Turbulence characterises many physical phenomena and is key to astrophysical and cosmological scenarios. Despite its success in simulating astrophysical scenarios, the smoothed particle hydrodynamics technique (SPH) has historically struggled when simulating turbulence.
Classical SPH implementations showed to be too dissipative and inaccurate to resolve the subtle scales and energy flow of the Kolmogorov cascade. Moreover, simulating turbulence is impossible without HPC, due to the large numbers of fluid elements and long simulated physical times. SPH-EXA uniquely combines novel SPH methodology with extreme scalability.
The project will use it to conduct the largest and most accurate to-date turbulence + self-gravity simulations. The objective is to study the formation of stellar cores and their initial mass function at unprecedented resolution. The team will also study turbulent mixing aiming to advance the knowledge regarding the origin of the apparent chemical homogeneity of stellar clusters. This study will contribute to the general theory of turbulence measuring the Lyapunov exponent, facilitated by the SPH Lagrangian nature. The project will also extract performance data to characterise the evolution of load imbalance and mitigate it with effective spatial and temporal load balancing techniques.
This study will impact (without being limited to) astrophysics, cosmology, fluid dynamics, and computer science.
University of Zurich, Switzerland;
University of Basel, Switzerland;
CSCS / ETH Zurich, Switzerland;
University of Zürich, Switzerland;
University of Heidelberg, Germany;