Subroutine Library Research Articles

Solving the Trust-Region Subproblem using the Lanczos Method

The approximate minimization of a quadratic function within an ellipsoidal trust region is an important subproblem for many nonlinear programming methods. When the number of variables is large, the most widely used strategy is to trace the path of conjugate gradient iterates either to convergence or until it reaches the trust-region boundary. In this paper, we investigate ways of continuing the process once the boundary has been encountered. The key is to observe that the trust-region problem within the currently generated Krylov subspace has a very special structure which enables it to be solved very efficiently. We compare the new strategy with existing methods. The resulting software package is available as HSL_VF05 within the Harwell Subroutine Library.

LOQO user's manual — version 3.10

LOQO is a system for solving smooth constrained optimization problems. The problems can be linear or nonlinear, convex or nonconvex, constrained or unconstrained. The only real restriction is that the functions defining the problem be smooth (at the points evaluated by the algorithm). If the problem is convex, LOQO finds a globally optimal solution. Otherwise, it finds a locally optimal solution near to a given starting point. This manual describes 1. how to install LOQO on your hardware. 2. how to use AMPL together with LOQO to solve general optimization problems, 3. how to use the subroutine library to formulate and solve optimization problems, and 4. how to formulate and solve linear and quadratic programs in MPS format.

Programming the finite element method

From the Publisher: In the third edition of this highly-successful book, readers are shown how to develop computer programs which use the finite element method to solve specific problems which they may encounter. This third edition is completely revised, and has all programs and subroutine libraries written in FORTRAN 90. The programs and most subroutines are listed in full in the text, and are all available on the Internet via anonymous ftp. A wide variety of new problem solving analyses are presented, including classical structural analysis, elasticity and plasticity, steady state and transient fluid flow, linear and non-linear solid dynamics and construction processes in geomechanics.

Performance of several nonlinear programming software packages on microcomputers

Nonlinear programming (NLP) problems arise often in science, engineering, and business applications. Due to the inherent mathematical limitations of the solution techniques, the existing solution methods are not guaranteed to solve all problems optimally. In the past years, a number of new commercial packages that solve NLP problems on microcomputers have been developed. To grasp some idea of the performance of the NLP packages on microcomputers from the viewpoints of effectiveness, efficiency, accuracy, and ease of use, six popular NLP packages are selected for comparison. The results provide some information regarding the suitable package to use under different considerations. This paper investigates the performance of six NLP packages, including MATLAB, IMSL, AMPL, GAMS, GINO, and SQP, by solving seventy test problems on a PC 486 machine. These packages are categorized as library subroutines, modeling languages, and the conventional NLP codes. The results show that AMPL and GINO have the best ability in solving NLP problems optimally, GAMS is the most efficient package in terms of the execution time, GINO has the most accurate solution, and MATLAB needs the least effort in modeling problems. In sum, GAMS and GINO have the best overall performance, followed by AMPL. All these three packages are of modeling language type.

Computational Physics: Problem Solving with Computers

Partial table of contents: GENERALITIES. Computing Software Basics. Errors and Uncertainties in Computations. APPLICATIONS. Data Fitting. Deterministic Randomness. Monte Carlo Applications. Differentiation. Differential Equations and Oscillations. Anharmonic Oscillations. Unusual Dynamics of Nonlinear Systems. Differential Chaos in Phase Space. Matrix Computing and Subroutine Libraries. Bound States in Momentum Space. Computing Hardware Basics: Memory and CPU. High-Performance Computing: Profiling and Tuning. Object-Oriented Programming: Kinematics. Thermodynamic Simulations: The Ising Model. Fractals. PARTIAL DIFFERENTIAL EQUATIONS. Heat Flow. Waves on a String. NONLINEAR PARTIAL DIFFERENTIAL EQUATIONS. Solitons, The KdeV Equation. Confined Electronic Wave Packets. Appendices. Glossary. References. Index.

Lot Sizing in General Assembly Systems with Setup Costs, Setup Times, and Multiple Constrained Resources

We introduce a heuristic method for finding good, feasible solutions for multiproduct lot sizing problems with general assembly structures, multiple constrained resources, and nonzero setup costs and setup times. We evaluate the performance of this heuristic by comparing its solutions to optimal solutions of small randomly generated problems and to time-truncated Optimization Subroutine Library (OSL) solutions of medium-sized randomly generated problems. In the first case, the heuristic locates solutions averaging 4 percent worse than optimal in less than 1 percent of time required by OSL. The heuristic solutions to medium-sized problems are approximately 26 percent better than solutions OSL finds after 10,000 CPU seconds, and the heuristic finds these solutions in approximately 10 percent of OSL time.

ReUse/Web: Web-based Ada reuse

In his article "Software Reuse Myths" from the January 1988 issue of ACM SIGSOFT Software Engineering Notes, Will Tracz discusses why software reuse has not grown to its full potential. Engineers first proposed the concept of a subroutine library at the University of Cambridge in 1949. Yet the concept has not extended much past this level.At the time Tracz wrote this article, interest had begun to reemerge in the concept of software reuse. Still, Tracz believes that nine myths surround software reuse that have prevented it from becoming more widespread. Specifically, three myths relate to my thesis:1. Software reuse is a technical problem.2. Special tools are needed for software reuse.3. Ada has solved the reuse problem.As far as the first myth goes, the only technical problem is the lack of search methods to find the right pieces of code (which my project should help alleviate). Beyond that, the problem is more psychological, sociological or economic. For example, there is the "Not From Here" attitude pervading industry that Icads people to believe that software developed outside of their company is unreliable.Tracz believes you only need special tools (the second myth) when you become truly serious about reuse. Most software libraries, he claims, are only 100-200 components in size and that programmers can easily navigate this many components without any tools. Also, most instances of reuse occur when a programmer modifies existing programs that either he or a coworker wrote, so again he does not need a tool to find what he needs. I believe that the reuse mentality is a difficult one to accept and any tool that can make the transition to this attitude easier should be explored and utilized, no matter how small the number of components in the software library.Tracz refutes the third myth by stating that "writing a generic package in Ada does not necessarily make it reusable any more than writing a Fortran subroutine or assembly language macro." Simply put, generic packages are not enough to encourage reuse because the packages must still be instantiated with actual data types. The attitudes of the programmers using the language are more important than the language itself in determining whether or not code will be reused. I hope that my project makes Ada more accessible to code reuse and can change some programmers' attitudes about reuse.I believe that this attitude can change if programmers have access to a simple tool to browse software libraries that runs on a variety of hardware platforms to maximize usage and potential. This idea gave rise to ReUSE-WWW, an Ada specification file browser running under Netscape. Since Netscape runs on many platforms (PC, Macintosh, Unix-based., etc.), an application using Netscape as its "operating system" would be ideal. Fortunately, recent developments in World Wide Web programming, such as JavaScript and Perl CGI (Common Gateway Interface) have made this goal extremely attainable.ReUSE-WWW may not run under Microsoft's Internet Explorer, though. It also does not provide any editing, compiling, linking, or running capabilities. I assume that a potential user of the system already has an editor of choice and knows where to get an Ada 95 compiler, such as GNAT.

The design, implementation, and evaluation of a symmetric banded linear solver for distributed-memory parallel computers

This article describes the design, implementation, and evaluation of a parallel algorithm for the Cholesky factorization of symmetric banded matrices. The algorithm is part of IBM's parallel engineering and scientific subroutine library version 1.2 and is compatible with ScaLAPACK's banded solver. Analysis, as well as experiments on an IBM SP2 distributed-memory parallel computer, shows that the algorithm efficiently factors banded matrices with wide bandwidth. For example, a 31-mode SP2 factors a large matrix more than 16 times faster than a single node would factor it using the best sequential algorithm, and more than 20 times faster than a single node would using LAPACK's DPBTRF. The algorithm uses novel ideas in the area of distributed dense-matrix computations that include the use of a dynamic schedule for a blocked systolic-like algorithm and the separation of the input and output layouts from the layout the algorithm uses internally. The algorithm alson uses known techniques such as blocking to improve its communication-to-computation ratio and its data-cache behavior.

Modeling the Multiphase Atmospheric Chemistry of Launch Clouds

Abstract : The adaptation is described of a widely available subroutine library, originally developed to model chemical vapor deposition to model a plume of steam and hydrogen chloride gas that is released during solid rocket booster launches such as the Space Shuttle and the Titan IV. Hydrogen chloride is a very hygroscopic gas, and it tends to form an aerosol cloud from the water vapor in the atmosphere as well as from the steam released by engine combustion and launch deluge water. The aerosol cloud is the most difficult feature to model because it involves phase changes, complex thermodynamic data for HCl/water solutions, and rapid changes in temperature, concentration, and relative humidity. The model predictions are compared to field data taken during a launch of the Space Shuttle.

The ASAD atmospheric chemistry integration package and chemical reaction database

The ASAD software package frees the atmospheric chemist from writing code to solve chemical time dependent equationsand can be used for a wide range of problems including stratospheric chemistry, tropospheric chemistry or pollution studies, and for 1D, 2D and 3D models. Since little programming is required, the effects of programming errors are greatly reduced and consistent results between different models ensured. Through the use of input files to list the desired chemical species and their properties and the chemical reactions, different chemical scenarios can be rapidly developed and tested without any additional programming. ASAD comprises a plug-in subroutine library to solve atmospheric chemistry time-dependent problems, a chemical reaction database and utility programs. ASAD has been designed to be coupled to any type of atmospheric model and be used as far as possible as a ‘black box’. The code has been carefully optimized for performance and it will vectorize well on machines with this capability. It includes a choice of integrators based on published methods and supports approximations often used in atmospheric chemistry modelling, such as chemical families, fractional products and the steady state approximation. ASAD also supports the inclusion of user-supplied schemes for wet and dry deposition, source emissions, heterogeneous chemistry and photolysis.

Smoothed Particle Hydrodynamics with GRAPE and Parallel Virtual Machine

We have developed Remote-GRAPE (for GRAvity PipE), a subroutine library to be used with the special purpose computer GRAPE-3A. GRAPE-3A can efficiently calculate gravitational forces between particles and construct neighbor lists. All other calculations are performed on the host workstation (WS), which is directly connected to GRAPE. The use of GRAPE for smoothed particle hydrodynamics (GRAPE-SPH) can in principle greatly speed up the calculations on the WS. However, the current bottleneck of GRAPE-SPH is that its performance is limited by the speed of the host WS. To solve this problem, we implement Remote-GRAPE; it allows us to run applications that use GRAPE-3A hardware on significantly faster computers than the physical host WS. Thus, we can take advantage of fast computers even though they cannot physically be connected to GRAPE. The Remote-GRAPE system is implemented on Parallel Virtual Machine. The details of implementation are described. We analyze the performance of Remote-GRAPE and obtain the following results. (1) When Remote-GRAPE is used to calculate gravitational forces only, the overhead due to the network decreases according to the number of particles (20%-40% of total time). (2) When Remote-GRAPE is used to calculate gravity and neighbor lists, the overhead due to the network does not occupy a large fraction of total time but only ~20%-30%, because the computation required on the slave machine is very large because of the properties of the GRAPE system. (3) We also compare the performance of Remote-GRAPE with the tree method. The tree method requires more time for a higher degree of clustering, while the required time for Remote-GRAPE does not much depend on it. We then analyze the performance of GRAPE-SPH with Remote-GRAPE. The performance of Remote-GRAPE is about 4 times faster than GRAPE-SPH with usual GRAPE for our configuration. Using Remote-GRAPE, we can calculate the GRAPE part and other parts in parallel. This parallel method leads to a further speedup of our SPH code. We estimate how the performance of Remote-GRAPE depends on the configuration. We also show that the performance of Remote-GRAPE can be further improved.

Evidence-based static branch prediction using machine learning

Correctly predicting the direction that branches will take is increasingly important in today's wide-issue computer architectures. The nameprogram-basedbranch prediction is given to static branch prediction techniques that base their prediction on a program's structure. In this article, we investigate a new approach to program-based branch prediction that uses a body of existing programs to predict the branch behavior in a new program. We call this approach to program-based branch predictionevidence-based static prediction, or ESP. The main idea of ESP is that the behavior of a corpus of programs can be used to infer the behavior of new programs. In this article, we use neural networks and decision trees to map static features associated with each branch to a prediction that the branch will be taken. ESP shows significant advantages over other prediction mechanisms. Specifically, it is a program-based technique; it is effective across a range of programming languages and programming styles; and it does not rely on the use of expert-defined heuristics. In this article, we describe the application of ESP to the problem of static branch prediction and compare our results to existing program-based branch predictors. We also investigate the applicability of ESP across computer architectures, programming languages, compilers, and run-time systems. We provide results showing how sensitive ESP is to the number and type of static features and programs included in the ESP training sets, and we compare the efficacy of static branch prediction for subroutine libraries. Averaging over a body of 43 C and Fortran programs, ESP branch prediction results in a miss rate of 20%, as compared with the 25% miss rate obtained using the best existing program-based heuristics.

Analysing neutron scattering data with the cambridge crystallographic subroutine library (CCSL)

The Cambridge Crystallographic Subroutine Library (CCSL) is a library designed to facilitate crystallographic calculations. It defines a data format for describing both nuclear and magnetic structures. Procedures contained in the library use these da

CIF Applications. V.CIFtbx2: extended tool box for manipulating CIFs

CIFtbx2 is a new version of a Fortran subroutine library for programmers developing CIF applications. The functions for reading and writing CIF data in CIFtbx [Hall (1993). J. Appl. Cryst. 26, 482–494] have been expanded and facilities for handling macromolecular CIF data and dictionaries have been added. The CIFtbx2 library facilitates applications involving all current versions of the dictionary definition language.

Solving the resource constrained project scheduling problem with optimization subroutine library

In this paper we present three new models for the resource constrained project scheduling problem, using two objective functions: minimizing the makespan and maximizing the net present value of the cash flows of the activities. In these models, compared to the models in the literature, we relax one of the most constraining assumptions about the integrality of the activity durations and the project time line. Furthermore resources are assumed to be partially-renewable. These models have multiple uses in Operations Management in that they can be used to solve resource constrained MRP problems. The models are found to be amenable to solution using the Optimization Subroutine Library (OSL). Extensive computational experiments with OSL on a total of 1400 problems with different parameter settings are reported. Detailed analysis is done to determine the factors that affect the computational efficiency of the code and the models. In general, results demonstrate that OSL is a powerful tool for solving the models with either objective function. Optimal solutions are obtained with reasonable computational effort. A discussion of the application of several special features of OSL is also given.

The design of a new frontal code for solving sparse, unsymmetric systems

We describe the design, implementation, and performance of a frontal code for the solution of large, sparse, unsymmetric systems of linear equations. The resulting software package, MA42, is included in Release 11 of the Harwell Subroutine Library and is intended to supersede the earlier MA32 package. We discuss in detail the extensive use of higher-level BLAS kernels within MA42 and illustrate the performance on a range of practical problems on a CRAY Y-MP, an IBM 3090, and an IBM RISC System/6000. We examine extending the frontal solution scheme to use multiple fronts to allow MA42 to be run in parallel. We indicate some directions for future development.

RefVal: a program implementing the recommendations of the International Federation of Clinical Chemistry on the statistical treatment of reference values

RefVal is a computer program that implements the recommendations of the International Federation of Clinical Chemistry on the statistical treatment of reference values. The program performs the following main tasks: graphical display of the distribution of reference values, identification or elimination of outliers, testing of the fit of the distribution to Gaussian shape (coefficients of skewness and kurtosis, Anderson-Darling's test, Cramér-von Mises' test, Kolmogorov-Smirnov's test), non-parametric and parametric estimation of reference limits (fractiles, percentiles). The parametric estimation method is based on a two-stage mathematical transformation of data: (1) Manly's exponential transformation (to remove skewness) and (2) John and Draper's modulus transformation (to adjust for remaining kurtosis). The program exists in different versions. The paper describes two of these: (1) a library of FORTRAN functions and subroutines and (2) a Pascal PC program that runs under MS-DOS.

An Arnoldi code for computing selected eigenvalues of sparse, real, unsymmetric matrices

Arnoldi methods can be more effective than subspace iteration methods for computing the dominant eigenvalues of a large, sparse, real, unsymmetric matrix. A code, EB12 , for the sparse, unsymmetric eigenvalue problem based on a subspace iteration algorithm, optionally combined with Chebychev acceleration, has recently been described by Duff and Scott and is included in the Harwell Subroutine Library. In this article we consider variants of the method of Arnoldi and discuss the design and development of a code to implement these methods. The new code, which is called EB13 , offers the user the choice of a basic Arnoldi algorithm, an Arnoldi algorithm with Chebychev acceleration, and a Chebychev preconditioned Arnoldi algorithm. Each method is available in blocked and unblocked form. The code may be used to compute either the rightmost eigenvalues, the eigenvalues of largest absolute value, or the eigenvalues of largest imaginary part. The performance of each option in the EB13 package is compared with that of subspace iteration on a range of test problems, and on the basis of the results, advice is offered to the user on the appropriate choice of method. — Author's Abstract

The genesis distributed memory benchmarks. Part 2: COMMS1, TRANS1, FFT1 and QCD2 benchmarks on the suprenum and IPSC/860 computers

AbstractThe Genesis benchmark suite has been assembled to evaluate the performance of distributed‐memory MIMD systems. The problems selected all have a scientific origin (mostly from physics or theoretical chemistry), and range from synthetic code fragments designed to measure the basic hardware properties of the computer (especially communication and synchronisation overheads), through commonly used library subroutines, to full application codes. This is the second of a series of papers on the Genesis distributed‐memory benchmarks, which were developed under the European ESPRIT research program. Results are presented for the SUPRENUM and iPSC/860 computers when running the following benchmarks: COMMS1 (communications), TRANS1 (matrix transpose), FFT1 (fast Fourier transform) and QCD2 (conjugate gradient kernel). The theoretical predictions are compared with, or fitted to, the measured results, and then used to predict (with due caution) how the performance might scale for larger problems and more processors than were actually available during the benchmarking.

Pointing geometry for low Earth orbit auroral observations

Introduction A LGORITHMS are developed to solve the geometry required to point telescopes mounted on a low Earth-orbiting satellite at features in the aurora (at height of typically 100 km above Earth's surface). It is based on requirements for the proposed AURIO telescopes, which may be flown on a polar platform of the European Space Agency (ESA). The idea of pointing an instrument and tracking a target from space is not new and the geometry has been solved for several cases (see, e.g., Jerkovsky, Burdick et al., and Hablani). The traditional approach is to use vector notation, but at the implementation stage this method usually needs to be supported by a library of subroutines to implement the basic vector operations of addition, cross and scalar products, and rotations of vectors about the coordinate axes, which raises difficulties in a small embedded computer system. In this Note a computational method is presented to find the line-of-sight (LOS) vector in the satellite-body-attached coordinate frame. This method uses homogeneous transformation matrices (normally used in robot kinematics) to provide efficient algorithms that can readily be implemented on-board the satellite. A spherical Earth is assumed in this work as the basic model. However, the spheroidal model of Earth has been used to assess errors due to this assumption. The software coding of the pointing equations can easily be implemented in any high level programming language, and they have been tested in the occam language.