Fermionic Programmable Quantum Simulators running Variational Quantum Algorithms

Understanding fermionic quantum many-body physics is important for materials and chemistry, but difficult classically due to large computational resources required. Programmable fermionic quantum simulators, such as Fermi-Hubbard tweezer arrays and optical lattices combined with local addressing, offer a promising approach to overcoming these limitations. We show a new way to employ these architectures, whereby running variational quantum algorithms on them. Compared to traditional analog quantum simulation with these platforms, this approach allows one to reach states colder than by cooling technique and to study Hamiltonians beyond those that can be natively created in the experiment. Compared to digital quantum computers, this approach eliminates the need for fermion encoding by qubits, and it may enhance the robustness of quantum noise. In this presentation, I will demonstrate the framework we propose to variationally search the ground state of strongly correlated quantum systems, including onsite-interacting and neighbor-site-interacting Fermi-Hubbard models, the latter of which cannot be natively realized in cold atoms[KH1] . Our methods suggest that the resources needed grow more slowly with increasing accuracy and system size than classical simulations. These findings underscore the potential of fermionic quantum simulations, paving the way for deeper insights into quantum matter.