Statistical Condition Estimation for the Sensitivity of Generalized Eigenvalue and Singular Value Problems
|Keywords||Statistical condition number estimates Eigenvalue Singular value Compressed subspace Singular subspace|
This paper proposed a new method for solving the condition number of the generalized eigenvalue and singular value problems Kenney and Laub - based statistical condition number estimation method . This method can estimate the condition number of the characteristic values ??( singular value ) and a characteristic space ( singular subspace ) . It can be applied to the special structure of the matrix on and applies to different types of disturbances . In particular , this method of component disturbance type is very effective this disturbance type does not apply to the other conditions of the estimate . Subsequent numerical experiments to verify the reliability and validity of this method .