Speaker
Description
Modern accelerator facilities generate heterogeneous data from diagnostics, sensors and simulations, making it difficult to manage, reproduce, and contextualize results as software and models evolve. While FAIR principles (Findable, Accessible, Interoperable, Reusable) are increasingly applied to research data, the iterative development of scientific software, with its rich metadata and benchmarks, rarely follows them. Small changes in compiler flags, dependencies, or hardware can alter outcomes, yet are often undocumented. We address this gap with BenchTune, a telemetry and reporting layer in C++ and Python that interfaces with the Kadi4Mat (Kadi) virtual research environment. BenchTune records metadata for each algorithm run, compiler information, parameters, metrics and stores it in Kadi as portable, traceable research artifacts. A project-view plugin in Kadi visualizes the evolution of a research project, linking code versions, datasets, and benchmark runs into coherent workflows.
As a result, our contribution provides a reproducible framework that supports FAIR principles for research data, enabling sustainable and collaborative research in accelerator physics and beyond.
| In which format do you inted to submit your paper? | LaTeX |
|---|