Robust Parallel Algorithm Research Articles

A robust parallel algorithm for Eulerian-Lagrangian method crossing sliding non-conformal interfaces

Numerical simulations of flows with dilute disperse phase play an important role in a wide range of industrial applications. The most commonly used technique for solving such problems is the Eulerian-Lagrangian approach, in which individual particles are tracked inside the domain. However, the use of sliding and moving mesh in complex engineering applications poses more challenges for the parallelization of the Eulerian-Lagrangian method in the two-way coupling. In this context, this manuscript presents the parallel algorithm for tracking Lagrangian particles crossing sliding non-conformal interfaces in a mixed Eulerian-Lagrangian CFD algorithm, which is based on a parallel supermesh method introduced recently for the handling of sliding mesh in large-scale parallel computing (Y. Xiao and P.J. Ming, Jour. Computat. Phys., 2022). In the proposed parallel algorithm, critical problems such as geometric calculation for particles across non-conformal mesh interfaces and the dynamic communication for particle data can be handled based on the one-to-one addressing constructed by the supermesh method introduced in the author's previous work. Further, a robust and easy-to-implement algorithm is also provided for particle communication caused by mesh moving. All methodologies are integrated and implemented for parallel computation obtaining good robustness and consistency. Specific examples are solved to examine the performance of this parallel strategy when applied to practical engineering problems.

A scalable, robust parallel algorithm on handling of sliding non-conformal interfaces with an efficient supermesh method

Handling of non-conformal interfaces communication is one of the challenges from which the parallelization of multi-zone unstructured grids technology suffers. Especially for dynamic mesh simulation, changing dynamic communication patterns induced by the grids relative motion requires data to be correctly transferred across discrete non-conformal interfaces at every time step. To overcome the challenges at extreme scales in this paper we attempt to design the parallel algorithm in such a way that it enhances the scalability of the sliding mesh method implemented in the simulation of practical engineering problems with complex geometries. In the proposed parallel algorithm, the issue of changing dynamic interface communication is handled by a scalable and robust parallel supermeshing algorithm which has no constraints on the domain partition schemes and is adapted to complex practical engineering simulation. Besides, a unified cell-based communication map adapting to parallelized matrixes solvers is dynamically reconstructed by assembling both the inner-boundary communication map and interface communication map, which greatly improves the robustness and scalability of the sliding mesh algorithm executed on the parallel CFD code. All methodologies are integrated and implemented for parallel computation obtaining good scalability and computational efficiency. Specific examples are solved to analyze the conservation, scalability and robustness of the parallelized interface approach and finally, the full-cycle unsteady simulation of a four-stroke engine is tested to evaluate the performance of this parallel strategy when applied to practical engineering problems.

A robust parallel algorithm of the particle swarm optimization method for large dimensional engineering problems

The application of the Particle Swarm Optimization (PSO) method to large engineering problems is strongly limited by the required computational cost. This limitation comes from the large number of particles needed to optimize the many-variable function, the high computational cost of its evaluation and the lack of an adequate criteria to early detect the approach of the global optimum. The first two cost sources can be mitigated by an efficient parallel implementation of the PSO method but the last one need the development of a robust convergence criterion for the algorithm. This work develops an efficient and robust optimization method by using a new convergence criterion in an asynchronous parallel implementation of PSO. In the optimization of benchmark test functions, this method showed very good performance, with parallel efficiency between 80% and 100%, and excellent robustness, always detecting the global optimum. Finally, the method was successfully applied to an actual estimation problem with 81 parameters.

ROBUST SCALABLE VISUALIZED CLUSTERING IN VECTOR AND NON VECTOR SEMI-METRIC SPACES

We describe an approach to data analytics on large systems using a suite of robust parallel algorithms running on both clouds and HPC systems. We apply this to cases where the data is defined in a vector space and when only pairwise distances between points are defined. We introduce improvements to known algorithms for functionality, features and performance but review state of the art as this is not broadly familiar. Visualization is valuable for steering complex analytics and we discuss it for both the non vector semi-metric case and for clustering high dimension vector spaces. We exploit deterministic annealing which is heuristic but has clear general principles that can give reasonably fast robust algorithms. We apply methods to several life sciences applications.

Vector Fitting Based Adaptive Frequency Sampling for Compact Model Extraction on HPC Systems

This paper proposes an accurate and robust parallel algorithm for the extraction of reduced order electromagnetic models for passive components, by taking into consideration high frequency field effects. It aims at an extraction time as low as possible while keeping the accuracy control of the extracted model. The core of the reduction procedure is based on vector fitting (VF) rational approximation of the circuit functions, which are computed by solving the field equations with the finite integration technique (FIT) in a minimal list of frequency samples. The computationally intensive part of algorithm is efficiently mapped on a nonhomogeneous hierarchical multiprocessor architecture.

Effective extraction of accurate reduced order models for HF-ICs using multi-CPU architectures

In this article a fast, accurate and robust parallel algorithm for the extraction of reduced-order models for passive parts of high-frequency integrated circuits is proposed. The core of the reduction procedure is based on vector fitting rational approximation of the circuit functions, which are obtained by finite integration technique. In order to reduce the global computational effort, an original adaptive frequency sampling is used to generate an optimal list of frequency samples. The computationally intensive part of the algorithm is efficiently mapped on a hierarchical multiprocessor architecture. The goal is to reduce the extraction time and to control the accuracy of the extracted reduced-order model.

Robust Parallel Algorithms for Solution of Riccati Equation

Robust and efficient parallel-vector algorithms are presented for the solution of the Riccati equations encountered in optimal control problems on shared-memory multiprocessor machines. The algorithms have been implemented on a Cray YMP 8/8128 and applied to three large problems resulting from a continuous bridge structure, a 21-story space truss structure, and a 12-story space moment-resisting building structure. Efficiency of the algorithms is presented in terms of millions of floating point operations per second (MFLOPS) and the speedup. The MFLOPS for the largest example resulting from the 12-story space frame structure is a high 206.0. The speedup due to parallel processing only (for the same example) using seven processors, is 6.33. When vectorization is combined with parallel processing a very significant speedup of 54.4 is obtained using seven processors. The algorithms developed in this research find applications in the complex integrated control/structural optimization problem. Further, the writers are currently using them to develop large adaptive/smart structures.

Comments on "parallel algorithms for finding a near-maximum independent set of a circle graph" [with reply

The authors refers to the work of Y. Takefuji et al. (see ibid., vol.1, pp. 263-267, Sept. (1990)), which is concerned with the problem of RNA secondary structure prediction, and draws the reader's attention to his own model and experiments in training the neural networks on small tRNA subsequences. The author admits that Takefuji et al. outline an elegant way to map the problem onto neural architectures, but suggests that such mappings can be augmented with empirical knowledge (e.g., free energy values of base pairs and substructures) and the ability to learn. In their reply, Y. Takefuji and K.-C. Lee hold that the necessity of the learning capability for the RNA secondary structure prediction is questionable. They believe that the task is to build a robust parallel algorithm considering more thermodynamic properties in the model.

An efficient parallel scheme for minimizing a sum of Euclidean norms

The problem of minimizing a weighted sum of Euclidean norms is considered. Applications include minimal-surface computations. A robust parallel algorithm, based upon a line-search Newton's method, is presented which takes full advantage of the structure of the problem in order to fully utilize vectorization and concurrency in the computations. The proposed method can achieve good performance, especially on a machine with an architecture that combines vector and parallel capabilities on a two-level shared memory structure such as that on the Alliant FX/8 system. Performance results are given on the Alliant to illustrate the efficiency of the algorithm.

Matrix Quadratic Solutions

Matrix quadratic equation solution derivation applied in finding steady state solutions of Riccati differential equations with constant coefficients