Abstract:This paper introduces a randomized Householder QR factorization (RHQR). This factorization can be used to obtain a well conditioned basis of a vector space and thus can be employed in a variety of applications. The RHQR factorization of the input matrix $W$ is equivalent to the standard Householder QR factorization of matrix $\Psi W$, where $\Psi$ is a sketching matrix that can be obtained from any subspace embedding technique. For this reason, the RHQR factorization can also be reconstructed from the Householder QR factorization of the sketched problem, yielding a single-synchronization randomized QR factorization (recRHQR). In most contexts, left-looking RHQR requires a single synchronization per iteration, with half the computational cost of Householder QR, and a similar cost to Randomized Gram-Schmidt (RGS) overall. We discuss the usage of RHQR factorization in the Arnoldi process and then in GMRES, showing thus how it can be used in Krylov subspace methods to solve systems of linear equations. Based on Charles Sheffield's connection between Householder QR and Modified Gram-Schmidt (MGS), a BLAS2-RGS is also derived. A finite precision analysis shows that, under mild probabilistic assumptions, the RHQR factorization of the input matrix $W$ inherits the stability of the Householder QR factorization, producing a well-conditioned basis and a columnwise backward stable factorization, all independently of the condition number of the input $W$, and with the accuracy of the sketching step. We study the subsampled randomized Hadamard transform (SRHT) as a very stable sketching technique.

VDR decomposition of Chebyshev-Vandermonde matrices with the Arnoldi Process

Fast Moving Window Algorithm for QR and Cholesky Decompositions

A randomized algorithm for the QR decomposition-based approximate SVD

Vandermonde with Arnoldi

Improving eigenvectors in Arnoldi's method

A REVERSE ORDER IMPLICIT Q-THEOREM AND THE ARNOLDI PROCESS

QR factorization of ill-conditioned tall-and-skinny matrices on distributed-memory systems

A Refined Iterative Algorithm Based on the Block Arnoldi Process for Large Unsymmetric Eigenproblems

A VARIANT OF THE REFINED ARNOLDI METHOD FOR LARGE SCALE MATRIX EIGENPROBLEMS

CholeskyQR with Randomization and Pivoting for Tall Matrices (CQRRPT)

An Efficient Randomized Algorithm for Computing the Approximate Tucker Decomposition

The Hadamard decomposition problem

Convergence and Near-optimal Sampling for Multivariate Function Approximations in Irregular Domains via Vandermonde with Arnoldi

Towards a more robust algorithm for computing the restricted singular value decomposition

Multivariate confluent Vandermonde with G-Arnoldi and applications

Biorthogonal polynomial bases and Vandermonde-like matrices

A Variation on the Block Arnoldi Method for Large Unsymmetric Matrix Eigenproblems

Randomized Householder QR

URV Factorization with Random Orthogonal System Mixing

UTV decomposition of dual matrices and its applications

Effective Methods of QR-Decompositions of Square Complex Matrices by Fast Discrete Signal-Induced Heap Transforms