Randomized matrix-vector algorithms need Ω(κ log(1/ε)) products for general linear systems (two-sided) and n products without the transpose, matching CGNE and GMRES.
Meyer and Haim Avron
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DS 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The matrix-vector complexity of $Ax=b$
Randomized matrix-vector algorithms need Ω(κ log(1/ε)) products for general linear systems (two-sided) and n products without the transpose, matching CGNE and GMRES.