This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: For decades, physicists have worked to prove the strange predictions of quantum mechanics with real experiments. As quantum computers have grown more powerful, researchers have devised increasingly sophisticated ways to test whether these machines are truly harnessing quantum effects—but every method so far has run into limits.
In new research published in Nature Communications, a team led by computer scientists Marcello Benedetti and Harry Buhrman at Quantinuum in the U.K. has proposed a new kind of test built around a simple game with a mathematically guaranteed outcome. Most previous tests have relied on checking for violations of "Bell inequalities," which are mathematical rules that describe how strongly entangled particles should behave if the universe obeyed classical logic. When quantum systems break these rules, it is taken as evidence of genuine quantum behavior.
These Bell tests have real drawbacks, however. Many depend on unproven assumptions about computational difficulty and are highly sensitive to the noise and errors that plague today's hardware, while verifying the results efficiently becomes harder as the systems scale up. The Quantinuum team's alternative approach is a game based on "complement sampling." Here, every possible answer to a problem is secretly split into two equal groups, A and B.
A computer is then handed one answer from group A and asked to produce one from group B. When attempting this task, a classical computer has almost nothing to go on: It can rule out the answer it was given but has no way of knowing which of the remaining answers belong to which group. As the possibilities grow, the task only becomes exponentially harder.
In contrast, a quantum computer can hold the entire set A in superposition and use a "swapper" circuit to transform it directly into its complement before measuring an answer from set B. Crucially, the ceiling on classical performance can be proven mathematically, not just assumed. To test their approach, the team ran the game on Quantinuum's H2 trapped-ion quantum computers, using thousands of circuits scaled up to 55 qubits.
Despite the inevitable noise of real hardware, the quantum system consistently outperformed the best possible classical strategy. As the problems grew larger, the gap between quantum and classical performance widened exponentially—closely matching the behavior theory predicted. The result offers a test that is efficient to verify, doesn't lean on unproven assumptions and appears to scale consistently as quantum computers grow.
Extract — continue reading at the source.
