RQC Seminar

208th RQC Seminar

• Speaker

  Mr. Thomas Lagoutte - Ferre
  ( ENS Paris-Saclay )

• Date

  16:00-17:00 (4:00 p.m. - 5:00 p.m.), July 30, 2025 (Wednesday)

• Venue

  Hybrid(ZOOM・ Wako Main Research 3F 345-347 Seminar Room / 研究本館3階 セミナー室 (345-347) (C01))

• Title

  Controllable Low-Rank Quantum State Reconstruction via Neural Density Matrices

• Inquiries

  norilab_rqc_assist[at]ml.riken.jp

Abstract
Quantum state reconstruction is essential for characterizing quantum systems, yet it remains particularly challenging for mixed states and large-scale systems. This study introduces a method for controllable low-rank quantum state reconstruction using neural density matrices. By embedding mixed quantum states into an enlarged Hilbert space through purification with auxiliary degrees of freedom, the approach enables precise control over the rank of the reconstructed density matrix via the number of ancillary units. The method employs various optimization techniques to train the neural network and is validated on experimental data from trapped-ion systems. Results show that the approach accurately reconstructs dominant eigenstates and offers a scalable solution for quantum state tomography. However, the recovery of dominant eigenstates is sensitive to the choice of loss function, underscoring its critical role in low-rank approximation accuracy. While the method demonstrates promise, challenges persist in recovering subdominant eigenstates and scaling to larger systems. Future work will focus on optimizing loss function selection and addressing scalability to further enhance low-rank quantum state reconstruction.