Synergy between Parallel Computing, Optimization and Simulation
• 大类 : 工程技术 - 4区
• 小类 : 计算机：跨学科应用 - 4区
• 小类 : 计算机：理论方法 - 3区
Plenty of hard optimization problems in a wide range of areas such as health, manufacturing and production, logistics and supply chain management, energy, engineering, etc. are increasingly large and complex in terms of number of input parameters, decision variables, objective functions and landscape complexity. These problems are often tackled using optimization approaches including greedy algorithms, exact methods (dynamic programming, Branch-and-X, constraint programming, A*, etc.) and metaheuristics (evolutionary algorithms, particle swarm, ant or bee colonies, simulated annealing, Tabu search, etc.). Solving efficiently and effectively large and complex problems requires the joint use of these approaches and massively parallel heterogeneous computing.
In addition, for many real-world problems (e.g. in engineering design) the evaluation of the objective function(s) often consist(s) of the execution of an expensive simulation as a black-box complex system. This is for instance typically the case in aerodynamics where a CFD-based simulation may require several hours. Optimization algorithms, even if combined with parallel computing, fail to solve these simulation-based optimization problems. A typical approach to deal with the computational burden is to use (instead of real values) cheaper data-driven approximations of the objective function(s) socalled surrogates or meta-models. A wide range of surrogates was applied during the last decade including: classical regression models such as polynomial regression or response surface methodology, support vector machines, artificial neural networks (ANN), radial basis functions, and kriging or Gaussian process. ANN is probably the most prevalent of them with the recent "explosion" of the deep learning popularized thanks to GPGPUs.
This special issue seeks to provide an opportunity for researchers to present their original contributions on the joint use of advanced single- or multi-objective optimization methods, simulation and/or its data-driven approximation, and distributed and/or parallel multi/many-core computing, and any related issues. The special issue topics include (but are not limited to) the following:
Parallel exact optimization/metaheuristics for solving complex problems on multi-core processors, accelerators/co-processors (e.g. GPU), clusters, grids/clouds, etc.
(Parallel) hybrid algorithms combining optimization and simulation.
(Parallel) surrogate-assisted optimization.
Implementation issues of methods combining optimization, simulation and/or parallel computing.
Software frameworks for the design and implementation of parallel and/or distributed simulation and/or surrogate-assisted techniques.
Simulation-based optimization applications including healthcare, manufacturing, logistics, biological applications, advanced big data analytics, engineering design, etc.
Computational/theoretical studies reporting results on solving complex problems using the joint use of parallel computing, optimization and simulation.