Efficient coordinate-wise leading eigenvector computation

02/25/2017
by   Jialei Wang, et al.
0

We develop and analyze efficient "coordinate-wise" methods for finding the leading eigenvector, where each step involves only a vector-vector product. We establish global convergence with overall runtime guarantees that are at least as good as Lanczos's method and dominate it for slowly decaying spectrum. Our methods are based on combining a shift-and-invert approach with coordinate-wise algorithms for linear regression.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset