Speaker
Rocco Barač
(University of Padova & University of Cyprus)
Description
Quadratic unconstrained binary optimization (QUBO) problems arise naturally in spin glass models like the Sherrington-Kirkpatrick Hamiltonian, where strongly frustrated interactions create rugged energy landscapes. In this talk, I explore how the density matrix renormalization group (DMRG), the famous tensor network method used as a ground-state search algorithm, can be adapted as a heuristic QUBO solver. By introducing a weak transverse field, DMRG gains the ability to escape local minima. I will also discuss how these results compare to those of Gurobi optimizer, a state-of-the-art classical solver that struggles with strongly frustrated dense problems, and whether DMRG can offer an advantage.
Author
Rocco Barač
(University of Padova & University of Cyprus)