Quantifying the fidelity of a quantum memory

In the emerging world of quantum technology, information is not stored on hard drives or in the cloud, but in the delicate states of individual particles like atoms or photons. To build a future network capable of sending this information across cities or continents, scientists need a way to pause the signal, hold it for a moment, and release it without losing its unique quantum character. This pause is performed by a device called a quantum memory. Unlike a standard computer memory that simply copies data, a quantum memory must preserve the fragile relationships between particles, a property known as entanglement. If the memory fails, the information degrades into ordinary noise, and the quantum advantage is lost. The central challenge for engineers and physicists has been how to measure exactly how well these devices are working. Traditional methods often rely on checking a few specific examples, which can give a misleading picture of the device's overall performance, much like judging a chef's entire menu by tasting only one dish.

A team of researchers at the University of Geneva has developed a comprehensive framework to solve this problem, offering a reliable way to certify the quality of any quantum memory. Their work moves beyond checking a few specific cases to provide a single, clear number that represents the average quality of the memory across all possible inputs. They demonstrated that by testing a carefully chosen, small set of input states, they can calculate the exact average performance of the device without needing to test every possible state, which would be impossible. This approach allows them to determine not just if the memory works, but how well it preserves the dimensionality of the information and its ability to transmit quantum data. The researchers applied their new methods to a real-world experiment using a solid-state crystal doped with ytterbium ions, a type of memory that stores light in the form of time-based pulses. They found that this specific memory could store quantum information with an average fidelity of 98.6% for a duration of 10 microseconds, and 98.5% for 125 microseconds.

The core of this new method relies on a concept called a "2-design," which is a specific collection of test states that are mathematically representative of all possible states. Imagine trying to understand the average color of a vast, continuous rainbow; testing every single shade is impossible. Instead, if you pick a specific, well-spaced set of colors that cover the entire spectrum, you can calculate the average color of the whole rainbow with perfect accuracy. The researchers showed that for quantum memories, a small set of probe states—such as six specific directions on a sphere representing the possible states of a particle—acts as this representative set. By preparing these states, sending them through the memory, and measuring how much they change, the team can calculate the average channel fidelity. This number tells the story of the memory's performance in a way that is independent of the specific physical platform used, whether it is based on light, atoms, or crystals.

The paper also addresses the practical reality that quantum memories are not perfect; they often lose some of the signal due to technical imperfections. The researchers developed techniques to separate the efficiency of the memory (how often it successfully stores a signal) from its fidelity (how well it preserves the information when it does succeed). They showed that even with limited data or less-than-ideal equipment, it is possible to establish a lower bound on the memory's quality. They introduced "witnesses," which are simpler tests using fewer states, such as just two or four, that can still guarantee the memory is performing above a certain threshold. While these simpler tests might slightly underestimate the true performance, they provide a robust and efficient way to certify that the device is functioning as a true quantum channel, capable of preserving the complex correlations required for future networks.

To prove the practical value of their framework, the team applied these methods to a solid-state quantum memory experiment. In this setup, a source generated pairs of entangled photons, with one photon serving as a signal to herald the arrival of the other. The second photon, carrying the quantum information encoded in its arrival time, was sent into the crystal memory. The memory stored the photon for a programmed duration before releasing it. The researchers then measured the photon to see how closely it matched its original state. Using their six-state protocol, which corresponds to the most accurate test, they confirmed an average fidelity of 98.6% for a storage time of 10 microseconds and 98.5% for 125 microseconds. These results were consistent across different storage times, indicating that the memory itself did not introduce significant noise, even as the efficiency of retrieving the photon dropped from 16.6% to 3.65% over the longer duration.

Beyond simply measuring how well the memory works, the researchers showed that their fidelity metric reveals deeper properties of the device. They demonstrated that a high fidelity score guarantees the memory is not "entanglement-breaking," meaning it can still be used to store and transmit entangled states, which is essential for quantum networks. Furthermore, they linked the fidelity to the quantum capacity of the memory, a measure of how much quantum information the device can transmit per use. Their analysis showed that the memory in their experiment had a quantum capacity of roughly 81.9% of the ideal maximum for the 10-microsecond storage time. This connection allows scientists to predict the potential of a memory for complex tasks based on a single, easily measured number.

The framework also extends to scenarios where the measurement equipment is not fully trusted or characterized. In many real-world applications, the devices used to read the output of a memory might be complex black boxes that are difficult to calibrate. The researchers developed a "semi-device-independent" approach that allows for certification even when the measurement device is treated as unknown. By relying only on the statistics of the input states and the final outcomes, they could still derive a lower bound on the memory's fidelity. In their experimental data, this more conservative method still certified a fidelity of 96.8%, proving that the memory's high performance is robust even when accounting for uncertainties in the measurement process.

This work provides a unified language for comparing different types of quantum memories, from those based on trapped ions to those using solid-state crystals. By establishing a standard for certification that is both rigorous and experimentally feasible, the researchers have removed a significant barrier to the development of quantum networks. Their methods allow engineers to quickly assess whether a new memory design is ready for integration into a larger system, ensuring that the quantum information remains intact as it travels from the past to the future. The successful application of these techniques to a real solid-state device demonstrates that high-fidelity quantum storage is not just a theoretical possibility, but a tangible reality that can be measured, understood, and improved with precision.