Computational Science Top Journals

 Computational science applications use a big amount of the available high performance computing (HPC) resources. A typical breakdown of the kinds of computational science research areas represented on HPC resources is presented in Fig. 10.1. This summary of HPC allocations originates from the acute Science and Engineering Discovery Environment (XSEDE) virtual system, which integrates 12 very large HPC resources to be utilized in peer-reviewed research. While these applications are utilized in a good sort of very different disciplines, their underlying computational algorithms are frequently very almost like each other. As a consequence, several software libraries are developed for HPC resources to fill a specific computing need, so application developers don't got to waste time redeveloping supercomputing software that has already been developed elsewhere. Subsequently, these libraries end up becoming required software dependencies across many user applications, and their performance and usage become critically important for an application's performance. Libraries targeting numerical linear algebra operations are the most common, given the ubiquity of linear algebra in scientific computing algorithms. Other libraries target operations like input/output (I/O), fast Fourier transform (FFT), the finite element method, and solving ordinary differential equations. These libraries have generally been highly tuned for performance, often for quite a decade, making it difficult for the casual application developer to match a library's performance using a homemade equivalent. On account of their simple use and their highly tuned performance across an honest range of HPC platforms, the use of scientific computing libraries as software dependencies in computational science applications has become widespread.