Learning how a physical system moves is not only a question of fitting data. A useful model should also respect the mathematical structure behind that motion. This connection between geometry and learning featured at the VIII Iberoamerican Meeting on Geometry, Dynamics and Field Theory, held in La Laguna, Tenerife, from 19 to 23 January 2026.

Miguel Vaquero presented Learning Hamiltonian Dynamics, exploring how geometry-aware learning and geometric numerical integration can work together to learn physical dynamics while preserving their underlying structure.

The programme also included IE Research Datalab researcher Alexandre Anahory, whose short talk examined the mechanical Hamiltonization of unreduced Chaplygin systems. Together, these contributions highlighted links between geometry, computation and the mathematical description of motion.

Official event details and programme.

Campus photograph: Emeterio Suárez, University of La Laguna Central Building, CC0.

Miguel Vaquero and Alexandre Anahory are members of IE Research Datalab at IE University.