Advancing the frontiers of quantum science and engineering

Abstract: In this talk, I will introduce the multi-armed bandit problem, a fundamental challenge in reinforcement learning and decision-making. It captures the trade-off between exploration—gathering information about an unknown environment—and exploitation—using what has already been learned to maximize rewards. I will then discuss a specific instance: the multi-armed quantum bandit problem, where we have sequential oracle access to an unknown quantum state. In each round, we choose an observable from a given set of actions to maximize its expectation value on the state—our notion of reward. I will outline strategies based on the principle of Optimism in the Face of Uncertainty (OFU) and adaptive confidence regions, which guide decision-making under uncertainty. This framework has practical applications, such as optimizing energy transfer in quantum batteries. Specifically, when we have sequential access to a finite number of copies of an unknown pure qubit state, the goal is to design measurement strategies that both maximize energy extraction and improve our knowledge of the state, enabling more efficient charging in future rounds.

Abstract: AI is rapidly transforming various domains, and quantum computing has the potential to solve certain structured issues efficiently. My research intersects these two intriguing and promising fields: (1) Quantum for AI: we design the first quantum algorithm for training classical sparse neural networks, providing potential quantum enhancement at the early stage of actual learning. (2) AI for Quantum: we develop a provably efficient adiabatic learning (PEAL) algorithm for quantum-classical dynamics, with logarithmic system size sampling complexity and favorable time scaling properties. References: [1]Towards provably efficient quantum algorithms for large-scale machine learning models.Nature Communications 15, 434 (2024) [2]Provably Efficient Adiabatic Learning for Quantum-Classical Dynamics.arXiv:2408.00276

Abstract: Entanglement distribution is essential for realizing large-scale quantum networks, enabling secure communication and distributed quantum computation. Silicon photonics, compatible with CMOS technology, offers compactness, scalability, and cost-effective manufacturing. While satellite-based and free-space links achieve long-distance distribution, fiber-based links are more practical for urban environments but often rely on bulky crystal sources, limiting scalability. Chip-based sources face challenges like low off-chip photon brightness and high photon collection losses. We report a highly efficient silicon nanophotonic chip that provides an off-chip entangled photon pair brightness of between 8,000 to 460,000 pairs per second. This exceeds previous reports by three orders of magnitude in brightness. The entanglement fidelity is 99.85(6)% and 97.90(3)% respectively. A novel edge coupler ensures stable coupling losses of 0.64 dB per facet. Tests over 155 km of deployed metropolitan fiber, configured in a symmetric distribution scheme with loopback paths through NTU and SUTD, achieved 0.7 cps entangled photon coincidences at 87.6% fidelity under a total channel loss of 66 dB. Using ultra-low-loss fiber spools (0.16 dB/km) in the laboratory, the distribution distance is further extended to 302 km, marking the longest nanophotonic chip-based entanglement transmission to date. These results validate silicon chip-based entangled photon sources can achieve competitive performance with bulk crystal sources while offering superior scalability and integration.

Abstract: Entangling quantum sensors, such as magnetometers or interferometers, can dramatically increase their sensitivity. In this talk, we will discuss how entanglement in a network of quantum sensors can be used to accurately measure one or more properties of spatially varying fields, including spatially correlated noise.

Abstract: Accurate estimation of quantum observables is essential for quantum applications in chemistry, optimization, and machine learning. We introduce a hybrid error reduction framework that combines probabilistic error cancellation (PEC) with quantum error detection (QED) using four-qubit CSS codes, aiming to improve accuracy and reduce resource overhead. A central contribution is our approach to learning error-detected channels, which exhibit nonstationary dynamics induced by post-selection. We develop a method to characterize these dynamically evolving channels, enabling PEC application within the post-selected subspace. We analyze this hybrid technique on benchmark quantum circuits, including Pauli evolution and QAOA. Numerical simulations show that applying PEC to error-detected qubits improves expectation value estimation compared to using error detection alone. We explore regimes where the combined sampling overhead of both techniques can be reduced for a fixed target accuracy, advancing practical quantum error mitigation strategies.

Abstract: Quantum theory has shaped our scientific world view like no other theory. As a foundation for understanding current and, in particular, future quantum technologies, quantum literacy is becoming increasingly important for society as a whole. Accordingly, it is also gaining significance in secondary education. In this context, Quantum Information Science and Technology offer opportunities for promising modern teaching approaches. In the first part of the talk, general aspects, ideas and contemporary approaches for teaching key aspects of quantum physics in secondary education will be discussed, taking into account the current state of quantum education research. Based on this analysis, a teaching concept with reference to quantum computing was developed with the aim of teaching key aspects of quantum physics to lower secondary students without relying on mathematical formalism. The approach was evaluated in a field study in collaboration with numerous German schools. The second part of the talk will outline the approach and discuss respective study results.

Abstract:QCamp is a one-week summer school for curious and inquisitive minds who wish to explore and understand the physical laws that govern the world's smallest scales: quantum physics. The scope of QCamp is to provide vivid demonstrations of these laws, offer career advice, as well as to help students build a technical foundation for independent exploration of this world. QCamp is organised by the Centre for Quantum Technologies (CQT) at the National University of Singapore.
 

The one-day Flash QCamp for curious and inquisitive minds to get a taster in quantum technologies.

Abstract: Absolutely maximally entangled (AME) states are multipartite quantum states that are maximally entangled across every possible bipartition. They find applications in quantum communication and in holographic quantum error-correcting codes, which serve as toy models for the AdS/CFT correspondence. In this talk, we explore the relationship between AME states and quantum error correction codes. We demonstrate how to identify stabilizers and logical operators corresponding to such codes. In particular, we apply this method to the concatenation of codes and examine its connection to entanglement swapping. We also investigate the behavior of AME states transmitted through various quantum channels, such as the depolarizing and dephasing channels. Specifically, we analyze how the amount of entanglement is affected across different bipartitions of the transmitted states.

Quantum ThermoDynamics (QTD) 2025 will cover the study of thermal processes at the quantum scale. This is the 2025 edition of the annual quantum thermodynamics conference: previous conferences have been held in Berlin (2013),  Mallorca (2014) and Porquerolles (2015), Erice (2016), Oxford (2017), Santa Barbara (2018), Espoo (2019), Barcelona (online, 2020), Geneva (online, 2021), Belfast (online, 2022), Vienna (2023), and Maryland (2024).