|
|
|
Program with Abstracts > FridayA mixed-species atom array for quantum computing Giulia Semeghini (Harvard University -- invited) — Fr 9:00-10:00 In this talk, we will explore recent advancements in quantum computing using Rydberg atom arrays and present new opportunities enabled by the use of a dual-species array based on a mixture of alkali and alkaline-earth atoms. Trapped arrays of interacting Rydberg atoms have become a leading platform for quantum information processing and quantum simulation due to their large system size and programmability. The addition of a second atomic species opens new possibilities for implementing selective qubit control, engineering asymmetric inter- and intra-species Rydberg interactions, and exploring novel architectures for quantum computing. Additionally, continuous replenishment of both atomic species is central to improving scalability and circuit depths, while decreasing cycle times. We will present our ongoing efforts toward creating a new continuously reloaded, programmable atom array based on a mixture of Rb and Yb, and discuss its applications—from combining coherent and dissipative dynamics for quantum information processing to simulating lattice gauge theories. Fast and Error-Correctable Quantum RAM Francesco Cesa, Hannes Bernien and Hannes Pichler (IQOQI, U. Innsbruck; U. Chicago) — Fr 10:00-10:30 Quantum devices can process data in a fundamentally different way than classical computers. To leverage this potential, many algorithms require the aid of a quantum Random Access Memory (QRAM), i.e. a module capable of efficiently loading large datasets onto the quantum processor. However, a realization of this fundamental building block is still outstanding due to many crucial challenges, including incompatibility with current quantum hardware and quantum error-correction. In this talk, I will present a novel QRAM design, that enables fast and robust QRAM calls, naturally allows for fault-tolerant and error-corrected operation, and can be integrated on present hardware. This places a long missing, fundamental component of quantum computers within reach of currently available technology. Our proposal employs a special quantum resource state that is consumed during the QRAM call, after being assembled efficiently in a dedicated module. I will explain the key points of our work, and show how the long standing challenges that prevented the deployment of QRAM so far are overcome within our scheme. Concretely, I will discuss detailed blueprints and quantitative estimations for modern neutral-atom processors, where our proposed QRAM module finds a particularly efficient implementation. Preprint at arXiv:2503.19172
When Quantum and Classical Models Disagree: Learning Beyond Minimum Norm Least Square Léo Monbroussou et al. (LIP6, Sorbonne Université) — Fr 11:00-11:30 Variational Quantum Circuits (VQCs) are central to many Quantum Machine Learning algorithms and are among the leading candidates for demonstrating useful quantum advantage. However, these methods face key theoretical challenges, including barren plateaus (vanishing gradients) and difficulties in rigorously proving model expressibility. In the second part of the talk, I will present an alternative approach to designing practical and expressive VQCs, illustrated through the Subspace-Preserving Quantum Convolutional Neural Network. By enforcing symmetry-preserving constraints in the computation, this architecture provides strong polynomial advantages and is supported by theoretical arguments regarding its expressibility, trainability, and scalability, while avoiding existing surrogate methods.
Towards Practical Quantum Neural Network Diagnostics with Neural Tangent Kernels Francesco Scala et al. (U.Basel, IBM Zurich, U.Pavia) — Fr 11:30-12:00 In this work, we propose a practical framework allowing to employ the Quantum Neural Tangent Kernel (QNTK) for approximating Quantum Neural Networks (QNNs) performance before training. We show how a critical learning rate and a characteristic decay time for the average training error can be estimated from the spectrum of the QNTK evaluated at the initialization stage. We then show how a QNTK-based kernel formula can be used to analyze, up to a first-order approximation, the expected inference capabilities of the quantum model under study. We validate our proposed approach with extensive numerical simulations, using different QNN architectures and datasets. Our results demonstrate that QNTK diagnostics yields accurate approximations of QNN behavior for sufficiently deep circuits, can provide insights for shallow QNNs. This approach enables detecting – hence also addressing – potential shortcomings in model design.
Roundtable featuring session speakers — Th 12:00-12:30
End of Conference |
| Online user: 3 | Privacy | Accessibility |
|
