A newly published study from the University of Southern California (USC) has provided strong evidence that quantum computers can outperform classical supercomputers in solving complex optimization problems, marking a significant milestone in the field of quantum computing known as quantum advantage.
The research, published in Physical Review Letters, focuses on quantum annealing, a specialized form of quantum computation that identifies low-energy states in a system—these states correspond to optimal or near-optimal solutions. While previous efforts have aimed to demonstrate quantum advantage in exact optimization, this study shifts focus to approximate optimization, where finding a solution close to the best possible one is often sufficient for practical purposes.
Rather than requiring exact answers, the study assessed how efficiently quantum annealing could identify solutions within a certain threshold—typically within 1% of the optimal value. This makes the findings especially relevant to real-world scenarios, such as financial portfolio optimization or logistics planning, where “good enough” solutions are often acceptable and more cost-effective than perfect ones.
For example, when constructing a mutual fund, the goal may be to outperform a market index rather than to find the absolute best-performing combination of stocks. In such cases, approximate optimization provides tangible value.
A longstanding challenge in the field has been proving quantum scaling advantage—the idea that a quantum computer’s advantage grows as problem size increases. While theoretical models have suggested this potential, experimental validation had remained elusive, especially in quantum annealing systems.
To address this, researchers at USC used the D-Wave Advantage quantum annealer, located at USC’s Information Sciences Institute. The device was tested against PT-ICM (parallel tempering with isoenergetic cluster moves), the most efficient known classical algorithm for these types of problems.
One major obstacle was quantum noise, which can degrade performance and obscure any computational advantage. To mitigate this, the team employed quantum annealing correction (QAC), a technique that suppresses errors and enables the creation of over 1,300 stable, logical qubits. This error correction proved essential in outperforming classical methods.
The performance of the quantum and classical systems was compared using a method known as “time-to-epsilon”, which measures the time required to reach a solution within a given margin of the optimal result. The study centered on two-dimensional spin-glass problems, which are complex optimization tasks originating from statistical physics and involving disordered magnetic systems.
By demonstrating faster time-to-epsilon results using quantum annealing, the research offers one of the clearest experimental validations of quantum advantage in approximate optimization to date.
The study opens the door for further exploration into higher-dimensional and denser problem types, with the goal of extending quantum advantage into more practical domains. Ongoing advancements in quantum hardware and improved error suppression techniques are expected to enhance these early successes and broaden the applicability of quantum computing to industrial and scientific problems.
This work represents a crucial step in moving quantum computing from theory to real-world utility, particularly in sectors where near-optimal solutions are both acceptable and beneficial.
By Impact Lab
