High-Performance Computational and Information Technologies for Numerical Models and Data Processing

This chapter discusses high-performance computational and information technologies for numerical models and data processing. In the first part of the chapter, the numerical model of the oil displacement problem was considered by injection of chemical reagents to increase oil recovery of reservoir. Moreover the fragmented algorithm was developed for solving this problem and the algorithm for high-performance visualization of calculated data. Analysis and comparison of parallel algorithms based on the fragmented approach and using MPI technologies are also presented. The algorithm for solving given problem on mobile platforms and analysis of computational results is given too. In the second part of the chapter, the problem of unstructured and semi-structured data processing was considered. It was decided to address the task of n-gram extraction which requires a lot of computing with large amount of textual data. In order to deal with such complexity, there was a need to adopt and implement parallelization patterns. The second part of the chapter also describes parallel implementation of the document clustering algorithm that used a heuristic genetic algorithm. Finally, a novel UPC implementation of MapReduce framework for semi-structured data processing was introduced which allows to express data parallel applications using simple sequential code.