Speaker
Description
We are developing an emulator for nuclear many-body calculations based on Eigenvector Continuation (EC), a reduced-basis method applicable to both ab initio Hamiltonians and Energy Density Functional (EDF) approaches.
The method exploits the observation that the parametric evolution of low-lying eigenvectors is confined to a low-dimensional manifold embedded within the full Hilbert space. A reduced variational subspace is therefore constructed from a set of representative eigenvectors computed at selected parameter values. The Hamiltonian can then be projected onto this subspace, allowing the rapid prediction of eigenvalues and eigenvectors for new parameter sets at a significantly reduced computational cost.
This approach provides an efficient surrogate model for nuclear structure calculations while preserving high accuracy, making it particularly attractive for uncertainty quantification, parameter optimization, and large-scale explorations of nuclear interactions.