An important aspect of any parallel programming tool is its ability to provide useful information that can help the user optimize a program for efficient parallel execution or debug a parallel program. In many paralle...
详细信息
An important aspect of any parallel programming tool is its ability to provide useful information that can help the user optimize a program for efficient parallel execution or debug a parallel program. In many parallel programming tools, data dependence information is a fundamental requirement for the implementation of the various utilities provided by the tool. Experience has shown that while data dependence is a powerful concept, it sometimes causes several complexities both in internal analysis and in programmer-tool interaction. These problems can be overcome by summarizing the effect of data dependences in parts of the program. This paper presents a mechanism for summarizing data accesses in numerical scientific programs that is easy to implement and manipulate in a programming tool. Data dependence is viewed as an intersection between data access summaries, which allows data dependence and data access to be treated in a unified manner.
We use two large simulations, the chemical reaction dynamics of H + H2 and the collision of two galaxies to show that current parallel machines are capable of large supercomputer level calculations. We contrast the di...
详细信息
ISBN:
(纸本)0897913418
We use two large simulations, the chemical reaction dynamics of H + H2 and the collision of two galaxies to show that current parallel machines are capable of large supercomputer level calculations. We contrast the different architectural tradeoffs for these problems and draw some implications for future production parallel supercomputers.
A deterministic annealing technique is proposed for the nonconvex optimization problem of clustering. Deterministic annealing is used in order to avoid local minima of the given cost function which trap traditional te...
详细信息
A deterministic annealing technique is proposed for the nonconvex optimization problem of clustering. Deterministic annealing is used in order to avoid local minima of the given cost function which trap traditional techniques. A set of temperature parametrized Gibbs probability density functions relate each data point to each cluster. An effective cost function is defined and minimized at each temperature. It is shown that as the temperature approaches zero, the algorithm becomes the basic ISODATA algorithm. The method is independent of the initial choice of cluster means. Simulation results are given and show how the method succeeds in avoiding local minima.
We describe a set of software utilities designed to facilitate the writing of parallel codes and porting sequential ones. Emphasis is placed on portability so that code can be developed simultaneously on a sequential ...
详细信息
We have measured the dynamical critical exponents for the Swendsen-Wang and the Wolff cluster update algorithms, as well as a number of variants of these algorithms, for the q = 2 and q = 3 Potts models in two dimensi...
Numerical simulations of Lattice QCD have been performed on practically every computer, since its inception almost twenty years ago. Lattice QCD is an ideal problem for parallel machines as it can be easily domain dec...
详细信息
The branch-And-bound technique is a common method for finding exact solutions to difficult problems in combinatorial optimization. This paper will discusss issues surrounding implementation of a particular branch-And-...
详细信息
We discuss the design of a class of shift-register sequence random number generators for the MlMD parallel computers, and particularly for the hypercube concurrent computers. The simplest implementation is to have eac...
详细信息
We are currently performing large-scale numerical simulations of dynamically triangulated random surfaces on several parallel computers. Herein we briefly explain the importance of random surface simulations and descr...
详细信息
Earlier simulations of dynamically triangulated random surfaces with a pure Gaussian (Polyakov) action have suggested that the incorporation of a term which is equivalent to the square of the scalar curvature, R 2 , i...
Earlier simulations of dynamically triangulated random surfaces with a pure Gaussian (Polyakov) action have suggested that the incorporation of a term which is equivalent to the square of the scalar curvature, R 2 , in the continuum can affect the properties of the surfaces, despite the fact that such a term appears to be irrelevant on dimensional grounds. However, simulations by the current authors and Catterall of dynamically triangulated random surfaces with extrinsic curvature produced essentially identical results despite differing coefficients for the R 2 term. In this short note we show that small (positive or negative) values of this coefficient have little effect but that large values to produce measurable effects. This explains the concordance of our previous results with Catterall’s and also provides evidence for non-universal behavior in the random surface model.
暂无评论