Quantum many-body physics
Critical phenomena, exactly solvable models, quantum geometry, and structural descriptions of many-body systems.
Physics · Machine Learning · Scientific Computing
Qian-Rui Lee
PhD Student in Physics
Department of Physics, National Tsing Hua University
I work on theoretical and computational many-body physics and machine learning methods for physics, with particular interests in critical phenomena, quantum and geometric structures, generative models for scientific sampling, and efficient numerical methods.
Latest preprint
Channel concentration of critical quantum geometry
Qian-Rui Lee, Daw-Wei Wang · arXiv:2609.24905 [cond-mat.stat-mech]
Research
Research overview →Machine learning for physics
Generative modeling, Flow Matching, equivariant neural networks, and physically constrained learning architectures.
Geometric & mathematical physics
Hamiltonian general relativity, connection formulations, gauge structure, and geometric methods across field theory.
Scientific computation
GPU Monte Carlo, tensor and linear-algebra tooling, and reproducible numerical workflows for physics.
Selected publications & notes
All publications →2026
Channel concentration of critical quantum geometry
Preprint · arXiv:2609.24905 [cond-mat.stat-mech]
2026
From ADM Geometrodynamics to Ashtekar Connection Dynamics: A Compact Pedagogical Abstract-Index Derivation
Technical note · Zenodo technical note
2026
Efficient identification of critical regions via Flow Matching–based Monte Carlo initialization
Journal article · Machine Learning: Science and Technology
2025
Automatic Characterization of Fluxonium Superconducting Qubits Parameters with Deep Transfer Learning
Preprint · arXiv:2503.12099
Selected research software
All software →Channel Concentration Companion
Versioned code, data, and figure regeneration for the channel-concentration preprint.
Flow-to-Field
Flow Matching as a reusable, physically informed initializer for Monte Carlo simulations of many-body systems.
Lattice-gauge-equivariant-CNN
A modular lattice-gauge-equivariant CNN framework with U(1) and SU(2) demonstrations and explicit mathematical structure.
Lattice-Field-MCMC-GPU
GPU-accelerated Markov-chain Monte Carlo notebooks for lattice field and spin models, including 2D phi^4, Ising, Potts, and XY systems.