Marie Gueguen (University of Western Ontario)
Simulations are a crucial part of evaluating the standard cosmological model. Understanding in which case a simulation can succeed in (dis)confirming a model is, however, still a challenge. When simulations do not match observations, it is not clear how to determine whether the discrepancy stems from numerical artifacts or constitutes a genuine failed prediction. In this talk, I evaluate the merits and pitfalls of robustness analysis and suggest an alternative to this way of assessing the reliability of simulations. By doing so, I aim to formalize and generalize the methodology developed by cosmologists themselves against numerical artifacts.
