study of certain generalized
solutions of this equation. |
|
|
F. V. Atkinson and L. A. Peletier, ``Bounds for Vertical
Points of Solutions of Prescribed Mean Curvature Type Equations,'' Proc. of the
Focused Research Program on Spectral Theory and Boundary Value Problems: Nonlinear
Differential Equations, Vol. 4, ANL-87-26 (August 1989), pp. 21-42. |
Conditions are obtained under which radial solutions of
generalized capillary-type equations exhibit a first vertical point, together with bounds
and asymptotic estimates for these points. |
|
|
F. V. Atkinson and L. A. Peletier, ``Large Solutions of
Elliptic Equations Involving Critical Exponents,'' Proc. of the Focused Research
Program on Spectral Theory and Boundary Value Problems: Nonlinear Differential Equations,
Vol. 4, ANL-87-26 (August 1989), pp. 43-72. |
This paper focuses on obtaining precise estimates of positive
radial solutions of particular elliptic equations. |
|
|
A. Bayliss, B. J. Matkowsky, and M. Minkoff, ``Bifurcation
and Pattern Formation in Combustion,'' Ordinary and Partial Differential Equations,
ed. B. D. Sleeman and R. J. Jarvis, John Wiley and Sons, N.Y., 1989, pp. 1-51. |
The authors have developed an adaptive pseudo-spectral method
that is both very accurate and very efficient. The algorithm allows them to describe the
solution on bifurcation branches in combustion, far beyond the region where analytical
methods work well. The analytical results that they obtain are used to choose appropriate
parameter values and initial conditions for the numerical computations. In addition, the
analytical results serve as benchmarks for the computations. The computations reveal new
and interesting behavior. Two problems are discussed, involving solid and gaseous fuel
combustion. |
|
|
C. H. Bischof, ``Incremental Condition Estimation,'' SIAM J.
Matrix Anal. Appl. 11, no. 2 (April 1990) 312-322. |
This paper introduces a technique for estimating the smallest
singular value of a dense triangular matrix as it is generated one row or column at a
time. Numerical experiments demonstrate that despite its small computational cost, this
condition estimator is very reliable. The paper also gives an example that shows the
advantage of incorporating the incremental condition estimation strategy into the QR
factorization algorithm with column pivoting to guard against failure of the pivoting
strategy going unnoticed. |
|
|
C. H. Bischof and J. J. Dongarra, ``A Linear Algebra Library
for High-Performance Computers,'' Parallel Supercomputing: Methods, Algorithms and
Applications, ed. G. F. Carey, John Wiley and Sons, N. Y., 1989, pp. 45-56. |
Argonne, the Courant Institute, and NAG are developing a
transportable linear algebra library in Fortran 77. The library (called LAPACK) will
provide routines for solving systems of simultaneous linear equations, least-squares
solutions of overdetermined systems of equations, and eigenvalue problems. The new library
will be based on the successful EISPACK and LINPACK libraries. It will be designed to be
efficient and transportable across a wide range of computing environments, with special
emphasis on modern high-performance computers. The new library will be distributed over a
system similar to netlib, at no or minimal cost. |
|
|
R. Butler, I. Foster, A. Jindal, and R. Overbeek, ``A
High-Performance Parallel Theorem Prover,'' Proc. 10th International Conference on
Automated Deduction, in Lecture Notes in Artificial Intelligence, Vol. 449, ed. M. E.
Stickel, Springer-Verlag, Berlin, 1990, pp. 649-650 (also MCS-P139-0290). |
This paper reports on the implementation of a parallel
theorem proving system that supports hyper-resolution, subsumption, and the set of support
strategy. The upper level of the program, written in the concurrent programming language
Strand, implements the general algorithm and describes the interconnectivity of processes
that perform the theorem-proving tasks such as resolution and subsumption. These
compute-intensive tasks are performed by procedures implemented in C. Speedups have been
demonstrated on such test problems as SAM's lemma. |
|
|
M. T. Chu and G. H. Guirguis, ``A Numerical Method for
Solving Interface Problems Arising in Two-Point Boundary Value Problems,'' Computer
Methods in Appl. Mech. and Engrg. 74 (1989) 99-113 (also MCS-P31-1288). |
A numerical approach iterating on the position of interface
points is suggested for solving interface problems arising in two-point boundary value
problems (BVPs). Given an interface problem, it is decomposed into several standard local
boundary value problems that are coupled at the interface points. Meanwhile, a nonlinear
equation involving the interface points is formulated from the interface conditions. The
advantages of this approach are that the boundary conditions for each local problem can
easily be selected by considering the natural physical requirements and that each of the
local problems can be solved independently by standard numerical BVP techniques. |
|
|
W. J. Cody, ELEFUNT Test Results under AST Fortran V1.8.0
on the Sequent Symmetry, MCS-TM-138 (July 1990). |
This report discusses testing of the floating-point
arithmetic and of the elementary function libraries under AST Fortran on a 24-processor
Sequent Symmetry computer. The programs MACHAR and PARANOIA were used to check the quality
of arithmetic, and the ELEFUNT suite of programs from the book Software Manual for the
Elementary Functions by Cody and Waite was used to check function performance. Two
complete sets of tests were run, one for each type of floating-point processor, Intel
80387 and Weitek 1167, on the machine. |
|
|
W. J. Cody, ``Performance Evaluation of Programs for the
Error and Complementary Error Functions,'' TOMS 16 (1990) 29-37 (also MCS-P13-0988). |
This paper presents methods for performance evaluation of
computer programs for the the error and complementary error functions. Accuracy estimates
are based on comparisons using power series expansions and an expansion in the repeated
integrals of erfc (x). Some suggestions for checking robustness are also given. Details of
a specific implementation of a test program are included. |
|
|
W. R. Cowell, ``An Introduction to VecPar_77,'' NAG
Newsletter '90, no. 1 (1990) 13-21. |
VecPar77 is a precompiler tool that analyzes and transforms
Fortran 77 programs intended for execution on machines with vector or shared-memory
multitasking capabilities. In particular, the tool provides support for the task of
modifying existing programs, written originally for serial execution, so that they perform
well on advanced-architecture machines. The operation of VecPar77 is divided into two
separately invoked phases. In Phase 1, two kinds of Fortran program analysis are
performed: syntactic/semantic analysis to detect errors and anomalous constructions in a
program, and Do-loop parallelization analysis. In Phase 2, VecPar77 acts as a specialized
Fortran-intelligent editor, enabling a user to apply source-to-source transformations
aimed at producing Fortran that compiles into code in which the number of references to
main memory is reduced. VecPar77 is a direct descendent of Toolpack. It was created by a
project jointly supported by the Numerical Algorithms Group, Inc., and the Department of
Commerce and Community Affairs of the State of Illinois. |
|
|
W. R. Cowell and C. P. Thompson, ``Tools to Aid in
Discovering Parallelism and Localizing Arithmetic in Fortran Programs,''
Software--Practice and Experience,20 (January 1990) 25-47. |
The authors describe a collection of software tools that
analyze and transform Fortran programs. The analysis tools detect parallelism in blocks of
code and are primarily intended to aid in adapting existing programs to execute on
multiprocessors. The transformation tools are aimed at eliminating data dependencies,
thereby introducing parallelism, and at localizing arithmetic in registers, of primary
interest in adapting programs to execute on machines that can be memory bound (common for
machines with vector architectures). The tools are unified conceptually by their use of a
set of conditions for data independence; these conditions have been implemented so as to
combine tool analysis with user/tool interaction. Included are timing results from
applying the tools to programs intended for execution on two machines with different
architectures--a Sequent Balance and a CRAY-2. The tools are written in Fortran in the
tool-writing environment provided by Toolpack and are easily incorporated into a Toolpack
installation. |
|
|
J. Dongarra and E. Lusk, ``Summer Institute in Parallel
Computing: September 5-15, 1989,'' MCS-TM-136 (December 1989). |
This report summarizes the objectives of the Summer Institute
in Parallel Computing held at Argonne in September 1989. |
|
|
J. J. Dongarra, D. C. Sorensen, and S. J. Hammarling, ``Block
Reduction of Matrices to Condensed Forms for Eigenvalue Computations,'' J. Comp. and Appl.
Math. 27 (1989) 215-227. |
This paper describes block algorithms for the reduction of a
real symmetric matrix to tridiagonal form and for the reduction of a general real matrix
to either bidiagonal or Hessenberg form by using Householder transformations. The approach
is to aggregate the transformations and to apply them in a blocked fashion, thus achieving
algorithms that are rich in matrix-matrix operations. These reductions typically are a
preliminary step in the computation of eigenvalues or singular values. The authors
demonstrate how the initial reduction to tridiagonal or bidiagonal form may be pipelined
with the divide-and-conquer technique for computing the eigensystem of a symmetric matrix
or the singular value decomposition of a general matrix to achieve algorithms that are
load balanced and rich in matrix-matrix operations. |
|
|
I. Foster, ``Automatic Generation of Self-Scheduling
Programs,'' IEEE Trans. on Parallel and Distributed Computing, to appear (also
MCS-P143-0390). |
The author describes techniques for the automatic generation
of self-scheduling parallel programs. A high-level language is used to express both
concurrent components of applications and scheduling algorithms. Partitioning and data
dependency information are expressed by simple control statements. An automatic
source-to-source transformation takes application code, control statements, and scheduling
routines and generates a new program that can schedule its own execution of a parallel
computer. The approach has several important advantages compared to previous proposals. It
generates programs that are portable over a wide range of parallel computers. There is no
need to embed special control structures in application programs. Finally, the use of a
high-level language to express applications, load-balancing algorithms, and
transformations facilitates the development, modification, and reuse of parallel programs.
|
|
|
I. Foster, ``A Multicomputer Garbage Collector for a
Single-Assignment Language,'' International J. Parallel Prog., to appear (also
MCS-P81-0689). |
This paper describes an asynchronous garbage collector for a
message-passing multiprocessor. This combines Weighted Reference Counting interprocessor
collection and tracing intraprocessor collection to permit individual processors to
reclaim local storage independently. A novel feature is the integration of Weighted
Reference Counting collection and the communication algorithms required to support a
global address space in a single assignment language. This significantly reduces
communication overhead and space requirements attributable to garbage collection. |
|
|
I. Foster and R. O. Overbeek, ``Experiences with Bilingual
Parallel Programming,'' Proc. DMCC5 '90, to appear. |
Parallel programming requires tools that simplify the
expression of complex algorithms, provide portability across different classes of machine,
and allow reuse of existing sequential code. The authors have previously proposed
bilingual programming as a basis for such tools. In particular, they have proposed the use
of a high-level concurrent programming language (such as Strand) to construct parallel
programs from (possibly pre-existing) sequential components. Here they report on an
applications study intended to evaluate the effectiveness of this approach. They describe
experiences developing both new codes and parallel versions of existing codes in
computational biology, weather modeling, and automated reasoning. The bilingual approach
encourages the development of parallel programs that perform well, are portable, and are
easy to maintain. |
|
|
I. Foster and R. Stevens, ``Parallel Programming with
Skeletons,'' Proc. ICPP '90, to appear (also MCS-P124-0190). |
The authors describe an approach to parallel program
development based on the use of algorithmic motifs: useful parallel program structures
that users can combine with application-specific routines to construct parallel programs.
Key features are the use of a high-level concurrent programming language and automated
source-to-source transformations to implement motifs. These features support the
construction of new motifs from old by both modification and composition. This is turn
encourages an exploratory programming style, in which programmers experiment to develop
good parallel implementations of applications. The use of a high-level language encourages
the user to view libraries implementing motifs as archives of expertise that can be
consulted, modified, and extended when developing parallel programs. Examples of the
approach are given. |
|
|
I. Foster, C. Kesselman, and S. Taylor, ``Concurrency: Simple
Concepts and Powerful Tools,'' Computer Journal, to appear Dec. 1990. |
This paper describes programming language concepts that we
have found useful in applying stepwise refinement to parallel programs. The concepts allow
decisions concerning program structure to be delayed until late in the design process.
This capability permits rapid experimentation with alternative structures and leads to
both portable and scalable code. Although simple, the concepts form a sufficient basis for
the construction of powerful programming tools. Both concepts and tools have been applied
successfully in a wide variety of applications and are incorporated in a commercial
concurrent programming system Strand. |
|
|
M. Garbey, H. G. Kaper, G. K. Leaf, and B. J. Matkowsky,
``Nonlinear Analysis of Condensed-Phase Surface Combustion,'' Euro. J. Appl. Math. 1
(1990) 73-89 (also MCS-P82-0589). |
This article is concerned with the structure and stability
properties of a combustion front that propagates in the axial direction along the surface
of a cylindrical solid fuel element. The fuel consists of a mixture of two finely ground
metallic powders, which combine upon ignition in a one-step chemical reaction. The
reaction is accompanied by a melting process, which in turn enhances the reaction rate.
The combustion products are in the solid state. The reaction zone, inside which the
melting occurs, is modeled as a front that propagates along the surface of the cylinder.
The different modes of propagation that have been observed experimentally (such as single-
and multiheaded spin combustion and multiple-point combustion) are explained as the
results of bifurcations from a uniformly propagating plane circular front. The stability
properties of the various modes are discussed. |
|
|
A. Griewank, ``On Automatic Differentiation,'' in Mathematical
Programming: Recent Developments and Applications, ed. M. Iri and K. Tanabe, Kluwer
Academic Publishers, Boston, 1989, pp. 83-107 (also MCS-P10-1088). |
In comparison to symbolic differentiation and numerical
differencing, the main rule-based technique of automatic differentiation is shown to
evaluate partial derivatives accurately and cheaply. In particular, it is demonstrated
that the reverse mode of automatic differentiation yields any gradient vector at no more
than five times the cost of evaluating the underlying scalar function. After developing
the basic mathematics, the author describes several software implementations and briefly
discuss the ramifications for optimization. |
|
|
C. P. Gupta, ``A Fourth-Order Nonlinear Boundary Value
Problem and an Integral Type Sign Condition,'' Applied Math. and Computation 35 (1990)
231-242 (also MCS-P72-0489). |
This paper focuses on a fourth-order nonlinear boundary value
problem associated to the unstable static equilibrium of an elastic beam which is
supported by sliding clamps when the nonlinearity satisfies an integral sign condition
instead of the usual asymptotic pointwise sign condition. A necessary and sufficient
condition is given for the existence of a solution when the nonlinearity is nondecreasing.
|
|
|
C. P. Gupta, ``A Nonlinear Boundary Value Problem Associated
with the Static Equilibrium of an Elastic Beam Supported by Sliding Clamps,'' Internat. J.
Math. & Math. Sci. 12, no. 4 (1989) 697-712 (also MCS-P24-1188). |
A fourth-order boundary problem describes the unstable static
equilibrium of an elastic beam that is supported by sliding clamps at both ends. This
paper concerns the nonlinear analogue of this boundary value problem. Some resonance and
nonresonance conditions on the asymptotic behavior are studied for the existence of
solutions of this nonlinear boundary value problem. |
|
|
C. P. Gupta, ``Nonlinear Second Order System of Neumann
Boundary Problems at Resonance,'' JAMS 2, no. 3 (Fall 1989) 169-184 (also MCS-P74-0489). |
This paper focuses on a nonlinear Neumann boundary value
problem at resonance. Asymptotic conditions on the nonlinearity are presented to give
existence of solutions for the nonlinear systems. The methods apply to the corresponding
system of Lienard-type periodic boundary value problems. |
|
|
H. G. Kaper and A. M. Krall, ``M(lambda)-Computation for
Singular Differential Systems,'' Proc. Roy. Soc. Edinburgh 112A (1989) 327-330. |
Depending on the initial data associated with the fundamental
matrix, the function M(lambda) may vary. It is shown that there is a matrix bilinear
transformation between such functions M(lambda) with different initial data. The authors
also show how the result can be used to simplify the calculation of a specific
M(lambda)-function for a scalar second-order problem. |
|
|
H. G. Kaper and Man Kam Kwong, ``Free Boundary Problems for
Emden-Fowler Equations,'' Diff. and Integral Eqs. 3, no. 2 (March 1990) 353-362 (also
MCS-P8-0988). |
This article is concerned with free boundary problems for
differential equations of the Emden-Fowler type. In particular, it addresses the questions
of existence and uniqueness of a finite point P and a solution u satisfying conditions
u'(0) = 0; u(x) > 0, 0 < P ; u (P)="u'(P)" |
|
|
H. G. Kaper and M. K. Kwong, ``A Non-Oscillation Theorem for
the Emden-Fowler Equation; Ground States for Semilinear Elliptic Equations with Critical
Exponents,'' J. Diff. Eqs. 75 (1988) 158-185 (also Proc. of the Focused Research Program
on Spectral Theory and Boundary Value Problems: Nonlinear Differential Equations, Vol. 4,
ANL-87-26 (August 1989), pp. 141-167). |
The authors show that the Emden-Fowler equation is
nonoscillatory at infinity under certain conditions. The result improves earlier
oscillation results of Nehari and Chiou. |
|
|
M. K. Kwong, ``On the Kolodner-Coffman Method for the
Uniqueness Problem of Emden-Fowler BVP,'' J. App. Math. Phys (ZAMP) 41 (January 1990)
79-104 (also MCS-P44-1089). |
This paper surveys the Kolodner-Coffman method which has been
applied very successfully by many authors to obtain uniqueness theorems for boundary value
problems for differential equations of Emden-Fowler type. The author recasts some of the
arguments in the language of Sturm's oscillation theory for linear second-order
differential equations. The clarification provided by the new perspective makes it
possible to improve on several known results. |
|
|
M. K. Kwong and M. B. Dever, ``Computer-aided Study of a
Problem in Hermitian Matrix Theory,'' J. Symbolic Computation 9 (1990) 87-112. |
|
|
|
E. Lusk et al., ``The Aurora Or-Parallel Prolog System,'' New
Generation Computing 7 (1990) 243-271 (a modified version of MCS-P122-0190). |
Aurora is a prototype or-parallel implementation of the full
Prolog language for shared-memory multiprocessors, developed as part of an informal
research collaboration known as the Gigalips Project. It currently runs on Sequent and
Encore machines. It has been constructed by adapting Sistus Prolog, a fast, portable,
sequential Prolog system. The techniques for constructing a portable multiprocessor
version follow those pioneered in a predecessor system, ANL-WAM. The SRI model was adopted
as the means to extend the Sistus Prolog engine for or-parallel operation. This paper
describes the design and main implementation features of the current Aurora system, and
presents some experimental results. For a range of benchmarks, Aurora on a 20-processor
Sequent Symmetry is 4 to 7 times faster than Quintus Prolog on a Sun 3/75. Good
performance is also reported on some large-scale Prolog applications. |
|
|
J. N. Lyness and I. H. Sloan, ``Some Properties of Rank-2
Lattice Rules,'' Math. Comp. 53, no. 188 (October 1989) 627-637. |
This paper discusses lattice rules in general. In particular,
the authors categorize a special subclass, whose leading one- and two-dimensional
projections contain the maximum feasible number of abscissas. They show that rules of this
subclass can be expressed in a simple tricycle form. |
|
|
J. N. Lyness and T. Sorevik, ``The Number of Lattice Rules,''
BIT 29 (1989) 527-534 (also MCS-P17-1088). |
A lattice rule is a quadrature rule for integration over an
s-dimensional hypercube that employs N abscissas located on a lattice, chosen to conform
to certain specifications. In this paper, the authors determine the number of distinct
N-point s-dimensional lattice rules. |
|
|
W. McCune, ``OTTER 2.0,'' Proc. 10th International
Conference on Automated Deduction, in Lecture Notes in Artificial Intelligence, Vol.
449, ed. M. E. Stickel, Springer-Verlag, Berlin, 1990, pp. 663-664. |
OTTER 2.0 is Argonne's current production theorem prover. It
is a descendent of AURA and ITP, and operates on clauses in batch mode. OTTER's strong
points include hyperresolution and UR-resolution in small (i.e., one with few axioms)
theories represented by Horn sets and near-Horn sets. Another strong point is
paramodulation with sets of unit equalities. The program is free and very portable; it is
available electronically as well as on tape or diskette. No formal support is provided,
however. |
|
|
W. W. McCune, ``OTTER 2.0 Users Guide,'' ANL-90/9 (March
1990). |
OTTER is a resolution-style theorem-proving program for
first-order logic with equality. It includes the inference rules binary resolution,
hyperresolution, UR-resolution, and binary paramodulation. Some of its other abilities are
conversion from first-order formulas to clauses, forward and back subsumption, factoring,
weighting, answer literals, term ordering, forward and back demodulation, and evaluable
functions and predicates. OTTER is coded in C and is portable to a wide variety of
computers. |
|
|
J. J. More', ``Gradient Projection Techniques for Large-Scale
Optimization Problems,'' Proceedings of the 28th IEEE Conference on Decision and Control,
Tampa, Florida, December 1990, Volume 1, pp. 378-381. |
We briefly describe the impact of the identification
properties of the gradient projection method on algorithms for the solution of large-scale
optimization problems with bound constraints. In particular, we describe recent results of
More' and Toraldo which show that the gradient projection method can be combined with the
conjugate gradient method to solve an important class of large-scale (number of variables
between 5000 and 15000) bound constrained quadratic programming problems with a few (less
than 15) iterations. |
|
|
J. J. More', ``A Collection of Nonlinear Model Problems,''
Lectures in Applied Math. 26 (1990) 723-762 (also MCS-P60-0289). |
This paper presents a collection of nonlinear problems. The
aim of this collection is to provide model problems of scientific interest to researchers
interested in algorithm development. The problems and their origin are described, and
references are given to other papers that have dealt with these problems. |
|
|
G. W. Pieper, ed., Activities and Operations of the
Advanced Computing Research Facility: January 1989 through January 1990, ANL-90/12
(February 1990). |
This report highlights the activities and operations that
have contributed to the success of the ACRF during the period January 1989 through January
1990. |
|
|
G. W. Pieper, ed., Research in Mathematics and Computer
Science at Argonne: January 1988 - August 1989, ANL-89/12 (August 1989). |
This report highlights the research activities of the MCS
Division during the period January 1988 through August 1989. |
|
|
G. W. Pieper, ed., Joint Japanese-American Workshop on
Future Trends in Logic Programming, ANL-89/43 (December 1989)(280 pages). |
This report summarizes the activities related to the Joint
Japanese-American Workshop on Future Trends in Logic Programming, held at Argonne on
October 11-13, 1989. Four main areas were considered: multiprocessing, constraint logic
programming, logical foundation of programming, and knowledge representation and database.
This report includes abstracts of the 16 presentations, copies of the papers presented,
and summaries of the discussion sessions. An important part of the workshop was the
demonstration of the Japanese fifth-generation computer; included in this report is a
detailed guide to the Japanese demonstration. |
|
|
J. K. Slaney and E. L. Lusk, ``Parallelizing the Closure
Computation in Automated Deduction,'' Proc. 10th International Conference on Automated
Deduction, in Lecture Notes in Artificial Intelligence, Vol. 449, ed. M. E. Stickel,
Springer-Verlag, Berlin, 1990, pp. 28-39 (also MCS-P123-0190). |
This paper presents a parallel algorithm for computing the
closure of a set under an operation. This type of computation has been used in automated
theorem proving, abstract algebra, and formal logic. The algorithm is particularly suited
for shared-memory parallel computers, where it makes possible economies of space.
Implementations of the algorithm in two application contexts are described and
experimental results given. |
|
|
I. H. Sloan and J. N. Lyness, ``Lattice Rules: Projection
Regularity and Unique Representations,'' Math. of Comp. 54, no. 190 (April 1990) 649-660
(also MCS-P75-0489). |
This paper introduces a unique characterization for lattice
rules that are projection regular. |
|
|
P. T. T. Tang, Some Software Implementations of the
Functions Sine and Cosine, ANL-90/3 (April 1990). |
This document presents several software implementations of
the elementary functions sin and cos designed to fit a large class of machines.
Implementation details are provided. Also provided is a detailed error analysis that
bounds the errors of these implementations, over the full range of input arguments, from
0.721 to 0.912 units in the last place. Tests performed on these codes give results that
are consistent with the error bounds. |
|
|
S. Winker, R. Overbeek, C. R. Woese, G. J. Olsen, and N.
Pfluger, An Automated Procedure for Covariation-based Detection of RNA Structure, ANL-89/42
(December 1989). |
This paper summarizes investigations into the computational
detection of secondary and tertiary structure of ribosomal RNA. The authors have developed
a new automated procedure that not only identifies potential bondings of secondary and
tertiary structure, but also provides the covariation evidence that supports the proposed
bondings, and any counterevidence that can be detected in the known sequences. A small
number of previously unknown bondings have been detected in individual RNA molecules
through the use of the new procedure. |
|
|
L. Wos, ``Applications of Automated Reasoning,'' Knowledge
Engineering, Vol. II: Applications, <\em> ed. H. Adeli, McGraw-Hill, New York, 1990, pp. 56-83. |
This article focuses on how the field of automated reasoning
was founded, why it was extended, what caused it to grow, and what successes have been
realized by using automated reasoning programs. |
|
|
L. Wos, S. Winker, W. McCune, R. Overbeek, E. Lusk, R.
Stevens, and R. Butler, ``Automated Reasoning Contributes to Mathematics and Logic,'' Proc.
10th International Conference on Automated Deduction, <\em> in Lecture Notes in Artificial
Intelligence, Vol. 449, <\em> ed. M. E. Stickel, Springer-Verlag, Berlin, 1990, pp. 485-499. |
This article presents results of research focusing on the use
of the automated reasoning program OTTER to prove theorems from Robbins algebra,
equivalential calculus, implicational calculus, combinatory logic, and finite semigroups.
Included are answers to open questions and new shorter and less complex proofs to known
theorems. To obtain these results, the authors used a paradigm that relies heavily on
demodulation, subsumption, set of support, weighting, paramodulation, hyperresolution, and
UR-resolution. Examples are given that show success when the paradigm is used, failure
when it is not. |
|
|
L. Wos, ``Meeting the Challenge of Fifty Years of Logic,''
JAR 6, no. 2 (1990) 213-232 (also MCS-P129-0290). |
This paper discusses how experiments with a new
representation and a new use of the weighting strategy resulted in obtaining proofs of 13
theorems in logic. The author presents these theorems as challenging problems to test the
power of a reasoning program or to evaluate the effectiveness of a new idea. In addition,
the author discusses a possible approach to finding shorter proofs and presents new
results. |
|
|
L. Wos, ``The Problem of Choosing between Logic Programming
and General-Purpose Automated Reasoning,'' JAR 6, no. 1 (1990) 77-78 (also MCS-P130-0290). |
This article focuses on finding criteria for correctly
choosing between using logic programming and a more general automated reasoning approach
to attack a given assignment. The problem proposed for research asks one to find criteria
that classify problems as solvable with a well-focused algorithm or as requiring a more
general search for new information. Included are suggestions for evaluating a proposed
solution to this research problem. |
|
|
L. Wos, ``The Problem of Finding a Mapping between Clause
Representation and Natural-Deduction Representation,'' JAR 6, no. 2 (1990) 211-212 (also
MCS-P133-0290). |
The focus of this paper is on the relationship between two
approaches to the automation of reasoning: the paradigm based on the clause language and
that based on natural deduction. The problem proposed for research asks one to find a
mapping between the two approaches such that reasoning programs based on either approach
perform identically on a specific assignment. For evaluating a proposed solution to this
research problem, the author includes suggestions concerning possible test problems. |
|
|
L. Wos, ``The Problem of Guaranteeing the Existence of a
Complete Set of Reductions,''JAR 5, no. 3 (1989) 399-401 (also MCS-P66-0389). |
This article focuses on finding criteria for guaranteeing the
existence of a complete set of reductions. Included is a suggestion for evaluating a
proposed solution to this research problem. |
|
|
|
|
|
|
C. H. Bischof and P. C. Hansen, ``Structure-Preserving and
Rank-Revealing QR-Factorization,'' MCS-P100-0989. |
The rank-revealing QR factorization is a special QR
factorization guaranteed to reveal the numerical rank of the matrix in consideration.
Thus, it is useful in the numerical treatment of rank-deficient least-squares problems.
This paper presents a new, efficient method for computing such a factorization, which
seeks to preserve the structure and sparsity of the matrix as much as possible while
retaining the ability to safely capture the numerical rank. The method uses incremental
condition estimation and restricted pivoting in a forward pass to compute an approximation
to the numerical rank and null space. This pass can be tailored for the zero/nonzero
structure of the problem at hand. In a second pass, the algorithm of Chan is used to
refine the estimates for the numerical rank and null space. Numerical experiments show
that the new algorithm compares favorably with existing algorithms, in terms of both
fill-in and the number of floating-point operations and memory accesses required. |
|
|
C. H. Bischof and G. M. Shroff, ``On Updating Signal
Subspaces,'' MCS-P101-0989. |
This paper presents an algorithm to adaptively solve signal
processing problems using the signal subspace approach. The noise subspace is estimated
using a rank-revealing QR factorization instead of the more expensive singular value or
eigenvalue decompositions. The authors show how to inexpensively update a rank-revealing
triangular factorization of a matrix when new rows are added and old rows deleted. An
incremental condition estimation technique is used to exploit the structure of past
solutions during the updating process while maintaining the rank-revealing property of the
factorization. |
|
|
J. J. More', ``Gradient Projection Techniques for Large-Scale
Optimization Problems,'' MCS-P102-0989. |
This paper briefly describes the impact of the identification
properties of the gradient projection method on algorithms for the solution of large-scale
optimization problems with bound constraints. In particular, the authors describe recent
results of More' and Toraldo that show that the gradient projection method can be combined
with the conjugate gradient method to solve an important class of large-scale (number of
variables between 5000 and 15000) bound-constrained quadratic programming problems with a
few (less than 15) iterations. |
|
|
A. K. Singh and R. Overbeek, ``Derivations of Efficient
Parallel Programs: An Example from Genetic Sequence Analysis,'' MCS-P104-0989. |
Implementation issues such as synchronization of shared data,
implementation of abstract data types, and scheduling of processes are usually not
addressed in the formal derivation of parallel programs. Here the authors seek to redress
the situation by considering these issues in the context of developing an efficient
implementation of an actual parallel program. The computational problem selected was
motivated by work in aligning sequences of genetic material. The authors first develop an
algorithm in Unity and then investigate the issues that arose in producing an efficient C
implementation. Additionally, the authors present some theorems about program refinements
and discuss the usefulness of the theorems in the context of refining the original Unity
program. |
|
|
C. H. Bischof and J. J. Dongarra, ``A Project for Developing
a Linear Algebra Library for High-Performance Computers,'' MCS-P105-0989. |
This paper summarizes the objectives of the LAPACK project,
its use of the BLAS and block algorithms, documentation issues, and the programming
language and style. The library is intended to be usable on all of the most powerful
computers currently available for general-purpose scientific computing. |
|
|
C. H. Bischof, J. G. Lewis, and D. J. Pierce, ``Incremental
Condition Estimation for Sparse Matrices,'' MCS-P106-0989. |
Incremental condition estimation provides an estimate for the
smallest singular value of a triangular matrix. In particular, it gives a running estimate
of the smallest singular value of a triangular factor matrix as the factor is generated
one column or row at a time. An incremental condition estimator for dense matrices was
originally suggested by Bischof. In this paper the scheme is generalized to handle sparse
triangular matrices, especially those that are factors of sparse matrices. Numerical
experiments on a variety of matrices demonstrate the reliability of this scheme in
estimating the smallest singular value. A partial description of its implementation in a
sparse matrix factorization code further illustrates its practicality. |
|
|
M. Garbey, ``Asymptotic Analysis of Singular Perturbation
Problems Governed by a Conservation Law,'' MCS-P107-1089. |
This paper presents an asymptotic analysis of a quasi-linear
parabolic-hyperbolic singular perturbation problem in one dimension. The main part of the
analysis concerns the construction of the transition layers and boundary layers associated
with the discontinuities of a conservation law. An example is given of an oil problem
modeled by Le Fur. |
|
|
G. H. Chisholm, J. Kljaich, B. T. Smith, and A. S. Wojcik,
``A Paradigm for Analyzing Independence in System Behavior by Using Formal Methods,''
MCS-P108-1089. |
This paper introduces a technique based on formal methods,
which analyzes dependence by generating dependency lists. Each list represents all the
components on which a replicated channel depends for proper operation. The analysis
manipulates these lists to determine independence. The technique treats hardware,
software, and the interface between them, with the result that the total system can be
analyzed with respect to independence properties. |
|
|
G. H. Chisholm, B. T. Smith, and A. S. Wojcik, ``Using Formal
Methods to Develop a Paradigm for Software Design,'' MCS-P109-1089. |
This paper is a shortened version of MCS-P108. |
|
|
R. L. Stevens, ``The Interesting Future of Supercomputing,''
MCS-P110-1189. |
This informal discussion of trends in supercomputing looks
briefly at trends in graphics, programming languages, and computer architectures. |
|
|
G. H. Chisholm, B. T. Smith, J. Kljaich, and A. S. Wojcik,
``Formal Analysis of Software in a Fault-Tolerant Environment,'' MCS-P111-1189. |
This paper describes the application of modeling and analysis
techniques to software that is designed to execute on the Charles Stark Draper laboratory
Fault-Tolerant Processor (FTP). The software performs sensor validation of independent
signals from the Experimental Breeder Reactor-II operated by Argonne. The authors describe
an abstraction of the data flow in terms of the specification of a generic application
program for the FTP. They prove the fault-tolerance of the application program to hardware
and sensor failures, provided that the validation algorithms have a certain generic
structure and functionality. Using a more detailed specification of the application
software, they also demonstrate that this program satisfies the sufficient conditions
developed for the generic program to claim system fault tolerance. |
|
|
G. H. Chisholm, B. T. Smith, and A. S. Wojcik, ``An Automated
Reasoning Problem Associated with Proving Claims about Programs Using Floyd-Hoare
Inductive Assertion Methods,'' MCS-P112-1189. |
Proving claims about behavior of software is essential for
the qualification of computer-based systems used in the control of nuclear reactors. Here
the authors select one of the verification conditions for a C program that initializes an
array to zero. They add assertions about the initial conditions and state of the program
and about the expected behavior of the program in terms of its state. The modeling and
specification technique is the inductive assertion technique of Floyd-Hoare. The program
with assertions is then transformed by the source-to-source program transformation system
TAMPR into a set of separate verification conditions to be proven by the automated
reasoning system. |
|
|
M. K. Kwong, P.-Y. Woo, and R. Wong, ``Using MAPLE for
Symbolic Computations in Robotics,'' MCS-P113-1289. |
Programs are presented for deriving forward kinematics,
Jacobians, and dynamic equations of robotic manipulators in symbolic closed forms by using
the algebraic manipulation software MAPLE. Examples are given, and efficiency of the
computations is discussed. |
|
|
M. Garbey, H. G. Kaper, G. K. Leaf, and B. J. Matkowsky,
``Using MAPLE for the Analysis of Bifurcation Phenomena in Condensed-Phase Surface
Combustion,'' MCS-P114-1189. |
This article describes the use of the symbolic computation
language MAPLE for the analysis of bifurcation phenomena in condensed-phase combustion.
The physical problem concerns the structure and stability properties of a combustion front
that propagates in the axial direction along the surface of a cylindrical solid fuel
element. Experimental observations suggest that the front may propagate in a number of
different ways; the objective of the investigation is to describe these different modes.
The analysis involves the study of a set of nonlinear partial differential equations which
describe the structure and evolution of the combustion front. Because the location of the
front is unknown and must be found as part of the solution, the problem is a free boundary
problem. |
|
|
J. Dongarra, O. Brewer, S. Fineberg, and J. A. Kohl, ``A Tool
to Aid in the Design, Implementation, and Understanding of Matrix Algorithms for Parallel
Processors,'' MCS-P115-1189. |
This paper discusses a tool that aids in the design,
development, and understanding of parallel algorithms for high-performance computers. The
tool provides a vehicle for studying memory access patterns, different cache strategies,
and the effects of multiprocessors on matrix algorithms in a Fortran setting. Such a tool
puts the user in a better position to understand where performance problems may occur and
enhances the likelihood of increasing the program's performance before actual execution on
a high-performance computer. |
|
|
H. G. Kaper and M. K Kwong, ``Ground States of Semi-Linear
Diffusion Equations,'' MCS-P116-1189. |
This article is concerned with the uniqueness of nontrivial
nonnegative solutions of semilinear diffusion equations in radially symmetric domains. |
|
|
M. K. Kwong and L. Zhang, ``Uniqueness of the Positive
Solution of DELTA u + f(u) = 0,'' MCS-P117-1289. |
This paper presents an extension of the recent result of
Kwong on the uniqueness of the positive radial solution of a semilinear elliptic equation.
When reduced to the special case considered by Kwong, the proof is shorter. |
|
|
M. K. Kwong, ``Uniqueness Results for Emden-Fowler Boundary
Value Problems,'' MCS-P118-0190. |
This paper establishes new uniqueness results for boundary
value problems of the superlinear Emden-Fowler type, with either a Dirichlet or Neumann
condition at each endpoint. The first result extends a known criterion to nonlinear terms
that may change sign. The proof uses the theory of differential inequalities, after
changing the independent variable to the quantity u. |
|
|
N. T. Karonis, ``Timing Parallel Programs That Use Message
Passing,'' MCS-P119-0190. |
A timing methodology is presented that is capable of
accurately determining the execution time of parallel programs, This timing methodology is
also capable of acquiring timing data when the number of processes exceeds the number of
physical CPUs. Through simulation this methodology also provides a means by which to study
the behavior of a parallel program under an arbitrary hardware/software communications
configuration. |
|
|
D. Solow and R. S. Womersley, ``A Piecewise Linear Concave
Minimization Algorithm for the Linear Complementarity Problem,'' MCS-P120-0190. |
This paper considers solving the Linear Complementary Problem
by finding a global minimizer of a linearly constrained piecewise linear concave function.
In contrast to Lemke's algorithm and the principal pivoting algorithms, the algorithm
described here finds a feasible point for the linearity constraints first, then works
toward satisfying the complementarity condition. Once an extreme point of the feasible
region is found, the algorithm moves from one extreme to another while reducing this
concave objective. |
|
|
A. Jindal, R. Overbeek, and W. C. Kabat, ``Exploitation of
Parallel Processing for Implementing High-Performance Deduction Systems,'' MCS-P121-0190. |
This paper describes a scheme for parallelizing first-order
logic deduction systems. This scheme has been used for parallelizing OTTER. The new
program, PARROT-II, has attained real speedups in excess of 20 over the best results of
current sequential deduction systems. |
|
|
L. M. Ewerbring and F. T. Luk, ``The HK Singular Value
Decomposition of Rank-Deficient Matrix Triplets,'' MCS-P125-0190 |
(to appear in proceedings, published by Springer-Verlag).
This paper focuses on a simultaneous reduction of three matrices. The described method is
extended from earlier work by Ewerbring and Luk to include rank-deficient data. it is
shown how, via an initial reduction, the problem becomes one of diagonalizing a product of
three matrices. Three different algorithms are compared. |
|
|
N. R. Nagarajan, A. S. Cullick, and A. Griewank, ``New
Strategy for Phase Equilibrium and Critical Point Calculations by Thermodynamic Energy
Analysis. Part I. Stability Analysis and Flash,'' MCS-P126-0190. |
The use of equations of state to compute phase behavior of
petroleum reservoir fluids and their mixtures with injected fluids is well established in
the oil industry. The first step prior to performing a two- or multiphase flash
calculation for a reservoir fluid is to determine the stable phase condition of the
mixture. The stability analysis presented here is based on work by Michelsen but uses a
different set of primary variables, the component molar densities in place of mole numbers
of mole fractions to improve the numerical stability of the Newton iterative scheme. The
new procedure is especially well suited to predict behavior in the critical region where
other methods sometimes face convergence problems. |
|
|
N. R. Nagarajan, A. S. Cullick, and A. Griewank, ``New
Strategy for Phase Equilibrium and Critical Point Calculations by Thermodynamic Energy
Analysis. Part II. Critical Point Calculations,'' MCS-P127-0190. |
In enhanced oil recovery processes, miscibility is developed
through a multiple contact process via a critical point. Prediction of these critical
points is important in modeling the phase behavior of the fluid mixtures for simulating
these processes. Tradition equation-of-state prediction of the critical points involves
complex computations and is often made difficult because of the extreme sensitivity of
fluid properties to small changes in pressure or temperature in the critical region. The
authors of this paper present a new strategy for critical point search based on the Gibbs
criteria for criticality and thermodynamic energy analysis. |
|
|
X. Yang, N. J. Zabusky, and I-L. Chern, ``Breakthrough via
Dipolar-Vortex/Jet Formation in Shock-Accelerated Density-Stratified Layers,''
MCS-P128-0290. |
In this paper the authors observe and quantify convective
``breakthrough'' after an inclined planar light fluid layer is struck by a shock wave. The
results of direct numerical simulations are interpreted, and a convective breakthrough
time is quantified. Initially, vorticity of opposite sign is deposited on each interface
of the layer. Breakthrough occurs when these vortex regions approach and bind into a
dipolar vortex that moves away from the wall. Variations of the breakthrough time with the
Mach number of the incoming shock wave are given. |
|
|
J. V. Burke, ``A Robust Trust Region Method for Constrained
Nonlinear Programming Problems,'' MCS-P131-0190. |
This paper presents a general framework for trust region
algorithms for constrained problems that does not require strong global regularity and
that allows very general constraints. The approach is modeled on the one given by Powell
for convex composite optimization problems and is driven by linear subproblems that yield
viable estimates for the value of an exact penalty parameter. These results are applied to
the Wilson-Han-Powell SQP algorithm and Fletcher's algorithm. Local convergence results
are also given. |
|
|
A. Griewank, ``Direct Calculation of Newton Steps without
Accumulating Jacobians,'' MCS-P132-0290. |
In computational practice, nonlinear vector functions whose
roots must be calculated are often specified in the form of an evaluation program supplied
by the user. This program usually generates a large number of intermediate quantities that
are local variables and thus ordinarily not visible to the calling routine. By considering
the intermediate quantities as additional independent variables, one obtains an extended
nonlinear system, which should be attacked directly for several reasons. First, the
extended Jacobian is extremely sparse and can be evaluated at little extra cost
simultaneously with the original vector function. Second, any method for computing Newton
steps on the original vector function is, in terms of complexity, equivalent to a scheme
for solving a linear system in the extended Jacobian. The conditioning of this large
sparse matrix reflects not only the stability of the original vector function but also the
accuracy of the evaluation algorithm. This allows the design of realistic stopping
criteria without user-supplied tolerances. |
|
|
I-Liang Chern, ``Multiple-Mode Diffusion Waves for Viscous
Nonstrictly Hyperbolic Conservation Laws,'' MCS-P134-0290. |
This paper focuses on the large-time behavior of solutions of
viscous conservation laws whose inviscid part is a nonstrictly hyperbolic system. The
initial data considered here is a perturbation of a constant state. It is shown that the
solutions converge to single-mode diffusion waves in directions of strictly hyperbolic
fields, and to multiple-mode diffusion waves in directions of nonstrictly hyperbolic
fields. The multiple-mode diffusion waves, which are the new elements here, are the
self-similar solutions of the viscous conservation laws projected to the nonstrictly
hyperbolic fields, with the nonlinear fluxes replaced by their quadratic parts. |
|
|
D. Jacobs, A. Langen, and W. Winsborough, ``Multiple
Specialization of Logic Programs with Run-Time Tests,'' MCS-P135-0290. |
This paper presents a framework for multiple specialization
of logic programs that incorporates run-time testing. This framework supports
specialization based on a compiler-generated ``wish list'' of requirements that enable
useful optimizations. In addition, entry mode declarations are used to restrict the class
of reachable activations. Our goal is to generate code containing tests at outer levels of
the call tree that guard high-performance specialized procedures that are likely to be
called. |
|
|
W. McCune, ``Skolem Functions and Equality in Automated
Deduction,'' MCS-P136-0290. |
This paper presents a strategy for restricting the
application of the inference rule paramodulation. The strategy applies to problems in
first-order logic with equality and is designed to prevent paramodulation into subterms of
Skolem expressions. A weak completeness result is presented (the functional reflexive
axioms are assumed). Experimental results on problems in set theory, combinatory logic,
Tarski geometry, and algebra show that the strategy can be useful when searching for
refutations and when applying Knuth-Bendix completion. The emphasis of the paper is on the
effectiveness of the strategy rather than on its completeness. |
|
|
A. Mulkers, W. Winsborough, and M. Bruynooghe, ``Analysis of
Shared Data Structures for Compile-Time Garbage Collection in Logic Programs,''
MCS-P137-0290. |
One of the central problems in program analysis for
compile-time garbage collection is detecting the sharing of term substructure that can
occur during program execution. The authors present an abstract domain for representing
possibly shared structures and an abstract unification operation based on this domain.
When supplied to an abstract interpretation framework, this domain induces a powerful
analysis of shared structures. |
|
|
I-Liang Chern, ``Large-Time Behavior of Solutions of
Lax-Friedrichs Finite Difference Equations for Hyperbolic Systems of Conservation Laws,''
MCS-P138-0290. |
This paper focuses on the large-time behavior of discrete
solutions of the Lax-Friedrichs finite-difference equations for hyperbolic systems of
conservation laws. The initial data considered here is small and tends to a constant state
at x = +- infinity. The author shows that the solutions tend to discrete diffusion waves
which can be constructed from the self-similar solutions of the heat equation and the
Burgers equation through an averaging process. |
|
|
J. J. More', ``On the Performance of Algorithms for
Large-Scale Bound-Constrained Problems,'' MCS-P140-0290. |
The paper discusses issues that affect the performance of
algorithms for the solution of large-scale bound-constrained problems on parallel
computers. The discussion centers on the solution of the elastic-plastic torsion problem
and the journal bearing problem. These two problems are model large-scale quadratic
programming problems that arise as finite element approximations to elliptic variational
inequalities. Performance issues are illustrated with the GPCG algorithm of More' and
Toraldo. This algorithm uses the gradient projection method to select an active set and
the conjugate gradient method to explore the active set defined by the current iterate.
Significant improvements in performance are obtained with the GPCG algorithm by using
partitioning techniques in a parallel environment. These partitioning techniques lead to
almost linear speedups on function-gradient evaluations and Hessian-vector products for
partially separable functions. |
|
|
G.-Q. Chen, ``Limit Behavior of Approximate Solutions to
Conservation Laws,'' MCS-P141-0390. |
This paper focuses on the limit behavior of approximate
solutions to hyperbolic systems of conservation laws. Several mathematical compactness
theories and their role are described. Some recent and ongoing developments are reviewed
and analyzed. |
|
|
H. G. Kaper, M. Knaap, and M. K. Kwong, ``Existence Theorems
for Second-Order Boundary Value Problems,'' MCS-P142-0390. |
This article is concerned with the existence of positive
solutions of boundary value problems for nonlinear second-order differential equations
that arise, for example, in the study of reaction -diffusion equations in annular regions.
|
|
|
P. T. P. Tang, ``Table-driven Implementation of the Expm1
Function in IEEE Floating-Point Arithmetic,'' MCS-P144-0390. |
Algorithms and implementation details for the Expm1 function
in both single and double precision of IEEE 754 arithmetic are presented. With a table of
moderate size, the implementations need only working-precision arithmetic and are provably
accurate to within 0.58 ulp. |
|
|
L. Wos, ``The Problem of Finding a Semantic Strategy for
Focusing Inference Rules,'' MCS-P145-0390. |
The problem posed for research asks one to find semantic
criteria for selecting the clauses needed to complete the application of an inference
rule. Suggestions for test problems are included. |
|
|
H. Zhang, ``Large-Time Behavior of the Maximal Solution of
[the Singular Diffustion] Equation,'' MCS-P146-0490. |
The singular diffusion equation arises in many areas of
application. This paper considers the large-time behavior of the solution of a more
general equation. It is shown that the solution of the Cauchy problem converges to its
corresponding similarity solution as t -> infinity. Moreover, an iteration process is
used to obtain the optimal rate of convergence. |
|
|
W. D. Gropp and D. E. Keyes, ``Parallel Performance of
Domain-decomposed Preconditioned Krylov Methods for PDEs with Adaptive Refinement,''
MCS-P147-0490. |
Numerical experiments are presented on both shared- and
distributed-memory computers for convection-diffusion problems at modest Peclet (or
Reynolds) numbers, without recirculation. Because of the development of boundary layers,
these problems benefit from local mesh refinement, which is straightforward to accommodate
within the domain decomposition framework in a locally uniform sense, but which introduces
load balancing as a further consideration in choosing the granularity of the
preconditioner. In spite of the tradeoffs, cumulative speedups are obtainable out to at
least medium-scale granularity (up to 64 processors in the tests reported). |
|
|
C. H. Bischof, ``A Block QR Factorization Algorithm for
Rank-Deficient Matrices,'' MCS-P148-0490. |
This paper presents a new algorithm for computing the QR
factorization of a rank-deficient matrix that is well suited for high-performance
machines. The traditional QR factorization algorithm with column pivoting is not well
suited for such environments, since it precludes the use of matrix-matrix operations. The
author suggests a restricted pivoting strategy based on incremental condition estimation
which allows one to formulate a block QE factorization logarithm where the bulk of the
work is in matrix-matrix operations. Performance results on the CRAY 2, CRAY X-MP, and
CRAY Y-MP show that the new algorithm performance significantly better than the
traditional scheme and can more than halve the cost of computing the QR factorization. |
|
|
H. G. Kaper and M. K. Kwong, ``A Free Boundary Problem
Arising in Plasma Physics,'' MCS-P149-0490. |
This paper is concerned with the existence and uniqueness of
free boundary problems for certain differential equations. |
|
|
C. Bischof, ``Fundamental Linear Algebra Computations on
High-Performance Computers,'' MCS-P150-0490. |
This paper presents an overview of the LAPACK library, a
portable, public-domain library to solve the most common linear algebra problems. This
library provides a uniformly designed set of subroutines for solving systems of
simultaneous linear equations, least-squares problems, and eigenvalue problems for dense
and banded matrices. We elaborate on the design methodologies incorporated to make the
LAPACK codes efficient on today's high-performance architectures. In particular, we
discuss the use of block algorithms and the reliance on the Basic Linear Algebra
Subprograms (BLAS). We present performance results that show the suitability of the LAPACK
approach for vector uniprocessors and shared-memory multiprocessors. We also discuss some
issues that have to be addressed in implementing such a library on other types of parallel
architectures. These issues involve efficient implementations of the BLAS on parallel
machines, the proper choice of blocking parameters, and the use of parallelism outside of
the BLAS. |
|
|
C. H. Bischof and P. G. Lacroute, ``An Adaptive Blocking
Strategy for Matrix Factorizations,'' MCS-P151-0490. |
This paper presents an adaptive blocking methodology for
determining in a systematic manner an optimal blocking strategy for a uniprocessor
machine. The technique is demonstrated on a block QR factorization routine on a
uniprocessor. After generating timing models for the high-level kernels of the algorithm,
the author shows how one can formulate the optimal blocking strategy in a recurrence
relation that can be solved inexpensively with a dynamic programming technique.
Experiments of one processor of a CRAY-2 show that this strategy is as good as any
fixed-width blocking strategy. Thus, while the optimum fixed-width blocking strategy
cannot be determined unless the problem is run several times, adaptive blocking provides
optimum performance in the very first run. |
|
|
D. C. Sorensen and P. T. P. Tang, ``On the Orthogonality of
Eigenvectors Computed by Divide-and-Conquer Techniques,'' MCS-P152-0490. |
A detailed analysis on the accuracy issues in calculating the
eigensystems of rank-1 perturbed diagonal systems is presented. Such calculations are the
core of the divide-and-conquer technique. Here the authors prove that the computed
eigenvectors are guaranteed orthogonality provided the secular equation is evaluated in a
precision that doubles the&127; working one. An efficient algorithm that simulates
such ``doubled precision'' in working precision is also provided. Numerical results that
confirm the analysis and implementation are presented. |
|
|
G.-Q. Chen, ``The Theory of Compensated Compactness and the
System of Isentropic Gas Dynamics,'' MCS-P154-0590. |
Properties of the system of isentropic gas dynamics,
especially the behavior of entropy, are analyzed and documented. The Lax-Friedrichs
approximations, the Godunov approximations, and the viscosity approximations to the Cauchy
problem for the system are constructed; and a compactness framework for these
approximations is established with the aid of the theory of compensated compactness.
Convergence for the approximations is shown. |
|
|
G.-Q. Chen, ``Global Solutions to the Compressible
Navier-Stokes Equations for a Reacting Mixture,'' MCS-P155-0590. |
In this note the author establishes existence theorems for
global generalized solutions to the compressible Navier-Stokes equations for a reacting
mixture, which describe dynamic combustion. Equivalence of the Navier-Stokes equations in
Euler coordinates and Lagrangian coordinates for the generalized solutions is verified.
The asymptotic behavior obtained from impermeable thermally insulated boundaries and
prescribed data boundaries is identified and proved. |
|
|
M. K. Kwong and Yi Li, ``Uniqueness of Radial Solutions of
Semilinear Elliptic Equations,'' MCS-P156-0590. |
E. Yanaida has proved that the classical Matukuma equation
with a given exponent has only one finite mass solution. In this paper the authors show
how similar ideas can be exploited to obtain uniqueness results for other classes of
equations as well as Matukuma equations with more general coefficients. The key
ingredients of the method are energy functions and suitable transformations. Also studied
are general boundary conditions, using an extension of a recent result by Bandle and
Kwong. Yanaida's proof does not extend to solutions of Matukuma's equation satisfying
other boundary conditions. |
|
|
Per Christian Hansen, ``Analysis of Discrete Ill-Posed
Problems by Means of the L-Curve,'' MCS-P157-0690. |
When discrete ill-posed problems are analyzed and solved by
various numerical regularization techniques, a convenient way to display information about
the regularized solution is to plot the norm or seminorm of the solution vs. the norm of
the residual vector. In particular, the graph associated with Tikhonov regularization
plays a central role. This paper advocates the use of this graph in the numerical
treatment of discrete ill-posed problems. The graph is characterized quantitatively, and
important relations between regularized solutions and the graph are derived. The authors
demonstrate that several methods for choosing the regularization parameter are related to
locating a characteristic L-shaped ``corner'' of the graph. |
|
|
M. T. Jones and M. L. Patrick, ``LANZ: Software for Solving
the Large Sparse Symmetric Generalized Eigenproblem,'' MCS-P158-0690. |
This paper describes the package LANZ for solving the large
sparse symmetric generalized eigenproblem. The package has been tested on four different
architectures: Convex 200, CRAY Y-MP, Sun-3, and Sun-4. The package uses a version of
Lanczos' method and is based on recent research into solving the generalized eigenproblem.
|
|
|
J. M. Boyle and T. J. Harmer, ``A Practical Functional
Program for the CRAY X-MP,'' MCS-P159-0690. |
This paper demonstrates how one can have all the advantages
of functional programming (correctness, clarity, simplicity, and flexibility) without any
sacrifice in performance, even for a scientifically significant computation on a
supercomputer. In the example chosen, pure Lisp is used to express the functional
specification for a PDE solver. From this specification, an executable Fortran program is
produced automatically by means of automated program transformations. The generated
program vectorizes on the CRAY X-MP and runs about 13% faster than a hand-written Fortran
program for the same problem. |
|
|
R. Veroff and L. Wos, ``The Linked Inference Principle, I:
The Formal Treatment,'' MCS-P160-0690. |
This paper presents a detailed, formal treatment of the
linked inference principle. The principle is applied to obtain the abstract formulations
of various linked inference rules. Included among such rules are linked UR-resolution,
linked hyperresolution, and linked binary resolution, each of which generalized the
corresponding standard and well-known inference rule. In addition to the formalism, the
paper focuses on the motivation and objectives for the formulation of linked inference
rules. Included are experimental results and numerous examples. |
|
|
G.-Q. Chen and A. Rustichini, ``Global Solutions to a System
of Conservation Laws and Nonzero Sum Dynamic Games,'' MCS-P161-0790. |
This paper uses techniques for solving solutions of systems
of conservation law to solve a problem in game theory. At the same time, it provides an
example of an interesting new application for the theory of systems of conservation laws.
Specifically, the paper extends the idea of viscosity solutions to nonzero sum
differential games. The extensions provide a sound way of defining the concept of value
for a nonzero sum game. They also provide a criterion for selecting among the possible
equilibria of the deterministic game. |
|
|
S. J. Wright, ``An Interior-Point Algorithm for Linearly
Constrained Optimization,'' MCS-P162-0790. |
The author describes an algorithm for optimization of a
smooth function subject to general linear constraints. An algorithm of the gradient
projection class will be used, with the important feature that the projection at each
iteration is performed by using a primal-dual interior point method for convex quadratic
programming. Convergence properties can be maintained even if the projection is done
inexactly in a well-defined way. Higher-order derivative information on the manifold
defined by the apparently active constraints can be used to increase the rate of local
convergence. |
|
|
I. Foster and R. Overbeek, ``Bilingual Parallel
Programming,'' MCS-P163-0790. |
This paper presents an approach to parallel programming based
on bilingual programming. The key is to construct the upper levels of applications in a
high-level language while coding selected low-level components in low-level languages.
This approach permits the advantages of a high-level notation--expressiveness, elegance,
and conciseness--to be obtained without the cost in performance normally associated with
high-level approaches. In addition, it provides a natural framework for reusing existing
code. |
|
|
I. Foster, C. Kesselman, and S. Taylor, ``Concurrency: Simple
Concepts and Powerful Tools,'' MCS-P164-0790. |
This paper describes programming language concepts that have
proved useful in applying stepwise refinement to parallel programs. The concepts allow
decisions concerning program structure to be delayed until late in the design process.
This capability permits rapid experimentation with alternative structures and leads to
both portable and scalable code. The concepts form a sufficient basis for the construction
of powerful programming tools. The concepts and the tools have been applied successfully
in a wide variety of applications and are incorporated in a commercial concurrent
programming system Strand. |
|
|
D. G. Szyld and M. T. Jones, ``Two-Stage and Multi-Splitting
Methods for the Parallel Solution of Linear Systems,'' MCS-P165-0790. |
A two-stage multi-splitting method for the parallel solution
of linear systems is presented. The method reduces to each of the others in specific
cases. Conditions for convergence are given. In the case of a multi-splitting method
related to block Jacobi, the method is equivalent to a two-stage method with only one
iteration per outer iteration. A fixed number p of iterations of this method is compared
with a two-stage method with a p inner iterations. The asymptotic rate of convergence of
the first method is faster, but--depending on the structure of the matrix and the parallel
architecture--it takes more time to converge. Numerical experiments are presented for
illustration. |
|
|
G. M. Shroff and C. H. Bischof, ``Adaptive Condition
Estimation for Rank-One Updates of QR Factorizations,'' MCS-P166-0790. |
In this paper the authors consider general (i.e.,
nonsymmetric) matrices undergoing rank-one changes and develop an adaptive condition
estimation algorithm GRACE to monitor the condition number of the matrices during the
update process. The algorithm requires only O(n) overhead beyond the cost of updating the
QR factorization. Potential numerical difficulties of the algorithm are analyzed, and
modifications are presented to overcome these difficulties. The paper includes
experimental results that demonstrate that GRACE works well in practice. |
|
|
C. H. Bischof, C.-T. Pan, and P. T. P. Tang, ``Stable
Cholesky Up- and Downdating Algorithms for Systolic and SIMD Architectures,''
MCS-P167-0790. |
This paper presents a stable algorithm for maintaining
Cholesky factors of symmetric positive definite matrices under arbitrary rank-one changes.
The algorithm is well suited for implementation on systolic arrays and SIMD architectures.
By exploiting the similarity between the downdating algorithm recently suggested by Pan
and the updating algorithm suggested by Carlson, one can derive an algorithm where up- and
downdates can be pipelined. Implementation results on the 1024-processor AMT DAP 510
emphasize the simplicity and practicality of the proposed scheme. |
|
|
S. Winker, R. Overbeek, C. R. Woese, G. J. Olsen, and N.
Pfluger, ``Structure Detection through Automated Covariance Search,'' MCS-P168-0890. |
This paper summarizes the authors' investigations into the
computational detection of secondary and tertiary structure of ribosomal RNA. They have
developed a new automated procedure that not only identifies potential secondary and
tertiary structural interactions, but also provides the covariation evidence that supports
the proposed bondings, and any counterevidence that can be detected in the known
sequences. A small number of previously unknown higher-order structural features have been
detected in individual RNA molecules (16S rRNA and 7S RNA) through the use of this
automated procedure. |
|
|
C. R. Woese, S. Winker, and R. R. Gutell, ``The Architecture
of Ribosomal RNA: Constraints on the Composition of Tetra-loops,'' MCS-P169-0890. |
The four base loops that cap many double helical structures
in ribosomal RNA (the so-called ``tetra-loops'') exhibit highly invariant to highly
variable compositions depending upon their location in the molecule. However, in the vast
majority of these cases the composition of a tetra-loop is independent of its location and
conforms to one of three general motifs: GNRA, UNCG, and (more rarely) CUUG. For the most
frequently varying of the 16S rRNA tetra-loops, that at position 83 (E. coli) numbering),
the three sequences CUUG, UUCG, and GCAA account for almost all the examples encountered;
and each of them has independently arisen at least a dozen times. The closing base pair of
tetra-loop hairpins reflects the loop composition: tending to be C:G for UUCG loops and
G:C for CUUG loops. |
|
|
M. Garbey, H. G. Kaper, and M. K. Kwong, ``Symbolic
Manipulation Software and the Study of Differential Equations,'' MCS-P170-0890. |
In this note, the authors describe some of their experiences
with symbolic manipulation software for the analysis of differential equations. They give
several examples from their work on semilinear diffusion equations and from a bifurcation
analysis of a nonlinear problem in combustion. |
|
|
A. W. Bojanczyk, M. Ewerbring, F. T. Luk, and P. Van Dooren,
``An Accurate Product SVD Algorithm,'' MCS-P171-0890. |
The authors propose a new algorithm for computing a singular
value decomposition of a product of three matrices. The algorithm is numerically desirable
in that all relevant residual elements are numerically small. |