A Search Space Exploration Framework for e-Science Applications

Abstract

High Performance Computing (HPC) has always been a fundamental component to conduct scientific experiments. Model calibrations/simulations often require several executions of scientific applications by changing their input parameters. This process is a common practice in research even though it represents a tedious and error-prone task. In this paper we propose Copper framework which employs a black-box strategy and contains a set of plugins to accelerate user experiments for exploring search spaces in HPC parametric applications. Copper has been used to conduct scientific experiments in different areas including, agriculture, oil gas, flood simulation, and bioinformatics.