Householder QR factorization with randomization for column pivoting (HQRRP)

DOI10.1137/16M1081270MaRDI QIDQ5738147zbMATH OpenFDO

Authors Gregorio Quintana-Ortí, Nathan Heavner, P. G. Martinsson, Robert A. van de Geijn

Publication date 31 May 2017

Published in SIAM Journal on Scientific Computing (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1512.02671

zbMATH Keywords

blocked algorithmHouseholder QR factorizationmatrix-matrix multiplicationsnumerical examplepivoting strategyrandomized algorithm

Mathematics Subject Classification ID

Direct numerical methods for linear systems and matrix inversion (65F05)

Abstract: A fundamental problem when adding column pivoting to the Householder QR factorization is that only about half of the computation can be cast in terms of high performing matrix-matrix multiplications, which greatly limits the benefits that can be derived from so-called blocking of algorithms. This paper describes a technique for selecting groups of pivot vectors by means of randomized projections. It is demonstrated that the asymptotic flop count for the proposed method is $2m{n}^2-(2/3)n^3$ for an $mimesn$ matrix, identical to that of the best classical unblocked Householder QR factorization algorithm (with or without pivoting). Experiments demonstrate acceleration in speed of close to an order of magnitude relative to the {sc geqp3} function in LAPACK, when executed on a modern CPU with multiple cores. Further, experiments demonstrate that the quality of the randomized pivot selection strategy is roughly the same as that of classical column pivoting. The described algorithm is made available under Open Source license and can be used with LAPACK or libflame.

Recommendations

• Randomized QR with column pivoting
• A BLAS-3 Version of the QR Factorization with Column Pivoting
• Computing rank-revealing QR factorizations of dense matrices
• A Parallel QR Factorization Algorithm with Controlled Local Pivoting
• A randomized blocked algorithm for efficiently computing rank-revealing factorizations of matrices

Cites work

• A BLAS-3 Version of the QR Factorization with Column Pivoting
• A note on parallel matrix inversion
• A randomized algorithm for the decomposition of matrices
• A randomized blocked algorithm for efficiently computing rank-revealing factorizations of matrices
• A set of level 3 basic linear algebra subprograms
• A Storage-Efficient WY Representation for Products of Householder Transformations
• Accumulating Householder transformations, revisited
• Algorithm 782
• Anatomy of high-performance matrix multiplication
• BLIS: a framework for rapidly instantiating BLAS functionality
• Computing rank-revealing QR factorizations of dense matrices
• Efficient Algorithms for Computing a Strong Rank-Revealing QR Factorization
• Elemental, a new framework for distributed memory dense matrix computations
• Fast Monte Carlo Algorithms for Matrices II: Computing a Low-Rank Approximation to a Matrix
• Fast monte-carlo algorithms for finding low-rank approximations
• Finding structure with randomness: probabilistic algorithms for constructing approximate matrix decompositions
• FLAME
• scientific article; zbMATH DE number 48972 (Why is no real title available?)
• scientific article; zbMATH DE number 1226426 (Why is no real title available?)
• scientific article; zbMATH DE number 961607 (Why is no real title available?)
• Hybrid Gauss-Trapezoidal Quadrature Rules
• LAPACK Users' Guide
• Numerical Linear Algebra
• Programming matrix algorithms-by-blocks for thread-level parallelism
• Randomized Algorithms for Matrices and Data
• Randomized algorithms for the low-rank approximation of matrices
• Sketching as a tool for numerical linear algebra
• The WY Representation for Products of Householder Matrices

Cited in

(23)

• Single-pass randomized QLP decomposition for low-rank approximation
• Randomized QLP decomposition
• A stable parareal-like method for the second order wave equation
• ALORA: affine low-rank approximations
• Nonnegative diagonals and high performance on low-profile matrices from Householder QR
• A BLAS-3 Version of the QR Factorization with Column Pivoting
• Randomized complete pivoting for solving symmetric indefinite linear systems
• Efficient randomized algorithms for the fixed-precision low-rank matrix approximation
• Computing localized representations of the Kohn-Sham subspace via randomization and refinement
• Randomized Projection for Rank-Revealing Matrix Factorizations and Low-Rank Approximations
• Analytical low-rank compression via proxy point selection
• Randomized QR with column pivoting
• Randomized numerical linear algebra: Foundations and algorithms
• Randomized low-rank approximation methods for projection-based model order reduction of large nonlinear dynamical problems
• Robust blockwise random pivoting: fast and accurate adaptive interpolative decomposition
• CholeskyQR with randomization and pivoting for tall matrices (CQRRPT)
• Sparse linear least-squares problems
• Structure-aware analyses and algorithms for interpolative decompositions
• Truncated QR factorization with pivoting in mixed precision
• Randomized Householder QR
• Adaptive, Matrix-Free Low-Rank Approximation
• Accelerating the Canonical Polyadic Alternating Least Squares Optimization via a Randomized Interpolative Decomposition
• Collect, commit, expand: efficient CPQR-based column selection for extremely wide matrices

Describes a project that uses

Uses Software

• toms/782
• Elemental
• BLIS
• LAPACK
• BLAS
• FLAME
• RRQR
• libflame

This page was built for publication: Householder QR factorization with randomization for column pivoting (HQRRP)

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5738147)