Ronald F. Boisvert

From scientific software libraries to problem-solving environments

As more scientists and engineers adopt computation as a primary tool, they will want more problem-solving help from easy-to-use, comprehensive software systems. A workshop discussed the long path to this vision of scientific softwares future, and the roadblocks in the way. In order to understand the findings of the workshop, the paper presents some background on software libraries and problem solving environments.

GAMS: a framework for the management of scientific software

The Guide to Available Mathematical Software (GAMS) provides a framework for both a scientist-end-user and a librarian-maintainer to deal with large quantities of mathematical and statistical software. This framework includes a classification scheme for mathematical and statistical software, a database system to manage information about this software, and both an on-line interactive consulting system and a printed catalog for providing users with access to this information. A description is given of GAMS and its use at the National Bureau of Standards.

Developing numerical libraries in Java

SUMMARY The rapid and widespread adoption of Java has created a demand for reliable and reusable mathematical software components to support the growing number of computationally intensive applications now under development, particularly in science and engineering. In this paper we address practical issues of the Java language and environment which have an effect on numerical library design and development. Benchmarks which illustrate the current levels of performance of key numerical kernels on a variety of Java platforms are presented. Finally, a strategy for the development of a fundamental numerical toolkit for Java is proposed and its current status is described. ©1998 John Wiley & Sons, Ltd.

Java and numerical computing

Java represents both a challenge and an opportunity to practitioners of numerical computing. The article analyzes the current state of Java in numerical computing and identifies some directions for the realization of its full potential. Many research projects have demonstrated the technology to achieve very high performance in floating-point computations with Java. Its incorporation into commercially available JVMs is more an economic and market issue than a technical one. The combination of Java programming features, pervasiveness, and performance could make it the language of choice for numerical computing. Furthermore, all Java programmers can potentially benefit from the techniques developed for optimizing Javas numerical performance. The authors hope the article will encourage more numerical programmers to pursue developing their applications in Java. This, in turn, will motivate vendors to develop better execution environments, harnessing Javas true potential for numerical computing.

A System for Performance Evaluation of Partial Differential Equations Software

This paper describes a system to systematically compare the performance of various methods (software modules) for the numerical solution of partial differential equations. We discuss the general nature and large size of this performance evaluation problem and the data one obtains. The system meets certain design objectives that ensure a valid experiment: 1) precise definition of a particular measurement; 2) uniformity in defimition of variables entering the experiment; and 3) reproducibility of results. The ease of use of the system makes it possible to make the large sets of measurements necessary to obtain confidence in the results and its portability allows others to check or extend the measurements. The system has four parts: 1) semiautomatic generation of problems for experimental input; 2) the ELLPACK system for actually solving the equation; 3) a data management system to organize and access the experimental data; and 4) data analysis programs to extract graphical and statistical summaries from the data.

Experimental study of high speed polarization-coding quantum key distribution with sifted-key rates over Mbit/s

We present a quantitative study of various limitations on quantum cryptographic systems operating with sifted-key rates over Mbit/s. The dead time of silicon APDs not only limits the sifted-key rate but also causes correlation between the neighboring key bits. In addition to the well-known count-rate dependent timing jitter in avalanche photo-diode (APD), the faint laser sources, the vertical cavity surface emission lasers (VCSELs) in our system, also induce a significant amount of data-dependent timing jitter. Both the dead time and the data-dependent timing jitter are major limiting factors in designing QKD systems with sifted-key rates beyond Mbit/s.

Matrix market: a web resource for test matrix collections

We describe a repository of data for the testing of numerical algorithms and mathematical software for matrix computations. The repository is designed to accommodate both dense and sparse matrices, as well as software to generate matrices. It has been seeded with the well-known Harwell-Boeing sparse matrix collection. The raw data files have been augmented with an integrated World Wide Web interface which describes the matrices in the collection quantitatively and visually. For example, each matrix has a Web page which details its attributes, graphically depicts its sparsity pattern, and provides access to the matrix itself in several formats. In addition, a search mechanism is included which allows retrieval of matrices based on a variety of attributes, such as type and size, as well as through free-text search in abstracts. The URL is http://math.nist.gov/MatrizMarket/.

The quality of numerical software: assessment and enhancement

In this chapter we provide some background on the conference which brought together the researchers whose work is described in this volume. Further information, including access to transparencies and audio recordings of some of the talks, are available at the conference World Wide Web site: http://renoir.csc.ncsu.edu/IFIP/WoCo7/.

A fourth-order-accurate Fourier method for the Helmholtz equation in three dimensions

We present fourth-order-accurate compact discretizations of the Helmholtz equation on rectangular domains in two and three dimensions with any combination of Dirichlet, Neumann, or periodic boundary conditions. The resulting systems of linear algebraic equations have the same block-tridiagonal structure as traditional central differences and hence may be solved efficiently using the Fourier method. The performance of the method for a variety of test cases, including problems with nonsmooth solutions, is presented. The method is seen to be roughly twice as fast as deferred corrections and, in many cases, results in a smaller discretization error.

JavaGrande - High Performance Computing with Java

The JavaGrande Forum is a group of users, researchers, and interested parties from industry. The Forum members are either trying to use Java for resource-intensive applications or are trying to improve the Java platform, making it possible to create large-sized applications that run quickly and efficiently in Java. In order to improve its floating point arithmetic, the Forum has suggested to add the keywords strictfp and fastfp to the Java programming language It has has worked on complex numbers, multidimensional arrays, fast object serializations, and a high-speed remote method invocation (RMI). Results about the new field of research that has been started by the JavaGrande Forum have been recognized both internationally and within Sun Microsystems.