Matrix Factorization Approach for Dynamic Rank/Select in Data Streams

minilek · x · 2026-08-03

The author shares their final project from Google, focusing on the dynamic rank/select data structure problem within the data streaming model. The research aims to maintain a database supporting insertions and deletions while allowing approximate rank queries.

By adopting a matrix factorization perspective, the team broadened the algorithmic design space to potentially outperform existing methods like Dyadic CountSketch, seeking more memory-efficient streaming algorithm solutions.

Original post →

More from Research

Research channel →