22 Innovation Studies have started

Commencing on February 1, Inno4scale initiated 22 Innovation Studies across Europe, focusing on computational fluid dynamics, materials science, artificial intelligence, and other critical domains. These studies, spanning a maximum of 12 months, aim to translate mathematical concepts into novel algorithms feasible with exascale systems, showcasing their potential through proof-of-concept demonstrations.

The EuroHPC Joint Undertaking has provided funding for the Inno4scale initiative, targeting the development of applications capable of fully exploiting the computational power of the forthcoming EuroHPC exascale supercomputers. Inno4scale has initiated the Innovation Studies across Europe, with the primary goal of translating mathematical concepts into innovative algorithms enabled by exascale computing.

The Inno4scale Innovation Studies focus on advancing scientific computing by developing novel concepts for exascale supercomputers. A key objective is to implement computational solutions for significant numerical problems in scientific computing. This involves translating mathematical concepts into new algorithms that leverage the heightened performance of exascale systems. The studies aim to conduct high-quality research and development, culminating in proof-of-concept demonstrations that highlight the transformative potential of exascale computing for critical applications.

The Inno4scale Innovation Studies primarily concentrate on applications in computational fluid dynamics and materials science. Additionally, they address topics in plasma physics, partial differential equations, molecular dynamics, graph neural networks, biomedicine, and wind farm simulation. Nearly half of the studies integrate artificial intelligence to achieve their research goals, demonstrating the interdisciplinary nature of the initiative, which goes beyond traditional scientific computing domains. The AI topics used for the research work in the studies consist mainly of machine learning and partly of neural networks.

Application areas for the 22 selected proposals.
Application areas for the 22 selected proposals
Technical approaches for the 22 selected proposals.
Technical approaches for the 22 selected proposals
AI topics for the 22 selected proposals for the Innovation Studies. 59% are non AI topics. 27% are machine learning, of which 9% are for linear algebra. 14% are neural networks.
AI topics for the 22 selected proposals for the Innovation Studies

Inno4scale received a total of 51 proposals from 19 European countries, reflecting the widespread interest and engagement across Europe. A panel of expert evaluators assessed the proposals, leading to the selection of 22 studies for funding. The high quality of the proposals exceeded expectations, resulting in more research studies being funded than initially anticipated. In total, the Inno4scale Innovation Studies take place in 15 European countries and are funded with a budget of approximately €4 million. 9 of these countries are coordinating the consortia of organisations across Europe that carry out the studies.

Map of participating and coordinating countries for all the proposals that where submitted to the Inno4scale project.
Map of participating and coordinating countries. For each country: # of coordinating org | # of participating org

The results of the innovation studies can make a significant step towards harnessing the potential of exascale supercomputers for scientific computing applications. Through them, Inno4scale aims to pioneer new algorithms and computational solutions, demonstrating the transformative capabilities of exascale systems.