Research

Research Program

My research develops high-order, efficient, and structure-preserving numerical methods for multiscale and multiphysics PDEs arising in wave propagation, charged-particle transport, liquid crystals, and quasiperiodic materials.

A central emphasis is the design of structure-preserving spectral and spectral-element methods for electromagnetic wave problems and plasma computation, together with asymptotic-preserving and geometric integrators that remain accurate across parameter regimes.

Unifying Question

How can one design computational methods that are simultaneously mathematically reliable, physically faithful, and practical for genuinely multiscale and multiphysics systems?

Structure preserving methods
Divergence constraints, Gauss's law, conservation laws, asymptotic limits, quasiperiodicity, and manifold geometry.
Fundamental tools
Spectral and spectral-element discretization, exact or semi-exact subflows, particle-in-cell methods, and geometric time integration.

Overview

Although the physical models I study are diverse, they share a common mathematical core: highly oscillatory and multiscale PDE systems involving transport, wave propagation, self-consistent field interactions, internal microstructure, geometric constraints, and in many cases relaxation or dissipation. My work aims to construct algorithms that remain accurate across regimes while preserving key analytical and physical structures.

Wave computation

Structure-preserving spectral and spectral-element methods with exact curl/divergence information and robust treatment of unbounded domains.

Transport and plasma

Analytically solvable and asymptotic-preserving schemes for kinetic systems with severe scale separation.

Complex materials

Projection and geometric-discrete-gradient methods for quasiperiodicity, anisotropy, and manifold constraints.

1. Structure-Preserving Spectral and Spectral-Element Methods for Electromagnetic Waves

I have developed structure-preserving spectral and spectral-element methods for curl-curl equations and Maxwell systems, with emphasis on divergence constraints, Gauss's law, high-order accuracy, and robust solvers for highly oscillatory regimes.

  • Divergence-free spectral bases in two and three dimensions for curl-curl problems.
  • Gauss's-law-preserving spectral methods for Maxwell double-curl source and eigenvalue problems.
  • High-order and accurate truncation techniques for scattering in unbounded domains, including exact perfect absorbing layers and time-domain nonreflecting boundary conditions.
  • Related work on incompressible magnetohydrodynamics using pointwise divergence-free spectral discretization.
This direction is one of the clearest signatures of my research profile: using spectral and spectral-element ideas to build high-order algorithms that preserve intrinsic field structure rather than treating it as a post-processing issue.
Representative Publications
  • SIAM Journal on Scientific Computing (2024): divergence-free spectral method for the curl-curl equation.
  • CSIAM Transactions on Applied Mathematics (2025): Gauss's-law-preserving spectral algorithms for Maxwell problems.
  • SIAM Journal on Scientific Computing (2021): exact and optimal perfect absorbing layer for time-harmonic acoustic wave scattering.
  • Journal of Scientific Computing (2021): time-domain nonreflecting boundary conditions for Maxwell's equations.

2. Electromagnetic-Kinetic Transport

A major part of my current work concerns multiscale kinetic plasma and charged-particle transport models, especially Vlasov-Poisson, Vlasov-Ampère, and Vlasov-Maxwell systems in regimes with severe scale separation.

In this direction I developed an ASAP-style viewpoint: rather than resolving dominant collective oscillations by brute-force over-discretization, identify the mechanisms producing stiffness and integrate them analytically whenever possible.

  • Analytically solvable, waveform asymptotic-preserving time-splitting for Vlasov-Poisson in the quasi-neutral regime.
  • Energy-conserving and asymptotic-preserving particle-in-cell algorithms for Vlasov-Ampère and Vlasov-Maxwell systems.
  • Parameter-oriented splitting with exact or variation-of-constants subflows for the multiscale relativistic Vlasov-Maxwell system.
Representative Publications
  • Journal of Computational Physics (2026): analytically solvable and energy-conserving splitting for quasi-neutral Vlasov-Poisson.
  • Journal of Computational Physics (2023): energy-conserving Fourier PIC method with asymptotic-preserving preconditioner for Vlasov-Ampère.
  • Journal of Mathematical Physics (2023): asymptotic-preserving and energy-conserving PIC method for Vlasov-Maxwell.
  • Submitted (2026): parameter-oriented splitting with exact subflows for the multiscale relativistic Vlasov-Maxwell system.

3. Liquid Crystals and Moiré Materials

I also work on computational methods for condensed-matter systems where microstructure, anisotropy, symmetry, and quasiperiodicity are central. These include liquid crystals and photonic moiré lattices.

  • Reduced projection methods for quasiperiodic Schrödinger eigenvalue problems in moiré lattices.
  • Numerical and theoretical discovery of low-dimensional compact states in 3D moiré lattices.
  • Second-order rotational discrete-gradient methods that preserve geometric constraints and dissipate energy for Oseen–Frank and biaxial liquid-crystal gradient flows.
  • Extension of the structure-preserving framework toward the full Ericksen–Leslie model.
Representative Publications
  • Nature Communications (2025): low-dimensional compact states in 3D moiré lattices.
  • Journal of Scientific Computing (2025): reduced projection method for photonic moiré lattices.
  • Communications in Computational Physics (2024): second-order length-preserving and energy-stable rotational discrete gradient method for Oseen–Frank gradient flows.
  • Journal of Scientific Computing (2024): SO(3)-preserving and energy-stable scheme for biaxial nematic liquid crystals.

Future Directions

Looking ahead, I plan to further develop the ASAP framework into a broader methodology for oscillatory multiscale computation, including quantum transport models and collisional kinetic equations.

A second direction is to combine reduced-order modeling and scientific machine learning with the structure-preserving philosophy, so that reduced models inherit conservation laws, asymptotic limits, and physical constraints.