Northeastern professor pushes computers to solve ‘unsolvable’ problems

Northeastern University professor Cristian Cassella is making waves in the world of computing by tackling a specific class of problems known as quadratic unconstrained binary optimization, or QUBO. At their core, these problems are vast collections of yes-or-no decisions. Whether you are managing thousands of ride-sharing driver assignments or attempting to map the complex folding patterns of proteins, the goal is to find the single most efficient outcome among billions of possibilities. Finding the absolute best solution has long been an exercise in frustration because standard methods frequently get stuck in local minima, which are essentially false summits that prevent machines from reaching the true, ideal result.

To solve this, researchers often look to Ising machines. These systems draw inspiration from physics, specifically how electrons in materials like iron align themselves. By mapping a problem to the spin state of electrons, computers can search for the lowest energy state, which correlates to the most efficient mathematical solution. However, the limitation of traditional Ising machines is their tendency to stop at those local minima, leaving the job unfinished.

Cassella and his team introduced the Analog Floquet Solver to bypass this limitation. By applying a periodic force to the system, the method acts as a nudge that pushes the computation out of those shallow ruts. It allows the system to continue searching until it finds the global minimum, representing the actual best solution. In testing, this approach solved equations with 10,000 variables at record speeds while cutting energy consumption by nine orders of magnitude.

The implications for this breakthrough extend far beyond the laboratory. By providing a reliable way to compute high-complexity variables, this method holds potential for fields ranging from pharmaceutical development to urban logistics and financial modeling. As researchers continue to refine these techniques, the barriers currently standing between us and the answers to previously unsolvable problems are beginning to drop.