{"id":"arxiv/2601.10248","url":"https://arxiv.org/abs/2601.10248","tags":["math.NA"],"score":0,"title":"Restoring similarity in randomized Krylov methods with applications to eigenvalue problems and matrix functions","domain":"math.NA","source":"arxiv","authors":["Laura Grigori","Daniel Kressner","Nian Shao","Igor Simunec"],"reviews":0,"subject":"math.NA","summary":"The randomized Arnoldi process has been used in large-scale scientific computing because it produces a well-conditioned basis for the Krylov subspace more quickly than the standard Arnoldi process. However, the resulting Hessenberg matrix is generally not similar to the one produced by the standard Arnoldi process, which can lead to delays or spike-like irregularities in convergence. In this paper, we introduce a modification of the randomized Arnoldi process that restores similarity with the Hessenberg matrix generated by the standard Arnoldi process. This is accomplished by enforcing…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-01-15","reactions":{}}