About

I am Josep Lumbreras, a Research Fellow at Nanyang Technological University (Singapore), where I work in the group of Prof. Mile Gu. I completed my Ph.D. at the Centre for Quantum Technologies, National University of Singapore, under the supervision of Prof. Marco Tomamichel. Before that, I obtained a double major in Physics and Mathematics as well as a Master’s degree in Particle Physics and Gravitation from the University of Barcelona.

Research interests

My research is at the intersection of quantum information and machine learning. I am broadly interested in how agents can learn, adapt, and make decisions when interacting with quantum systems or other complex environments with hidden structure.

A recurring theme in my work is adaptivity: how measurements, actions, and protocols should be chosen sequentially in order to extract information, optimize performance, or manipulate quantum resources. This connects naturally with reinforcement learning, bandits, partially observable systems, and world models, as well as with adaptive protocols for quantum information processing and resource distillation.

More generally, I am interested in developing theoretical tools for understanding learning and decision-making in quantum settings, especially when memory, partial observability, and sequential interaction play an essential role.

Publications

1. Learning Pure Quantum States in Any Dimension (Almost) Without Regret
   J. Lumbreras, M. Tomamichel.
   [Nature Partner Journal: quantum information][arXiv (2026)]

2. Quantum Tilted Loss in Variational Optimization: Theory and Applications
   Y. Qiu, J. Lumbreras, X. Li, P. Rebentrost
   [arXiv (2026)]

3. Reinforcement learning for quantum processes with memory
   J. Lumbreras, R.C. Cheng, Y. Hu, M. Fanizza, M. Gu.
   [arXiv (2026)]

4. Bandits roaming Hilbert space
   J. Lumbreras.
   [arXiv (2025)]

5. Quantum state-agnostic work extraction (almost) without dissipation
   J. Lumbreras, R.C. Cheng, Y. Hu, M. Gu, M. Tomamichel.
   [QTML2025][AQIS25][Quantum Resources (2025)][arXiv (2025)]

6. Learning pure quantum states (almost) without regret
   J. Lumbreras, M. Terekhov, M. Tomamichel.
   [Nature Partner Journal: quantum information][QTML2025][INFORMS2025][AQIS24 (long talk)][arXiv (2024)]

7. Linear bandits with polylogarithmic minimax regret
   J. Lumbreras, M. Tomamichel.
   [COLT(2024)][arXiv (2024)]

8. Learning finitely correlated states: stability of the spectral reconstruction
   M. Fanizza, N. Galke, J. Lumbreras, C. Rouzé, A. Winter.
   [QuARC workshop 2025][Beyond IID 2024][arXiv (2024)]

9. Quantum Theory in Finite Dimension Cannot Explain Every General Process with Finite Memory
   M. Fanizza^†, J. Lumbreras^† (co-first author), A. Winter.
   [Communications in Mathematical Physics][TQC 23][Beyond IID 2022][arXiv (2023)]

10. Quantum contextual bandits and recommender systems for quantum data
    S. Brahmachari^†, J. Lumbreras^† (co-first author), M. Tomamichel.
    [Quantum Machine Intelligence][arXiv (2023)]

11. Multi-armed quantum bandits: Exploration versus exploitation when learning properties of quantum states
    J. Lumbreras, E. Haapasalo, M. Tomamichel.
    [Quantum][AQIS 21 (long talk)][Beyond IID 2021][QTML 21][arXiv (2022)]

12. Scaling of variational quantum circuit depth for condensed matter systems
    C. Bravo^†, J. Lumbreras^† (co-first author), L. Tagliacozzo, J. I. Latorre.
    [Quantum][arXiv (2020)]