Researchers at Johns Hopkins Applied Physics Laboratory in Laurel, Maryland have cracked a stubborn puzzle in quantum computing: how to predict what will go wrong with quantum computers before you run your most important calculations. Their breakthrough, published in PRX Quantum, achieves sevenfold better accuracy than existing methods—a leap that could accelerate the shift from experimental quantum machines to ones reliably solving real-world problems.
Quantum computers are extraordinarily sensitive instruments. The qubits that power them—quantum bits that exploit the strange rules of quantum mechanics to process information—are so finely tuned that they're vulnerable to the slightest environmental disturbances: stray electrical and magnetic fields, temperature wobbles, vibrations. This noise doesn't just occasionally produce wrong answers; it systematically corrupts quantum calculations in ways that are notoriously difficult to predict. Building computers resilient enough to handle this noise is one of the field's central challenges.
Gregory Quiroz, senior physicist at the lab and associate research professor at Johns Hopkins University, led a team that took an unconventional approach. Rather than trying to model all the fundamental physics underlying quantum interactions—a theoretical exercise that often fails to match real hardware—they built a practical framework that captures the essential noise patterns with far fewer parameters. "We need models that can predict a wide range of behavior while utilizing a small number of parameters," Quiroz explained.
The team's innovation lay in their experimental method. They gained cloud access to 39 qubits spread across seven superconducting quantum devices, specifically transmons—a qubit type prized for its relative resistance to electrical noise and popular in mainstream quantum architectures. But they faced a distinctly modern constraint: they had no low-level access to the hardware itself. This wasn't a bug in their experiment; it was a feature. As Quiroz noted, actual quantum computer users won't have direct hardware access either—they'll simply run applications and need confidence those applications work correctly. The team designed their approach around these real-world conditions.
Yasuo Oda, the paper's first author and a postdoctoral researcher who had studied under Quiroz at Johns Hopkins, explained the creative workaround. Rather than studying single quantum operations in detail, they ran repeated computations on the processors, deliberately accumulating errors. By observing how often those errors appeared and how far they deviated from expected results, they could infer what was actually happening inside the quantum system—a form of reverse-engineering constrained by realistic limits.
The framework's greatest strength is its comprehensiveness. The team accomplished something no one had done before: unifying two fundamentally different error types in a single predictive model. Incoherent errors represent information irretrievably lost, while coherent errors—often fixable flaws in control hardware calibration—can be engineered away once identified. Most existing research treats these separately. "We were able to put a wide variety of errors together into one model, which is simple in terms of parameters but also comprehensive in the types of phenomena it can describe," Oda said. The model even predicts the performance of small quantum algorithms, not just isolated quantum gates.
For the field, the implications are clear. Accurate noise models are essential for creating fault-tolerant quantum computers—machines that can run long, complex calculations without collapsing into error. With this framework now in hand, Quiroz's team is already moving toward the next phase: using these insights to actually improve hardware performance. The path from understanding quantum noise to controlling it is finally becoming visible.