Tensor cross interpolation and other algorithms for learning tensor networks
6-10 oct. 2025
House of Quantum Grenobles Alpes.
Y.SPOT building. 5 pl. Nelson Mandela, 38000 Grenoble - Grenoble (France)
Tensor network techniques are well-established numerical tools for solving seemingly exponentially hard problems in quantum many-body physics with important successes in the simulation of model correlated electronic systems, quantum magnetisms and quantum computers. In this tensor network historical use, the quantum many-body wave function is compressed, using tensor networks, allowing for a considerable speed-up of an otherwize impossible calculation.
In the last few years, a crucial piece of learning algorithm (e.g. the Tensor Cross Interpolation algorithm) has emerged. These techniques can learn a tensor network representation of a mathematical object a priori unrelated to tensors. These new algorithms provide a path to transfer all the technology that have been developped for the quantum many-body wave function to an entirely new set of mathematical problems: Integrals in high dimensions, Partial differential equations, function optimization. New applications for e.g. computational fluid dynamics, basis set in chemistry, Feynman diagram calculations, plasma physics…are quickly emerging.
This workshop will gather important players in this emerging field, from both the applied mathematics and theoretical physics communities, to discuss these exciting new applications and algorithms. The schedule will be kept relatively light to leave as much room for scientific discussions as possible.
Discipline scientifique :
Algèbres quantiques - Physique Quantique
Lieu de la conférence