James Keeble, "Topical Seminar: Neural Quantum States for Nuclei: Testing the Limits of Classical Computing"

por Dr. James Keeble (Bielefeld University)

lunes 27 abr. 2026 12:00 → 13:30 Europe/Madrid

1001-Primera-1-1-1 - Paterna. Seminario (Universe)

In this talk, I will discuss the application of so-called Neural Quantum States, which use machine-learning techniques to represent quantum many-body wave functions. These approaches have been shown to represent tensor-network states and states with volume-law entanglement efficiently. However, the key property underlying the limitations of Neural Quantum State techniques remains an open question.
Within quantum information theory, among all possible unitary transformations that can be applied to an 
N-qubit state, there exists a subgroup known as the Clifford unitaries. Although these transformations can generate entanglement, they are entirely classically simulable and therefore provide no quantum advantage. As a result, quantifying the non-stabilizerness of a given computational approach is important for assessing its expressibility in representing arbitrary quantum many-body states.
I will focus in particular on recent results highlighting the limitations of Neural Quantum States in representing the ground-state wave functions of atomic nuclei in the nuclear shell model within variational Monte Carlo calculations, thereby testing the limits of classical computing in strongly correlated quantum systems.