{"id":"arxiv/2609.23332","url":"https://arxiv.org/abs/2609.23332","tags":["stat.ME","math.ST"],"score":0,"title":"Streaming PCA: averaging from a geometric perspective","domain":"stat.ME","source":"arxiv","authors":["Tuan Pham","Alessandro Rinaldo","Purnamrita Sarkar"],"reviews":0,"subject":"stat.ME","summary":"We study principal component analysis (PCA) under memory constraints, a setting that is increasingly important in large-scale data analysis. Our focus is on Oja's algorithm, which is a one-pass, memory-efficient algorithm requiring only $O(p)$ storage in the rank-one case and $O(pk)$ storage for $k$- PCA. The main goal is to develop a procedure based on Oja's algorithm that is optimal without prior knowledge of the eigengap, which is otherwise needed to tune the learning rates effectively. We do this by first introducing a geometric perspective on the convergence of $k$-PCA: by embedding the…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-09-20","reactions":{}}