Supercomputational environmental acoustics

摘要

In the language of computer hardware engineers, "supercomputing" means using advanced, highly specialized technologies, to build super fast machines; one can call the super fast computing machines "supercomputers". In the language of computer software developers, "supercomputing" may have another meaning. It may mean using advanced, highly specialized knowledge and techniques, to develop high-performance computer codes, or in another terminology, computer models. Such computer models could be designed, at will, to solve different real problems arising in all branches of science. This paper centers the attention at solving environmental acoustics problems by high-poerformance computer codes involving supercomputing. There are a number of phases to go through in order to produce an applicable high-performance computer code. One world that can be used to describe the whole task is MODELLING. The initial phase involved in modelling is mathematical. It requires the formulation of environmental phenomena into a mathematical model. The next phase is to develop efficient solution techniques for the mathematical model which are implementable into a computer code. These efficient solution techniques are usually numerical; thus, a lot of computations are required. The next phase is the construction of accurate and applicable computer models. Desirably, by using the computer model to solve large-scale environmental acoustics problems the required computations are not only accurate but also the computation speed is fast. This is where "supercomputing" can play an important role. Mathematical modelling considerations and its requirements are called to the modellers attention and discussions are also given to emphasize the importance of developing accurate solutions. How can "supercomputing" play an important role? The question is answered by using a realistic large-scale ocean acoustic propagation problem for discussion. It can be easily seen that this realistic problem motivated meaningful mathematical modelling, called for the development of efficient numerical solutions, challenged the construction of high-performance computer codes, and required supercomputing. Briefly discussed will be supercomputing by different supercomputers.