We support innovation and research by providing multi-scale models of complex systems, integrating electronic and atomistic properties to nanoscale, mesoscale, microscale and device-scale features. Multi-scale integration relies on the definition of the interrelationships connecting hierarchically different time and scale domains. In each domain, realistic models are defined through state-of-the-art computational techniques, including electronic structure (density functional theory) calculations, molecular dynamics, coarse-grained simulations and finite-element methods. The interconnection among domains is realized by developing a modelling ontology, developed upon chemico-physical criteria. We apply this approach to understand and develop innovative platforms in materials science and technology. In the research news, applications of our approach in fields ranging from energy, health, manufacturing, are showcased.

Recent developments of Artificial Intelligence and high-level implementations on high-performance computing infrastructures have allowed significant progresses in several fields. We apply machine learning to predictive modelling of materials and devices. Machine learning and neural networks boost our ability to develop and predict the properties of novel materials and to optimize processes and devices.  The automatic elaboration of vast amounts of modelling data implies strict requirements for the representation of the information. We work at the definition and implementation of ontologies of materials and devices, which allow a consistent representation of modelling data across different scales and simulation methods. This step enables the development of simulation frameworks in several application domains.