Computational Complexity of Highly Nonlinear Approximations
Grant period
2025-07-01 - 2030-06-30
Funding body
European Union
CORDIS
Call number
ERC-2024-COG
Grant number
101170147
Identifier
G:(EU-Grant)101170147
Note:
Performance and accuracy guarantees of numerical methods for high-complexity problems. Partial differential equations are essential to describing virtually all processes from environmental to industrial ones. Numerical methods are techniques to solve them approximately. However, as the number of independent variables (dimension) increases, specialised numerical methods are required to achieve tractable computational costs. The ERC-funded COCOA project aims to enable better performance and accuracy in solving such highly complex problems. Researchers will focus on mathematically rigorous methods, based in particular on highly non-linear low-rank tensor representations, neural networks, and linear combinations of arbitrary Gaussian functions. The team will also attempt to identify the most suitable methods for specific partial differential equation problems; construct methods to avoid numerical instabilities; and ensure reliability of results in high dimensions.
Recent Publications
There are no publications
Datensatz erzeugt am 2026-06-25, letzte Änderung am 2026-06-25