Chaotic effects in cosmological hydrodynamical simulations and archive

This PhD thesis addresses two distinct challenges in the field of astrophysical simulations: quantifying variability in simulations and developing standards- compliant data-sharing frameworks for simulation outputs. In the first part, I investigate the chaotic effects inherent in galaxy cluster simulations. To understand the range of variability that arises from minor differences in initial conditions and numerical precision, I ran a set of identical simulations on the HotCat computing cluster using the OpenGadget3 code. I focused on quantifying the variation in galaxy properties by matching galaxies across runs and analyzing the variations using two statistical methods. My findings for low resolution galaxy cluster simulation indicate that noise, primarily due to Poisson sampling, is the dominant source of variability in galaxy properties, with only minimal run-to- run differences observed in baseline and feedback tests. Stronger stellar feedback result in an increase in stochastic noise within given simulation and this correlation depends on galaxies within specific mass bins that are more sensitive to the respective feedback processes. These insights set the first foundational basis for understanding the limitations of numerical simulations and set the stage for studying the effects of chaos for accurately interpreting results from cosmological simulations. The second part of this thesis focuses on enhancing the accessibility and interoperability of simulation data by developing a standards-compliant archive for astrophysical simulations. Recognizing the need for efficient data sharing, the primary objectives for the part 2 of my thesis were established. For this work, I implemented a custom database and interface as a foundational step, followed by the integration of key Virtual Observatory (VO) data standards, such as VOTable for tabular data representation, Universal Worker Service (UWS) for managing long-running processes, and preliminary templates based on the Simulation Data Model (SimDM). However, there were several challenges encountered with VO standards, particularly regarding the SimDM and SimDAL (Simulation Data Access Layer) standards, which lack explicit guidelines for managing large, distributed datasets typical of cosmological simulations. In collaboration with VO standards authors, a systematic review will be initiated to address these limitations, with a specific focus on adapting SimDM templates for complex simulation datasets. This part of my work contributes to the broader goals of the FAIR (Findable, Accessible, Interoperable, and Reusable) data principles by refining the standards and practices for astrophysical simulation data sharing, ultimately advancing the capabilities of researchers to analyze, interpret, and build on existing simulation results.