Cassella Team Solves 10,000-Variable QUBO Problems With Floquet Method as Energy Use Drops 9 Orders
Updated
Updated · Northeastern University · Jul 17
Cassella Team Solves 10,000-Variable QUBO Problems With Floquet Method as Energy Use Drops 9 Orders
1 articles · Updated · Northeastern University · Jul 17
Summary
Cristian Cassella’s team said its Analog Floquet Solver found accurate solutions to QUBO optimization problems with as many as 10,000 variables, including cases previously considered unsolved.
The method targets a key weakness in Ising machines, which often get trapped in local minima—good but not best answers—when searching for the global optimum.
Using Floquet theory, the solver periodically injects energy so the system can keep “jumping” out of shallow solution valleys instead of settling too early.
Cassella said the approach improved energy consumption by nine orders of magnitude and could broaden optimization work in drug discovery, logistics, finance, biology, engineering and wireless communications.