Eigopt : Software for Eigenvalue Optimization

1-dimensional and 2-dimensional Matlab implementations of the algorithm (described in [1]) are available at the moment. The user is expected to write the routines calculating the eigenvalue function and its derivative (or gradient). But in the tar files below these routines are already provided for some common eigenvalue optimization problems, specifically for the distance to instability, numerical radius, H-infinity norm, distance to uncontrollability, distance to defectiveness, distance to a triple eigenvalue.

For details on how to use the routines please see the brief guide below.

1-dimensional version : eigopt1D.tar
2-dimensional version : eigopt2D.tar
Brief users' guide

Please send questions to emengi.at.ku.edu.tr

References:

M. Kilic, E. Mengi and E.A. Yildirim, Numerical Optimization of Eigenvalues of Hermitian Matrix Functions, Math ArXiv, 2011