Mathematics – Numerical Analysis
Scientific paper
2011-09-09
Mathematics
Numerical Analysis
32 pages, 7 figures
Scientific paper
The unordered eigenvalues of a Hermitian matrix function depending on one parameter analytically is analytic. Ordering these eigenvalues yields piece-wise analytic functions. For multi-variate Hermitian matrix functions depending on d parameters analytically the ordered eigenvalues are piece-wise analytic along lines in the d-dimensional space. These classical results imply the boundedness of the second derivatives of the pieces defining the eigenvalue functions along any direction. We derive an algorithm based on the boundedness of these second derivatives for the global minimization of an eigenvalue of an analytic Hermitian matrix function. The algorithm is globally convergent. It determines the globally minimal value of a piece-wise quadratic under-estimator for the eigenvalue function repeatedly. In the multi-variate case determining this globally minimal value is equivalent to the solution of a quadratic program. The derivatives of the eigenvalue functions are used to construct quadratic models yielding rapid global convergence as compared to traditional global optimization algorithms. The applications that we have in mind include the H-infinity norm of a linear system, numerical radius, distance to uncontrollability and distance to the nearest defective matrix.
Kilic Mustafa
Mengi Emre
Yildirim Emre Alper
No associations
LandOfFree
Numerical Optimization of Eigenvalues of Hermitian Matrix Functions does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Numerical Optimization of Eigenvalues of Hermitian Matrix Functions, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Numerical Optimization of Eigenvalues of Hermitian Matrix Functions will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-414240