王定宇 (Dingyu Wang) I'm a PhD candiate in theoretical computer science at University of Michigan. My advisor is Seth Pettie. I'm currently looking for jobs next year (2025). Check out my CV. Contact me by email: wangdy[at]umich[dot]edu

Research Interest

I am broadly interested in the design and analysis of randomized algorithms with tools and intuition from information theory, statistical inference, and theory of random processes.

Publications

Sketching, Moment Estimation, and the Lévy-Khintchine Representation Theorem (ITCS25), with Seth Pettie (arxiv)

Universal Perfect Samplers for Incremental Streams (SODA25), with Seth Pettie (arxiv)

Multi-dimensional Approximate Counting (SOSA25) (arxiv)

Better Cardinality Estimators for HyperLogLog, PCSA, and Beyond (PODS23), with Seth Pettie (link)

Non-Mergeable Sketching for Cardinality Estimation (ICALP21), with Seth Pettie and Longhui Yin (arxiv)

Information Theoretic Limits of Cardinality Estimation: Fisher meets Shannon (STOC21), with Seth Pettie (arxiv)

Optimal Protocols for 2-Party Contention Resolution (SSS21) (link)

Manuscripts/In Submission

Harmonic Decomposition in Data Sketches (submitted to STOC25) (arxiv)

Fraud Detection in Sketching and Streaming, with Seth Pettie

Fun Stuff

• A small platform game: Moment Master