Eflows4HPC

Recently, the simulation of complex engineering problems has become possible due to modern simulation techniques. However, this capability comes at the cost of high computational requirements that are often incompatible with the deployment in computationally constrained environments.

Reduced Order Models (ROMs) provide a possible solution to this limitation by taking the output of the “Full Order Models” (FOMs) and identifying the common patterns allowing similar problems to be solved at a much-reduced computational cost.

SISSA mathLab is part of “Pillar 1” of the Eflows4HPC project that aims to enable the effective usage of large-scale HPC hardware. With other partners, we are responsible for speeding up the generation of ROMs enabling the execution of large industrial problems hence providing a flexible workflow for the construction of reduced order models to be used in defining Digital Twins for manufacturing applications.

Keywords: Reduced Order Models (ROM); Machine Learning (ML); High-Performance Computing (HPC); Parallel Computing; Digital Twins

Project Duration: 1 January 2021 / 31 December 2023

WEB: https://eflows4hpc.eu

Funding Scheme: European High-Performance Computing Joint Undertaking (EUROHPC JU)

Grant Agreement ID: 955558

Overall Budget: € 7,66M

Partners: