A Short Note on Improved ROSETA

where Ωt is a selection operator that specifies the observable subset of entries at time t, at ∈ R r are the coefficients specifying the linear combination of the columns of Ut, and st ∈ R n is a sparse outlier vector. When the subspace Ut is stationary, we drop the subscript t from Ut and the problem reduces to robust matrix completion or robust principal component analysis where the task is to separate a matrix B ∈ R into a low rank component UA and a sparse component S using incomplete observations