Simulation Verification in Practice

Abstract

AbstractWith the increased use of simulations as investigative tools in various scientific fields, the question naturally arises as to how these simulations are epistemically justified. One natural approach is to insist that the numerical aspects of simulation justification be performed separately from the physical aspects, but Winsberg (2010) has argued that this is impossible for highly complex simulations. Based on a survey and close examination of a range of astrophysical MHD codes and their attendant literature, I argue that insisting on a strict separation of these aspects of simulation justification is neither epistemically necessary nor advisable.