Search

Abstract: Parallel code presents a non-trivial problem of load balancing computational workload throughout a system of hardware and software resources. The task of load balancing is further complicated when the number of allowable processors changes through time. This paper presents a two-component load-balancing mechanism using optimal initial workload distribution and dynamic load maintenance. The initial guess is provided by inversion of the workload distribution function. Workload distribution inversion enables efficient domain decomposition for arbitrary workloads and easily compensates for changes in system resources. Dynamic load balancing is provided by process feedback control as used, for example, in control mechanisms of physical processes. Proportional, integral, and differential (PID) feedback readily allows controls to compensate for runtime-changes of the workload distribution function. This paper demonstrates a one-dimensional realization of the ideas presented here. We apply this load-balancing technique to our gridless direct simulation Monte Carlo algorithm. We demonstrate that the method does indeed maintain uniform workload distribution across available resources as the workload and usable system resources undergo change through time. [ABSTRACT FROM AUTHOR]

Abstract: The development of a new parallel framework for integrated modeling of tokamak plasmas is a primary objective of the SciDAC Framework Architecture for Core-Edge Transport Simulations (FACETS) project. The FACETS code will be used to predict the performance of tokamak discharges and to optimize tokamak discharge scenarios. Novel parallel numerical algorithms and solvers have been developed in the FACETS project in order to simulate the multi-scale dynamics of tokamak plasmas. The status of development of modules for anomalous transport in the FACETS code is described in this paper. Mechanisms that are used for coupling 1D anomalous transport in the plasma core together with 2D transport in the plasma edge (in near separatrix and scrape-off-layer regions) are considered. Results of the first verification studies based on predictive modeling of several analytical and experimental equilibria are presented. [Copyright &y& Elsevier]

Abstract: We describe an algorithm and a C++ implementation that we have written and made available for calculating the fully nonlinear evolution of 5D braneworld models with scalar fields. Bulk fields allow for the stabilization of the extra dimension. However, they complicate the dynamics of the system, so that analytic calculations (performed within an effective 4D theory) are usually only reliable for static bulk configurations or when the evolution of the extra dimension is negligible. In the general case, the nonlinear 5D dynamics can be studied numerically, and the algorithm and code we describe are the first ones of that type designed for this task. The program and its full documentation are available on the Web at http://www.cita.utoronto.ca/~jmartin/BRANECODE/. 1 [1] We also maintain a mirror of the BRANECODE website at http://www.cita.utoronto.ca/~kofman/BRANECODE/. In this paper we provide a brief overview of what the program does and how to use it. Program summary: Title of program: BRANECODE Catalogue identifier: ADVX Program summary URL: http://cpc.cs.qub.ac.uk/summaries/ADVX Program obtainable from: CPC Program Library, Queen''s University of Belfast, N. Ireland Licensing provisions: none Operating systems under which the program has been tested: Linux Programming language used: C++ Memory required to execute with typical data: less than 1 MB Has the code been vectorized?: no Peripherals used: none No. of lines in distributed program, including test data, etc.: 8277 No. of bytes in distributed program, including test data, etc.: 74 939 CPC Program Library subprograms used: none Nature of physical problem: Dynamics of two co-dimension one branes in a five-dimensional spacetime with a bulk scalar field and arbitrary potentials. The dynamics is governed by the five dimensional Einstein equations of gravity and the junction conditions at the position of the branes. Method of solution: Leapfrog algorithm to solve system of -dimensional partial differential equations; Initial and boundary value problem. Restrictions on the complexity of the problem: Assumption of homogeneity along three spatial dimensions parallel to the branes. Typical running time: Depending on the grid size and length of the time evolution: from ∼1 s to ∼1 h or longer. Unusual features of the program:none [Copyright &y& Elsevier]