Abstract of the talk// Quantum computers offer new opportunities for simulating quantum many-body systems, but near-term devices remain strongly limited by noise and computational resources. Developing approaches that extract useful information from imperfect quantum simulations is therefore an important challenge.
In this talk recent work exploring two complementary directions is presented. First, Stefano discusses how quantum simulations can serve as a source of data for classical machine-learning models that learn physically meaningful representations of electronic structure. Second, he introduces algorithmic strategies aimed at reducing resource requirements, including methods that target robust topological properties and fragment-based approaches for decomposing molecular systems. Together, these ideas outline possible routes toward practical quantum-assisted simulations of complex quantum systems.