Climate data transformation for optimization

Abstract

Climate change is the defining environmental challenge facing our planet. One challenge that climate researchers face is the scalability issue since the climate data are normally colossal and complex. Our research goal is taking on the scalability challenge - helping the climate group at the U transform their code for optimization. The climate data research group at the U wrote their programs mainly in Matlab and C/C++. Our proceeding is parsing their programs written in Matlab and automatically generating parallel C code. We use Silver and Copper to parse the language and generate new C code and use OpenMP to enable paralleling computing. Silver and Copper are the language tool developed by the MELT software group at the U that is used to specify the language grammar and support language extensions. OpenMP is a programming tool that enables different section of code/data running concurrently on multiple processors to execute a parallel program. Upon generating the parallel C code, we are able to run the original program faster and solve larger problems, thus making the problem more scalable.