Distributed localized bi-objective search

We propose a new distributed heuristic for approximating the Pareto set of bi-objective optimization problems. Our approach is at the crossroads of parallel cooperative computation, objective space decomposition, and adaptive search. Given a number of computing nodes, we self-coordinate them locally, in order to cooperatively search different regions of the Pareto front. This offers a trade-off between a fully independent approach, where each node would operate independently of the others, and a fully centralized approach, where a global knowledge of the entire population is required at every step. More specifically, the population of solutions is structured and mapped into computing nodes. As local information, every node uses only the positions of its neighbors in the objective space and evolves its local solution based on what we term a ‘localized fitness function  ’. This has the effect of making the distributed search evolve, over all nodes, to a high quality approximation set, with minimum communications. We deploy our distributed algorithm using a computer cluster of hundreds of cores and study its properties and performance on ρ$\rho$MNK-landscapes. Through extensive large-scale experiments, our approach is shown to be very effective in terms of approximation quality, computational time and scalability.     

SPEW: Synthetic Populations and Ecosystems of the World

Agent-based models (ABMs) simulate interactions between autonomous agents in constrained environments over time and are often used for modeling the spread of infectious diseases. ABMs use information about agents and their environments as input, together referred to as a “synthetic ecosystem.” Previous approaches for generating synthetic ecosystems have some limitations: they are not open-source, cannot be adapted to new or updated input data sources, or do not allow for alternative methods for sampling agent characteristics and locations. We introduce a general framework for generating synthetic ecosystems, called “Synthetic Populations and Ecosystems of the World” (SPEW). SPEW lets researchers choose from a variety of sampling methods for agent characteristics and locations and is implemented as an open-source R package. We analyze the accuracy and computational efficiency of SPEW, given different sampling methods for agent characteristics and locations, and provide a suite of statistical and graphical tools to screen our generated ecosystems. SPEW has generated over five billion human agents across approximately 100,000 geographic regions in over 70 countries available online.     

A Hybrid Domain Decomposition and Multigrid Method for the Acceleration of Compressible Viscous Flow Calculations on Unstructured Triangular Meshes

This paper is concerned with the formulation and the evaluation of a hybrid solution method that makes use of domain decomposition and multigrid principles for the calculation of two-dimensional compressible viscous flows on unstructured triangular meshes. More precisely, a non-overlapping additive domain decomposition method is used to coordinate concurrent subdomain solutions with a multigrid method. This hybrid method is developed in the context of a flow solver for the Navier-Stokes equations which is based on a combined finite element/finite volume formulation on unstructured triangular meshes. Time integration of the resulting semi-discrete equations is performed using a linearized backward Euler implicit scheme. As a result, each pseudo time step requires the solution of a sparse linear system. In this study, a non-overlapping domain decomposition algorithm is used for advancing the solution at each implicit time step. Algebraically, the Schwarz algorithm is equivalent to a Jacobi iteration on a linear system whose matrix has a block structure. A substructuring technique can be applied to this matrix in order to obtain a fully implicit scheme in terms of interface unknowns. In the present approach, the interface unknowns are numerical fluxes. The interface system is solved by means of a full GMRES method. Here, the local system solves that are induced by matrix-vector products with the interface operator, are performed using a multigrid by volume agglomeration method. The resulting hybrid domain decomposition and multigrid solver is applied to the computation of several steady flows around a geometry of NACA0012 airfoil.     

Large scale numerical simulations for multi-phase fluid dynamics with moving interfaces

This short communication presents our recent studies to implement numerical simulations for multi-phase flows on top-ranked supercomputer systems with distributed memory architecture. The numerical model is designed so as to make full use of the capacity of the hardware. Satisfactory scalability in terms of both the parallel speed-up rate and the size of the problem has been obtained on two high rank systems with massively parallel processors, the Earth Simulator (Earth simulator research center, Yokohama Kanagawa, Japan) and the TSUBAME (Tokyo Institute of Technology, Tokyo, Japan) supercomputers.     

无网格Galerkin法GPU加速并行计算及其应用

针对无网格Galerkin法计算耗时的问题,采用逐节点对法来组装刚度矩阵、共轭梯度法求解基于CSR格式存储的稀疏线性方程组,提出了一种利用罚函数法施加本质边界条件的EFG法GPU加速并行算法,给出了刚度矩阵和惩罚刚度矩阵的统一格式,以及GPU加速并行算法的流程图。编写了基于CUDA构架平台的GPU程序,且在NVIDIA GeForce GTX 660显卡上通过数值算例对所提算法进行了性能测试与分析比较,探讨了影响加速比的因素。算例结果验证了所提算法的可行性,并在满足计算精度的前提下,其加速比最大可达17倍;同时线性方程组的求解对加速比起决定性影响。     

从几何代数到高级不变量计算

综述近几年来几何代数和高级不变量计算两方面的主要进展,重点是共形几何代数的背景、思路、发展和对经典几何的高级不变量理论发展的重要作用.     

An algebraic multigrid method for elasticity based on an auxiliary topology with edge matrices

This article introduces a novel approach to algebraic multigrid methods for large systems of linear equations coming from finite element discretizations of certain elliptic second-order partial differential equations. Based on a discrete energy made up of edge and vertex contributions, we are able to develop coarsening criteria that guarantee two-level convergence even for systems of equations such as linear elasticity . This energy also allows us to construct prolongations with prescribed sparsity pattern that still preserve kernel vectors exactly. These allow for a straightforward optimization that simplifies parallelization and reduces communication on coarse levels. Numerical experiments demonstrate efficiency and robustness of the method and scalability of the implementation.     

Improvements in the treatment of stress constraints in structural topology optimization problems

Topology optimization of continuum structures is a relatively new branch of the structural optimization field. Since the basic principles were first proposed by Bendsøe and Kikuchi in 1988, most of the work has been dedicated to the so-called maximum stiffness (or minimum compliance) formulations. However, since a few years different approaches have been proposed in terms of minimum weight with stress (and/or displacement) constraints.These formulations give rise to more complex mathematical programming problems, since a large number of highly non-linear (local) constraints must be taken into account. In an attempt to reduce the computational requirements, in this paper, we propose different alternatives to consider stress constraints and some ideas about the numerical implementation of these algorithms. Finally, we present some application examples.     

A parallel implementation of overlapping Schwarz method for the dual reciprocity method

We present a parallel algorithm for the overlapping domain decomposition boundary integral equation method for two dimensional partial differential equations. In addition to the improvement of the ill-conditioning and the computational efficiency achieved by domain partitioning, using a parallel computer with p processors can offer up to p times efficiency. Assuming direct solution is used throughout, partitioning the domain into p subregions and employing a processor for each subproblem, overall, result in p² times efficiency over using a single domain and a single processor, taking into account that a sequential algorithm of the underlying method can improve the computational efficiency at least p times over using a single domain. Some numerical results showing the efficiency of the parallel technique will be presented.     

A MPI PARALLEL PRECONDITIONED SPECTRAL ELEMENT METHOD FOR THE HELMHOLTZ EQUATION

Spectral element method is well known as high-order method, and has potential better parallel feature as compared with low order methods. In this paper, a parallel preconditioned conjugate gradient iterative method is proposed to solving the spectral element approximation of the Helmholtz equation. The parallel algorithm is shown to have good performance as compared to non parallel cases, especially when the stiffness matrix is not memorized. A series of numerical experiments in one dimensional case is carried out to demonstrate the efficiency of the proposed method.