Modeling and simulation

Updated on 02/05/2024

Automobile construction, weather forecasting, traffic management, environmental sciences, finance, space… Since the calculation of ephemerides several centuries ago, mathematical modelling has developed immeasurably to facilitate, through abstraction, the understanding of the world around us. In recent decades, with the advent of the computer, digital simulation has made it possible to calculate solutions to models and to use them to reproduce physical reality, typically making forecast of reality. Today, the power of computers is such that the concept of the digital twin has been born, making it possible to use simulation to massively and brutally optimise the parameters of the initial model and thus at least partially replace real experiments. We have thus moved from the era of fine-tuned understanding to that of generalised anticipation.
But these digital models, which are based on increasingly complex algorithms, also have a cost: they require an ever-increasing amount of resources and energy to make them work efficiently. This has given rise to new perspectives for the men and women of Inria, who are interested in so-called reduced (simplified) or hybrid models (coupled with artificial intelligence, for example) with the arduous but stimulating objective of combining performance and sobriety.