Beicheng Lou

Currently:

Researcher at Jump Trading (2024-present).

Previously:

PhD in Applied Physics from Stanford University (2018-2024). Advised by Prof. Shanhui Fan. Research on novel theory, numerical methods and optimization algorithms for photonic structures and optical information processing.

Interested in:

Artificial intelligence at scale
Modeling interactions in complex systems

"More is different"
— PW Anderson

Background

Education

PhD in Applied Physics*, minor in Statistics** @ Stanford University

2018.09 - 2024.06

* Advisor: Prof. Shanhui Fan.
** and unofficially, Computer Science and Physical Education too by coursework : )
Bachelor in Engineering Science, and Physics, Minor in Computer Science @ National University of Singapore

2015.09 - 2018.06

Work

Researcher @ Jump Trading

2024.09 - present
Research Scientist @ Nvidia Research

2024.06 - 2024.09
Quantitative Researcher @ Jump Trading

2023.06 - 2023.08
Quantitative Researcher @ Citadel

2022.06 - 2022.08

Past Projects

Twisted Bilayer Photonic Crystals

An exotic structure that is ordered but complex and disorder-like, which could not be generally solved until our work

Pseudomodular Group

Answered the open problem "Are there infinitely many commensurability classes of pseudomodular groups"

Latest Posts

Mar 22, 2023	Architecture Optimization with GNN

Selected Publications

PRL

Theory for Twisted Bilayer Photonic Crystal Slabs

Beicheng Lou, Nathan Zhao, Momchil Minkov, and 3 more authors

Phys. Rev. Lett., 2021

Abs HTML

We analyze scattering properties of twisted bilayer photonic crystal slabs through a high-dimensional plane wave expansion method. The method is applicable for arbitrary twist angles and does not suffer from the limitations of the commonly used supercell approximation. We show strongly tunable resonance properties of this system which can be accounted for semianalytically from a correspondence relation to a simpler structure. We also observe strongly tunable resonant chiral behavior in this system. Our work provides the theoretical foundation for predicting and understanding the rich optical physics of twisted multilayer photonic crystal systems.
NeurIPS

Meta-learning from sparse recovery

Beicheng Lou, Nathan Zhao, and Jiahui Wang

In Fifth Workshop on Meta-Learning at the Conference on Neural Information Processing Systems, 2021

Abs HTML

Meta-learning aims to train a model on various tasks...
IMRN

Hyperbolic Jigsaws and Families of Pseudomodular Groups II

Beicheng Lou, Ser Peow Tan, and Anh Duc Vo

International Mathematics Research Notices, 2021

Abs HTML

In our previous paper, we introduced a hyperbolic jigsaw construction...
Geom. Topol.

Hyperbolic jigsaws and families of pseudomodular groups, I

Beicheng Lou, Ser Tan, and Anh Duc Vo

Geometry & Topology, 2018

Abs HTML

We show that there are infinitely many commensurability classes of pseudomodular groups...