Quantum Sensing Leverages ML to Track Three-Level System Phase

Researchers affiliated with the Dipartimento di Fisica e Astronomia ”Ettore Majorana”, Università di Catania, Italy have successfully trained a multi-layer perceptron (MLP) to estimate the plaquette phase within a three-level system, demonstrating a new method for extracting information using artificial intelligence.

The team utilized STImulated Raman Adiabatic Passage (STIRAP) population transfer efficiencies as the data source for the machine learning model, establishing a direct link between a specific quantum control technique and AI-driven analysis. This plaquette phase profoundly affects system dynamics by breaking coherent population trapping and inducing a non-trivial phase dependence, according to the work. The results highlight how combining coherent control and machine learning enables effective phase identification, potentially opening new perspectives for quantum technologies, specifically quantum sensing applications including synthetic gauge fields.

Plaquette Phase Impacts Coherent Population Trapping The subtle interplay of quantum phases can dramatically alter system behavior, and recent work demonstrates this with the identification of a phase in three-level quantum systems that profoundly affects the system dynamics, breaking coherent population trapping.

The team’s findings reveal that accurately estimating this plaquette phase is now possible through a combination of established quantum control methods and machine learning. STIRAP is a well-established technique for efficiently moving quantum populations between states, but the presence of the plaquette phase introduces complexities. The researchers discovered that the efficiency of STIRAP is affected by the phase, creating a measurable signature that a machine learning algorithm can interpret. Specifically, a multi-layer perceptron (MLP), a type of machine learning, was successfully trained to estimate the plaquette phase, demonstrating a novel way to extract information from quantum systems using AI. This is not simply a case of machine learning assisting quantum physics; the MLP accurately estimates a previously inaccessible parameter from observable data. The implications extend beyond mere measurement. Coherent population trapping is a quantum interference effect that normally allows for stable population storage in a specific state; breaking this with the plaquette phase introduces a sensitivity that can be exploited. “This phase sensitivity degrades the performance of the protocol, it provides a valuable resource encoding information about the non-directly accessible phase into measurable observables,” the paper explains. The ability to accurately estimate the plaquette phase, previously a hidden parameter, provides a new degree of control over these systems and expands the possibilities for manipulating quantum states. STIRAP Efficiency as Observable for Phase Sensitivity The pursuit of increasingly precise quantum sensors is driving innovation in how we extract information from delicate quantum states. Current approaches often rely on directly measuring a property affected by the target signal, but a growing body of work explores indirect methods, leveraging subtle changes in established quantum control protocols. Researchers affiliated with Dipartimento di Fisica e Astronomia ”Ettore Majorana”, Università di Catania, Italy are now demonstrating that the efficiency of a standard technique, STImulated Raman Adiabatic Passage (STIRAP), can serve as a sensitive indicator of a previously hidden system parameter, the plaquette phase, through the application of machine learning. This work, centered on systems composed of three quantum states, mutually coupled to form a fully connected network, reveals a surprising link between a quantum control method and the ability to discern a phase that profoundly affects the system dynamics, breaking coherent population trapping.

The team focused on systems where a closed-loop configuration gives rise to a gauge-invariant phase associated with the triangular plaquette formed by the couplings. While traditionally considered a hindrance to efficient population transfer, this sensitivity is now being repurposed as a resource for sensing. Critically, the researchers reconstructed the plaquette phase from observables. The MLP was trained in a supervised-learning framework, learning to accurately estimate the plaquette phase from the observed STIRAP efficiencies under the driving conditions. The success of this method hinges on the fact that the identified phase doesn’t simply degrade performance; it encodes information. Researchers affiliated with the Dipartimento di Fisica e Astronomia ”Ettore Majorana”, Università di Catania are applying artificial intelligence to extract hidden information from quantum systems. This achievement demonstrates a significant step toward utilizing AI not simply as a tool for analyzing quantum data, but as an integral component of quantum sensing itself.

The team’s approach bypasses the need for directly measuring the plaquette phase, instead leveraging existing control infrastructure to gather the necessary data for training the model. Rather than attempting to directly observe the plaquette phase, the researchers utilized STIRAP population transfer efficiencies as the experimentally accessible data to train the MLP. This is crucial because direct measurement of the phase is challenging; by focusing on the effects of the phase on a well-controlled process like STIRAP, they created a pathway for indirect inference. The significance of identifying this plaquette phase lies in its impact on system dynamics. Machine Learning in Quantum Sensing Applications The pursuit of ever-more-sensitive quantum sensors often focuses on maximizing the intrinsic properties of the quantum system itself. However, a growing body of work demonstrates that machine learning offers a complementary path, not by altering the quantum hardware, but by cleverly extracting information from existing signals. Researchers affiliated with Dipartimento di Fisica e Astronomia ”Ettore Majorana”, Università di Catania, Italy are now demonstrating that sophisticated algorithms can bypass direct measurement of subtle quantum phenomena, instead inferring their presence through indirect observation of system behavior. The researchers employed a supervised-learning framework, effectively teaching the MLP to estimate the plaquette phase from STIRAP population transfer efficiencies measured under varying driving conditions. This approach extends beyond simply identifying the phase; the MLP actively estimates a previously inaccessible parameter from observable data and offers a route towards controlling quantum systems in novel ways. The ability to accurately determine this phase has implications for creating potentially enabling new functionalities in quantum materials and devices. The success of this method suggests that machine learning can act as a powerful tool for unlocking hidden information within complex quantum systems. 👉 More information🗞 Machine-Learning-Empowered Quantum Sensing of the Plaquette Phase in a Three-Level Delta System✍️ Lorenzo Vitale, Shreyasi Mukherjee, Dario Fasone, Enrico Martello, Elisabetta Paladino, Luigi Giannelli and Giuseppe Falci🧠 ArXiv: https://arxiv.org/abs/2607.15040 Stay currentSee today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals. Tags: