Parallel node placement method by bubble simulation
Introduction
With the rapid development of parallel computers and the increasing scale of the finite element computing, parallel techniques for the finite element method (FEM) have been given unprecedented attention [1], [2], [3], [4], [5], [6], [7], [8]. At present, the parallel aspects of FEM primarily focus on each function block such as mesh generation (referred to as pre-processing) [1], [2], [3], [4], [5], global stiffness matrix formation and solving systems of linear equations (referred to as main-processing) [6], [7], [8]. The structural analysis of FEM cannot be done until the completion of the FEM mesh generation, inevitably, this serial characteristic seriously restricts the parallel efficiency and becomes one of the bottlenecks in large-scale parallel FEM analysis [9], [10], [11], [12].
In recent years, much effort has been devoted to improve the parallel efficiency of the FEM analysis based on nodes. The Free Mesh Method (FMM) and the Node-based Local Finite Element Method (NLFEM) have been developed by Yagawa et al. [9] and Nie et al. [12], respectively. The FMM and the NLFEM are parallel node-based finite element methods featuring node-based local mesh generation and node-based finite element calculation. These new parallel mechanisms achieve naturally the seamless link between pre-processing and main-processing, and get rid of the original serialization process. Nevertheless, it is worth noting that the FMM and the NLFEM begin by appropriately distributing the nodes in the analysis domain [9], [12], [13], i.e. node coordinates and nodal density information are given as input information. However, how to generate nodes in parallel is not mentioned.
In fact, the quality of the node set has a great influence on the accuracy and convergence properties of finite element solution for node-based parallel finite element methods. Therefore, how to generate node sets properly and efficiently is attracting much research interest, and some research results have already been reported in Refs. [14], [15], [16], [17], [18] and references therein. Li et al. [16]construct an advancing front-based sphere packing process. Based on a centroid Voronoi structure, Ju et al. [17] use probabilistic methods to generate centroidal voronoi tessellations (CVTs) and the parallel implementations are also presented. Zhang et al. [18] propose a node placement approach using Monte Carlo simulation to minimize system potential energy, and thereby to find a near-equilibrium configuration of nodes. Shimada et al. [19] describe a scheme to pack circles by defining proximity-based interacting forces among circles and finding a force-balancing configuration using dynamic simulation. However, the existing node placement methods mentioned above are serial algorithms except for Ju’s method [17]. When performing the node-based parallel finite element calculation, these serial node distribution methods are usually not beneficial to improving the parallel efficiency. For the parallel CVTs method presented by Ju et al., the method may require special handling when generating boundary nodes of the whole domain. Furthermore, the number of iterations required is usually huge, and in an iterative process, global communication is needed between processors. This kind of communication will inevitably result in idle time, influencing the whole parallel efficiency.
Recently, the Node Placement method by Bubble Simulation (NPBS) has been developed [20], [21], [22], and it has also been successfully applied to node-based local mesh generation [23] and anisotropic triangular meshing problems [24]. In the NPBS method, due to the short-range interaction force between bubbles, for two relatively distant bubbles, their positions and velocities can be updated simultaneously and independently during simulation, making the NPBS method suitable for parallel environments.
Furthermore, domain decomposition (DD) techniques have been employed in parallel node placement methods in order to decompose a large, complex problem into many simpler subproblems which can be solved in parallel. Within the context of parallel mesh generation, the DD method, such as the Medial Axis Domain Decomposition (MADD) [25], the coarse-grained parallel harness method [26] and the METIS-based DD technique [27], [28] have been developed in recent years. Within the context of parallel molecular dynamics simulations, there are a large variety of DD methods proposed in the literature, see Refs. [29], [30], [31], [32] and references therein.
In this paper, we focus on developing a parallel, scalable, robust node placement method by bubble simulation (PNPBS) using MPI communication for an arbitrary number of processors, and the METIS-based DD technique is used. The METIS-based DD technique [27] addresses the issue of load imbalance among processors in the process of node placement. Furthermore, the average speed of bubbles could decrease quickly during dynamic simulation [21], thus the PNPBS method requires much fewer iterations for convergence. Communication is performed every steps among geometrically neighboring processors, such that the proportion of the communication time in the total time is reduced greatly, so parallel efficiency can be improved effectively. Furthermore, an adjacency list related to each node is also provided which stores the information of neighboring nodes, this can be directly used for node-based local mesh generation [12], [23] and node-based finite element calculation [9], [10] when solving partial differential equations system in parallel.
The rest of this paper is organized as follows: the NPBS method is described briefly in Section 2, and its parallel features based on the METIS DD scheme are investigated in detail in Section 3. In Section 4, the numerical results of the PNPBS method are presented. Finally, conclusions are described in Section 5.
Section snippets
Outline of node placement method by bubble simulation
In this section we give a brief overview of the node placement method by bubble simulation (NPBS) (refer to Refs. [20], [21] for more details). The main steps of the NPBS method are given as follows: First, an initial node set is positioned in the domain. It is important to obtain a good initial bubble configuration for speeding up the simulation. Then nodes are considered as the centers of bubbles, and bubbles are driven by their interacting forces, until a force-balancing configuration of
The parallel algorithm
A parallel version of the node placement method by bubble simulation based on domain decomposition is introduced in this section. Efficient parallel algorithms require a balance of work between the processors while keeping inter-processor communication to a minimum. In this paper, the software package is used as a DD tool in the PNPBS, which is well suited to be used for a load-balancing approach [27], [28]. The flowchart of the PNPBS method is shown in Fig. 2, and the detailed
Results and discussions
In this section, parallel numerical experiments are performed in 2D to test the parallel performances of the PNPBS method. The parallel experiment environment is a HP rx2600 cluster system. It is composed of 40 computing nodes and 2 managing nodes (each node is a HP rx2600 framework server), connected by a high-speed Myrinet network.
Conclusions
In this paper, an efficient parallel node placement method PNPBS has been introduced on a memory-distributed PC-cluster system, using MPI as the communication protocol. In the PNPBS method, a multi-level graph-partitioning library is used to decompose the computational domain into subdomains which is expected to keep the number of nodes in each processor equal. The benefit of our approach is that it is very simple compared with other existing methods, after performing bubble simulation in
Acknowledgments
This research was supported by National Natural Science Foundation of China (No: 11071196, 90916027), and the Doctorate Foundation of Northwestern Polytechnical University. We are grateful to the referees for their valuable suggestions which greatly improved the paper.
References (34)
- et al.
Parallel unstructured mesh generation by an advancing front method
Math. Comput. Simul.
(2007) - et al.
Domain decomposition method for unstructured meshes in an OpenMP computing environment
Comput. & Fluids
(2011) - et al.
Large-scale finite element fluid analysis by massively parallel processors
Parallel Comput.
(1997) - et al.
A distributed memory parallel element-by-element scheme based on Jacobi-conditioned conjugate gradient for 3D finite element analysis
Finite Elem. Anal. Des.
(2007) - et al.
Probabilistic methods for centroidal Voronoi tessellations and their parallel implementations
Parallel Comput.
(2002) - et al.
Automatic triangular mesh generation of trimmed parametric surfaces for finite element analysis
Comput. Aided Geom. Design
(1998) - et al.
MDEC: METIS-based domain decomposition for parallel 2D mesh generation
- et al.
Parallel implementation of molecular dynamics simulation for short-ranged interaction
Comput. Phys. Comm.
(2005) - N. Chrisochoides, A survey of parallel mesh generation methods, Brown University, Providence RI,...
- et al.
Parallel guaranteed quality Delaunay uniform mesh refinement
SIAM J. Sci. Comput.
(2006)
Algorithms and data structures for massively parallel generic adaptive finite element codes
ACM Trans. Math. Software
A survey of parallel solvers for the finite element method in computational electromagnetics
Int. J. Comput. Math. Electr. Electron. Eng.
Node-by-node parallel finite elements: a virtually meshless method
Int. J. Numer. Methods Eng.
Parallel computing of high-speed compressible flows using a node-based finite-element method
Int. J. Numer. Methods Eng.
The parallel mechanism of node-based seamless finite element method
Comput. Model. Eng. Sci.
Node-based local mesh generation algorithm
Chin. J. Comput. Mech.
Fast searching algorithm for candidate satellite-node set in NLMG
Comput. Model. Eng. Sci.
Cited by (11)
Acceleration strategies based on bubble-type adaptive mesh refinement method
2020, Mathematics and Computers in SimulationParallel adaptive mesh refinement method based on bubble-type local mesh generation
2018, Journal of Parallel and Distributed ComputingCitation Excerpt :Finally, the centers of bubbles form a well-designed node set. The parallel node placement by bubble simulation method (PNPBS) has been developed in [16] and is described as follows. After domain decomposition, some bubbles, called boundary bubbles, distributed near the interface of two neighboring subdomains, may interact with the bubbles located in neighboring subdomains.
Adaptive finite element analysis of elliptic problems based on bubble-type local mesh generation
2015, Journal of Computational and Applied MathematicsCitation Excerpt :Instead of utilizing Delaunay triangulation to define the neighborhood of a bubble topologically, an adjacency list is introduced [26] for each bubble which stores the information of neighbor bubbles. Then the parallel version [27] and a series of acceleration strategies [28] for this node placement method have been developed subsequently. Further, based on the generated high-quality nodes set and the information of adjacency lists, the node-based local mesh generation method BLMG has also been presented in [29].
Implementation of Different Types of Meshfree Technique in Computational Solid Mechanics: A Comprehensive Review Across Nano, Micro, and Macro Scales
2024, Archives of Computational Methods in EngineeringResearch on improved node placement method by bubble simulation
2023, Jisuan Lixue Xuebao/Chinese Journal of Computational MechanicsPoint Cloud Generation for Meshfree Methods: An Overview
2023, Archives of Computational Methods in Engineering