Journal
JOURNAL OF SCIENTIFIC COMPUTING
Volume 91, Issue 2, Pages -Publisher
SPRINGER/PLENUM PUBLISHERS
DOI: 10.1007/s10915-022-01824-9
Keywords
Least squares; QR; Sparse; Hierarchical solver; Linear time
Categories
Ask authors/readers for more resources
In this work, a fast hierarchical solver is developed for solving large, sparse least squares problems. The solver utilizes a multifrontal QR approach and low-rank approximation to reduce computational complexity and memory usage, resulting in an approximate factorization of the matrix stored as a sequence of sparse orthogonal and upper-triangular factors.
In this work, we develop a fast hierarchical solver for solving large, sparse least squares problems. We build upon the algorithm, spaQR (sparsified QR Gnanasekaran and Darve in SIAM J Matrix Anal Appl 43(1):94-123, 2022), that was developed by the authors to solve large sparse linear systems. Our algorithm is built on top of a Nested Dissection based multifrontal QR approach. We use low-rank approximations on the frontal matrices to sparsify the vertex separators at every level in the elimination tree. Using a two-step sparsification scheme, we reduce the number of columns and maintain the ratio of rows to columns in each front without introducing any additional fill-in. With this improvised scheme, we show that the runtime of the algorithm scales as O(M log N) and uses O(M) memory to store the factorization. This is achieved at the expense of a small and controllable approximation error. The end result is an approximate factorization of the matrix stored as a sequence of sparse orthogonal and upper-triangular factors and hence easy to apply/solve with a vector. Numerical experiments compare the performance of the spaQR algorithm with direct multifrontal QR, an inner-outer iterative method, and CGLS iterative method with a diagonal preconditioner and Robust Incomplete Factorization (RIF) preconditioner.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available