Building a Reproducibility Check Service for Computational Research

Reproducibility remains a persistent challenge in computational research, despite increasing requirements from journals for authors to share underlying software and data. While such policies are a step forward, they do not guarantee that published results can be independently reproduced. In this talk, we present the Reproducibility Check initiative, developed by the TU Delft Digital Competence Center (DCC) in collaboration with 4TU.ResearchData repository. The initiative targets TU Delft researchers who deposit the software and data underlying their publications in 4TU.ResearchData. Deposits are screened and authors are invited to participate in a reproducibility check, in which RSEs from the DCC attempt to reproduce the published results using the CODECHECK process (Nüst and Eglen, 2021). We will share insights from the pilot phase, running since spring 2025, highlighting common challenges and strategies to address them, including approaches to engaging researchers and lowering participation barriers. Based on these experiences, we provide recommendations for institutions and RSE teams interested in implementing similar initiatives.