Yanlin Qu        

Modern Applied Probabilist

About

I am a postdoctoral research scholar in the Decision, Risk, and Operations Division at Columbia Business School, working with Hongseok Namkoong and Assaf Zeevi. I earned my PhD in Management Science and Engineering from Stanford University, where I had the privilege of being advised by Peter Glynn and Jose Blanchet. I completed my bachelor's degree in Mathematics at the University of Science and Technology of China (USTC).

Research

I am interested in leveraging scalable tools to analyze complex systems including stochastic models in operations research and stochastic algorithms in machine learning. To facilitate this analysis in great generality, I develop theories and algorithms in the realm of general state-space Markov chains, as summarized in my thesis Markov Chain Convergence Analysis: From Pen and Paper to Deep Learning. In addition, I am fascinated by a particular discrete state-space Markov chain, for which I prove that Rubik’s Cube Scrambling Requires at Least 26 Random Moves.

The Trilogy

Deep Learning for Computing Convergence Rates of Markov Chains

with Jose Blanchet and Peter Glynn, NeurIPS (spotlight), 2024

  • The first general-purpose algorithm to bound the convergence

Computable Bounds on Convergence of Markov Chains in Wasserstein Distance

with Jose Blanchet and Peter Glynn, submitted

  • The theoretical fondation of the above algorithm
  • Applied Probability Society Best Student Paper Prize, 2023
  • Applied Probability Society Conference Best Poster Award, 2023

Estimating the Convergence Rate to Equilibrium of a Markov Chain via Simulation

with Jose Blanchet and Peter Glynn, preprint

  • A consistent estimator for the exact convergence rate

Bias of Markov Chain Sample Quantile

with Peter Glynn, preprint

Uniform Edgeworth Expansion for Markov Chains

with Peter Glynn, preprint

On a New Characterization of Harris Recurrence for Markov Chains and Processes

with Peter Glynn, Mathematics, 2023

Strong Limit Interchange Property of a Sequence of Markov Processes

with Jose Blanchet and Peter Glynn, work in progress

  • Verifying \(X_n(t)\Rightarrow X_\infty(\infty)\) as \(n,t\rightarrow\infty\)

Double Distributionally Robust Bid Shading for First Price Auctions

with Ravi Kant, Yan Chen, Brendan Kitts, San Gultekin, Aaron Flores, Jose Blanchet, preprint, slides

  • "Canceling" two sources of uncertainty
Sunset Panorama