Speaker
Description
Assessing the systematic performance of molecular simulation methods requires carefully curated experimental reference data and, ideally, unbiased evaluation procedures such as blind challenges. QMBENCh (qmbench.net) is a web platform developed in cooperation between the Faculty of Chemistry and the Göttingen State and University Library (SUB) that aims to support both. It hosts curated benchmark databases and organizes blind test challenges in which theoretical groups predict experimental observables without prior knowledge of the target values. In this talk, we discuss the motivation behind the platform, its development process as a collaboration between domain scientists and research software developers at the SUB, and the practical design choices made to ensure long-term accessibility and community participation. We present experiences from challenges conducted so far and outline planned extensions of the platform. The initiative is supported by the DFG-funded RTG2455 and NFDI4Chem.