Parallel Computation Method Research Articles

Parallel computation in biological sequence analysis

A massive volume of biological sequence data is available in over 36 different databases worldwide, including the sequence data generated by the Human Genome project. These databases, which also contain biological and bibliographical information, are growing at an exponential rate. Consequently, the computational demands needed to explore and analyze the data contained in these databases is quickly becoming a great concern. To meet these demands, we must use high performance computing systems, such as parallel computers and distributed networks of workstations. We present two parallel computational methods for analyzing these biological sequences. The first method is used to retrieve sequences that are homologous to a query sequence. The biological information associated with the homologous sequences found in the database may provide important clues to the structure and function of the query sequence. The second method, which helps in the prediction of the function, structure, and evolutionary history of biological sequences, is used to align a number of homologous sequences with each other. These two parallel computational methods were implemented and evaluated on an Intel IPSC/860 parallel computer. The resulting performance demonstrates that parallel computational methods can significantly reduce the computational time needed to analyze the sequences contained in large databases.

Parallel radiative transport using a unified matrix approach

This paper describes the use of parallel computational methods applied to the solution of radiative transfer problems involving participating media. A unified matrix approach, involving a range of operations (ray tracing, numerical integration, and solution of large, dense linear systems), is employed. Several issues associated with massively parallel architectures which affect performance are examined and applied to problems in two and three dimensions.

Conformational Characteristics of Polymethylene Chains in Melts and in Various Phantom Chains from Explicit Atom Molecular Dynamics Simulations

Conformer populations, mean-square chain dimensions, and end-to-end distance distributions for polymethylene (PM) chains, H(CH2)nH with n = 13, 44, and 100, in melts and in various phantom chains have been studied using a parallel computational method of molecular (and Brownian) dynamics simulations with a well-calibrated explicit atom force field. The specific volume−temperature properties of the simulated melt systems are in excellent agreement with experiments. Conformational characteristics of PM chains in melts are identical to those of 1,5 phantom chains, characterized by short-range intramolecular interactions between the atoms separated by up to four skeletal bonds, which correspond to unperturbed chains in ϑ solvents according to recent implicit solvent model results of Sariban et al. and Destrée et al. Moreover, a Gaussian approximation of end-to-end distance distributions is found to be valid only for long PM chains with n greater than 100. According to our simulation results, the characteristic ratio C∞ for unperturbed polyethylene chains is found to be ca. 7.9 at 413 K. This value is larger than the generally accepted value, 6.8 (derived from intrinsic viscosity data for relatively polydisperse polyethylene samples), but is corroborated by recent small angle neutron scattering results (7.8 ± 0.4) for melt chains of nearly monodisperse samples. The fact that PM chains exhibit identical conformational characteristics in melts, 1,5 phantom chains, and ϑ solvents is attributed to the lack of any correlation between local chain conformations and (polymer−polymer or polymer−solvent) intermolecular attractions. Departures from such an ideal behavior seen for other polymers are also discussed.

Parallel computational methods for 3D simulation of a parafoil with prescribed shape changes

In this paper we describe parallel computational methods for 3D simulation of the dynamics and fluid dynamics of a parafoil with prescribed, time-dependent shape changes. The mathematical model is based on the time-dependent, 3D Navier-Stokes equations governing the incompressible flow around the parafoil and Newton's law of motion governing the dynamics of the parafoil, with the aerodynamic forces acting on the parafoil calculated from the flow field. The computational methods developed for these 3D simulations include a stabilized space-time finite element formulation to accommodate for the shape changes, special mesh generation and mesh moving strategies developed for this purpose, iterative solution techniques for the large, coupled nonlinear equation systems involved, and parallel implementation of all these methods on scalable computing systems such as the Thinking Machines CM-5. As an example, we report 3D simulation of a flare maneuver in which the parafoil velocity is reduced by pulling down the flaps. This simulation requires solution of over 3.6 million coupled, nonlinear equations at every time step of the simulation.

Computation efficiency of parallel systems related to granularity and application form in power system calculations

Many techniques for enhancing the calculation speed of power system analysis by utilizing parallel systems have been proposed. However, the speedups gained in these systems have been small compared to other applications such as circuit simulation, mainly due to the smallness of the problem size of power calculations. The abovementioned studies concentrated on “space parallel computing,” and the parallel implementation granularity is therefore fine due to the inherent size of power system calculations. In this paper, we first study the effect that the ratio of data transmission time to calculation time per grain has on the speedup of total calculation. We show that the smaller ratio raises the efficiency of parallel implementation. Coarser granularity of the problem generally yields smaller ratios. Thus, it is shown that attaining a coarser granularity is important. In power system calculations, the method of time-domain parallel computing produces larger granularity. Another application form which results in much coarser granularity is multiple case analysis. We studied the application of these parallel approaches to transient stability analysis using an MIMD (multiple instruction stream, multiple data stream) distributed memory parallel processor system. On a 16 PE system, we attained a speedup of 6.3 in time-domain parallel computing and a speedup of up to 15.8 in multiple case parallel analysis. These values are much larger than the speedup in space parallel computing, which was less than 2.0. © 1997 Scripta Technica, Inc. Electr Eng Jpn, 119(3): 14–26, 1997

Methods for parallel computation of SCF NMR chemical shifts by GIAO method: Efficient integral calculation, multi-Fock algorithm, and pseudodiagonalization

We implemented our gauge-including atomic orbital (GIAO) NMR chemical shielding program on a workstation cluster, using the parallel virtual machine (PVM) message-passing system. On a modest number of nodes, we achieved close to linear speedup. This program is characterized by several novel features. It uses the new integral program of Wolinski that calculates integrals in vectorized batches, increases efficiency, and simplifies parallelization. The self-consistent field (SCF) step includes a multi-Fock algorithm, i.e., the simultaneous calculation of several Fock matrices with the same integral set, increasing the efficiency of the direct SCF procedure. The SCF diagonalization step, which is difficult to parallelize, has been replaced by pseudodiagonalization. The latter, widely used in semiempirical programs, becomes important in ab initio type calculations above a certain size, because the ultimate scaling of the diagonalization step is steeper than that of integral computation. Examples of the calculation of the NMR shieldings in large systems at the SCF level are shown. Parallelization of the density functional code is underway. © 1997 by John Wiley & Sons, Inc. J Comput Chem 18: 816–825, 1997

A class of ebe time integration algorithms for transient finite element structural dynamical analysis

A new class of algorithms for transient finite element structural dynamical analysis which is amenable to an efficient implementation in parallel computers (especially Massively Parallel Computers) is proposed. The suitability of the method for parallel computation stems from the fact that, given an arbitrary partition of the finite element mesh, each element in the partition can be processed over a time step independently and simultaneously with the rest, and no global equation solving effort is involved. Although the proposed EBE time integration algorithms are shown to have the structure of an explicit scheme, they are unconditionally stable over a certain range of the algorithmic parameter.

Parallel Computing Concepts and Methods for Floquet Analysis of Helicopter Trim and Stability

Floquet analysis is widely used for small-order systems (say, order M 100); the run time on a sequential computer is simply prohibitive, Accordingly, a massively parallel Floquet analysis is developed with emphasis on large systems, and it is implemented on two SIMD or single-instruction, multiple-data computers with 4096 and 8192 processors, The focus of this development is a parallel shooting method with damped Newton iteration to generate trim results; the Floquet transition matrix (FTM) comes out as a byproduct, The eigenvalues and eigenvectors of the FTM are computed by a parallel QR method, and thereby stability results are generated, For illustration, flap and flap-lag stability of isolated rotors are treated by the parallel analysis and by a corresponding sequential analysis with the conventional shooting and QR methods; linear quasisteady airfoil aerodynamics and a finite-state three-dimensional wake model are used, Computational reliability is quantified by the condition numbers of the Jacobian matrices in Newton iteration, the condition numbers of the eigenvalues and the residual errors of the eigenpairs, and reliability figures are comparable in both the parallel and sequential analyses, Compared to the sequential analysis, the parallel analysis reduces the run time of large systems dramatically, and the reduction increases with increasing system order; this finding offers considerable promise for design and comprehensive-analysis applications.

3-D field computation: the near-triumph of commercial codes

In recent years, more and more of those who design and analyze magnets and other devices are using commercial codes rather than developing their own. This paper considers the commercial codes and the features available with them. Other recent trends with 3-D field computation include parallel computation and visualization methods such as virtual reality systems.

Studies on the Efficiency of Massively Parallel Computational Method for Large-Scale Computation based on Unstructured Grid

Studies on the Efficiency of Massively Parallel Computational Method for Large-Scale Computation based on Unstructured Grid

Operator Training Simulator with real-time transient stability analysis

The operator training simulator (OTS) system, which is integrated with a real-time transient stability analysts program, has been in operation at the training center of Hokuriku Electric Power Company since July, 1990. The developed OTS system achieves real-time transient stability analysis with approximately 100 generator and 400 node power systems by parallel processing performed on a computer system with multiprocessor architecture. This paper outlines the developed OTS system and describes the requirements, design concepts, modeling and parallel calculation method of transient stability analysis and the verification result of the developed transient stability analysis program. >

Implementation of a data parallel, logical domain decomposition method for interparticle interactions in molecular dynamics of structured molecular fluids

AbstractIn this article, we describe a domain decomposition method for the efficient parallel computation of nonbonded forces and energies in condensed‐phase molecular systems. This decomposition is based upon the monotonic logical grid (MLG) approach of Boris [J. Boris, J. Comp. Phys., 66, 1 (1986)] and yields an efficient, scalable algorithm for interparticle interaction computation on private‐memory, single‐instruction multiple‐data (SIMD) parallel computers. We illustrate the application of this technique in a molecular dynamics kernel for rigid molecular solvents by simulating the structural and thermodynamic properties of water and methanol. The performance of this algorithm on the Thinking Machines' CM‐2, private‐memory SIMD computer, is demonstrated to be good compared to conventional vector/parallel supercomputers. However, as the fluid becomes less structured performance slightly degrades. © 1994 by John Wiley & Sons, Inc.

Part III Matrix decomposition methods for parallel computers

Part III Matrix decomposition methods for parallel computers

A Krylov multisplitting algorithm for solving linear systems of equations

We consider the practical implementation of Krylov subspace methods (conjugate gradients, Gmres, etc.) for parallel computers in the case where the preconditioning matrix arises from a multisplitting. We show that the algorithm can be efficiently implemented by dividing the work into tasks that generate search directions and a single task that minimizes over the resulting subspace. Each task is assigned to a subset of processors. It is not necessary for the minimization task to send frequent information to the direction generating tasks, and this leads to high utilization with a minimum of synchronization. We study the convergence properties of various forms of the algorithm and present results of numerical examples on a sequential computer.

Explicit, high-order Runge-Kutta-Nyström methods for parallel computers

The paper describes the construction of explicit Runge-Kutta-Nyström (RKN) methods of arbitrarily high order. The order is borrowed from an underlying implicit RKN method. For the approximate solution of this method, an iteration scheme is defined. Prescribing a fixed number of iterations, the resulting scheme is an explicit RKN method. The iteration scheme is defined in such a way that many of the right-hand side evaluations can be done concurrently. As a result, explicit RKN schemes of order p are obtained which require, on a parallel computer, approximately p/2 right-hand side evaluations per step. Both in fixed- and variable-step mode, the schemes are compared with existing (sequentiasl) high-order RKN methods from the literature and are shown to demonstrate superior behaviour.

Massively parallel computational methods for finite element analysis of transient structural responses

With the emphasis on the finitely damped system (e.g. control structure interaction) case, two fully implicit and two semi-implicit sets of finite element method-based numerical algorithms are formulated for transient response analysis of space frame and truss structures in a massively parallel processing (MPP) environment. All algorithm sets use an implicit force calculation/vector equation of motion assembly procedure. The semi-implicit algorithms are based on the explicit central difference (CD) and the fourth-order Runge-Kutta (RK4) schemes, respectively, in conjunction with the use of mass lumping techniques so that solution of the recurrence equations for unknown displacements is reduced to a trivial diagonal matrix inversion operation. The fully implicit algorithm sets are based on the Newmark Beta (NB) and CD schemes, respectively, in conjunction with use of the (iterative) preconditioned conjugate gradient (PCG) method for solving the linear algebraic recurrence equations. The semi-implicit algorithm sets are fully implemented and assessed on an MPP CM-2 computer. A preliminary assessment of the fully implicit sets of algorithms is made on a Sun Workstation. These numerical study results show that the newly formulated MPP algorithms are, to a varying degree, superefficient (or potentially superefficient) on the CM-2 compared with, and even highly competitive against, the conventional sequential algorithms on an advanced serial computer.

A-stable diagonally implicit Runge-Kutta-Nyström methods for parallel computers

In this paper, we study diagonally implicit Runge-Kutta-Nyström methods (DIRKN methods) for use on parallel computers. These methods are obtained by diagonally implicit iteration of fully implicit Runge-Kutta-Nyström methods (corrector methods). The number of iterations is chosen such that the method has the same order of accuracy as the corrector, and the iteration parameters serve to make the method at least A-stable. Since a large number of the stages can be computed in parallel, the methods are very efficient on parallel computers. We derive a number of A-stable, strongly A-stable and L-stable DIRKN methods of orderp withs * (p) sequential, singly diagonal-implicit stages wheres *(p)=[(p+1)/2] ors * (p)=[(p+1)/2]+1,[°] denoting the integer part function.

Parallelization of 3-D Radiative Heat Transfer Analysis in Non-gray Gas.

A parallel computational method is developed for the Monte Carlo analysis of radiative heat transfer in three-dimensional nongray gas enclosed by gray walls. The highly parallel computer AP1000 is used for the analysis. For the nongray gas, a mixture of water vapor and carbon dioxide is chosen as the absorbing-emitting medium. Three types of parallelization are studied: type (1), event parallelization; type (2), combination of event parallelization and algorithm parallelization, and type (3), memory-saving algorithm of type (2). When 512 cells are used, values of the speed-up ratios of 395, 430, and 370 are obtained for types (1), (2), and (3), respectively. The maximum number of elements which can be treated is increased from 1200 for type (1) to 140000 for type (3).

Combination method for parallel computation in ODE

In this paper, a 4th order parallel computation method with four processes for solving ODEs is discussed. This method is the Runge-Kutta method combined with a linear multistep method, which overcomes the difficulties of the 4th order parallel Runge-Kutta method discussed in [1]. The concept of critical speedup for parallel methods is also defined, and speedups of some methods are analyzed by using this concept.

Second symposium on parallel computational methods for large-scale structural analysis and design

Second symposium on parallel computational methods for large-scale structural analysis and design