Use of innovative HPC techniques for faster than real time simulations

Abstract: Faster than real time simulations are mandatory for the integration of the numerical models in the core of Early warning systems for natural disasters. Achieving this goal could be possible combing domain decomposition techniques with dynamic or static load balancing algorithms and efficient implementation on new computational architectures, as GPUs. Moreover, the design of low cost Riemann solvers also plays an important role to improve the final efficiency of the numerical models. 

Researchers: EDANYA group

Partner: University of Málaga with collaborations with researchers from the University of Granada and Valencia. 

Related projects: 

Highlighted articles:

  • de la Asunción et al. Numerical simulation of tsunamis generated by landslides on multiple GPUs. Advances in Engineering Software, 99, 59-72, 2016.
  • de la Asunción et al. Simulation of tsunamis generated by landslides using adaptive mesh refinement on GPU. Journal of Computational Physics, 345, 91-110, 2017
  • Escalante et al. Non-hydrostatic pressure shallow flows: GPU implementation using finite volume and finite difference scheme. Applied Mathematics and Computation, 338, 631-659, 2018.
  • J. M. Mantas et al. An introduction to GPU computing for numerical simulation. Numerical simulation in physics and engineering. SeMA-SIMAI series vol 9. Pp. 219-251, 2016. Springer.

iMAT research lines:    RL3. Modeling Environmental Systems & Risk analysis.     RL7. Numerical Analysis.