Skip to main content
Go to the home page of the European Commission (opens in new window)
English en
CORDIS - EU research results
CORDIS

Learning Dynamical Systems with Linearizing Representations

Objective

Our ability to understand scientific dynamical systems is failing to keep pace with our power to simulate them. From protein folding to the Earth's climate, the properties of key physical, chemical, and biological systems are determined by dynamical patterns that are increasingly hard to identify and understand from simulations and observations alone.

Project LINDY proposes a paradigm shift to meet this challenge, with the central thesis that characterizing the collective behaviours of complex dynamical systems is, fundamentally, a representation learning problem. LINDY will study machine-learning dynamical models that map the system's state into a latent representation where the dynamics becomes globally linear. This linearization is key, as it enables the decomposition of the dynamical evolution into physically meaningful modes, unlocking powerful analysis, forecasting, and decision-making capabilities. LINDY's approach is underpinned by the theory of linear evolution operators — also known as Koopman or transfer operators — for which powerful statistical learning guarantees have recently been established.

This project is first and foremost an effort to break new ground on machine learning for dynamical systems, leveraging modern learning theory and methods to produce reliable and interpretable dynamical models. LINDY's overarching goal is to speed up and automate the identification and understanding of complex dynamical patterns, ultimately serving as a catalyst for scientific discovery.

Fields of science (EuroSciVoc)

This project has not yet been classified with EuroSciVoc.
Be the first one to suggest relevant scientific fields and help us improve our classification service

Keywords

Programme(s)

Topic(s)

Funding Scheme

HORIZON-ERC - HORIZON ERC Grants

See all projects funded under this funding scheme

Call for proposal

(opens in new window) ERC-2026-STG

See all projects funded under this call

Host institution

FONDAZIONE ISTITUTO ITALIANO DI TECNOLOGIA
Net EU contribution
€ 1 499 625,00
Address
VIA MOREGO 30
16163 Genova
Italy
Region
Nord-Ovest Liguria Genova
Activity type
Research Organisations
Links
Total cost
€ 1 499 625,00

Beneficiaries (1)