How and where the TeraGrid supercomputing infrastructure benefits science

Abstract

We investigate how the benefits of the TeraGrid supercomputing infrastructure are distributed across the scientific community. Do mostly high-impact scientists benefit from the TeraGrid? Are some scientific domains more strongly represented than others in TeraGrid-supported work? To answer these questions, we examine the relation between TeraGrid usage and scientific impact for a set of scientists whose projects relied to varying degrees on the TeraGrid infrastructure. For each scientist we measure TeraGrid usage expressed in terms of allocated Service Units (SU) vs. various indicators of their scientific impact such as the h-index, total citations, and citations per article. Our results show a significant correlation between scientific impact and TeraGrid usage. We furthermore examine the distribution of TeraGrid-related publications across various scientific journals. A superposition of these journals over an existing large-scale map of science shows how TeraGrid-supported work is mostly concentrated in Physics and Chemistry, with a lesser focus on biology.