Clouds and Reproducibility: A Way to Go to Scientific Experiments?

Abstract

Scientific research is supported by computing techniques and tools that allow for gathering, management, analysis, visualization, sharing, and reproduction of scientific data and its experiments. The simulations performed in this type of research are called in silico experiments, and they are commonly composed of several applications that execute traditional algorithms and methods. Reproducibility plays a key role and gives the ability to make changes in the data and test environment of a scientific experiment to evaluate the robustness of the proposed scientific method. By verifying and validating generated results of these experiments, there is an increase in productivity and quality of scientific data analysis processes resulting in the improvement of science development and production of complex data in various scientific domains. There are many challenges to enable experimental reproducibility in in silico experiments. Many of these challenges are related to guaranteeing that simulation programs and data are still available when scientists need to reproduce an experiment. Clouds can play a key role by offering the infrastructure for long-term preserving programs and data. The goal of this chapter is to characterize terms and requirements related to scientific reproducibility and show how clouds can aid the development and selection of reproducibility approaches in science.