A parallel implementation of One-Sided Jacobi SVD for non-symmetric squared matrices on a high-performance GPU.
R. I. Acosta-QuiñonezDeni Torres RománR. Rodriguez-AvilaDaniel Robles-ValdezPublished in: CONIELECOMP (2016)
Keyphrases
- parallel implementation
- singular value decomposition
- graphics processing units
- least squares
- distributed memory
- parallel computers
- singular values
- embedded processors
- symmetric matrices
- parallel computation
- hurwitz radon
- low rank approximation
- positive definite
- singular vectors
- euclidean distance
- parallel implementations
- data matrix
- principal component analysis
- gpu implementation
- graphics hardware
- low rank
- dimensionality reduction
- scientific computing
- message passing interface
- symmetric matrix
- linear algebra
- positive semidefinite matrices
- positive semidefinite
- real time
- optical flow