“…In a diverse array of real-world applications such as collaborative filtering (Rao et al, 2015), quantum-state tomography (Gross, 2011), spectrum sensing (Corroy et al, 2011) and recommender system (Ramlatchan et al, 2018), we are interested in recovering a large-scale low-rank data matrix from noisy and highly incomplete observations. This problem, usually termed as matrix completion, has attracted a lot of attention over a decade (Candes and Plan, 2010;Keshavan et al, 2010;Candes and Plan, 2011;Ma et al, 2017;Chi et al, 2019;Chen et al, 2020a,b).…”