Research

My work sits at the interface of theoretical physics, computational physics, and machine learning. A recurring theme is to use structure—symmetry, geometry, locality, constraints, and scaling—to make difficult physical problems more interpretable or computationally tractable.

Quantum many-body physics and critical phenomena

I am interested in critical behavior, exactly solvable many-body models, lattice systems, and geometric quantities that characterize quantum states and responses. The goal is to connect formal structures to concrete observables and scalable numerical tests.

Machine learning for physical systems

I develop and study generative and equivariant learning methods for physics, including Flow Matching and lattice-gauge-equivariant neural networks. The emphasis is on models whose inductive biases reflect physical structure rather than treating physical data as generic high-dimensional arrays.

Geometric and mathematical physics

I am interested in geometric formulations of field theory and gravity, including constrained Hamiltonian systems, ADM geometrodynamics, connection variables, gauge structure, and related differential-geometric ideas.

Scientific computation

My computational work includes GPU-accelerated Monte Carlo, numerical simulation of lattice systems, tensor and linear-algebra utilities, and reproducible PyTorch-based research workflows.

Related work: Lattice-Field-MCMC-GPU · EfficientSVD · ncon-torch