Yuda Bi
  • About
  • Publications
  • Projects
  • Notes
  • 小说 Novels
Portrait photo of Yuda Bi

Yuda Bi

Ph.D. in Computer Science
TReNDS Center (Georgia State University, Georgia Tech, Emory)

Academic homepage of Yuda Bi, featuring research in statistical physics, information geometry, theoretical neuroscience, and spectral methods.
Keywords

Yuda Bi, statistical physics, information geometry, theoretical neuroscience, spectral graph theory, nonequilibrium thermodynamics, TReNDS Center, Georgia State University

Email Google Scholar GitHub LinkedIn

Homepage Visits Homepage visit counter

Research Program

I received my Ph.D. from the TReNDS Center (GSU, Georgia Tech, Emory), advised by Prof. Vince D. Calhoun.

In 2025 I pivoted from engineering-driven multimodal neuroimaging toward a theory-first research program grounded in statistical physics, nonequilibrium thermodynamics, and information geometry.

Research Program I study how latent structure leaves faint but principled signatures in data: hidden forcing in stochastic systems, spectral structure in biological molecules, and low-rank geometry in large-scale brain organization.

Methodological Style My work sits at the intersection of statistical manifolds, stochastic thermodynamics, spectral graph theory, and matrix geometry, with an emphasis on compact theory that still leads to concrete empirical tests.

Research Directions

Statistical Physics Information Geometry Nonequilibrium Thermodynamics Theoretical Neuroscience Spectral Graph Theory Financial Physics Structural Biology

Academic Path

Ph.D. in Computer Science, Georgia State University, 2020–2025

M.S. in Computer Science, University of Georgia, 2017–2020

B.S. in Management, Shandong University of Science and Technology, 2013–2017

Working Snapshot

Current axis Hidden forcing and detectability

Core toolkit Geometry, spectra, and thermodynamics

Application domains Brains, molecules, and market-like collective systems

Output style Papers, technical notes, and open methods

Featured Work

The Geometry of Invisible Forces: A Quartic Detection Theory

\(D_{\mathrm{KL}}^{\min}(\lambda) = C\,\lambda^4 + O(\lambda^6)\)

Hidden variables are orders of magnitude harder to detect than classical statistics predicts. When a reduced model absorbs the leading perturbation through reparametrization, only the normal residual survives — producing a quartic, not quadratic, detection law. A single probe is provably blind; two probes sharing the hidden driver break the impossibility via cross-spectral geometry.

Four papers under review at Physical Review Letters and Physical Review E.

Theory Note Project Page Tutorial PDF

Recent Notes

Spectral Methods in Brain Network Analysis

Notes on how graph Laplacian spectra reveal community structure in functional brain networks, with key derivations.

Apr 7, 2026
SpecRNA-QA: Why Spectral Graph Features See What Local Metrics Miss

How graph Laplacian spectra capture global RNA topology for reference-free 3D structure quality assessment — outperforming geometry baselines by +0.224 on CASP16.

Apr 7, 2026
The Geometry of Invisible Forces: A Quartic Detection Theory

Why hidden variables are orders of magnitude harder to detect than classical statistics predicts — and how cross-spectral geometry breaks the impossibility.

Apr 7, 2026
No matching items

“The sad truth is that most evil is done by people who never make up their minds to be good or evil.” — Hannah Arendt

 

Yuda Bi  |  Built with Quarto